Chapter 17. Pancreatic Hormones and Diabetes Mellitus

• ABCC8 ATP-binding cassette transporter sub-family C member 8
• ACE Angiotensin-converting enzyme
• ADA American Diabetes Association
• ADH Antidiuretic hormone (vasopressin)
• AGE Advanced glycation end product
• AGPAT 1-acylglycerol-3-phosphate-O-acyltransferase 2
• AKT/PKB AKR mouse tumor 8 kinase/protein kinase B
• AIRE Autoimmune regulator
• ALT Alanine aminotransferase
• AMPK Adenosine monophosphate-activated protein kinase
• APS1 Autoimmune polyendocrinopathy syndrome type 1
• ATF6 Activating transcription factor 6
• BMIy Body mass index
• cAMP Cyclic adenosine monophosphate
• CCK Cholecystokinin
• CEL Carboxyl-ester lipase
• cGMP Cyclic 3′,5′-guanosine monophosphate
• CIDEC Cell death-inducing DFFA-like effector C
• CSII Continuous subcutaneous insulin infusion
• CST Contraction stress test
• DCCT Diabetes Control and Complications Trial
• DIDMOAD Diabetes insipidus, diabetes mellitus, optic atrophy, deafness (Wolfram syndrome)
• DPP Diabetes Prevention Program
• DPP-4 Dipeptidyl peptidase 4
• DPT-1 Diabetes Prevention Trial-1
• EIF2AK3 Eukaryotic translation initiation factor 2-α kinase 3
• ER Endoplasmic reticulum
• FDA Food and Drug Administration
• FFA Free fatty acid
• FHR Fetal heart rate
• Foxo1 Forkhead box, subfamily O, member 1
• FOXP3 Forkhead box, subfamily P, member 3
• GABA Gamma-aminobutyric acid
• GAD Glutamic acid decarboxylase
• GCK Glucokinase
• GH Growth hormone
• GHb Glycohemoglobin
• GHSR Growth hormone-secretagogue receptor
• GI Glycemic index
• GIP Gastric inhibitory polypeptide
• GLIS3 Glioma-associated oncogene homolog—similar 3
• GLP-1 Glucagon-like peptide-1
• GLP-2 Glucagon-like peptide-2
• GLUT Glucose transporter
• GPCR G protein-coupled receptor
• GRPP Glicentin-related polypeptide
• GWAS Genome-wide association study
• HDL High-density lipoprotein
• HLA Human leukocyte antigen
• HNF Hepatocyte nuclear factor
• hPL Human placental lactogen
• IA2 Insulinoma antigen 2
• IAA Insulin autoantibody
• IAPP Islet amyloid polypeptide
• ICA Islet cell antibody
• IFG Impaired fasting glucose
• IGF-1 Insulin-like growth factor 1
• IGT Impaired glucose tolerance
• IL-6 Interleukin-6
• INS Insulin
• IP3 Inositol 1,4,5-triphosphate
• IPEX Immunodysregulation polyendocrinopathy enteropathy, X-linked
• IPF-1 Insulin promoter factor-1
• IRE1 Inositol-requiring enzyme 1
• IRS Insulin receptor substrate
• ISL1 Islet transcription factor 1
• IUGR Intrauterine growth restriction
• KPD Ketosis prone diabetes
• KCNJ11 Potassium inwardly-rectifying channel, subfamily J, member 11
• Kir6.2 Potassium channel, inwardly- rectifying 6.2
• LADA Latent autoimmune diabetes of adulthood
• LBK1 Liver kinase B1
• LDL Low-density lipoprotein
• MAPK Mitogen-activated protein kinase
• MODY Maturity-onset diabetes of the young
• MHC Major histocompatibility complex
• mTOR Mammalian target of rapamycin
• NeuroD1 Neural differentiation factor 1
• Neurog3 Neural genesis factor 3
• NIH National Institutes of Health
• NPH Neutral protamine hagedorn
• NST Nonstress test
• PAI-1 Plasminogen activator inhibitor-1
• PAX4 Paired homeobox 4
• PCSK Proprotein convertase, subtilisin/kexin
• PDE5 Phosphodiesterase type 5
• PDX1 Pancreatic duodenal homeobox-1
• PERK Protein kinase R-like endoplasmic reticulum kinase
• PGC1α PPAR gamma coactivator-1α
• PNDM Permanent neonatal diabetes mellitus
• POEMS Polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes
• PP Pancreatic polypeptide
• PPAR Peroxisome proliferator-activated receptor
• PTF1A Pancreatic transcription factor 1α
• PTP1b Protein tyrosine phosphatase 1b
• RFX6 Regulatory factor X-6
• RXR 9-cis-retinoic acid receptor
• SCL19A2 Solute carrier family 19 (thiamine transporter), member 2
• SOCS Suppressor of cytokine signaling
• SREBP1c Sterol regulatory element-binding protein 1c
• SSTR Somatostatin receptor
• SUR1 Sulfonylurea receptor 1
• TCF7L2 Transcription factor 7-like 2
• TGF-a Transforming growth factor-β
• TNDM Transient neonatal diabetes mellitus
• TNF-α Tumor necrosis factor-α
• UGDP University Group Diabetes Program
• UKPDS United Kingdom Prospective Diabetes Study
• VEGF Vascular endothelial growth factor
• VLDL Very low density lipoprotein
• VNTR Variable number of tandem repeats
• WFS1 Wolfram syndrome protein 1
• ZAC Zinc finger protein-inducing apoptosis and cell cycle arrest
• ZnT8 Zinc transporter 8
• ZFP57 Zinc finger protein 57

The pancreas comprises two functionally distinct organs: the exocrine pancreas, the major digestive gland of the body; and the endocrine pancreas, the source of insulin, glucagon, somatostatin, pancreatic polypeptide (PP), and ghrelin. Whereas the major role of the products of the exocrine pancreas (the digestive enzymes) is the processing of ingested foodstuffs so that they become available for absorption, the hormones of the endocrine pancreas modulate every other aspect of cellular nutrition from rate of adsorption of foodstuffs to cellular storage or metabolism of nutrients. Dysfunction of the endocrine pancreas or abnormal responses to its hormones by target tissues cause serious disturbances in nutrient homeostasis, including the important clinical syndromes grouped under the name diabetes mellitus.

Anatomy and Histology

The endocrine pancreas consists of approximately 1 million small endocrine glands—the islets of Langerhans—scattered throughout the glandular substance of the exocrine pancreas. The exocrine pancreas consists of the enzyme-producing cells organized into acini, and the duct system that delivers those enzymes to the lumen of the duodenum. The islet volume comprises 1% to 1.5% of the total mass of the pancreas and weighs about 1 to 2 g in adult humans. At least five cell types—α, β, δ, ε, and PP—have been identified in the islets (Table 17–1). Each of these islet cell types produces a distinguishing peptide hormone: glucagon, insulin, somatostatin, ghrelin, and PP, respectively. Within individual islets, the different cell types are scattered throughout. A typical human islet is depicted in Figure 17–1.

These cell types are not distributed uniformly throughout the pancreas. The PP cells reside primarily in islets in the posterior portion (posterior lobe) of the head, a discrete lobe of the pancreas separated from the anterior portion by a fascial partition. This lobe originates in the primordial ventral bud as opposed to the dorsal bud. The posterior lobe receives its blood supply from the superior mesenteric artery; the remainder of the pancreas derives most of its blood flow from the celiac artery. The islets themselves are richly vascularized, receiving five to ten times the blood flow of the surrounding exocrine pancreatic tissue. Each islet is surrounded by a lattice of astroglial cells and innervated by sympathetic, parasympathetic, and sensory neurons.

Hormones of the Endocrine Pancreas

Insulin

Biosynthesis

The human insulin gene resides on the short arm of chromosome 11. A unique set of transcription factors found in the β cell nucleus activates the transcription of the preproinsulin mRNA from the insulin gene (Figure 17–2). A precursor molecule, preproinsulin, a peptide of MW 11,500, is translated from the preproinsulin messenger RNA in the rough endoplasmic reticulum of pancreatic β cells (see Figure 17–2). Microsomal enzymes cleave preproinsulin to proinsulin (MW ∼9000) almost immediately after synthesis. Proinsulin is transported to the Golgi apparatus, where packaging into clathrin-coated secretory granules takes place. Maturation of the secretory granule is associated with loss of the clathrin coating and conversion of proinsulin into insulin and a smaller connecting peptide, or C peptide, by proteolytic cleavage at two sites along the peptide chain. Mature secretory granules contain insulin and C peptide in equimolar amounts and only small quantities of proinsulin, a small portion of which consists of partially cleaved intermediates.

Biochemistry

Proinsulin (Figure 17–3) consists of a single chain of 86 amino acids, which includes the A and B chains of the insulin molecule plus a connecting segment of 35 amino acids. Two proteins—the prohormone-converting enzymes type 1 and 2 (PCSK1 and PCSK2)—are packaged with proinsulin in the immature secretory granules. These enzymes recognize and cut at pairs of basic amino acids, thereby removing the intervening sequence. After the two pairs of basic amino acids are removed by carboxypeptidase E, the result is a 51 amino acid insulin molecule and a 31 amino acid residue, the C peptide, as shown in Figure 17–3.

A small amount of proinsulin produced by the pancreas escapes cleavage and is secreted intact into the bloodstream, along with insulin and C peptide. Most anti-insulin sera used in the standard immunoassay for insulin cross-react with proinsulin; about 3% to 5% of immunoreactive insulin extracted from human pancreas is actually proinsulin. Because proinsulin is not removed by the liver, it has a half-life three to four times that of insulin. Its long half-life allows proinsulin to accumulate in the blood, where it accounts for 12% to 20% of the circulating immunoreactive insulin in the basal state in humans. Human proinsulin has about 7% to 8% of the biologic activity of insulin. The kidney is the principal site of proinsulin degradation.

Of the two major proinsulin split products present in plasma, the one split at arginine 32-33 far exceeds in amount the barely detectable 65-66 split product. In control subjects, concentrations of proinsulin and 32-33 split proinsulin after an overnight fast averaged 2.3 and 2.2 pmol/L, respectively, with corresponding postprandial rises to 10 and 20 pmol/L.

C peptide, the 31 amino acid peptide (MW 3000) released during cleavage of insulin from proinsulin, has no known biologic activity. β Cells release C peptide in equimolar amounts with insulin. It is not removed by the liver but is degraded or excreted chiefly by the kidney. It has a half-life three to four times that of insulin. In the basal state after an overnight fast, the average concentration of C peptide may remain as high as 1000 pmol/L.

Insulin is a protein consisting of 51 amino acids contained within two peptide chains: an A chain, with 21 amino acids; and a B chain, with 30 amino acids. The chains are connected by two disulfide bridges as shown in Figure 17–3. In addition, an intra chain disulfide bridge links positions 6 and 11 in the A chain. Human insulin has a molecular weight of 5808.

Endogenous insulin has a circulatory half-life of 3 to 5 minutes. It is degraded chiefly by insulinases in liver, kidney, and placenta. A single pass through the liver removes approximately 50% of the plasma insulin.

Secretion

The human pancreas secretes about 30 units of insulin per day into the portal circulation of normal adults in distinct pulses with a period of approximately 5 minutes. The basal concentration of insulin in the peripheral blood of fasting humans averages 10 μU/mL (0.4 ng/mL, or 61 pmol/L). In normal control subjects, insulin seldom rises above 100 μU/mL (610 pmol/L) after standard meals. After ingestion of food, peripheral insulin concentration increases within 8 to 10 minutes, reaches peak concentrations by 30 to 45 minutes, and then rapidly declines to baseline values by 90 to 120 minutes postprandially.

Basal insulin secretion occurs in the absence of exogenous stimuli, in the fasting state. Plasma glucose levels below 80 to 100 mg/dL (4.4-5.6 mmol/L) do not stimulate insulin release, and most other physiologic regulators of insulin secretion only function in the presence of stimulatory levels of glucose. Stimulated insulin secretion occurs in response to exogenous stimuli. In vivo, ingested meals provide the major stimuli for insulin secretion. Glucose is the most potent stimulant of insulin release. The perfused pancreas releases insulin in two phases in response to glucose stimulation (Figure 17–4). When the glucose concentration increases suddenly, an initial short-lived burst of insulin release occurs (the first phase); if the glucose elevation persists, the insulin release gradually falls off and then begins to rise again to a steady level (the second phase). However, sustained levels of high glucose stimulation (∼4 hours in vitro or >24 hours in vivo) result in a reversible desensitization of the β cell response to glucose but not to other stimuli.

The β cell senses glucose through its metabolism (Figure 17–5). Indeed, agents such as 2-deoxyglucose that inhibit the metabolism of glucose block the release of insulin. Glucose enters the pancreatic β cell by passive diffusion, facilitated by membrane proteins termed glucose transporters (see later). Because the transporters function in both directions, and the β cell has an excess of glucose transporters, the glucose concentration inside the β cell is in equilibrium with the extracellular glucose concentration. The low-affinity enzyme glucokinase catalizes the subsequent, and rate-limiting, step in glucose metabolism by the pancreatic β cell, the phosphorylation of glucose to glucose-6-phosphate. Glucose catabolism in the β cell causes a rise in the intracellular ATP-ADP ratio. Acting through the sulfonylurea receptor (SUR1), the nucleotide-sensing subunit of the ATP-sensitive potassium channels on the surface of the β cell, the rise in ATP-ADP ratio closes the potassium channels and depolarizes the cell, thereby activating the voltage-sensitive calcium channels and allowing the entry of calcium ions into the cell.

Insulin release requires calcium ion signaling. In addition to the voltage-dependent entry of extracellular Ca2+ into the β cell as described earlier, glucose also retards Ca2+ efflux from the β cell and releases Ca2+ from intracellular compartments (predominantly the endoplasmic reticulum) into the cytosol. Some nonglucose stimuli of insulin release also function through increases in cytoplasmic Ca2+. The sulfonylurea and meglitinide (such as repaglinide) medications act by closing the ATP-sensitive potassium channels. Secretagogues such as acetylcholine that act through G protein–coupled receptors of the Gαq class stimulate the release of intracellular Ca2+ stored in the endoplasmic reticulum by activating phospholipase C and releasing the intracellular signaling molecule inositol 1,4,5-triphosphate (IP3).

Glucose metabolism in the β cell also generates additional signals that amplify the secretory response to elevations in cytoplasmic Ca2+ concentration. The exact mechanisms of these amplifying signals remains unknown but involve multiple pathways and include increases in the intracellular signaling molecules diacylglycerol and cAMP. Secretagogues such as the gut hormones glucagon-like peptide-1 (GLP-1) and gastric inhibitory polypeptide (also known as glucose-dependent insulinotropic peptide, GIP) that act via G protein–coupled receptors of the Gαs class also stimulate insulin secretion through elevations in cAMP.

Other factors involved in the regulation of insulin secretion are summarized in Table 17–2. These factors can be divided into three categories: direct stimulants, which directly raise cytoplasmic calcium ion concentrations and thus can act in the absence of stimulatory glucose concentrations; amplifiers, which potentiate the response of the β cell to glucose; and inhibitors. Many of the amplifiers are incretins: gastrointestinal hormones that are released in response to the ingestion of meals and stimulate insulin secretion. The action of the incretins explains the observation that orally ingested glucose provokes a greater insulin secretory response than does the same amount of intravenously administered glucose.

Insulin Receptors and Insulin Action

Insulin action begins with the binding of insulin to a receptor on the surface of the target cell membrane. Most cells of the body have specific cell surface insulin receptors. In fat, liver, and muscle cells, binding of insulin to these receptors is associated with the biologic response of these tissues to the hormone. These receptors bind insulin rapidly, with high specificity and with an affinity high enough to bind picomolar amounts.

Insulin receptors, members of the growth factor receptor family (see Chapter 1), are membrane glycoproteins composed of two protein subunits encoded by a single gene. The larger alpha subunit (MW 135,000) resides entirely extracellularly, where it binds the insulin molecule. The alpha subunit is tethered by disulfide linkage to the smaller beta subunit (MW 95,000). The beta subunit crosses the membrane, and its cytoplasmic domain contains a tyrosine kinase activity that initiates specific intracellular signaling pathways.

Downstream Signaling. On binding of insulin to the alpha subunit, the beta subunit activates itself by autophosphorylation. The activated beta subunit then recruits additional proteins to the complex and phosphorylates a network of intracellular substrates, including insulin receptor substrate-1 (IRS-1), insulin receptor substrate-2 (IRS-2), and others (Figure 17–6). These activated substrates each lead to subsequent recruitment and activation of additional kinases, phosphatases, and other signaling molecules in a complex pathway that generally contains two arms: the mitogenic pathway, which mediates the growth effects of insulin and the metabolic pathway, which regulates nutrient metabolism.

In the metabolic signaling pathway, activation of phosphatidylinositol-3-kinase leads to the activation of serine/threonine kinase Akt. Akt activation drives the movement of glucose transporter (GLUT) 4–containing vesicles to the cell membrane, increases glycogen and lipid synthesis, and stimulates protein synthesis through the activation of mTOR. In the mitogenic signaling pathway, activation of Ras initiates a cascade of activating phosphorylations via the MAP kinase pathway, leading to cell growth and proliferation.

Transcriptional regulation. In addition, the insulin–signaling pathway regulates the activity of several nuclear transcription factors that in turn control the expression of genes involved in metabolism and growth. These include members of the forkedhead family of transcription factors, including Foxo1, which is inactivated by phosphorylation by Akt downstream of insulin signaling. Foxo1 coordinates the expression of gene networks involved in nutrient metabolism in multiple tissues, generally activating genes involved in the response to fasting. In this process, Foxo1 works with several other transcriptional regulators including the lipogenic transcription factor SREBP1c, members of the PPAR family of nuclear receptors and the PPAR coactivator PGC1α (Figure 17–7). Foxo1 also inhibits β cell proliferation and survival.

The three members of the PPAR family of nuclear hormone receptors play pleiotropic roles in regulating genes involved in metabolism in many tissues. They may function as targets of insulin signaling, modulators of insulin signaling, or both. Despite overlap in the tissue expression and gene targets of the three PPARs, some general conclusions can be drawn about the function of each. PPARα regulates genes involved in fatty acid catabolism and gluconeogenesis and is most highly expressed in brown fat, heart, liver, kidney, and intestine. PPARβ/Δ is broadly expressed and activates gene programs involved in fatty acid oxidation. PPARγ is most highly expressed in adipose tissue, intestine, and immune cells, but also at lower levels in many other tissues. PPARγ drives white adipocyte differentiation and lipid storage and inhibits production of many of the pro-resistance adipokines and pro-inflammatory cytokines in adipose tissue (see section on insulin resistance later). In macrophages, PPARγ acts to promote their alternative activation to the anti-inflammatory M2 state, rather than the pro-inflammatory M1 state.

The PPARs bind to DNA as heterodimers with the 9-cis-retinoic acid receptor (RXR), and recruit a variety of coactivators and corepressors. PGC1α was originally identified as a coactivator interacting with PPARγ, but the interaction is not exclusive. On different genes PPARγ works with different coactivators, and PGC1α interacts with the other PPARs and many other transcription factors. In collaboration with a variety of different transcription factors in various tissues, PGC1α orchestrates the expression of a set of genes involved in metabolism. PGC1α itself is highly regulated by several signaling pathways including insulin signaling, which inhibits PGC1α activity via phosphorylation by AKT.

A number of natural and synthetic lipids and related compounds can act as PPAR ligands, but the endogenous ligands acting in vivo remain a mystery. The fibrate class of lipid-lowering drugs, used clinically to lower circulating triglyceride levels, act as PPARα ligands. The thiazolidinedione class of insulin-sensitizing drugs, used for the treatment of type 2 diabetes (see later), act as PPARγ ligands.

Deactivation of Insulin Signaling. Once activated by binding to insulin, the insulin receptor and downstream signaling cascades to rapidly deactivate again by several mechanisms. Insulin can simply disengage from the receptor, or the receptor can be internalized and degraded. The receptor and its tyrosine–phosphorylated substrates can be deactivated by specific protein tyrosine phosphatases such as PTP1b. In addition, inhibitory SOCS (suppressor of cytokine signaling) proteins block interactions between the phosphorylated receptor and interacting IRS proteins, direct the ubiquitination and degradation of the IRS proteins, and terminate the activation of downstream components of the signaling pathway. Finally, serine phosphorylation of the insulin receptor and its active substrates by several different serine/threonine kinases, including components of the insulin-signaling pathway such as AKT, blocks insulin signaling. Many of these mechanisms may play a role in the development of insulin resistance (see later).

Metabolic Effects of Insulin

The major function of insulin is to promote storage of ingested nutrients. Although insulin directly or indirectly affects the function of almost every tissue in the body, the discussion here will be limited to a brief overview of the effects of insulin on the major tissues specialized for energy metabolism: liver, muscle, adipose tissue, and brain. In addition, the paracrine effects of insulin will be discussed briefly.

Paracrine Effects.

The effects of the products of endocrine cells on surrounding cells are termed paracrine effects, in contrast to actions that take place at sites distant from the secreting cells, which are termed endocrine effects (see Chapter 1). Paracrine effects of the β and Δ cells on the nearby α cells (see Figure 17–1) are of considerable importance in the endocrine pancreas. Insulin directly inhibits α cell secretion of glucagon. In addition, somatostatin, which Δ cells release in response to most of the same stimuli that provoke insulin release, also inhibits glucagon secretion.

Because glucose stimulates only β and Δ cells (whose products then inhibit α cells) whereas amino acids stimulate glucagon as well as insulin, the type and amounts of islet hormones released during a meal depend on the ratio of ingested carbohydrate to protein. The higher the carbohydrate content of a meal, the lower the amount of glucagon released by any amino acids absorbed. In contrast, a predominantly protein meal results in relatively greater glucagon secretion, because amino acids are less effective at stimulating insulin release in the absence of concurrent hyperglycemia but are potent stimulators of α cells.

Endocrine Effects

(Table 17–3)

1. Liver—The first major organ reached by insulin via the bloodstream is the liver. Insulin exerts its action on the liver in two major ways:

   Insulin promotes anabolism—Insulin promotes glycogen synthesis and storage while inhibiting glycogen breakdown. These effects are mediated by changes in the activity of enzymes in the glycogen synthesis pathway (see below). The liver has a maximum storage capacity of 100 to 110 g of glycogen, or approximately 440 kcal of energy.

   Insulin increases both protein and triglyceride synthesis and very low density lipoprotein (VLDL) formation by the liver. It also inhibits gluconeogenesis and promotes glycolysis through its effects on the function and expression of key enzymes of both pathways.

   Insulin inhibits catabolism—Insulin acts to reverse the catabolic events of the postabsorptive state by inhibiting hepatic glycogenolysis, ketogenesis, and gluconeogenesis.

2. Muscle—Insulin promotes protein synthesis in muscle by increasing amino acid transport, as well as by stimulating ribosomal protein synthesis. In addition, insulin promotes glycogen synthesis to replace glycogen stores expended by muscle activity. This is accomplished by increasing glucose transport into the muscle cell, enhancing the activity of glycogen synthase, and inhibiting the activity of glycogen phosphorylase. Approximately 500 to 600 g of glycogen are stored in the muscle tissue of a 70-kg man, but because of the lack of glucose 6-phosphatase in this tissue, it cannot be used as a source of blood glucose, except for a small amount produced when the debranching enzyme releases unphosphorylated glucose from branch points in the glycogen polymer, and the glucose indirectly produced via the liver from lactate generated by muscle.

3. Adipose tissue—Fat, in the form of triglyceride, is the most efficient means of storing energy. It provides 9 kcal/g of stored substrate, as opposed to the 4 kcal/g generally provided by protein or carbohydrate. In the typical 70-kg man, the energy content of adipose tissue is about 100,000 kcal.

   Insulin acts to promote triglyceride storage in adipocytes by a number of mechanisms. (1) It induces the production of lipoprotein lipase in adipose tissue (this is the lipoprotein lipase that is bound to endothelial cells in adipose tissue and other vascular beds), which leads to hydrolysis of triglycerides from circulating lipoproteins, thereby yielding fatty acids for uptake by adipocytes. (2) By increasing glucose transport into fat cells, insulin increases the availability of α-glycerol phosphate, a substance used in the esterification of free fatty acids into triglycerides. (3) Insulin inhibits intracellular lipolysis of stored triglyceride by inhibiting intracellular lipase (also called hormone-sensitive lipase). This reduction of fatty acid flux to the liver is a key regulatory factor in the action of insulin to lower hepatic gluconeogenesis and ketogenesis.

4. Central nervous system—Although the brain is traditionally not considered an insulin-sensitive tissue, and overall glucose utilization by the brain is not acutely regulated by insulin, key regions of the brain can respond to insulin. Insulin signaling via PI3 kinase in key cells in the hypothalamus functions with leptin signaling to decrease appetite and increase energy expenditure (see Chapter 20).

AMPK and Insulin-Independent Regulation of Nutrient Metabolism

Insulin, along with the counter-regulatory hormones and other circulating enhancers and inhibitors of their actions, coordinates nutrient metabolism in response to the overall needs of the organism. At the level of the individual cell, however, additional mechanisms sense and respond to the local energy state. Among these mechanisms, adenosine monophosphate protein kinase (AMPK) plays a central role. When energy availability falls, the drop in cellular ATP concentration and rise in AMP trigger a conformational change in the trimeric AMPK complex and the subsequent activation of the catalytic domain by the serine/threonine kinase LBK1/STK11. AMPK then drives the production of ATP by activating catabolic pathways and inhibiting synthetic pathways in the cell (Figure 17–8). In muscle, in response to the rise in AMP during exercise, AMPK increases fatty acid oxidation and insulin-independent glucose uptake while inhibiting mTOR and protein synthesis. In the long term, AMPK also drives mitochondrial biogenesis. In liver cells, AMPK blocks fatty acid and triglyceride synthesis while activating fatty acid oxidation, and also inhibits the gluconeogenic program by blocking cAMP activation of gene expression and inhibiting Foxo1/PGC1α-driven expression of the gluconeogenic genes. In brain, AMPK also functions as an energy sensor and plays a role in the regulation of appetite and energy expenditure by the hypothalamus. AMPK has also been implicated in the regulation of insulin secretion by β cells.

While predominantly an intracellular energy sensor, AMPK increases the sensitivity of cells to insulin, although the mechanisms remain uncertain. AMPK also responds to extracellular signals, and contributes to the regulation of metabolism by many of the adipokines and cytokines (see later) as well as cannabinoids. The biguanide drugs, including metformin, which is used in the treatment of type 2 diabetes, activate AMPK by reducing mitochondrial production of ATP and raising intracellular levels of AMP, and thereby lower blood glucose levels by inhibiting gluconeogenesis.

Glucose Transporter Proteins

Glucose oxidation provides energy for most cells and is critical for brain function. Because cell membranes are impermeable to hydrophilic molecules such as glucose, all cells require carrier proteins to transport glucose across the lipid bilayers into the cytosol. Whereas the intestine and kidney have an energy-dependent Na+-glucose cotransporter, all other cells utilize non-energy-dependent transporters that facilitate diffusion of glucose from a higher concentration to a lower concentration across cell membranes. Facilitative glucose transporters (GLUTs) comprise a large family including at least 13 members, although some of the recently identified members of the family have not yet been shown to transport glucose. The first four members of the family are the best characterized, and they have distinct affinities for glucose and distinct patterns of expression.

GLUT 1 is present in all human tissues. It mediates basal glucose uptake, because it has a very high affinity for glucose and, therefore, can transport glucose at the relatively low concentrations found in the fasted state. For this reason, its presence on the surface of the endothelial cells of the brain vascular system (blood–brain barrier) ensures adequate transport of plasma glucose into the central nervous system.

GLUT 3, which is also found in all tissues, is the major glucose transporter on neurons. It also has a very high affinity for glucose and is responsible for transferring glucose into neuronal cells at the lower concentrations found in the central nervous system.

In contrast, GLUT 2 has a lower affinity for glucose and thus increases glucose transport when plasma glucose levels rise, such as postprandially. It is a major transporter of glucose in hepatic, intestinal, and renal tubular cells. The low affinity of GLUT 2 for glucose reduces hepatic uptake of glucose during fasting, while its ability to transport glucose equally efficiently in both directions assists in the export of glucose from hepatocytes. GLUT 2 is also expressed on the surface of the β cells in rodents, but it is not detected at significant levels on human β cells.

GLUT 4 is found in two major insulin target tissues: skeletal muscle and adipose tissue. It is sequestered mainly within an intracellular compartment of these cells and thus does not function as a glucose transporter until insulin signaling causes translocation of GLUT 4 to the cell membrane, where it facilitates glucose entry into these tissues after a meal (see Figure 17–6). In muscle, exercise also drives GLUT 4 translocation to the cell surface by activating AMPK.

Islet Amyloid Polypeptide

Islet amyolid polypeptide (IAPP), or amylin, is a 37 amino acid peptide produced and stored with insulin in pancreatic β cells but only at a low ratio of approximately one molecule of IAPP to 100 of insulin. β Cells cosecrete IAPP with insulin in response to glucose and other β cell secretagogues. Although it plays a role in regulating gut physiology by decreasing gastric emptying and gut motility after meals, the full physiologic functions of IAPP remain uncertain. A soluble analog of IAPP called pramlintide has been approved for use in patients with type 1 diabetes and insulin-treated type 2 diabetes (see later).

IAPP produces amyloid deposits in pancreatic islets of most patients with type 2 diabetes of long duration. These amyloid deposits are insoluble fibrillar proteins generated from IAPP oligomers that encroach on and may even occur within pancreatic β cells. Islets of nondiabetic elderly persons may contain less extensive amyloid deposits. Whether amyloid fibrils and deposition contributes to the islet dysfunction and β cell loss seen in type 2 diabetes or is simply a consequence of disordered and hyperstimulated islet function remains an unresolved question.

Glucagon

Biochemistry

Pancreatic glucagon, along with several other biologically active peptides, derives from the large proglucagon peptide encoded by the preproglucagon gene located on human chromosome 2. Tissue-specific proteases (the prohormone convertases) cleave different sets of peptide products from the proglucagon molecule in the endocrine l-cells of the gut and the α cells in the islet (Figure 17–9). The activity of prohormone convertase 2 in α cells generates the glucagon peptide, along with the amino–terminal glicentin-related peptide, a small central hexapeptide, and a large carboxyl-terminal fragment.

Glucagon consists of 29 amino acids in a single-chain polypeptide with a molecular weight of 3485. In healthy humans, the average fasting plasma immunoreactive glucagon level is 75 pg/mL (25 pmol/L). Only 30% to 40% of this is actually pancreatic glucagon, the remainder being a heterogeneous composite of higher-molecular-weight molecules with glucagon immunoreactivity such as proglucagon, glicentin, and oxyntomodulin. Circulating glucagon has a half-life of 3 to 6 minutes due to removal by the liver and kidney.

Secretion

In contrast to its stimulation of insulin secretion, glucose inhibits glucagon secretion. Conflicting data surround the question of whether glucose directly inhibits secretion from the α cell or whether it only acts via release of insulin and somatostatin from the β and Δ cells, both of which inhibit the α cell directly. In addition, because β cells release gamma-aminobutyric acid (GABA) and α cells express inhibitory GABA receptors, GABA also may participate in the inhibition of α cells during β cell stimulation.

Many amino acids stimulate glucagon release, although they differ in their ability to do so. Some, such as arginine, release both glucagon and insulin; others (eg, alanine) stimulate primarily glucagon release. Leucine, an effective stimulant of insulin release, does not stimulate glucagon. Other substances that promote glucagon release include catecholamines, gastrointestinal hormones (cholecystokinin [CCK], gastrin, and gastric inhibitory polypeptide [GIP]), and glucocorticoids. Both sympathetic and parasympathetic (vagal) stimulation promote glucagon release, especially in response to hypoglycemia. High levels of circulating fatty acids suppress glucagon secretion.

Action of Glucagon

In contrast to insulin, which promotes energy storage in a variety of tissues in response to feeding, glucagon provides a humoral mechanism for delivering energy from the liver to the other tissues between meals. The ratio of insulin to glucagon affects key target tissues by regulating the expression and activity of key enzymes controlling nutrient metabolism and, thereby, controlling the flux of these nutrients into or out of storage.

The liver, because of its connection to the pancreas via the portal vein, represents the major target organ for glucagon, with portal vein glucagon concentrations reaching as high as 300 to 500 pg/mL (100-166 pmol/L) during fasting. It is unclear whether physiologic levels of glucagon affect tissues other than the liver. Glucagon signals through the glucagon receptor, a G protein–coupled receptor (GPCR) of the Gαs class found predominantly on the surface of hepatocytes. Binding of glucagon to its receptor in the liver activates adenylyl cyclase and the generation of cAMP, which in turn mediates the phosphorylation or dephosphorylation of key enzymes regulating nutrient metabolism. In addition, like insulin, glucagon receptor signaling modifies the activity of a set of cAMP responsive transcriptional regulators that in turn control the expression of the genes encoding these same enzymes.

Glucagon signaling in the liver stimulates the breakdown of stored glycogen, maintains hepatic output of glucose from amino acid precursors (gluconeogenesis), and promotes hepatic output of ketone bodies from fatty acid precursors (ketogenesis). Glucagon facilitates the uptake of the gluconeogenic substrate alanine by liver, and directs fatty acids away from reesterification to triglycerides and toward ketogenic pathways. In sum, glucagon signaling results in the net release of readily available energy stores from the liver in the form of glucose and ketones.

Glucagon-Related Peptides

In the intestinal l-cells, found predominantly in the distal ileum and colon, prohormone convertase 1 generates a different set of peptides from the proglucagon molecule, including glicentin, glicentin-related polypeptide (GRPP), oxyntomodulin, and the two glucagon-like peptides GLP-1 and GLP-2 (see Figure 17–9). Several biological activities have been attributed to glicentin and oxyntomodulin based on studies using high concentrations of the peptides, but all these actions can be explained by low-affinity interactions with the receptors for glucagon, GLP-1 and GLP-2. Specific receptors for glicentin and oxyntomodulin have not been identified, and it remains uncertain whether these peptides play any biological role at physiologic concentrations. GRPP also has no clearly established biological activity. The other two gut–derived glucagon-related peptides, GLP-1 and GLP-2, however, play important roles in nutrient metabolism and gastrointestinal physiology (Table 17–4).

There are two active forms of GLP-1: GLP-1(7-36) amide, and GLP-1(7-37). The intestinal l-cells secrete GLP-1 in response to meals, through dietary glucose and lipids and parasympathetic stimulation. The l-cells sense dietary fat in the gut lumen in part through the GPR119 receptor, which binds the long chain fatty acid derivative oleoylethanolamide. GPR119 is also expressed on the surface of the β cells. GLP-1 binds to the GLP-1 receptor, a GPCR similar to the glucagon receptor. The ubiquitous protease dipeptidyl peptidase 4 (DPP-4) rapidly inactivates circulating GLP-1 (half-life <2 min) by removing the two amino-terminal amino acids. Pancreatic islets are major targets of GLP-1 action. GLP-1 directly stimulates the production and secretion of insulin and somatostatin, and thereby indirectly inhibits the secretion of glucagon. In addition, GLP-1 protects the β cells from destruction and stimulates β cell growth. Other targets of GLP-1 include the stomach, where the peptide inhibits gastric emptying and gastric acid secretion; the brain, where it inhibits appetite and induces weight loss; and the heart, where it has some protective effects.

Along with GLP-1, intestinal l-cells cosecrete GLP-2 in response to eating; and like GLP-1, GLP-2 binds to a specific GPCR closely related to the glucagon and GLP-1 receptors. DPP-4 also inactivates GLP-2. GLP-2 signaling predominantly targets the intestine, where it stimulates mucosal growth and nutrient absorption and inhibits motility.

The K cells in the duodenum and jejunum produce a related 42 amino acid incretin peptide, GIP, that has both functional and sequential similarity to GLP-1, but is the product of a distinct gene and binds to a distinct receptor, GIPR, which also belongs to the family of glucagon-related Gαs-linked receptors. The K cells secrete GIP in response to glucose—via the same pathway used by the β cell (see Figure 17–5)—and lipids. Interestingly, the GIP prepropeptide is also expressed in α cells, but prohormone convertase 2 in α cells produces a shorter peptide, GIP1-30, which lacks the 12 carboxyl amino acids present in intestinal GIP1-42. The two forms of GIP appear to function identically. GIP signaling through its receptor has similar effects to those of GLP-1 on the stomach and β cells. α Cells also express the GIP receptor, through which GIP directly stimulates glucagon secretion; but GIP concomitantly suppresses glucagon secretion indirectly through its stimulation of insulin secretion. The GIP receptor is also expressed in adipose tissue and bone. In adipose tissue, GIP plays an important role in the differentiation of new adipocytes, and also drives lipogenesis and adipokine production in mature adipocytes. In bone, GIP stimulates the osteoblasts and increases bone density.

Somatostatin

The pancreatic Δ cells transcribe the gene for somatostatin on the long arm of chromosome 3. It codes for a 116 amino acid peptide, preprosomatostatin, from whose carboxyl end is cleaved the hormone somatostatin, a 14 amino acid cyclic polypeptide with a molecular weight of 1640 (Figure 17–10). First identified in the hypothalamus, it owes its name to its ability to inhibit the release of growth hormone (GH; pituitary somatotropin). Since that time, somatostatin has been found in a number of tissues, including many areas of the brain and peripheral nervous system, the endocrine D cells in the epithelial lining of the stomach and intestine, and the Δ cells in the pancreatic islets. In neurons, gastric D cells and the islet, somatostatin-14 predominates, but approximately 5% to 10% of the somatostatin-like immunoreactivity in the brain consists of a 28 amino acid peptide, somatostatin-28. Somatostatin-28 consists of an amino terminal region of 14 amino acids and a carboxyl terminal segment containing somatostatin-14. In small intestine, the larger molecule predominates, with 70% to 75% of the hormone in the 28 amino acid form and only 25% to 30% as somatostatin-14. Somatostatin-28 is 10 times more potent than somatostatin-14 in inhibiting growth hormone and insulin secretion, whereas somatostatin-14 is more effective in inhibiting glucagon release.

Most known stimulators of insulin release also promote somatostatin release from Δ cells. This includes glucose, arginine, gastrointestinal hormones, and sulfonylureas. The importance of circulating somatostatin is unclear; the major action of this peptide appears to be paracrine regulation of the pancreatic islet and the gastrointestinal tract. Physiologic levels of somatostatin in humans seldom exceed 80 pg/mL (49 pmol/L). The metabolic clearance of exogenously infused somatostatin in humans is extremely rapid; the half-life of the hormone is less than 3 minutes.

Molecular cloning has identified five somatostatin receptors (SSTR1-5), all of which are GPCRs. They vary in size from 364 to 418 amino acids (with 105 amino acids invariant) and function in the central nervous system and a wide variety of peripheral tissues, including the pituitary gland, the small intestine, and the pancreas. All five receptors belong to the Gαi class and inhibit the activity of adenylate cyclase, thereby lowering intracellular levels of cAMP and inhibiting cAMP-activated secretion. In addition, however, each of the different somatostatin receptors interacts with additional distinct downstream effectors that modify the cellular consequences of receptor activation. Binding of ligand to SSTR5 on β cells mediates the inhibition of insulin secretion, whereas inhibition of GH release from pituitary somatotrophs as well as glucagon release from α cells of the pancreas works through SSTR2. This explains why an analog of somatostatin, octreotide, which has a much greater affinity for SSTR2 than for SSTR5, is effective in correcting GH excess without having much effect on carbohydrate tolerance when used to treat acromegaly.

Somatostatin acts in several ways to restrain the movement of nutrients from the intestinal tract into the circulation. It prolongs gastric emptying time, decreases gastric acid and gastrin production, diminishes pancreatic exocrine secretion, decreases splanchnic blood flow, and retards xylose absorption.

Pancreatic Polypeptide

PP is found in PP cells located chiefly in islets in the posterior portion of the head of the pancreas. Similar to the other islet hormones, PP derives from a larger prepropeptide of 85 amino acids that is cleaved to a single 36 amino acid peptide with a molecular weight of 4200. Circulating levels of the peptide increase in response to a mixed meal; however, intravenous infusion of glucose or lipid does not produce such a rise, and intravenous amino acids induce only a small increase. In contrast, vagotomy abolishes the response to an ingested meal, demonstrating that PP secretion responds predominantly to neural, rather than nutrient signals.

In healthy subjects, basal levels of PP average 24 ± 4 pmol/L and may become elevated owing to a variety of factors including old age, alcohol abuse, diarrhea, chronic renal failure, hypoglycemia, or inflammatory disorders. Values above 300 pmol/L are found in most patients with pancreatic endocrine tumors such as glucagonoma or vasoactive intestinal polypeptide-secreting tumor and in all patients with tumors of the pancreatic PP cell. As many as 20% of patients with insulinoma and one-third of those with gastrinomas also have plasma concentrations of PP that are greater than 300 pmol/L.

Although it has been implicated in the regulation of exocrine pancreatic secretion and gall bladder contraction, the physiologic actions of PP remain uncertain.

Ghrelin

The peptide hormone ghrelin was originally identified in extracts from the stomach based on its ability to bind to and activate the growth hormone secretagogue receptor (GHSR) and stimulate growth hormone release from the pituitary. The P/D1 endocrine cells in the gastric mucosa and the ε cells in the islet make ghrelin, as do a few cells in the heart, lung, kidney, immune system, hypothalamus, and pituitary. The human GHRELIN gene comprises four exons, and the major splice product encodes the 117 amino acid preproghrelin peptide. Processing in the ε cells yields the active form of ghrelin: a 28 amino acid peptide (amino acids 24-51 of preproghrelin) with the serine in position 3 modified by the attachment of an octanoyl side chain. Full biological activity requires the n-octanoyl modification. In addition, protease cleavage generates a second peptide, obestatin (amino acids 76-98 of preproghrelin) of less certain biological function.

Initially identified as a stimulator of growth hormone secretion, ghrelin signals through its receptor, the previously identified GHSR, which is a GPCR found in a variety of tissues, including the hypothalamus, pituitary, intestine, and islet. Ghrelin signaling stimulates growth hormone secretion directly through its receptor on pituitary somatotrophs, and also through its stimulation of hypothalamic GHRH secretion. In addition, Ghrelin induces gastric emptying and acid secretion and regulates appetite and energy balance via neurons in the arcuate nucleus of the hypothalamus (see Chapter 20). The role of ghrelin signaling in the pancreas, and the relative contribution of islet-derived ghrelin to the overall actions of ghrelin remains unresolved.

Clinical diabetes mellitus is a syndrome of disordered metabolism with inappropriate hyperglycemia due to an absolute or relative deficiency of insulin. There may also be a defect in insulin action (insulin resistance).

Classification

Diabetes is classified into four main groups based on known pathological and etiologic mechanisms—type 1, type 2, other specific types, and gestational diabetes (Table 17–5). Type 1 diabetes (previously referred to as juvenile-onset or insulin dependent diabetes mellitus [IDDM]) results from pancreatic islet β cell destruction most commonly by an autoimmune process. These patients are prone to developing ketoacidosis and require insulin replacement. Type 2 diabetes (previously referred to as adult-onset or non-insulin-dependent diabetes mellitus [NIDDM]), the most prevalent form of diabetes, is a heterogeneous disorder most commonly associated with insulin resistance in the presence of an associated impairment in compensatory insulin secretion.

Type 1 Diabetes Mellitus

Type 1 diabetes is immune-mediated in more than 95% of cases (type 1a) and idiopathic in less than 5% (type 1b). The rate of pancreatic β cell destruction may vary, but in most cases the process is prolonged, extending over months or years, since evidence for an immune response can be detected long in advance of hyperglycemia in patients that eventually develop type 1 diabetes. It is a catabolic disorder in which circulating insulin is virtually absent, plasma glucagon is elevated, and the pancreatic β cells fail to respond to all known insulinogenic stimuli. In the absence of insulin, the three main target tissues of insulin (liver, muscle, and fat) not only fail to appropriately take up absorbed nutrients but continue to deliver glucose, amino acids, and fatty acids into the bloodstream from their respective storage depots. Furthermore, alterations in fat metabolism lead to the production and accumulation of ketones. This inappropriate persistence of the fasted state postprandially can be reversed by the administration of insulin.

The incidence of type 1 diabetes varies widely in different populations. Scandinavia and northern Europe have the highest incidence of type 1 diabetes: the yearly incidence per 100,000 youngsters 14 years of age or less is as high as 40 in Finland, 31 in Sweden, 22 in Norway, 27 in Scotland, and 20 in England. The incidence of type 1 diabetes generally decreases across the rest of Europe to 11 in Greece and 9 in France. Surprisingly, the island of Sardinia has as high an incidence as Finland, even though in the rest of Italy, including the island of Sicily, the incidence is only 11 per 100,000 per year. The United States averages 16 per 100,000. The lowest incidence of type 1 diabetes worldwide is less than 1 per 100,000 per year in China and parts of South America.

Worldwide incidence of type 1 diabetes continues to increase steadily. In Finland, the incidence has more than tripled since 1953, when it was 12/100,000/year, with an average increase of 2.4% per year. The EURODIAB study group reported recently 0.6 % to 9.3 % annual increases in incidence of type 1 diabetes in children younger than 15 years in various European countries. The most rapid increases have occurred in low-prevalence countries and in younger patients. Changes in environmental factors most likely explain this increased incidence.

Latent autoimmune diabetes of adulthood (LADA): Type 1 diabetes can present at any age, although peaks in incidence occur before school age and again at around puberty. Older adults often present with a more indolent onset that sometimes leads to misdiagnosis and has led to the use of the term latent autoimmune diabetes of adulthood (LADA) to distinguish these patients. These initially unrecognized patients may retain enough β cell function at the outset to avoid ketosis, but develop increasing dependence on insulin therapy over time as their β cell mass diminishes. Islet cell antibody surveys among northern Europeans indicate that up to 15% of patients previously diagnosed with type 2 diabetes may actually have LADA.

Autoimmunity and Type 1 Diabetes

Most patients with type 1 diabetes at diagnosis have circulating antibodies against β cell proteins: islet cell antibodies (ICA), insulin autoantibodies (IAA), and antibodies to glutamic acid decarboxylase 65 (GAD), tyrosine phosphatase IA2 (ICA512), and zinc transporter 8 (ZnT8) (Table 17–6). These autoreative antibodies can often be detected well before the onset of frank hyperglycemia, even decades earlier, providing evidence that the autoimmune process may be prolonged. After diagnosis, autoantibody levels often decline with increasing duration of the disease. Also, once patients are treated with insulin, low levels of IAA develop, even in patients that do not have an autoimmune etiology for their diabetes.

Although useful for diagnosing and predicting type 1 diabetes, antibodies against β cell proteins do not directly cause the destruction of β cells in type 1 diabetes. Instead, it is the cellular immune system, the T lymphocytes, that infiltrate the islets (a process called insulitis) and destroy the β cells. At the time of diagnosis, the islets of patients with type 1 diabetes are extensively infiltrated with both helper and cytotoxic T lymphocytes.

Normally, the thymus deletes autoreactive T cells during development so that the immune system becomes tolerant of self–antigens. In addition, certain specialized T cells, the regulatory T cells, further prevent attacks against healthy tissues by retraining the activity of any autoreactive cytotoxic and helper T cells that escape the thymus. Type 1 diabetes results from a breakdown in these processes of self-tolerance in the immune system.

Type 1b diabetes: Approximately 5% of patients with the clinical features of type 1 diabetes lack serum evidence of auto-immunity. Some of these individuals have high risk HLA haplotypes (see later) and may have T-cell–mediated β cell destruction in the absence of measurable levels of the known autoantibodies. Others in this group have low-risk HLA haplotypes, and appear to have a nonautoimmune cause for loss of β cell function. Such nonautoimmune type 1 diabetes has been referred to as type 1b diabetes, but a variety of terms has been used. This probably represents a heterogeneous group of disorders that lead to profound β cell dysfunction or loss, absolute insulin deficiency and a syndrome clinically similar to autoimmune type 1a diabetes. Under the accepted classification system, as specific disorders within this subgroup become defined and the genetic or environmental causes are identified, these disorders become reclassified within the group of "Other Specific Types of Diabetes."

Included within this group are patients that present with a course of relapsing diabetic ketoacidosis with intervening normoglycemia that eventually progresses to permanent insulin-deficient diabetes. This disorder, ketosis prone diabetes (KPD, see later), has also been referred to as type 1b diabetes, and may result from unknown environmental insults combined with genetic defects in the β cell.

Autoimmune diabetes and stiff person syndrome: GAD antibodies, the first identified in type 1 diabetes, remain among the most clinically useful. Human pancreatic β cells produce GAD65, which functions as an enzyme that catalyzes the synthesis of GABA from glutamate. GAD65 and the closely related isoform GAD67 are also found in central nervous system inhibitory neurons that secrete GABA. Some patients with GAD antibodies develop a rare neurologic condition, stiff person syndrome, caused by the depletion of GABA in the central nervous system and characterized by progressive rigidity and fluctuating muscle spasms. Approximately half of the patients with stiff person syndrome develop type 1 diabetes.

The vast majority of patients with type 1 diabetes do not develop symptoms of stiff person syndrome, despite the presence of GAD antibodies. The rare patients that develop the syndrome usually have much higher titers of GAD antibodies than typical patients with type 1 diabetes alone.

Genetics of Type 1 Diabetes

Family members of patients with type1 diabetes have an increased lifetime risk of developing type 1 diabetes. The offspring of a mother with type 1 diabetes have a risk of 3%, whereas the risk is 6% for children of affected fathers. The risk in siblings of affected individuals is related to the number of human leukocyte antigen (HLA) haplotypes (see later) that the sibling shares. If one haplotype is shared, the risk is 6% and if two haplotypes are shared, the risk increases to 12% to 25%. For monozygotic twins, the concordance rate reaches 25% to 50%. Although these data demonstrate a strong genetic contribution to the risk of type 1 diabetes, genetics plays an even larger role in type 2 diabetes, and environment also clearly contributes substantially to the risk of type 1 diabetes.

Genes in the major histocompatibility (MHC) locus on the short arm of chromosome 6 explain at least half of the familial aggregation of type 1 diabetes. Within the MHC locus lie a number of closely packed genes involved in the function and regulation of the immune response. Although a number of genes within the MHC locus have been linked to the risk of developing type 1 diabetes, the most important of these are the genes encoding the HLA class II molecules DQ and DR. The professional antigen-presenting cells—dendritic cells, macrophages and B lymphocytes—use the class II molecules on their cell surface to present peptide antigens to T lymphocytes through the T-cell receptor. T cells activated by antigen-presenting cells carry out the β cell destruction that leads to type 1 diabetes. Although exact mechanisms remain uncertain, the variations in the amino acid sequence of individual HLA class II molecules may impact their ability to present specific self-peptides to T cells either in the process of central or peripheral tolerization or later during the development of the autoimmune response, thereby contributing to the risk of developing type 1 diabetes.

The DR haplotypes DR3 and DR4 are major susceptibility risk factors for type 1 diabetes. As many as 95% of type 1 diabetic patients have a DR3 or a DR4 haplotype—or both—compared with 45% to 50% of Caucasian nondiabetic controls. Individuals who express both a DR3 and a DR4 allele carry the highest risk for type 1 diabetes in the United States.

The high-risk DR genes are generally in linkage disequilibrium with DQ genes that themselves confer high risk, particularly DQA1*0501, DQB1*0201 (coupled with DR3), and DQA1*0301, DQB1*0302 (coupled with DR4). DQ alleles are associated not only with risk for type 1 diabetes but also with dominant protection, often in linkage with HLA-DR2. The most protective of these—and a quite common allele—is DQA1*0102, DQB1*0602. It occurs in over 20% of individuals in the United States but in less than 1% of children who develop type 1 diabetes.

An independent genetic link to chromosome 11 has also been identified in type 1 diabetes. Studies of a polymorphic DNA locus flanking the 5′ region of the insulin gene on chromosome 11 revealed a small but statistically significant linkage between type 1 diabetes and this genetic locus in a Caucasian population with type 1 diabetes. This polymorphic locus, which consists of a variable number of tandem repeats (VNTRs) with two common sizes in Caucasians, small (26-63 repeats) or large (140-243 repeats), does not encode a protein. An intriguing proposal to explain how the VNTR might influence susceptibility to type 1 diabetes was based on findings that insulin gene transcription is facilitated in the fetal thymus gland by the presence of the large allele of the VNTR locus flanking the insulin gene. The large VNTR allele might produce a dominant protective effect by promoting negative selection (deletion) by the thymus of insulin-specific T lymphocytes that play a critical role in the immune destruction of pancreatic β cells.

The established genetic association with the MHC region of chromosome 6 contributes much more (about 50%) to the genetic susceptibility to type 1 diabetes than does this locus flanking the insulin gene on chromosome 11, which contributes about 10%. Both candidate gene studies and genome-wide association studies (GWAS) have identified a number of additional risk loci that make smaller contributions to the genetic risk of type 1 diabetes. Many of the genes linked to these additional loci also play important roles in the function and regulation of the immune response.

Mutations in two genes involved in T-cell tolerance cause rare syndromes of type 1 diabetes together with other autoimmune diseases. In the autosomal recessive disease autoimmune polyglandular syndrome type 1 (APS1; see Chapter 2), homozygous mutations in the gene encoding the autoimmune regulator (AIRE) prevent the expression of certain self-proteins in the thymus, thus allowing mature autoreactive T cells to leave the thymus. In addition to other autoimmune diseases and mucocutaneous candidiasis, approximately 20% of patients with APS1 develop type 1 diabetes. The second gene, FOXP3, found on the X chromosome, encodes a transcription factor required for the formation of regulatory T cells. Mutations in FOXP3 cause immunodysregulation polyendocrinopathy enteropathy X-linked (IPEX) syndrome. IPEX presents in male patients with very early onset type 1 diabetes, often neonatal, combined with other autoimmune endocrinopathies, autoimmune skin disorders, diarrhea secondary to autoimmune enteropathy, and frequent severe infections.

Environmental Factors in Type 1 Diabetes

While genetic inheritance may play an important role in causing type 1 diabetes, the monozygotic twin studies demonstrate that other causes, stochastic or environmental, are at least as important. Most individuals with type 1 diabetes do not have other family members with the disease. Environmental factors associated with increased risk of type 1 diabetes include viruses (mumps, congenital rubella, Coxsackie virus B4), toxic chemical agents such as vacor (a nitrophenylurea rat poison), and other destructive cytotoxins such as hydrogen cyanide from spoiled tapioca or cassava root. How these environmental insults lead to type 1 diabetes is unknown; they may directly damage β cells in some cases, or may act as initiators or accelerators of the autoimmune attack on the β cells. In some cases, molecular mimicry, wherein the immune system mistakenly targets β cell proteins that share homologies with certain viral or other foreign peptides may play a role.

Epidemiological studies have demonstrated an association between breast-feeding in the first 6 months of life and protection from type 1 diabetes. While it has been suggested that proteins in cow's milk may be the culprits, the strongest evidence supports the idea that human breast milk may reduce the risk of autoimmune disease.

Accumulating evidence shows that in the process of modernizing and improving public health, the risk of type 1 diabetes has increased, possibly due to the removal of some protective factors. Type 1 diabetes is almost unheard of in many third-world countries, and has its highest incidence in countries with the best public health systems, such as the Scandinavian countries. In addition, the incidence of the disease has been steadily increasing over the past century in western and westernizing countries and is especially high among the more affluent. This has led to the suggestion that a dirty environment, one with more infections (especially more parasitic diseases) and more antigen exposure, may reduce the risk of type 1 disease.

Type 2 Diabetes

Type 2 diabetes mellitus—previously called non-insulin–dependent diabetes or adult-onset diabetes mellitus—results from relative insulin deficiency, in contrast to the absolute insulin deficiency of patients with type 1 diabetes. Type 2 diabetes is a heterogeneous disorder and probably represents a large number of different primary genetic and environmental insults leading to relative insulin deficiency—a mismatch between insulin production and insulin requirements. Clinically, patients with type 2 diabetes can range from those with severe insulin resistance and minimal insulin secretory defects to those with a primary defect in insulin secretion.

Type 2 diabetes accounts for 80% to 90% of cases of diabetes in the United States. These patients commonly present as adults with some degree of obesity, although increasing rates of obesity are leading to earlier onset of the disease in adolescents and children. At onset, most patients with type 2 diabetes do not require insulin to survive, but over time their insulin secretory capacity tends to deteriorate, and many eventually need insulin treatment to achieve optimal glucose control. Ketosis seldom occurs spontaneously, and if present, it is a consequence of severe stress from trauma or infection.

Most patients with type 2 diabetes, irrespective of weight, have some degree of tissue insensitivity to insulin attributable to several interrelated factors (Table 17–7). These include putative (mostly as yet undefined) genetic factors, which are aggravated in time by further enhancers of insulin resistance such as aging, a sedentary lifestyle, and abdominal visceral obesity. Not all patients with obesity and insulin resistance develop hyperglycemia, however. An underlying defect in the ability of the β cells to compensate for the increased demand determines which patients will develop diabetes in the setting of insulin resistance. Furthermore, both the tissue resistance to insulin and the impaired β cell response to glucose appear to be further aggravated by sustained hyperglycemia, which may impede both insulin signaling and β cell function. Treatment that reduces the blood glucose levels toward normal reduces this acquired defect in insulin resistance and may also improve glucose-induced insulin release to some degree, although the long-term decline in β cell function continues.

Type 2 diabetes frequently goes undiagnosed for many years because the hyperglycemia may develop quite gradually and without initial symptoms. Despite this mild presentation, these patients develop microvascular, and, especially, macrovascular complications. Furthermore, as noted above, patients with type 2 diabetes suffer from a progressive decline in β cell capacity, leading to worsening hyperglycemia over time.

Obesity in Type 2 Diabetes

The majority of people with type 2 diabetes have excess adiposity, although the prevalence of obesity in association with type 2 diabetes varies among different racial groups. Sixty to eighty percent of North Americans, Europeans, or Africans with type 2 diabetes and close to 100% of individuals with type 2 disease among Pima Indians or Pacific Islanders from Nauru or Samoa have obesity as defined by body mass index (BMI, see Chapter 20), while as few as 30% of Chinese and Japanese patients with type 2 diabetes are obese. However, many of those individuals with type 2 diabetes who do not meet BMI criteria for obesity have a predominantly abdominal distribution of fat, producing an abnormally high waist to hip ratio. Increases in visceral adiposity correlate with increased insulin resistance.

Insulin Resistance in Type 2 Diabetes

Insulin resistance can be broadly defined as a decrease in tissue responsiveness to insulin. Clinically it can be assessed directly by measuring the ability of a fixed dose of insulin to promote total body glucose disposal. It can be assessed indirectly by measuring fasting insulin levels. An increase in insulin levels with normal plasma glucose indicates insulin resistance.

As adiposity increases, especially abdominal visceral fat deposits, total body insulin sensitivity decreases. Since adipose tissue only removes a small fraction of plasma glucose, clearly the increased adipose fat stores impact total body insulin sensitivity through effects on other tissues, especially muscle and liver, causing them to decrease insulin-stimulated glucose disposal. The exact means by which fat storage in adipocytes affects the insulin sensitivity of other cells remains uncertain, but experimental evidence suggests several possible mechanisms.

Abnormalities of insulin receptors—in concentration, affinity, or both—affect insulin action. Target tissues downregulate the number of insulin receptors on the cell surface in response to chronically elevated circulating insulin levels, probably by increased intracellular degradation. When insulin levels are low, on the other hand, receptor binding is upregulated. Conditions associated with high insulin levels and lowered insulin binding to the receptor include obesity, high intake of carbohydrates, and chronic exogenous overinsulinization. Conditions associated with low insulin levels and increased insulin binding include exercise and fasting. The insulin receptor itself is probably not the major determinant of insulin sensitivity under most circumstances, however. Clinically relevant insulin resistance most commonly results from defects in postreceptor intracellular signaling pathways.

Adipokines. Adipose tissue can affect the insulin sensitivity of other tissues through the secretion of signaling molecules, adipokines, that inhibit (TNF-α, IL-6, leptin, resistin, and others) or enhance (adiponectin) insulin signaling locally or in distal target tissues (see Chapter 20). Levels of fat storage in adipocytes, along with insulin signaling itself, regulate the production and secretion of many of the adipokines. Some of these mediators of insulin resistance may reduce insulin signaling by blocking access of insulin to target tissues through reduced transendothelial transit. However, most evidence suggests that the secreted adipokines influence insulin signaling in distant tissues through effects on postreceptor intracellular signaling pathways. Potential intracellular effectors include protein tyrosine phosphatases that dephosphorylate the receptor and pathway components, inhibitors such as the SOCS proteins that block receptor–IRS interactions, and serine/threonine kinases that inhibit the receptor and substrates through serine phosphorylation.

Free fatty acids and ectopic lipid storage. The release of fatty acids by the engorged adipocytes (especially visceral adipocytes, from which fatty acids are more readily mobilized) may play a role in the development of insulin resistance as well. Oxidation of fatty acids by muscle and other tissues could inhibit glycolysis and reduce insulin-stimulated glucose removal (the Randle hypothesis, named after its original proponent). Increased fat storage in adipocytes and release of fatty acids may also eventually cause a shift in lipid storage, increasing lipid uptake and storage in nonadipose tissues such as muscle, liver, and β cells. Ectopic lipid storage in these tissues may lead to a decrease in their insulin sensitivity. In addition, free fatty acids may function directly in a signaling role both locally within the adipose tissue and systemically.

Inflammation. In addition to adipocytes, adipose tissue contains a variety of other cell types including inflammatory/immune cells, such as macrophages and lymphocytes. Recent evidence implicates these cells in obesity-induced insulin resistance. As adipocyte lipid stores rise, the increased release of free fatty acids and proinflammatory adipokines recruits macrophages to the adipose tissue and activates them. The activated macrophages then release a variety of molecules (TNF-α, IL-6, nitric oxide, and others) that decrease the insulin sensitivity of the adipocytes and further increase their release of proinflammatory fatty acids and peptides, creating a positive feedback loop that maintains a chronic state of local inflammation and insulin resistance. Release of these adipokines and proinflammatory cytokines, along with the increased release of free fatty acids and the development of ectopic lipid accumulation, promotes the development of inflammation and insulin resistance in the other key insulin-target tissues, such as muscle and liver. Similar mechanisms could also lead to inflammation in the islets and contribute to β cell failure.

PPARγ activity in the adipose tissue generally has beneficial effects on systemic insulin signaling through several mechanisms: (1) Promotion of adipose lipid storage, which thereby decreases ectopic lipid storage in nonadipose tissues; (2) Inhibition of the production of adipokines and proinflammatory cytokines, which promote insulin resistance, by adipocytes; (3) Promotion of the alternative activation of macrophages to the anti-inflammatory M2 state, rather than the proinflammatory M1 state; (4) Inhibition of the release of proinflammatory and proresistance cytokines by macrophages. Although it is expressed at much lower levels in muscle than in adipose tissue, PPARγ in myocytes might also have a direct role in controlling muscle insulin sensitivity; however, these findings remain controversial.

Tissue heterogeneity in insulin resistance. Finally, it must be kept in mind that not all tissues necessarily develop insulin resistance in parallel. The combination of local and systemic contributors to obesity-induced insulin resistance may explain the different levels of insulin resistance in different tissues of the same patient. Even in the same cell, insulin resistance may impact different arms of the insulin-signaling pathway discordantly. This heterogeneity leads to changes in tissue energy storage and insulin sensitivity that could explain unique syndromes associated with insulin resistance like hepatic steatosis and polycystic ovary syndrome.

Other causes of insulin resistance. Visceral obesity is not the only cause of insulin resistance, although it is by far the most common cause in most populations. Other causes of insulin resistance include a variety of genetic and acquired defects that impact the insulin receptors or postreceptor signaling pathways (see Table 17–7).

Clinical consequences of insulin resistance. In addition to the impact on glucose metabolism, severe insulin resistance and the resulting elevation in circulating insulin levels can cause other clinical consequences including acanthosis nigricans, pseudoacromegaly, and hyperandrogenism. Acanthosis nigricans appears to be a consequence of very high circulating insulin levels that cross over to bind to IGF receptors on epidermal and melanin–containing cutaneous cells. This leads to local skin hyperplasia with papillomatosis, hyperkeratosis, and hyperpigmentation. The dark, velvety patches of skin most commonly appear on the back of the neck, axillae, and anticubital fossae. In extreme and prolonged cases of insulin resistance, the secondary increase in signaling through the IGF-1 receptor, or possibly residual signaling through the mitogenic arm of the insulin signaling pathway, can cause pseudoacromegaly, a syndrome with all of the bone and soft tissue changes of acromegaly (see Chapter 4), but no elevation in growth hormone or IGF-1. A similar action of extremely high insulin levels on ovarian hilar cells has been implicated in women with insulin resistance who develop hyperandrogenism and hirsutism associated with menstrual irregularities, enlarged cystic ovaries and infertility (polycystic ovary syndrome).

β Cell Defects in Type 2 Diabetes

Although the majority of people with type 2 diabetes have insulin resistance, most people with insulin resistance do not have diabetes because their β cells compensate for the insulin resistance by producing and secreting more insulin. Those individuals with insulin resistance who develop type 2 diabetes have a defect in the compensatory response of their β cells to insulin resistance. Functionally, this defect is revealed by a reduction in first phase insulin secretion and the maximal insulin secretion stimulated by glucose.

While increased insulin secretion per β cell may contribute to the compensatory response to insulin resistance, increases in the number of β cells play a role as well. In the setting of obesity, hyperplasia of pancreatic β cells is often present and probably accounts for the normal or exaggerated insulin responses to glucose and other stimuli seen in obese individuals without type 2 diabetes. Assessment of total β cell mass at autopsy has revealed that β cell mass increases in obesity, but the individuals with type 2 diabetes have decreased β cell mass when compared to nondiabetic individuals with the same BMI.

Several possible defects could contribute to the failure of β cell mass compensation in people with type 2 diabetes. Underlying genetic differences in the pathways that drive β cell expansion appear to limit compensation in individuals with high genetic risk of diabetes. In susceptible individuals with obesity, ectopic fat deposition in the islets, local obesity-induced inflammation in the islets, and local and circulating adipokines and inflammatory cytokines may accelerate β cell loss. As β cell failure progresses, levels of glucose and free fatty acids start to rise, which in turn can cause further β cell toxicity. Increased demand on a decreased β cell mass may cause further damage through ER stress and the increased formation of toxic IAPP oligomers. Then, once diabetes is established, all of these mechanisms may further contribute to the progressive decline in β cell function that characterizes type 2 diabetes.

Metabolic Syndrome

Patients with visceral obesity and insulin resistance often present with a cluster of abnormalities commonly termed the metabolic syndrome. Hyperglycemia in these patients is frequently associated with hyperinsulinemia, dyslipidemia, and hypertension, which together lead to coronary artery disease and stroke. It has been suggested that this aggregation results from a genetic defect producing insulin resistance, particularly when obesity aggravates the degree of insulin resistance. In this model, impaired action of insulin predisposes to hyperglycemia, which in turn induces hyperinsulinemia. If this hyperinsulinemia is of insufficient magnitude to correct the hyperglycemia, type 2 diabetes is manifested. The excessive insulin level could also increase sodium retention by renal tubules, thereby contributing to or causing hypertension. Increased VLDL production in the liver, leading to hypertriglyceridemia (and consequently a decreased high-density lipoprotein [HDL] cholesterol level), has also been attributed to hyperinsulinism. Moreover, it has been proposed that high insulin levels can stimulate endothelial and vascular smooth muscle cell proliferation—by virtue of the hormone's action on growth factor receptors—to promote atherosclerosis.

Although there is full agreement on an association of these disorders, the mechanism of their interrelationship remains speculative and open to experimental investigation. Controversy persists about whether or not hypertension is caused by the hyperinsulinism that results from insulin resistance. Moreover, patients with hyperinsulinism due to an insulinoma are generally normotensive, and there is no reduction of blood pressure after surgical removal of the insulinoma restores normal insulin levels.

An alternative unifying hypothesis could be that visceral obesity directly induces the other components of this syndrome. Visceral obesity is an independent risk factor for all of the other components of the metabolic syndrome. In addition to the metabolic effects of visceral obesity, the adipokines and inflammatory cytokines generated from overloaded and inflamed adipose tissue may contribute to the pathophysiology of the syndrome. Although the full details of the role of these molecules in causation of the metabolic syndrome remain under investigation, the adipocytes and associated macrophages clearly are not just innocent bystanders but play active roles in the development of systemic insulin resistance, hypertension, and hyperlipidemia. Furthermore, thrombi in atheromatous vessels may be more hazardous in patients with visceral obesity because of an associated increase in plasminogen activator inhibitor-1 (PAI-1), a circulating factor produced by omental and visceral adipocytes that inhibits clot lysis. This model emphasizes the importance of measures such as diet and exercise that reduce visceral adiposity in the management of patients with metabolic syndrome and obese type 2 diabetes.

The main value of grouping these disorders as a syndrome, regardless of its nomenclature, is to remind physicians that the therapeutic goals in these patients must not only correct hyperglycemia but also manage the elevated blood pressure and hyperlipidemia that result in considerable cardiovascular morbidity as well as cardiovascular deaths. In addition, it reminds physicians that when choosing antihypertensive agents or lipid-lowering drugs to manage one of the components of this syndrome, their possible untoward effects on other components of the syndrome should be carefully considered. For example, physicians aware of this syndrome are less likely to prescribe antihypertensive drugs that raise lipids (diuretics, beta- blockers) or that raise blood glucose (diuretics). Likewise, they may refrain from prescribing drugs that correct hyperlipidemia, but increase insulin resistance with aggravation of hyperglycemia (nicotinic acid).

Genetics of Type 2 Diabetes

Type 2 diabetes has a strong genetic link. Depending on the population studied, monozygotic twins have lifetime concordance rates for type 2 diabetes exceeding 90%. In contrast, concordance rates for type 1 diabetes in monzygotic twins are 25% to 50%. Most individuals with type 2 diabetes have other family members with the disease, but the inheritance rarely fits Mendelian patterns, supporting the conclusion that multiple genes with varying degrees of penetrance contribute. Because of the heterogeneous nature of type 2 diabetes, and its complex inheritance, efforts to identify the genes that contribute to the disease have had very limited success in the vast majority of affected patients. There has been considerable success, however, in identifying small subsets of patients with unique monogenic forms of the disease. When the etiologic defect has been defined, these patients have been reclassified within a group designated "Other Specific Types of Diabetes" (see Table 17–5).

Efforts to identify the genes involved in polygenic type 2 diabetes have focused on two approaches: candidate gene testing and genome-wide association studies (GWAS). To date candidate gene and GWAS approaches have identified 19 loci with common variants linked to type 2 diabetes. Although statistically significant and validated in additional populations, these loci independently make very small contributions to type 2 diabetes risk. The highest risk of these common variants is found at a locus adjacent to the gene encoding TCF7L2, a transcription factor involved in Wnt signaling and implicated in β cell turnover. Inheritance of the high-risk TCF7L2 allele increases the probability of developing diabetes by 1.5-fold.

Among the genes identified so far, most are involved in β cell function and turnover. When combined with the predominance of β cell genes implicated in mongenic forms of diabetes (see Other Specific Types of Diabetes), these results reinforce the critical role of the β cell in controlling blood glucose and its involvement in the pathophysiology of type 2 diabetes. Hopefully, with the advent of high-throughput whole genome sequencing technologies, the identification of rarer, but higher risk, variants will further expand our understanding of the genetics of type 2 diabetes in the near future.

Environmental Factors in Type 2 Diabetes

Despite the critical role of genetics in type 2 diabetes, environment contributes as well, especially in determining the age of onset and severity of the disease. There is generally a low incidence of type 2 diabetes in underdeveloped countries, especially in rural areas. Western countries and westernizing countries suffer from a much higher incidence. Over the past half-century, the incidence of type 2 diabetes has increased rapidly in almost all world populations but especially in emerging third-world countries. This increase correlates with increasing rates of obesity in the same populations and reflects increased access to food with high caloric content and decreased physical activity. This combination inevitably leads to increased adiposity, especially in the more readily mobilized fat stores surrounding the viscera in the abdomen.

One of the most dramatic recent changes in the epidemiology of diabetes has been the growing incidence of type 2 diabetes in children. While rarely seen in children a generation ago, type 2 diabetes is now as common as type 1 diabetes in teenagers in the United States and is seen with increasing frequency, even in younger children. Again, this increase is directly related to increasing visceral adiposity.

Other Specific Types of Diabetes

Autosomal Dominant Genetic Defects of Pancreatic a Cells

MODY: This subgroup of monogenic disorders is characterized by the onset of diabetes in late childhood or before the age of 25 years as a result of a partial defect in glucose-induced insulin release and accounts for up to 5% of diabetes in North American and European populations. A strong family history of early-onset diabetes occurring in one parent and in one-half of the parent's offspring suggests autosomal dominant transmission. In contrast to most patients with type 2 diabetes, these patients are generally nonobese and lack associated insulin resistance. Instead they exhibit predominantly a defect in glucose-stimulated insulin secretion. However, because they are not ketosis-prone and may initially achieve good glycemic control without insulin therapy, their disease has been called maturity-onset diabetes of the young (MODY). Several distinct types have been described with single-gene defects, and all have been shown to produce a defect in glucose-induced insulin release. MODY 2 results from an abnormal glucokinase enzyme. Most of the other forms of MODY are due to mutations of nuclear transcription factors that regulate the expression of genes in β cells or β cell precursors (Table 17–8).

MODY 1 includes multiple members of a large pedigree known as the R-W family, descendants of a German couple who immigrated to Michigan in 1861. They were studied prospectively since 1958, and in 1996 the genetic defect was shown to be a nonsense mutation of a nuclear transcription factor found in liver as well as in pancreatic β cells. This gene has been termed hepatocyte nuclear factor 4α (HNF4α) and is found on chromosome 20. Mutations of this gene are among the rarest of the MODY groups, with very few reported in families outside the Michigan pedigree. These patients display a progressive decline in β cell function and eventually develop chronic complications of diabetes including microangiopathy at a rate approaching that of people with type 1 diabetes. They often fare better with insulin therapy.

MODY 2 was first described in French families but has now been found in racial groups from most parts of the world. Multiple different mutations of the glucokinase gene (GCK) on chromosome 7 have been identified and characterized. In pancreatic β cells, glucokinase controls the rate-limiting step in glycolysis and thereby determines the rate of ATP production from glucose and the insulin secretory response to glucose (see Figure 17–5). Reduced glucokinase activity resets the sensitivity of the β cell to glucose so that it requires higher plasma glucose levels to stimulate insulin secretion, resulting in fasting hyperglycemia and mild diabetes. Although some of these mutations can completely block the enzyme's function, others interfere only slightly with its action. In contrast to all the other forms of MODY, most patients with one mutated GCK allele (heterozygotes) have a benign course with few or no chronic complications and respond well to diet therapy or oral antidiabetic drugs without the need for insulin treatment. On the other hand, rare individuals who inherit two mutated GCK alleles have permanent neonatal diabetes (see later), a nonimmune form of absolute insulin deficiency that presents at birth.

In contrast to the mutations that reduce glucokinase enzyme activity and cause MODY 2, rare mutations in GCK that increase the enzymatic activity of glucokinase can cause increased insulin secretion and hypoglycemia (see Chapter 18), demonstrating the key role of this enzyme in determining the sensitivity of the β cell to glucose.

MODY 3 is caused by mutations of hepatocyte nuclear factor 1α (HNF1α), whose gene is located on chromosome 12. This is the most common form of MODY in European populations, with many different mutations having been reported. Like HNF4α, the HNF1α transcription factor is expressed in pancreatic β cells as well as in liver. Also similar to HNF4α, mutations in HNF1α cause a progressive form of diabetes with declining β cell function that often leads to dependence on insulin therapy and high rates of microvascular complications. Noteworthy, early in the course of the disease, these patients may display an exaggerated response to sulfonylureas.

Together, HNF1α and HNF4α, along with several other β cell transcription factors including PDX1 (discussed later), form an interacting network of transcription factors. This transcriptional network regulates genes involved in multiple β cell functions, including glucose-sensing and insulin secretion. Target genes include GCK, as well as genes implicated in the formation, maturation and expansion of β cells. Impairment in β cell formation and regeneration may explain the progressive nature of the MODY transcription factor syndromes and reinforces the importance of β cell mass in preventing hyperglycemia.

MODY 4 results from mutation of a pancreatic nuclear transcription factor known as pancreatic and duodenal homeobox-1 (PDX1), whose gene is on chromosome 13. It mediates insulin gene transcription and regulates expression of other β cell-specific genes including GCK. When both alleles of this gene are nonfunctioning, agenesis of the entire pancreas results; but in the presence of a heterozygous mutation of PDX1, a mild form of MODY has been described in which affected individuals developed diabetes at a later age (mean onset at 35 years) than occurs with the other forms of MODY, in whom onset generally occurs before the age of 25 years.

MODY 5 was initially reported in a Japanese family with a mutation of HNF1β, a hepatic nuclear transcription factor closely related in structure and molecular function to HNF1α. The two HNF1 factors, however, are expressed by different cells. HNF1β is expressed early in the development of the liver, pancreas, kidneys, and genitourinary system, and is not found in mature β cells.

Mutations in this gene cause a moderately severe form of MODY with progression to insulin treatment and severe diabetic complications in those affected. Consistent with its expression pattern early in development, HNF1β mutations also frequently cause reduction in the overall size of the pancreas, decreased insulin production, and congenital defects in the kidney and urogenital tract. Patients may also suffer from varying degrees of cholestatic jaundice, hyperuricemia, nephropathy, and hypomagnesemia secondary to renal magnesium wasting.

MODY 6, a milder form of MODY similar to MODY 4, results from mutations in the gene encoding the islet transcription factor NeuroD1. Like PDX1, NeuroD1 plays an important role in the expression of insulin and other β cell genes, and in the formation and maintenance of β cells.

Other MODY genes: The six MODY genes listed above explain the majority of cases of MODY in patients of European ancestry, but less than half of those in non-European populations. Several rare variants in other genes have been implicated in autosomal dominant diabetes in a few families (see Table 17–8); however, a consensus has not yet been reached that these families fit the criteria for MODY and that the reported variants cause the disorder. The causative genes in most non-Europeans with MODY remain unknown.

The identification of mutations in multiple genes encoding pancreatic transcription factors in patients with MODY has led to the screening of other genes encoding pancreatic transcription factors in patients with diabetes. Heterozygous mutations in genes encoding several transcription factors, including Isl1, Pax6, and Pax4, have been identified in patients with later onset diabetes. The association of diabetes with heterozygous mutations in so many β cell genes highlights the critical importance of optimal β cell function in metabolic regulation. Even modest defects in glucose-induced insulin secretion can result in hyperglycemia.

Insulin mutations: Sequencing of the insulin gene more than 30 years ago led to the first descriptions of heterozygous mutations in the coding sequence of the insulin gene that produce abnormal circulating forms of insulin. Most of these initial cases presented with high circulating levels of insulin, but normal insulin sensitivity and normal glucose levels. Because the abnormal insulins in these cases bind to receptors poorly, they have very low biologic activity and are cleared at a slower rate, leading to accumulation in the blood at higher levels than normal insulin and a subnormal molar ratio of C peptide to immunoreactive insulin. They typically are not associated with hyperglycemia.

However, a mutation in the insulin B chain in one family was associated with decreased circulating levels of both the mutant and normal insulins, and diabetes. Subsequent extensive sequencing of the INS gene has identified several other mutant insulins that produce a similar heterozygous form of diabetes. Modeling of these dominant insulin mutations in mice demonstrates that they lead to the accumulation of aberrantly folded proteins in the endoplasmic reticulum, activation of the unfolded protein response in the ER, and β cell apoptosis. Patients with diabetes secondary to INS gene mutations usually present at a younger age than most patients with MODY, often developing the disease as neonates (see later). Because of the profound β cell loss, these patients follow a disease course similar to type 1 diabetes with absolute insulin deficiency and ketosis, and they require insulin therapy.

This syndrome highlights the sensitivity of the β cells to ER stress, which may explain why β cells often fail when presented with the increased insulin demands associated with insulin resistance, the toxicity of IAPP oligomers, or mutations in genes involved in the unfolded protein response pathway (see later).

Mutations in the subunits of the ATP-sensitive potassium channel: β cells sense rising blood glucose concentrations by increasing the production of ATP from glucose. The rising intracellular ATP levels cause the closure of ATP-sensitive potassium channels on the cell surface, which depolarizes the cell and sets off a cascade of events that leads to the secretion of insulin (Figure 17–5). Rare dominant activating mutations in either of the two units of the channel, SUR1 and Kir6.2 (gene names ABCC8 and KCNJ11, respectively), can cause the channels to remain open and prevent glucose-induced depolarization and insulin secretion. Children heterozygous for these mutations present with early-onset diabetes, commonly as neonates, and may have associated neurologic deficits implying a role for these channels in the central nervous system. Depending on the exact mutation, some of these children can still respond to treatment with sulfonylureas, which may also ameliorate the neurologic symptoms.

Other Genetic Defects of Pancreatic β Cells

Autosomal recessive genetic defects: Although less common than the autosomal dominant β cell disorders, mutations in several genes causing autosomal recessive syndromes with defects in β cell function have been identified in patients with diabetes (Table 17–9). Due to the severity of the β cell defect, many of these present with neonatal diabetes. This group of disorders includes homozygous mutations in the MODY genes GCK and PDX1. Homozygous mutations in GCK cause a much more severe syndrome than the mild glucose-sensing defect seen in MODY 2. Patients with homozygous GCK mutations present at birth with severe hyperglycemia and require insulin therapy.

In patients with mutations in both alleles of PDX1, the pancreas fails to form, and they have pancreatic exocrine deficiency as well as diabetes. Homozygous mutations in several other pancreatic transcription factor genes have been described as well, including PTF1A, NEUROG3, RFX6, and GLIS3. Like PDX1, homozygous mutation of PTF1A causes diabetes and pancreatic agenesis, but it also causes cerebellar atrophy as well. The transcription factor Neurog3 drives the formation of the endocrine cells in both the pancreas and gut. In addition to diabetes onset prior to puberty, infants born with homozygous NEUROG3 mutations have severe malabsorption associated with a lack of gut endocrine cells from birth.

Homozygous mutations in RFX6, which encodes a transcription factor that functions downstream of NEUROG3 and upstream of PDX1 in β cell development, cause Mitchell–Riley syndrome in which neonates present with diabetes in association with complete absence of all islet cell types except PP cells, hypoplasia of the pancreas and gall bladder, intestinal atresia, and severe malabsorption.

The zinc finger transcription factor GLIS3 is expressed broadly in many tissues and plays a role in the transcription of the insulin gene. Homozygous mutations in GLIS3 cause congenital hypothyroidism in addition to neonatal diabetes.

In the autosomal recessive Wolcott-Rallison syndrome, affected children present with neonatal diabetes, epiphyseal dysplasia, and growth retardation together with a variety of progressive hepatic, renal, cardiac, and pancreatic exocrine defects and developmental delay. The causative gene, EIF2AK3, encodes a kinase (PKR-like ER kinase [PERK]) activated by the presence of unfolded proteins in the ER. PERK controls one of the three parallel arms of the unfolded protein response; inositol-requiring enzyme 1 (IRE1) and activating transcription factor 6 (ATF6) activate the other two arms. Together these three molecules activate signaling pathways that protect the cell from ER stress but lead to apoptosis when these protective mechanisms fail. Mice lacking PERK have an inadequate response to ER stress, which leads to accelerated β cell apoptosis. With the high load of insulin production in the ER, β cells are uniquely sensitive to ER stress, and this sensitivity probably underlies the damage caused by mutant insulins, IAPP oligomers, and Wolfram syndrome as well.

Wolfram syndrome is an autosomal recessive neurodegenerative disorder first evident in childhood. Patients present with diabetes insipidus, diabetes mellitus, optic atrophy, and deafness—hence the acronym DIDMOAD. Diabetes mellitus usually develops in the first decade together with the optic atrophy, followed by central diabetes insipidus and sensorineural deafness during the second decade in 60% to 75% of patients. Ureterohydronephrosis, neurogenic bladder, cerebellar ataxia, peripheral neuropathy, and psychiatric illness develop later in many patients. The diabetes mellitus is nonimmune and not linked to specific HLA antigens, but on autopsy these patients have selective loss of β cells in the pancreas. Genetic studies mapped the causative mutations to a gene called WFS1, which encodes a 100.3-kDa transmembrane protein localized to the endoplasmic reticulum membranes of all cells. WFS1 is expressed at particularly high levels in β cells. Studies in mice have shown that the WFS1 protein forms part of the unfolded protein response downstream of PERK and IRE1 and helps protect the β cells from ER stress and apoptosis, especially during periods of high insulin demand.

Children with thiamine-responsive megaloblastic anemia syndrome carry mutations in the high-affinity thiamine transporter SLC19A2 found on cell and mitochondrial membranes. They develop megalobastic anemia, diabetes, and sensorineuronal deafness. The diabetes usually presents in the first decade of life. In the absence of SLC19A2, cells and mitochondria can still transport thiamine through lower affinity transporters, and both the anemia and the diabetes respond to pharmacologic treatment with thiamine. However, all patients eventually require insulin replacement despite thiamine therapy. It remains unclear how partial defects in cellular and mitochondrial thiamine transport cause β cell failure.

Mitochondrial DNA mutations: Because sperm do not contain mitochondria, only the mother transmits mitochondrial genes to her offspring. Diabetes due to a mutation of mitochondrial DNA that impairs the transfer of leucine into mitochondrial proteins has now been described in a large number of families, and results from impaired β cell function. The incidence of this disorder is as high as 1% to 3% in patients with diabetes in Japan and Korea, but lower in European populations. Most patients have a mild form of maternally transmitted diabetes with insulin deficiency that responds to oral hypoglycemic agents; however, some patients have a more severe clinical picture similar to type 1a diabetes. As many as 63% of patients with this subtype of diabetes have hearing loss and a smaller proportion (15%) have a syndrome of myopathy, encephalopathy, lactic acidosis, and stroke-like episodes.

Ketosis-Prone Diabetes

First described in young adult African American males from the Flatbush neighborhood in New York City, but since described in a number of populations of both African and non-African ancestry, these patients typically present with diabetic ketoacidosis and absolute insulin deficiency, followed by extended clinical remission off insulin. When tested during remission, however, maximal insulin secretory capacity remains markedly reduced, and these patients follow a relapsing course of DKA and hyperglycemia with eventual permanent insulin-deficient diabetes. These patients do not have islet cell autoantibodies or increased frequencies of HLA haplotypes associated with risk of autoimmune type 1a diabetes. Although there is often a family history of similar diabetes, the inheritance is not clearly Mendelian. Although KPD patients have been distinguished based on the history of relapsing β cell dysfunction, it remains controversial as to whether KPD represents a clinical entity distinct from other forms of nonautoimmune type 1b diabetes.

In a study of KPD patients of West African ancestry, linkage was established with coding variants in the PAX4 gene. PAX4 encodes for a transcription factor that functions downstream of NEUROG3 in the formation of β cells. Fourteen to twenty-one percent of individuals of West African ancestry carry a coding variant that substitutes tryptophan for arginine at position 133 in Pax4 and reduces its ability to repress transcription of target genes. Interestingly, the R133W variant is unique to people of West African ancestry. The R133W allele is approximately twice as frequent in individuals with KPD, and all individuals homozygous for R133W have KPD. An additional Pax4 coding variant with reduced ability to bind to DNA was identified in one patient with KPD. Different coding variants in PAX4 have also been identified in Thai families with MODY (MODY 9) and in Japanese patients (both heterozygous and homozygous) with early onset insulin-deficient diabetes similar to KPD. Taken together, these data suggest that coding variants in the PAX4 gene predispose to insulin-deficient diabetes, but the clinical phenotype may depend on the exact nature of the variant and interactions with other genes and environmental factors.

Genetic Defects of Insulin Action

These are rare and unusual causes of diabetes that result from mutations of the insulin receptor or from other genetically determined postreceptor abnormalities of insulin action. Metabolic abnormalities associated with these disorders may range from hyperinsulinemia and modest hyperglycemia to severe diabetes. Many individuals have acanthosis nigricans, polycystic ovaries with hyperandrogenism, and, in exceptional cases, pseudoacromegaly (see earlier).

Familial forms of insulin resistance associated with acanthosis nigricans have been termed type A insulin resistance, and are often associated with heterozygous mutations in the insulin receptor or homozygous mutations that retain some insulin-signaling capacity. Because of the capacity of the β cell to compensate for insulin resistance, these patients often do not present with hyperglycemia, but have severely elevated levels of circulating insulin.

If both copies of the insulin receptor gene carry mutations that nearly or completely abrogate signaling, the affected children present at birth with a syndrome known as Leprechaunism (or Donohue syndrome) that includes growth retardation, multiple developmental defects, lipoatrophy, severely elevated insulin levels, and hyperglycemia that does not respond to insulin therapy. These patients generally do not survive beyond a few weeks. Patients with Rabson-Mendenhall syndrome also have homozygous or compound heterozygous mutations in the insulin receptor, but with some small amount of residual signaling that results in a slightly less severe syndrome associated with abnormalities in the nails, teeth, and pineal gland. They may survive into adolescence.

Familial forms of type A insulin resistance without mutations in the insulin receptor have been described, and some of these may result from mutations in downstream components of the insulin-signaling cascade, although to date few such mutations have been found in humans. Variants in IRS1 have been identified in patients with diabetes, but also occur in people with normal insulin sensitivity. Rare coding mutations in the genes encoding the p85α subunit of PI3 kinase, in AKT2, and in PPARγ have been associated with severe insulin resistance in a very small number of patients. The one patient with homozygous mutation of AKT2 also had mild atrophy of subcutaneous limb adipose tissue.

Congenital and acquired forms of lipodystrophy can cause severe insulin resistance. Except in newborns with lipoatrophy secondary to complete loss of insulin receptors (Leprechaunism), alterations in the structure and function of the insulin receptor cannot be demonstrated in patients with insulin-resistant lipodystrophic diabetes, suggesting that the cause of the insulin resistance in these patients must reside in postreceptor pathways. Replacement of the adipokines leptin and adiponectin can reverse insulin resistance in mouse models of severe lipoatrophy, and leptin has been helpful in human cases of generalized lipodystrophy, demonstrating the importance of the adipocyte in regulating insulin function. Also, the loss of adipose storage depots in lipoatrophy leads to very high levels of circulating triglyceride-rich lipoprotein particles and increased deposition of fat in nonadipose tissues such as liver and muscle, which may contribute to dysfunction and insulin resistance in these tissues. The increase in ectopic fat deposition leads to profound hepatic steatosis in affected patients and can progress to cirrhosis and liver failure.

The congenital syndromes can be divided into generalized and partial lipodystrophies with or without dystrophic features in other tissues. The generalized syndromes are often identified in neonates, have severe insulin resistance at diagnosis and rapidly develop hyperglycemia, acanthosis nigricans, hyperandrogenism, and pseudoacromegaly. Recessive mutations causing generalized congenital lipodystrophy have been identified in three genes encoding proteins involved in the formation of lipid droplets (seipin and caveolin-1) and triglyceride (1-acylglycerol-3–phosphate-O-acyltransferase 2 [AGPAT2]).

Two syndromes of familial partial lipodystrophy without associated dysmorphic defects have been described in which the lipoatrophy usually first appears late in childhood, but may be proceeded by evidence of insulin resistance. The type 1 syndrome consists of atrophy of limb, gluteal and subcutaneous abdominal fat with sparing or increases of the abdominal visceral, upper trunk, head, and neck fat. In the type 2 syndrome, truncal and visceral fat are also affected and only vulval and head and neck depots are spared.

To date no genetic causes of the type 1 syndrome have been identified. In the type 2 syndrome, dominant mutations have been found in LMNA, the gene encoding the nuclear intermediate filament lamin A/C. In addition, one individual with the type 2 syndrome has been identified with a homozygous nonsense mutation in the gene encoding the lipid droplet protein CIDEC, and one individual was identified with a homozygous truncation of the gene encoding lipase maturation factor 1, a protein required for the maturation of both lipoprotein lipase and hepatic lipase.

Consistent with its known role in adipocytes differentiation, dominant mutations in the gene encoding PPARγ have also been described in patients with familial partial lipodystrophy and insulin resistance. These patients have a pattern of lipoatrophy similar to the type 2 syndrome with LMNA mutations, but with less severe subcutaneous fat loss, and have been labeled type 3 familial partial lipodystrophy.

Several syndromes of lipodystrophy associated with other dysmorphic features have been described. Autosomal recessive mutations in ZMPSTE24, which encodes a metalloproteinase that cleaves prolamin A to produce mature lamin A, cause generalized lipodystrophy associated with mandibulo-acral dysplasia. In patients with partial lipodystrophy combined with mandibulo-acral dysplasia, autosomal recessive mutations have been identified in LMNA, but these mutations are distinct from those that cause familial partial lipodystrophy type 2. Remarkably, yet a different set of mutations in LMNA cause the autosomal recessive syndrome Hutchinson–Gilford progeria, which includes severe early onset generalized lipodystrophy. Additional distinct mutations in LMNA cause several other congenital dysmorphic syndromes, including muscular dystrophies, familial dilated cardiomyopathy, and Charcot-Marie-Tooth disease.

Acquired forms of partial and generalized lipodystrophy with insulin resistance and diabetes can develop secondary to infections, autoimmunity, paraneoplastic syndromes, collagen vascular disorders, drugs, or unknown causes. One common form is seen in patients with HIV infection following treatment with protease inhibitors.

Neonatal Diabetes

Neonatal diabetes, defined as diabetes diagnosed before 6 months of age, is rare, occurring in fewer than 1 in 200,000 live births. Children with neonatal diabetes often present with decreased birth weight (intrauterine growth retardation, IUGR) and decreased fat stores in addition to hyperglycemia. Most commonly these children have reduced circulating insulin and C-peptide levels caused by inherited β cells defects, although rare inherited defects in insulin signaling can also present in neonates. In approximately half of cases, neonatal diabetes is transient: the diabetes goes into remission (normoglycemia with no therapy) before 18 months of age, although it usually returns at puberty. In the remainder of cases, the diabetes is permanent.

Transient neonatal diabetes (TNDM): Defects in imprinting underlie most cases of TNDM. The most common genetic defect is paternal uniparental isodysomy (replacement of the maternal copy of the region with the paternal copy) of an imprinted region at chromosome 6q24. In this region, the maternal copy of the chromosome is normally silenced by methylation. Simple duplication of the paternal copy of this region, or mutations in ZFP57, a zinc-finger protein required globally for the methylation of imprinted regions, can give the same TNDM phenotype. The affected region of 6q24 contains two genes of uncertain function, but one of them, ZAC, has been identified as a tumor suppressor gene and may activate the expression of an inhibitor of cell proliferation. An increase in hypomethylated copies of ZAC may lead to the inhibition of β cell proliferation and inadequate β cell mass.

Some autosomal dominant activating mutations in KCNJ11 and ABCC8, the genes encoding the two subunits of the ATP-sensitive potassium channels in β cells (Figure 17–5) and listed among autosomal dominant forms of heritable diabetes (see earlier) can cause TNDM. Autosomal dominant mutations in the MODY 5 gene, HNF1B, can also cause TNDM.

Permanent neonatal diabetes (PNDM): Autosomal dominant activating mutations in KCNJ11 and ABCC8 can also cause PNDM. Some of the autosomal dominant insulin mutations that cause rapid β cell apoptosis will result in permanent diabetes in neonates. In addition, most of the syndromes caused by autosomal recessive genetic defects in β cells (see Table 17–9) present in neonates. These include mutations in the MODY genes GCK and PDX1, the transcription factor genes PTF1A, GLIS3, and RFX6, and the Wolcott–Rallison syndrome gene EIF2AK3. The IPEX syndrome (see Genetics of Type 1 Diabetes above), caused by mutations in FOXP3, can also present with accelerated autoimmune type 1 diabetes in neonates. Finally, mutations that cause complete, or nearly complete, loss of insulin signaling (Leprechaunism) also present with PNDM.

Several other syndromes of neonatal diabetes of unknown etiology associated with a variety of other developmental defects have been described. With the advent of rapid whole genome sequencing, the genetic defects that cause these syndromes, as well as later onset diabetes, may soon be identified, providing further insights into the pathogenesis of diabetes and identifying potential therapeutic targets.

Diabetes Due to Diseases of the Exocrine Pancreas

Any process that diffusely damages or substantially displaces the pancreas can cause diabetes, although individuals with a predisposition to type 2 diabetes are probably more susceptible to developing diabetes with lesser degrees of pancreatic involvement. Because glucagon-secreting α cells are also damaged or removed by these processes, less insulin is usually required for replacement—as compared with most other forms of diabetes that leave α cells intact.

Acquired causes include pancreatitis, trauma, infection, pancreatic carcinoma, and pancreatectomy. Fibrocalculous pancreatopathy, a form of acquired pancreatitis with extensive fibrosis and ductal calculi seen commonly in tropical regions, may result from both dietary and genetic contributors, although the exact cause remains obscure. Like chronic pancreatitis from a variety of causes, fibrocalculous involvement of the pancreas may be accompanied by abdominal pain radiating to the back and associated with pancreatic calcifications on x-ray. When extensive enough, hemochromatosis and cystic fibrosis can also displace β cells and reduce insulin secretion. Autosomal dominant mutations in carboxyl-ester lipase (CEL), an exocrine enzyme, cause accelerated exocrine pancreatic damage and diabetes at a young age, and have been designated as MODY 8 (see Table 17–8).

Endocrinopathies

Excess production of certain hormones—GH (acromegaly), glucocorticoids (Cushing syndrome or disease), catecholamines (pheochromocytoma), thyroid hormone (thyrotoxicosis), glucagon (glucagonoma), or pancreatic somatostatin (somatostatinoma)—can produce relative insulin deficiency and diabetes by a number of mechanisms. In all but the last instance (somatostatinoma), peripheral responsiveness to insulin is impaired. In addition, excess of catecholamines or somatostatin decreases insulin release from β cells. Diabetes mainly occurs in individuals with underlying defects in insulin secretion, and hyperglycemia typically resolves when the hormone excess is corrected.

Drug- or Chemical-Induced Diabetes

Many drugs are associated with carbohydrate intolerance or frank diabetes mellitus. Some act by interfering with insulin release from the β cells (thiazides, phenytoin, cyclosporine), some by inducing insulin resistance (glucocorticoids, oral contraceptive pills, niacin, and antiviral protease inhibitors), and some by causing β cell destruction (intravenous pentamidine). Patients receiving β interferon have been reported to develop diabetes associated with β cell antibodies and in certain instances severe insulin deficiency. Atypical antipsychotic medications can provoke substantial weight gain and insulin resistance, but the high reported incidence of diabetic ketoacidosis in patients on these drugs suggests that they may also impair β cell function.

Adrenergic drugs impact glucose metabolism in complex and often opposing ways because of differing effects on insulin secretion, glucagon secretion, hepatic glucose output, peripheral insulin sensitivity, and weight gain. In clinical practice, the first generation, nonselective β-blockers such as propanolol tend to modestly increase glucose levels, at least in part due to increases in insulin resistance, but potentially by decreasing insulin secretion as well. The second generation, selective β1 blockers also tend to increase blood glucose, but the third generation drugs with combined α and β blockade have minimal effects on blood glucose. In contrast, nonselective α agonist and α2 agonists tend to raise blood glucose, probably due to their combined effects on insulin secretion and hepatic glucose output. However, although β blockers and α2 agonists like clonidine, as well as calcium channel blockers, inhibit glucose-induced insulin release from in vitro preparations of pancreatic β cells, these drugs have minimal or modest effects on blood glucose control at the levels used in standard antihypertensive therapy in humans.

Finally it must be kept in mind that the most common toxin causing diabetes is ethanol. Chronic alcoholic pancreatitis with secondary loss of β cells accounts for approximately 1% of diabetes in the United States.

Infections Causing Diabetes

Certain viruses have been associated with direct pancreatic β cell destruction in animals. Diabetes is also known to develop frequently in humans who had congenital rubella, although most of these patients have HLA and immune markers characteristic of type 1 diabetes. In addition, coxsackievirus B, cytomegalovirus, adenovirus, and mumps have been implicated in inducing certain cases of diabetes.

Uncommon Forms of Immune-Mediated Diabetes

A severe form of insulin resistance has been reported in patients who developed high titers of antibodies that bind to the insulin receptor and block the action of insulin in its target tissues. As in other states of extreme insulin resistance, these patients often have acanthosis nigricans. In the past, this form of immune-mediated diabetes was termed type B insulin resistance. Most commonly these antibodies are of idiopathic origin, but they have also been described in monoclonal gamopathies and multiple myeloma and in patients with ataxia telangectasia syndrome.

Several syndromes of altered immune function with multiple endocrine gland involvement have been described. The APS1 and IPEX syndromes are described in the section on the genetics of type 1 diabetes and in Chapter 2. Patients with POEMS, a syndrome of plasma cell dyscrasia associated with polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes, have an increased incidence of diabetes as well as other endocrine disorders. The cause of the diabetes in these patients has not been established.

Other Genetic Syndromes Sometimes Associated with Diabetes

More than 50 distinct genetic syndromes involve an increased incidence of diabetes mellitus. These include the chromosomal abnormalities of Down syndrome, Klinefelter syndrome, and Turner syndrome. In addition, a number of complex syndromes associated with neuromuscular pathologies (Freidriech ataxia, Huntington chorea, porphyria, muscular dystrophies) or severe obesity (Laurence-Moon-Biedl, Bardet-Biedl, and Prader-Willi syndromes) have been associated with diabetes.

Clinical Features of Diabetes Mellitus

The principal clinical features of the two major types of diabetes mellitus are listed for comparison in Table 17–10.

Type 1 Diabetes

Patients with type 1 diabetes present with symptoms and signs related to hyperglycemia and hyperketonemia. The severity of the insulin deficiency and the acuteness with which the catabolic state develops determine the intensity of the osmotic and ketotic excess.

Symptoms

Increased urination is a consequence of osmotic diuresis secondary to sustained hyperglycemia. This results in a loss of glucose as well as free water and electrolytes in the urine. Nocturnal enuresis due to polyuria may signal the onset of diabetes in very young children. Thirst is a consequence of the hyperosmolar state, as is blurred vision, which often develops as the lenses and retinas are exposed to hyperosmolar fluids.

Weight loss, despite normal or increased appetite, is a common feature of type 1 diabetes when it develops subacutely over a period of weeks. The weight loss is initially due to depletion of water, glycogen, and triglyceride stores. Chronic weight loss due to reduced muscle mass occurs as amino acids are diverted to form glucose and ketone bodies.

Lowered plasma volume produces dizziness and weakness due to postural hypotension when sitting or standing. Total body potassium loss and the general catabolism of muscle protein contribute to the weakness.

Paresthesias may be present at the time of diagnosis of type 1 diabetes, particularly when the onset is subacute. They reflect a temporary dysfunction of peripheral sensory nerves and usually clear, as insulin replacement restores glycemic levels closer to normal; thus, their presence suggests neurotoxicity from sustained hyperglycemia.

When insulin deficiency is severe and of acute onset, the above symptoms progress in an accelerated manner. Ketoacidosis exacerbates the dehydration and hyperosmolality by producing anorexia, nausea, and vomiting, thus interfering with oral fluid replacement. As plasma osmolality exceeds 330 mOsm/kg (normal, 285-295 mOsm/kg), impaired consciousness ensues. With progression of acidosis to a pH of 7.1 or less, deep breathing with a rapid ventilatory rate (Kussmaul respiration) occurs as the body attempts to eliminate carbonic acid. With worsening acidosis (to pH 7.0 or less), the cardiovascular system may be unable to maintain compensatory vasoconstriction; severe circulatory collapse may result.

Signs

The patient's level of consciousness can vary depending on the degree of hyperosmolality. When insulin deficiency develops relatively slowly and sufficient water intake is maintained to permit renal excretion of glucose and appropriate dilution of extracellular sodium chloride concentration, patients remain relatively alert and physical findings may be minimal. When vomiting occurs in response to worsening ketoacidosis, dehydration progresses and compensatory mechanisms become inadequate to keep plasma osmolality below 330 mOsm/kg. Under these circumstances, stupor or even coma may occur. Evidence of dehydration in a stuporous patient, with rapid deep breathing and the fruity breath odor of acetone, suggests the diagnosis of diabetic ketoacidosis.

Postural hypotension indicates a depleted plasma volume; hypotension in the recumbent position is a serious prognostic sign. Loss of subcutaneous fat and muscle wasting are features of more slowly developing insulin deficiency. In occasional patients with slow, insidious onset of insulin deficiency, subcutaneous fat may be considerably depleted. An enlarged liver, eruptive xanthomas on the flexor surface of the limbs and on the buttocks, and lipemia retinalis indicate that chronic insulin deficiency has resulted in chylomicronemia, with elevated circulating triglycerides, usually to over 2000 mg/dL (Chapter 19).

Type 2 Diabetes

Patients with type 2 diabetes usually have less severe insulin deficiency than type 1 patients and the symptoms and signs at presentation reflect this difference.

Symptoms

Many patients with type 2 diabetes have an insidious onset of hyperglycemia and may be relatively asymptomatic initially. The diagnosis may be made only after glycosuria or hyperglycemia is noted during routine laboratory studies. Chronic skin infections are common. Generalized pruritus and symptoms of candidal vaginitis are frequently the initial complaints of women with type 2 diabetes. Men may complain of an itchy rash of the prepuce. Some patients can remain undiagnosed for many years and the initial presentation may be due to complications such as visual disturbance due to retinopathy or foot pain or infection due to a peripheral neuropathy. Patients with a more severe insulin deficiency have the classical symptoms of polyuria, thirst, blurred vision, paresthesias, and fatigue. This is especially true in individuals who consume large amounts of carbohydrate-rich fluids in response to the thirst.

Signs

Many patients are obese or overweight. Even those patients who are not significantly overweight often have a characteristic fat distribution with more fat in the upper part of the body (particularly the abdomen, chest, neck, and face) and relatively less fat on the appendages, which may be quite muscular (the metabolically obese). This centripetal fat distribution has been termed android and is characterized by a high waist circumference. It differs from the more centrifugal gynecoid form of obesity, in which fat is localized more in the hips and thighs and less in the upper parts of the trunk. A larger waist circumference increases the risk for diabetes for any given body mass index (BMI). Thus in patients with the metabolic syndrome, a waist circumference >40 in (102 cm) in men and >35 in (88 cm) in women is associated with an increased risk of diabetes. MRI and CT scans reveal that these patients with increased waist circumference have accumulation of fat in the omental and mesenteric distributions. This visceral fat correlates with insulin resistance, whereas fat predominantly in subcutaneous tissues of the abdomen has little, if any, association with insulin insensitivity.

Some patients, especially the obese, may have acanthosis nigricans—hyperpigmented, hyperkeratotic skin in the axilla, groin, and back of neck. This sign is associated with significant insulin resistance. Hypertension may be present especially in the obese patient. Eruptive xanthomata on the flexor surface of the limbs and on the buttocks and lipemia retinalis due to hyperchylomicronemia can occur in patients with uncontrolled type 2 diabetes who also have a familial form of hypertriglyceridemia. In women, candidal vaginitis with a reddened, inflamed vulvar area and a profuse whitish discharge may herald the presence of diabetes. In men, candidal infection of the penis may lead to reddish appearance of the penis and/or prepuce with eroded white papules and a white discharge. The occasional patient who has had undiagnosed diabetes for some time may present with retinopathy or peripheral neuropathy.

Patients can also present in hyperglycemic hyperosmolar coma—profoundly dehydrated, hypotensive, lethargic, or comatose without Kussmaul respirations.

Laboratory Testing in Diabetes Mellitus

Tests of urine glucose and ketone bodies, as well as whole blood or plasma glucose measured in samples obtained under basal conditions and after glucose administration, are very important in evaluation of the patient with diabetes. Tests for glycosylated hemoglobin have proved useful in both initial evaluation and in assessment of the effectiveness of therapeutic management. In certain circumstances, measurements of insulin or C-peptide levels and levels of other hormones involved in carbohydrate homeostasis (eg, glucagon, GH) may be useful. In view of the increased risk of atherosclerosis in patients with diabetes, determination of serum cholesterol, HDL—cholesterol, triglycerides, and LDL-cholesterol may be helpful.

Urine Glucose

Several problems are associated with using urine glucose as an index of blood glucose, regardless of the method employed. First of all, the glucose concentration in bladder urine reflects the blood glucose at the time the urine was formed. Therefore, the first voided specimen in the morning contains glucose that was excreted throughout the night and does not reflect the morning blood glucose at all. Some improvement in the correlation of urine glucose to blood glucose can be obtained if the patient double voids—that is, empties the bladder completely, discards that sample, and then urinates again about one-half hour later, testing only the second specimen for glucose content. However, difficulty in completely emptying the bladder (large residual volumes), problems in understanding the instructions, and the inconvenience impair the usefulness of this test. Self-monitoring of blood glucose has replaced urine glucose testing in most patients with diabetes (particularly those receiving insulin therapy).

Several commercial products are available for determining the presence and amount of glucose in urine. The older and more cumbersome bedside assessment of glycosuria with Clinitest tablets has generally been replaced by the dipstick method, which is rapid, convenient, and glucose specific. This method consists of paper strips (Clinistix, Tes-Tape) impregnated with enzymes (glucose oxidase and hydrogen peroxidase) and a chromogenic dye that is colorless in the reduced state. Enzymatic generation of hydrogen peroxide oxidizes the dye to produce colors whose intensity depends on the glucose concentration. These dipsticks are sensitive to as little as 0.1% glucose (100 mg/dL) but do not react with the smaller amounts of glucose normally present in nondiabetic urine. The strips are subject to deterioration if exposed to air, moisture, and extreme heat and must be kept in tightly closed containers except when in use. False-negative results may be obtained in the presence of alkaptonuria and when certain substances such as salicylic acid or ascorbic acid are ingested in excess. All these false-negative results occur because of the interference of strong reducing agents with oxidation of the chromogen.

Although glycosuria reflects hyperglycemia in more than 90% of patients, two major classes of nondiabetic glycosuria must be considered:

Alterations in Renal Handing of Glucose

Disorders associated with abnormalities in renal glucose handling include Fanconi syndrome, a group of disorders characterized by combined renal wasting of multiple solutes including amino acids, uric acid, phosphate, and bicarbonate as well as glucose and caused by both genetic and acquired defects of the proximal renal tubule. Familial renal glycosuria, a benign inherited disorder manifest only by persistent glycosuria in the setting of euglycemia is caused by mutations in SGLT2, the sodium-glucose cotransporter responsible for the bulk of glucose reabsorption in the proximal tubule.

In addition, glycosuria is relatively common in pregnancy as a consequence of the increased load of glucose presented to the tubules by the elevated glomerular filtration rate during pregnancy. As many as 50% of pregnant women normally have demonstrable sugar in the urine, especially after the first trimester.

Excretion of Sugars Other Than Glucose in the Urine

Occasionally, a sugar other than glucose is excreted in the urine. Lactosuria during the late stages of pregnancy and the period of lactation is the most common example. Much rarer are other conditions in which inborn errors of metabolism allow fructose, galactose, or a pentose (1-xylose) to be excreted in the urine. Testing the urine with glucose-specific strips helps differentiate true glucosuria from other glycosurias.

Microalbuminuria and Proteinuria

Urinary albumin can now be detected in microgram concentrations using high-performance liquid chromatography or immunoassay methodology that is more sensitive than previous available tests. Conventional 24-hour urine collections, in addition to being inconvenient for patients, also show wide variability of albumin excretion, because several factors such as sustained upright posture, dietary protein, and exercise tend to increase albumin excretion rates. For these reasons, it is preferable to measure the albumin-creatinine ratio in an early morning spot urine collected on awakening—prior to breakfast or exercise—and brought in by the patient for laboratory analysis. A ratio of albumin (μg/L) to creatinine (mg/L) of less than 30 is normal, and a ratio of 30 to 300 indicates abnormal microalbuminuria. Values greater than 300 are referred to as macroalbuminuria.

The minimal detection limit of protein on a standard urine dipstick is 10 to 20 mg/dL. If the dipstick is positive then it is likely that the patient has microalbuminuria and this should be specifically tested. The information from a spot urine sample is adequate for diagnosis and treatment and it is not usually necessary to perform a 24-hour urine collection for protein loss and creatinine clearance.

Blood Glucose Testing

Venous Blood Samples

Venous glucose samples should be collected in tubes containing sodium fluoride, which inhibits enolase and prevents glycolysis in the blood sample that would artifactually lower the measured glucose level. In the absence of fluoride, the rate of disappearance of glucose in the presence of blood cells has been reported to average 10 mg/dL/h—the rate increases with glucose concentration, temperature, and white blood cell count. Fluoride takes about 1 hour to effectively stop glycolysis. Therefore, the rate of decline during the first hour is the same in tubes with or without fluoride. A very high white blood cell count will lower glucose levels even in the presence of fluoride. Ideally the blood should be collected in a sodium fluoride/potassium oxalate tube, placed on ice and the plasma separated from the cells within 60 minutes.

Plasma or serum from venous blood samples has the advantage over whole blood of providing values for glucose that are independent of hematocrit and reflect levels in the interstitial spaces to which body tissues are exposed. For these reasons—and because plasma and serum lend themselves to automated analytic procedures—they are used in most laboratories. The glucose concentration is 10% to 15% higher in plasma or serum than in whole blood because structural components of blood cells are absent. Whole blood glucose determinations are seldom used in clinical laboratories, but are used by patients using home blood glucose monitors.

Glucose levels can be measured in the laboratory using enzymatic methods (such as glucose oxidase or hexokinase), condensation methods (such as o-toluidine), or reducing methods. The reducing methods take advantage of the reducing properties of glucose to change the redox state of a metal ion; however, the method is nonspecific and any strong reducing agent can cross-react to yield spuriously elevated glucose values. In condensation methods, the aldehyde group of glucose undergoes condensation with aromatic compounds to yield a colored product. In the most commonly used condensation reaction, o-toluidine reacts with glucose to form a glucosamine that has an intense green color. The colour is measured spectrophotometrically to estimate the glucose concentration. o-Toluidine, however, has the drawback of being highly corrosive and toxic. In the enzymatic method, glucose oxidase reacts with glucose, water, and oxygen to form gluconic acid and hydrogen peroxide. The hydrogen peroxide can then be used to oxidize a chromogen or the consumption of oxygen measured to estimate the amount of glucose present. Current laboratories use enzymatic methods to determine glucose levels.

The range of normal fasting plasma or serum glucose is 70 to 100 mg/dL (3.9-5.5 mmol/L). A plasma glucose level of 126 mg/dL (7.0 mmol/L) or higher on more than one occasion after at least 8 hours of fasting is diagnostic of diabetes mellitus (Table 17–11). Fasting plasma glucose levels of 100 mg/dL (5.6 mmol/L) to 125 mg/dL (6.9 mmol/L) are associated with increased risk for diabetes (impaired fasting glucose).

Capillary Blood Samples

Capillary blood glucose measurements performed by patients themselves, as outpatients, are extremely useful. In type 1 patients in whom tight metabolic control is attempted, they are indispensable. There are several paper strip (glucose oxidase; hexokinase; glucose dehydrogenase with nicotinamide adenine dinucleotide, glucose dehydrogenase with flavin-adenine dinucleotide, glucose dehydrogenase with pyrroloquinoline quinine) methods for measuring glucose on capillary blood samples. A reflectance photometer or an amperometric system is then used to measure the reaction that takes place on the reagent strip. A large number of blood glucose meters are now available. All are accurate, but they vary with regard to speed, convenience, size of blood samples required, and cost. Popular models include those manufactured by Life-Scan (One Touch), Bayer Corporation (Breeze, Contour), Roche Diagnostics (Accu-Chek), and Abbott Laboratories (Precision, FreeStyle). A Freestyle Flash meter, for example, which requires only 0.3 microliter of blood and gives a result in 5 seconds, is illustrative of the continued progress in this technologic area. Various glucometers appeal to a particular consumer need and are relatively inexpensive, ranging from $50.00 to $100.00 each. Test strips remain a major expense, costing 50 to 75 cents each. The meters have memories and can compute blood glucose averages. The data can be downloaded into a computer. In self-monitoring of blood glucose, patients must prick a finger with 26- to 33-gauge lancets. This can be facilitated by a small plastic trigger device such as an Accu-chek multiclix (Roche Diagnostics), Microlet (Boehringer-Mannheim), or one touch lancing device (Lifescan, Inc.). Some meters such as the FreeStyle (Abbott Laboratories) have been approved for measuring glucose in blood samples obtained at alternative sites such as the forearm and thigh. There is, however, a 5- to 20-minute lag in the glucose response on the arm with respect to the glucose response on the finger. Forearm blood glucose measurements could therefore result in a delay in detection of rapidly developing hypoglycemia.

The clinician should be aware of the limitations of the self-glucose monitoring systems. First, some meters require input of a code for each batch of strips—failure to enter the code can result in misleading results. Many of the newer meters no longer require this step. Second, increases or decreases in hematocrit can decrease or increase the measured glucose values, respectively. The mechanism underlying this effect is not known, but presumably it is due to the impact of red cells on the diffusion of plasma into the reagent layer. Third, the meters and the test strips are calibrated over glucose concentrations ranging from 60 to 160 mg/dL, and the accuracy is not as good for higher and lower glucose levels. Thus, when the glucose is less than 60 mg/dL, the difference between the meter and the laboratory value may be as much as 20%. Fourth, glucose-oxidase-based amperometric systems underestimate glucose levels in the presence of high oxygen tension. This may be important in critically ill patients who are on supplemental oxygen, and under these circumstances, a glucose dehydrogenase-based system may be preferable. Fifth, glucose-dehydrogenase pyrroloquinoline quinine (GDH-PQQ) systems may report falsely high glucose levels in patients who are receiving parenteral products containing nonglucose sugars such as maltose, galactose, or xylose or their metabolites. Patients have been given insulin based on the falsely high glucose values resulting in life-threatening hypoglycemia. The accuracy of data obtained by glucose monitoring requires education of the patient in sampling and measuring procedures as well as in proper calibration of the instruments. Bedside glucose monitoring in a hospital setting requires rigorous quality control programs and certification of personnel to avoid errors.

Interstitial Fluid Glucose Samples

A number of continuous glucose monitoring (CGM) systems are currently available for clinical use. The systems manufactured by Medtronic MiniMed, Abbott Diagnostics, and Dex Com involve inserting a subcutaneous biosensor (rather like an insulin pump cannula) that measures glucose concentrations in the interstitial fluid for 3 to 7 days. The glucose values are available for review by the patient at time of measurement. The systems also display directional arrows indicating rate and direction of change and alarms can be set for dangerously low or high glucose values. Patients still have to calibrate the devices with periodic fingerstix glucose levels; and since there are concerns about the reliability of the measurements, it is still necessary to perform a confirmatory capillary blood glucose measurement before intervening. The individual glucose levels are not that critical—what matters are the direction and the rate at which glucose is changing and the low glucose alerts that allow the user to take corrective action. The user also gains insight into the way particular foods or activities affect their glucose levels. Clinical trials with these systems show that they do enable some patients with type 1 diabetes to improve control without increasing the risk for hypoglycemia. The MiniMed insulin pump marketed in Europe can be programmed to automatically suspend insulin delivery for two hours when the glucose levels on their CGM device falls to a preset level. This feature may help reduce the risk for dangerous hypoglycemia especially when the patient is asleep. They are increasingly being approved by insurance companies. The out of pocket cost of the systems is about $4000 annually.

There is great interest in using the data obtained from these CGM systems to automatically deliver insulin by continuous subcutaneous insulin infusion pump. Algorithms have been devised to link CGM to insulin delivery and preliminary clinical studies appear promising.

Urine and Serum Ketone Determinations

In the absence of adequate insulin, three major ketone bodies are formed and excreted into the urine: β-hydroxybutyrate (often the most prevalent in diabetic ketoacidosis), acetoacetate, and acetone. Acetone and acetoacetate react with sodium nitroprusside (nitroferricyanide) in the presence of alkali to produce a purple-colored complex. Acetest tablets, Ketostix, and Keto-Diastix strips utilize this nitroprusside reaction to quantify acetone and acetoacetate levels in urine and plasma. When a few drops of serum are placed on a crushed Acetest tablet, the appearance of a purple color indicates the presence of ketones. A strongly positive reaction in undiluted serum correlates with a serum ketone concentration of at least 4 mmol/L. Although these tests do not detect β-hydroxybutyric acid, which lacks a ketone group, the semiquantitative estimation of the other ketone bodies is nonetheless usually adequate for clinical assessment of ketonuria. Ketostix and Keto-Diastix have short shelf-lives (90 days) once the containers are opened and using expired strips can give false-negative results. It is better therefore to buy individually foil wrapped strips.

Specific enzymatic techniques are available to quantitate each of the ketone acids, but these techniques are cumbersome and are not necessary in most clinical situations. There is also a paper strip method that patients can use to measure capillary blood β-hydroxybutyrate levels (Precision Xtra, Abbott Diagnostics). This technology uses hydroxybutyrate dehydrogenase to catalyse the oxidation of β-hydroxybutyrate to acetoacetate with concomitant reduction of NAD+ to NADH. The NADH is reoxidized to NAD+ by a redox mediator and a current is generated that is directily proportional to β-hydroxybutyrate concentration. β-Hydroxybutyrate levels >0.6 nmol/L require evaluation. Levels >3.0 nmol/L, which is equivalent to very large urine ketones, will require hospitalization.

Other conditions besides diabetic ketoacidosis may cause ketone bodies to appear in the urine; these include starvation, high-fat diets, alcoholic ketoacidosis, fever, and other conditions in which metabolic requirements are increased.

Glycated Hemoglobin Assays

Ketoamine reactions between glucose and other sugars and free amino groups on the alpha and beta chain lead to glycated forms of hemoglobin. Only glycation of the N-terminal valine of the beta chain imparts sufficient negative charge to the hemoglobin molecule to allow separation by charge-dependent techniques. The charge-separated hemoglobins are collectively referred to as hemoglobin A1 (HbA1). The major form of HbA1 is hemoglobin A1c (HbA1c), where glucose is the carbohydrate. This form comprises 4% to 6% of total hemoglobin. The remaining HbA1 species contain fructose 1,6-diphosphate (HbA1a1), glucose 6-phosphate (HbA1a2), and an unknown carbohydrate moiety (HbA1b). The hemoglobin A1c fraction is abnormally elevated in diabetic patients with chronic hyperglycemia. Some laboratories measure the sum of these glycohemoglobins (GHbs) and report the total as hemoglobin A1, but most laboratories have converted to the highly specific HbA1c assay. Methods for measuring HbA1c include electrophoresis, cation exchange chromatography, boronate affinity chromatography, and immunoassays.

Office-based immunoassays using capillary blood give a result in about 9 minutes, and this allows for immediate feedback to the patients regarding their glycemic control. Because GHbs circulate within red blood cells whose life span lasts up to 120 days, they generally reflect the state of glycemia over the preceding 8 to 12 weeks, thereby providing an improved method of assessing diabetic control. The HbA1c value, however, is weighted to more recent glucose levels (previous month) and this explains why significant changes in HbA1c are observed with short term (1 month) changes in mean plasma glucose levels. Measurements should be made in patients with either type of diabetes mellitus at 3- to 4-month intervals so that adjustments in therapy can be made if GHb is either subnormal or if it is more than 1% above the upper limits of normal for a particular laboratory. In patients monitoring their own blood glucose levels, GHb values provide a valuable check on the accuracy of monitoring. In patients who do not monitor their own blood glucose levels, GHb values are essential for adjusting therapy.

The various HbA1c assays have been standardized to the assay used in the Diabetes Control and Complications Trial (DCCT) allowing the results to be related to the risks of developing microvascular complications. There is a linear relationship between the HbA1c value and average glucose. A recent study (A1c-Derived Average Glucose Study) collected 3 months of blood glucose data on 507 subjects—normals, type 1 and type 2 diabetics. The estimated average glucose was calculated by combining weighted results from 2 days of continuous glucose monitoring per month and seven point capillary blood glucose profiles (preprandial, postprandial, and bedtime) for at least 3 d/wk. The HbA1c was measured at the end of the 3 months. The relationship between average glucose and HbA1c based on liner regression analysis was:

• Average glucose = (28.7 × HbA1c) − 46.7

(see also Table 17–12)

The accuracy of HbA1c values can be affected by hemoglobin variants or derivatives, the effect depending on the specific hemoglobin variant or derivative and the specific assay used. Immunoassays that use an antibody to the glycated amino terminus of β globin do not recognize the terminus of the γ globin of hemoglobin F, and so in patients with high levels of hemoglobin F, immunoassays give falsely low estimates of HbA1c. Cation exchange chromatography separates hemoglobin species by charge differences. Therefore, hemoglobin variants that coelute with HbA1c can lead to an overestimation of the HbA1c value. Chemically modified derivatives of hemoglobin such as carbamylated (in renal failure) or acetylated (high-dose aspirin therapy) hemoglobin can, in some methods, coelute with HbA1c.

Any condition that shortens erythrocyte survival or decreases mean erythrocyte age (eg, recovery from acute blood loss, hemolytic anemia) falsely lower HbA1c irrespective of the assay method used. Alternative methods such as fructosamine (see later) should be considered for these patients. Vitamins C and E are reported to falsely lower test results, possibly by inhibiting glycation of hemoglobin. The National Glycohemoglobin Standardization Program website (www.ngsp.org) has information on the impact of frequently encountered hemoglobin variants and derivatives on the results obtained with the commonly used HbA1c assays.

The American Diabetes Association (ADA) has now endorsed using the HbA1c as a diagnostic test for diabetes (see Table 17–11). A cutoff value of 6.5% was chosen because the risk for retinopathy increases substantially above this value. The advantages of using the HbA1c to diagnose diabetes is that there is no need to fast; it has lower intra-individual variability than the fasting glucose test and the oral glucose tolerance test; and it gives a better picture of glucose control for 2 to 3 months. People with HbA1c levels of 5.7 % to 6.4% should be considered at high risk for developing diabetes (prediabetes). The diagnosis should be confirmed with a repeat HbA1c test, unless the patient is symptomatic with plasma glucose levels >200 mg/dL. This test would not be appropriate to use in populations with high prevalence of hemoglobinopathies or in conditions with increased red cell turnover. Also, the testing should be performed using a National Glycohemoglobin Standardization Program certified method and standardized to the Diabetes Control and Complications Trial assay.

Serum Fructosamine

Serum fructosamine is formed by nonenzymatic glycosylation of serum proteins (predominantly albumin). Because serum albumin has a much shorter half-life (14-21 days) than hemoglobin, serum fructosamine generally reflects the state of glycemic control for the preceding 2 or 3 weeks. Reductions in serum albumin (eg, nephrotic state or hepatic disease) lower the serum fructosamine value. When abnormal hemoglobins or hemolytic states affect the interpretation of GHb or when a narrower time frame is required, such as for ascertaining glycemic control at the time of conception in a woman with diabetes who has recently become pregnant, serum fructosamine assays offer some advantage. Normal values vary in relation to the serum albumin concentration and are 200 to 285 umol/L when the serum albumin level is 5 g/dL. HbA1c values and fructosamine are highly correlated. The following relationship between fructosamine levels and HbA1c has been reported based on linear regression analysis:

Thus fructosamine levels of 317, 375, and 435 are equivalent to HbA1c values of 7%, 8%, and 9%, respectively.

In most circumstances, however, glycohemoglobin assays remain the preferred method for assessing long-term glycemic control in patients with diabetes.

Oral Glucose Tolerance Test

It is easy to screen for diabetes using an HbA1c or a fasting plasma glucose level (see Table 17–11). The oral glucose tolerance test, therefore, is mostly performed for research studies or when there is a suspicion of the diagnosis but the fasting plasma glucose is less than 126 mg/dL or the HBA1c level is below 6.5%. The test might be considered, for example, in a woman with a history of delivering an infant above 9 lb (4.1 kg).

In order to optimize insulin secretion and effectiveness, especially when patients have been on a low-carbohydrate diet, a minimum of 150 to 200 g of carbohydrate per day should be included in the diet for 3 days preceding the test. The patient should eat nothing after midnight prior to the test day. Adults are given 75 g of glucose in 300 mL of water; children are given 1.75 g of glucose per kilogram of ideal body weight. The glucose load is consumed within 5 minutes. The test should be performed in the morning because there is some diurnal variation in oral glucose tolerance and patients should not smoke; drink coffee, tea, or alcohol; or be active during the test.

Blood samples for plasma glucose are obtained at 0 and 120 minutes after ingestion of glucose. A fasting plasma glucose value of 126 mg/dL (7 mmol/L) or higher or a 2-hour value of greater than 200 mg/dL (11.1 mmol/L) is diagnostic of diabetes mellitus (see Table 17–11). An oral glucose tolerance test is normal if the fasting venous plasma glucose value is less than 100 mg/dL (5.6 mmol/L) and the 2-hour value falls below 140 mg/dL (7.8 mmol/L). Patients with 2-hour values of 140 to 199 mg/dL have impaired glucose tolerance. False-positive results may occur in patients who are malnourished at test time, bedridden, or afflicted with an infection or severe emotional stress. Diuretics, oral contraceptives, glucocorticoids, excess thyroxine, phenytoin, nicotinic acid, and some of the psychotropic drugs may also cause false-positive results.

Insulin Levels

Normal immunoreactive insulin levels range from 5 to 20 μU/mL in the fasting state. During an oral glucose tolerance test, they reach 50 to 130 μU/mL at 1 hour, and usually return to levels below 30 μU/mL by 2 hours. Insulin measurements are rarely of clinical usefulness. They are principally used in research studies to determine insulin sensitivity.

The homeostasis model of insulin resistance (HOMAIR) estimates insulin sensitivity using the following formula:

The higher the HOMAIR value the more resistant the individual. Data from the oral glucose tolerance test can also be used to estimate insulin sensitivity. The Matsuda & DeFronzo Insulin Sensitivity Index is calculated as:

The lower the ISI the more insulin resistant the subject.

Intravenous Glucose Tolerance Test

The intravenous glucose tolerance test (IVGTT) is performed by giving a bolus of 50 g of glucose per 1.7 m2 body surface area (or 0.5 g/kg of ideal body weight) as a 25% or 50% solution over 2 to 3 minutes after an overnight fast. Timing begins with injection and samples for plasma glucose determination are obtained from an indwelling needle in the opposite arm at 0, 10, 15, 20, and 30 minutes. The plasma glucose values are plotted on semilogarithmic paper against time. K, a rate constant that reflects the rate of fall of blood glucose in percent per minute, is calculated by determining the time necessary for the glucose concentration to fall by one-half (t1/2) and using the following equation:

The average K value for a nondiabetic patient is approximately 1.72% per minute; this value declines with age but remains above 1.3% per minute. Patients with diabetes almost always have a K value of less than 1% per minute. The disappearance rate reflects the patient's ability to dispose of a glucose load. Perhaps its most widespread present use is to screen siblings at risk for type 1 diabetes to determine if autoimmune destruction of β cells has reduced first phase insulin responses (at 1-5 minutes after the glucose bolus) to levels below the normal lower limit of 40 μU/mL.

The IVGTT has been modified by giving a glucose dose of 0.3 g/kg, with more frequent plasma sampling and extending the test to 3 to 4 hours. Also at 20 minutes, a 5-minute infusion of insulin is given (0.03 U/kg for the subject who is likely to be insulin sensitive and 0.06 U/kg for the likely resistant subject). Plasma glucose is sampled at 3, 4, 5, 6, 8, 10, 14, 19, 22, 25, 27, 30, 40, 50, 60, 80, 100, 140, and 180 minutes. Analysis of the time course of glucose and insulin during this frequently sampled IVGTT (FSIVGTT) allows for measurements of insulin sensitivity (Si), that is, fractional glucose clearance per unit insulin concentration; the first phase insulin response (AIRglucose); and glucose effectiveness (SG), the ability of glucose itself to enhance its own disappearance independent of any change in insulin. The data analysis requires use of specific software (Minmod).

Lipoproteins in Diabetes

Levels of circulating lipoproteins are dependent on normal levels and action of insulin, just as is the plasma glucose. In type 1 diabetes, moderately deficient control of hyperglycemia is associated with only a slight elevation of low-density lipoprotein (LDL) cholesterol and serum triglycerides and little if any changes in HDL cholesterol. Once the hyperglycemia is corrected, lipoprotein levels are generally normal. However, patients with type 2 diabetes frequently have a dyslipidemia that is characteristic of the insulin resistance syndrome. Its features are a high serum triglyceride level (300-400 mg/dL), a low HDL cholesterol (<30 mg/dL), and a qualitative change in LDL particles producing a smaller dense LDL whose membrane carries supranormal amounts of free cholesterol. Because low HDL cholesterol is a major feature predisposing to macrovascular disease, the term dyslipidemia has preempted the previous label of hyperlipidemia, which mainly described the elevated triglycerides. Measures designed to correct obesity and hyperglycemia, such as exercise, diet, and hypoglycemic therapy substantially correct the dyslipidemia but most patients require pharmacotherapy. Chapter 19 discusses these matters in detail.

Clinical Trials in Diabetes

Findings of the Diabetes Control and Complications Trial (DCCT) and of the United Kingdom Prospective Diabetes Study (UKPDS) have confirmed the beneficial effects of intensive therapy to achieve improved glycemic control in both type 1 and type 2 diabetes, respectively (see later). In addition, with increased understanding of the pathophysiology of both type 1 and type 2 diabetes, large prospective studies—Diabetes Prevention Trial-1 (DPT-1) and the Diabetes Prevention Program (DPP) have been conducted to identify interventions that prevent the onset of these disorders.

Clinical Trials in Type 1 Diabetes

The Diabetes Control and Complications Trial (DCCT)

This long-term randomized prospective study of patients with type 1 diabetes reported that lowering of blood glucose levels with intensive insulin therapy delayed the onset and slowed the progression of microvascular and neuropathic complications of diabetes.

More than 1000 patients from 29 medical centers ranging in age from 13 to 39 were divided into two study groups with equal numbers of subjects. Approximately half of the total group had no detectable diabetic complications (primary prevention group), whereas mild background retinopathy was present in the other half (secondary prevention group). Some patients in the latter group also had slightly elevated microalbuminuria and mild neuropathy, but no one with serious diabetic complications was enrolled in the trial. Multiple insulin injections (66%) or insulin pumps (34%) were used in the intensively treated group, and those subjects were trained to modify their therapy in response to frequent glucose monitoring. The conventionally treated group used no more than two insulin injections, and clinical well-being was the goal with no attempt to modify management based on glycated hemoglobin or glucose results.

In the intensively treated subjects, mean glycated hemoglobin of 7.2% (normal, <6%) and a mean blood glucose of 155 mg/dL were achieved, whereas in the conventionally treated group glycated hemoglobin averaged 8.9% and the average blood glucose was 225 mg/dL. Over the study period, which averaged 6.5 years, there was an approximately 60% reduction in risk of diabetic retinopathy, nephropathy, and neuropathy in the intensively treated group. The benefits were seen in both the primary and secondary prevention groups. The intensively treated group also had a nonsignficant reduction in the risk of macrovascular disease of 41% (95% CI, -10%-68%).

Intensively treated patients had a three-fold greater risk of serious hypoglycemia as well as a greater tendency toward weight gain. However, there were no deaths from hypoglycemia in any subjects in the DCCT study, and no evidence of posthypoglycemic neurologic damage was detected.

Subjects participating in the DCCT study were subsequently enrolled in a follow-up observational study (Epidemiology of Diabetes Interventions and Complications [EDIC]). Even though the between group differences in HbA1c narrowed within 4 years, the group assigned to intensive therapy had a lower risk for retinopathy at 4 years and microalbuminuria at 7 to 8 years of post study follow-up. Moreover, by the end of the 11 year follow-up period, the intensive therapy group had significantly reduced risk of any cardiovascular disease events by 42% (95% CI, 9%-23%; p = 0.02). Thus it seems that the benefits of good glucose control persist even if control deteriorates at a later date.

The general consensus of the ADA is that intensive insulin therapy associated with comprehensive self-management training should become standard therapy in most patients with type 1 diabetes after the age of puberty. Exceptions include those with advanced renal disease and the elderly, because the detrimental risks of hypoglycemia outweigh the benefit of tight glycemic control in these groups. In children under age 7 years, the risk of developing brain damage from hypoglycemia contraindicates attempts at tight glycemic control, particularly because diabetic complications do not seem to occur until some years after the onset of puberty.

Diabetes Prevention Trial-1 (DPT-1)

This NIH-sponsored multicenter study was designed to determine whether the development of type 1 diabetes could be prevented or delayed by immune intervention therapy. Daily low-dose insulin injections were administered for up to 8 years in first-degree relatives of individuals with type 1 diabetes who were selected as being at high risk for development of type 1 diabetes because of detectable ICAs and reduced early-insulin release. Unfortunately, this immune intervention failed to affect the onset of type 1 diabetes, which was approximately 15% per year in both the treated and control groups. A related study using oral insulin in lower risk first-degree relatives, who had ICAs but whose early insulin release remained intact also failed to show an effect on the onset of type 1 diabetes. After an average of 4.3 years of observation, type 1 diabetes developed in about 35% of persons in both the oral insulin and the placebo groups.

Immune Intervention Trials in New-Onset Type 1 Diabetes

At time of diagnosis of type 1 diabetes, patients still have significant β cell function. This explains why soon after diagnosis, patients go into a partial clinical remission (honeymoon) requiring little or no insulin. This clinical remission is short-lived, however, and eventually patients lose all β cell function and have more labile glucose control. Attempts have been made to prolong this partial clinical remission using drugs such as cyclosporine, azathioprine, prednisone, and antithymocyte globulin. These agents, however, have had limited efficacy, and there were concerns about their toxicity and the need for continuous treatment.

More specific strategies for immunosuppression, such as the use of monoclonal antibodies against particular T-cell products, may reduce the hazards of long-term immunotherapy. Phase I/II clinical trials using humanized monoclonal antibodies against CD3, hOKT3γ1 (Ala-Ala) (teplizumab), and ChAglyCD3 (otelexizumab) showed efficacy in reducing the decline in insulin production in patients newly diagnosed with type 1 diabetes. The CD3 complex is the major signal transducing element of the T-cell receptor. The anti-CD3 antibodies are believed to modulate the autoimmune response by selectively inhibiting pathogenic T cells and/or by inducing regulatory T cells. Patients were treated with the antibody for 14 days within 6 weeks of diagnosis of type 1 diabetes. One year later, the majority of patients in the treated group had maintained or increased insulin production and improved glycemic control relative to the control group. Larger phase II/III clinical trials are currently in progress.

These and other approaches that selectively modulate the autoimmune T-cell response hold the promise that type 1 diabetes may eventually be preventable without prolonged immunosuppression.

Clinical Trials in Type 2 Diabetes

Kumamoto Study

The Kumamoto study involved a relatively small number of patients with type 2 diabetes (n = 110) who were nonobese and only slightly insulin resistant, requiring less than 30 units of insulin per day for intensive therapy. Over a 6-year period it was shown that intensive insulin therapy, achieving a mean HbA1c of 7.1%, significantly reduced microvascular end points compared with conventional insulin therapy, which achieved a mean HbA1c of 9.4%. Cardiovascular events were neither worsened nor improved by intensive therapy, and weight changes were likewise not influenced by either form of treatment.

The United Kingdom Prospective Diabetes Study (UKPDS)

This multicenter study was designed to determine whether the risk of macrovascular or microvascular complications in patients with type 2 diabetes could be reduced by intensive blood glucose control with oral hypoglycemic agents or insulin and whether any particular therapy was better than the others. Patients aged 25 to 65 years who were newly diagnosed with type 2 diabetes were recruited between 1977 and 1991, and a total of 3867 were studied over 10 years. The median age at baseline was 54 years; 44% were overweight (>20% over ideal weight), and baseline HbA1c was 9.1%. Therapies were randomized to include a control group on diet alone and separate groups intensively treated with insulin, chlorpropamide, glyburide, or glipizide. Metformin was included as a randomization option in a subgroup of 342 overweight patients, and—much later in the study—an additional subgroup of both normal-weight and overweight patients, who were responding unsatisfactorily to sulfonylurea therapy, were randomized to either continue on their sulfonylurea therapy alone or to have metformin combined with it.

After the study was initiated, a further modification was made to evaluate whether tight control of blood pressure with stepwise antihypertensive therapy would prevent macrovascular and microvascular complications in 758 hypertensive patients among this UKPDS population—compared with 390 patients whose blood pressure was treated less intensively. The tight control group was randomly assigned to treatment with either an angiotensin–converting enzyme (ACE) inhibitor (captopril) or a beta blocker (atenolol). Both drugs were stepped up to maximum doses of 100 mg/d, and then, if blood pressure remained higher than the target level of less than 150/85 mm Hg, more drugs were added in the following stepwise sequence—a diuretic, slow-release nifedipine, methyldopa, and prazosin—until the target level of tight control was achieved. In the control group, hypertension was conventionally treated to achieve target levels less than 180/105 mm Hg, but these patients were not given either ACE inhibitors or beta blockers.

Intensive glycemic therapy in the entire group of 3897 patients newly diagnosed with type 2 diabetes patients followed over 10 years showed that intensive treatment with either sulfonylureas, metformin, combinations of these, or insulin achieved mean HbA1c levels of 7.0%. This level of glycemic control decreased the risk of microvascular complications in comparison with conventional therapy (mostly diet alone), which achieved mean levels of HbA1c of 7.9%. Weight gain occurred in intensively treated patients except when metformin was used as monotherapy. There was a trend towards reduction in cardiovascular events (fatal or nonfatal MI; sudden death) with intensive treatment but this did not reach statistical significance (16% reduction, p = 0.052). Hypoglycemic reactions occurred in the intensive treatment groups, but only one death from hypoglycemia was documented over 27,000 patient-years of intensive therapy.

When therapeutic subgroups were analyzed, some unexpected and paradoxical results were noted. Among the obese patients, intensive treatment with insulin or sulfonylureas did not reduce microvascular complications compared with diet therapy alone. This was in contrast to the significant benefit of intensive therapy with these drugs in the total group. Furthermore, intensive therapy with metformin was more beneficial in the overweight and obese persons than diet alone with regard to reducing myocardial infarctions, strokes, and diabetes-related deaths, but there was no significant reduction of diabetic microvascular complications with metformin as compared with the diet group. Moreover, in the subgroup of obese and nonobese patients in whom metformin was added to sulfonylurea failures, rather than showing a benefit, there was a 96% increase in diabetes-related deaths compared with the matched cohort of patients with unsatisfactory glycemic control on sulfonylureas who remained on sulfonylurea therapy. Chlorpropamide also came out poorly on subgroup analysis in that those receiving it as intensive therapy did less well as regards progression to retinopathy than those conventionally treated with diet. The University Group Diabetes Program (UGDP) study was designed to evaluate the effects of glucose-lowering therapies on vascular complications in type 2 diabetes and reported that tolbutamide may increase the risk for cardiovascular deaths. The UKPDS study refutes this concern by failing to confirm any cardiovascular hazard with sulfonylurea treatment.

Intensive antihypertensive therapy to a mean of 144/82 mm Hg had beneficial effects on microvascular disease as well as on all diabetes-related end points, including virtually all cardiovascular outcomes, in comparison with looser control at a mean of 154/87 mm Hg. In fact, the advantage of reducing hypertension by this amount was substantially more impressive than the benefit that accrued by improving the degree of glycemic control from a mean HbA1c of 7.9% to 7.0%. More than half of the patients needed two or more drugs for adequate therapy of their hypertension, and there was no demonstrable advantage of ACE inhibitor therapy over beta blockers as regards diabetes end points. Use of a calcium channel blocker added to both treatment groups appeared to be safe over the long term in this population with diabetes, despite some controversy in the literature about its safety in such individuals.

The UKPDS researchers, like the DCCT group, performed posttrial monitoring to determine if there were long-term benefits of having been in the intensively treated glucose and blood pressure arms of the study. The between group differences in HbA1c were lost within the first year of follow-up but the reduced risk of development or progression of microvascular complications in the intensively treated group persisted for 10 years (24%, p = 0.001). The intensively treated group also had significantly reduced risk for myocardial infarction (15%, p = 0.01) and death from any cause (13%, p = 0.007) during the follow-up period. The subgroup of overweight or obese subjects who were initially randomized to metformin therapy showed sustained reduction in risk of myocardial infarction and death from any cause in the follow-up period. The between group blood pressure differences disappeared within 2 years of the end of the trial. Unlike the sustained benefits seen with glucose control, there was no sustained benefit from having been in the more tightly controlled blood pressure group. Both blood pressure groups had similar risks for microvascular events and diabetes related end-points in the follow-up period.

Thus, the follow-up of the UKPDS type 2 diabetes cohort showed that, as in type 1 diabetes, the benefits of good glucose control persist even if control deteriorates at a later date. Blood pressure benefits, however, last only as long as the blood pressure is well controlled.

Diabetes Prevention Program (DPP)

This was a randomized clinical trial in 3234 overweight men and women, aged 25 to 85 years, who showed impaired glucose tolerance. Results from this study indicated that intervention with a low-fat diet and 150 minutes of moderate exercise (equivalent to a brisk walk) per week reduces the risk of progression to type 2 diabetes by 58% as compared with a matched control group. Another arm of this trial demonstrated that use of 850 mg of metformin twice daily reduced the risk of developing type 2 diabetes by 31% but was relatively ineffective in those who were either less obese or in the older age group.

The Steno-2 Study

The Steno-2 study was designed in 1990 to validate the efficacy of targeting multiple concomitant risk factors for both microvascular and macrovascular disorders in type 2 diabetes. A prospective, randomized, open, blinded end-point design was used in which 160 patients with type 2 diabetes and microalbuminuria were assigned to conventional therapy with their general practitioner or to intensive care at the Steno Diabetes Center. In the intensively treated group, stepwise introduction of lifestyle and pharmacologic interventions was aimed at keeping glycated hemoglobin less than 6.5%, blood pressure less than 130/80 mm Hg, total cholesterol less than 175 mg/dL, and triglycerides less than 150 mg/dL. All subjects in the intensively treated group received ACE inhibitors and if intolerant, an angiotensin II receptor blocker. The lifestyle component of intensive intervention included reduction in dietary fat intake to less than 30% of total calories, a smoking cessation program, light to moderate exercise, and a daily vitamin-mineral supplement (vitamins C and E and chromium picolinate). Initially, aspirin was given as secondary prevention to patients with a history of ischemic cardiovascular disease, but later all patients received aspirin.

After a mean follow-up of 7.8 years, 44% of patients in the conventional arm and 24% in the intensive multifactorial arm developed cardiovascular events (myocardial infarction, angioplasties, coronary bypass grafts, strokes, amputations, vascular surgical interventions)—a 53% reduction. Rates of nephropathy, retinopathy, and autonomic neuropathy were also lower in the multifactorial intervention arm—61%, 58%, and 63% of rates in the conventional arm, respectively.

The subjects who participated in this trial were subsequently enrolled in an observational follow-up study for an average of 5.5 years. Even though the significant differences in glycemic control and levels of risk factors for cardiovascular disease between the groups had disappeared by the end of the follow-up period, the interventional group continued to have a lower risk for retinal photocoagulation, renal failure, cardiovascular endpoints, and cardiovascular mortality.

The data from the UKPDS and this study thus provide support for guidelines recommending vigorous treatment of concomitant microvascular and cardiovascular risk factors in patients with type 2 diabetes.

Accord, Advance, and VADT Studies

The ACCORD study was a randomized controlled study designed to determine whether normal HbA1c levels would reduce the risk of cardiovascular events in middle-aged or older individuals with type 2 diabetes. About 35% of the 10,251 recruited subjects had established cardiovascular disease at study entry. The intensive arm of the study was discontinued after 3.5 years of follow-up because of more unexplained deaths in the intensive arm when compared to the conventional treatment arm (22%, p = 0.020). Analysis of the data at time of discontinuation showed that the intensively treated group (HbA1c 6.4%) had a 10% reduction in cardiovascular event rate compared to the standard treated group (HbA1c 7.5%), but this difference was not statistically significant. The ADVANCE trial randomly assigned 11,140 patients in their 60s with type 2 diabetes to standard or intensive glucose control. The primary outcomes were major macrovascular (nonfatal myocardial infarction or stroke or death from cardiovascular causes) or microvascular events. Thirty-two percent of the subjects had established cardiovascular disease at study entry. After a median follow-up of 5 years, there was a nonsignificant reduction (6%) in major macrovascular event rate in the intensively treated group (HbA1c 6.5%) compared to the standard therapy group (HbA1c 7.3%).

The Veteran Administration Diabetes Trial (VADT) randomly assigned 1791 patients in their 50s and 60s with type 2 diabetes to standard or intensive glucose control. Ninety seven percent of the subjects were men. The primary outcome was a composite of myocardial infarction, death from cardiovascular causes, congestive heart failure, vascular surgery, inoperable coronary artery disease, and amputation for gangrene. All patients had optimized blood pressure and lipid levels. After a median follow-up of 5.6 years, there was no significant difference in the primary outcome in the intensively treated group (HbA1c 6.9%) compared to the standard therapy (HbA1c 8.4%). Within this larger study, there was an embedded study evaluating the impact of intensive therapy in patients who were categorized as having low, moderate, and high coronary calcium scores on CT scans. Patients with low coronary calcium score showed reduced number of cardiovascular events with intensive therapy.

Thus, the ACCORD, ADVANCE, and VADT results do not provide support for the hypothesis that near-normal glucose control in type 2 diabetes will reduce cardiovascular events. It is, however, important not to over-interpret the results of these three studies. The results do not exclude the possibility that cardiovascular benefits might accrue with longer duration of near-normal glucose control. In the UKPDS, risk reductions for myocardial infarction and death from any cause were only observed during 10 years of posttrial follow-up. Specific subgroups of type 2 diabetes patients may also have different outcomes. The ACCORD, ADVANCE, and VADT studies recruited patients who had diabetes for 8 to 10 years, and a third of them had established cardiovascular disease. Patients in the UKPDS, in contrast, had newly diagnosed diabetes, and only 7.5% had a history of macrovascular disease. It is possible that the benefits of tight glycemic control on macrovascular events are attenuated in patients with longer duration of diabetes or with established vascular disease. Specific therapies used to lower glucose may also affect cardiovascular event rate or mortality. Severe hypoglycemia occurred more frequently in the intensively treated groups of the ACCORD, ADVANCE, and VADT studies; and the ACCORD investigators were not able to exclude undiagnosed hypoglycemia as a potential cause for the increased death rate in their intensive arm group.

A formal meta-analysis performed on the raw trial data from the ACCORD, ADVANCE, VADT, and UKPDS studies found that allocation to more intensive glucose control reduced the risk of MI by 15% (hazard ratio 0.85, 95% CI 0.76-0.94). The benefit appeared to be in patients who did not have preexisting macrovascular disease.

Treatment of Diabetes Mellitus

Diet

A well-balanced, nutritious diet remains a fundamental element of therapy for diabetes. However, in more than half of cases, patients with diabetes fail to follow their diets. The American Diabetes Association (ADA) recommends about 45% to 65% of total daily calories in the form of carbohydrates; 25% to 35% in the form of fat (of which less than 7% are from saturated fat), and 10% to 35% in the form of protein. In prescribing a diet, it is important to relate dietary objectives to the type of diabetes. In patients with type 2 diabetes, limiting the carbohydrate intake and substituting some of the calories with monounsaturated fats, such as olive oil, rapeseed (canola) oil, or the oils in nuts and avocados, can lower triglycerides and increase HDL cholesterol. In addition, in those patients with obesity and type 2 diabetes, weight reduction by caloric restriction is an important goal of the diet. Patients with type 1 diabetes or type 2 diabetes who take insulin should be taught carbohydrate counting, so they can administer their insulin bolus for each meal based on its carbohydrate content.

The current recommendations for both types of diabetes continue to limit cholesterol to 300 mg daily, and individuals with LDL cholesterol more than 100 mg/dL should limit dietary cholesterol to 200 mg daily. High protein intake may cause progression of renal disease in patients with diabetic nephropathy; for these individuals, a reduction in protein intake to 0.8 kg/d (or about 10% of total calories daily) is recommended.

Exchange lists for meal planning can be obtained from the American Diabetes Association and its affiliate associations or from the American Dietetic Association, 216 W. Jackson Blvd., Chicago, IL 60606 (312-899-0040 or http://www.eatright.org).

Special Considerations in Dietary Control

Dietary Fiber

Plant components such as cellulose, gum, and pectin are indigestible by humans and are termed dietary fiber. Insoluble fibers such as cellulose or hemicellulose increase stool bulk and decrease transit time. Soluble fibers such as gums and pectins, found in beans, oatmeal, or apple skin, can delay glucose absorption and so diminish postprandial hyperglycemia. Although the ADA diet does not require insoluble fiber supplements such as added bran, it recommends foods such as oatmeal, cereals, and beans with relatively high soluble fiber content as stable components of the diet in patients with diabetes. High soluble fiber content in the diet may also have a favorable effect on blood cholesterol levels.

Glycemic Index

Quantitation of the relative glycemic contribution of different carbohydrate foods has formed the basis of a glycemic index in which the area of blood glucose (plotted on a graph) generated over a 3-hour period following ingestion of a test food containing 50 g of carbohydrate is compared with the area plotted after giving a similar quantity of reference food such as glucose or white bread:

White bread is preferred to glucose as a reference standard because it is more palatable and has less tendency to slow gastric emptying by high tonicity, as happens when glucose solution is used. Eating foods with low glycemic index will result in lower glucose levels after meals. Low glycemic index foods have values of 55 or less and include many fruits (apples, oranges) and vegetables, grainy breads, pasta, legumes, milk, and yoghurt. High glycemic index foods have values of 70 and over and include baked potato, white bread, and most white rice. Glycemic index is lowered by the presence of fats and protein when the food is consumed in a mixed meal. Cooking methods can also affect the glycemic index—thus mashed potatoes have a higher glycemic index than baked potato. Since you have to have 50 g of available carbohydrate to measure the glycemic index, you cannot assign glycemic indices to foods which have very little carbohydrate. Even though it may not be possible to accurately predict the impact of the glycemic index of a particular food in the context of a meal, it is still reasonable to choose foods with low glycemic index.

Sweeteners

The nonnutritive sweetener saccharin is widely used as a sugar substitute (sweet and low). Aspartame (NutraSweet) consists of two major amino acids, aspartic acid and phenylalanine, which combine to produce a nutritive sweetener 180 times as sweet as sucrose. A major limitation is its heat lability, which precludes its use in baking or cooking. Sucralose (Splenda) and acesulfame potassium (Sunett, Sweet One, DiabetiSweet) are two other nonnutritive sweeteners that are heat stable and can be used in cooking and baking.

Fructose represents a natural sugar substance that is a highly effective sweetener. It induces only slight increases in plasma glucose levels and does not require insulin for its utilization. However, because of potential adverse effects of large amounts of fructose (up to 20% of total calories) on raising serum cholesterol, triglycerides, and LDL cholesterol, it does not have any advantage as a sweetening agent in the diabetic diet. This does not preclude, however, ingestion of fructose-containing fruits and vegetables or fructose-sweetened foods in moderation.

Sugar alcohols, also known as polyols or polyalcohol, are commonly used as sweeteners and bulking agents. They occur naturally in a variety of fruits and vegetables but are also commercially made from sucrose, glucose, and starch. Examples are sorbitol, xylitol, mannitol, lactitol, isomalt, maltitol, and hydrogenated starch hydrolysates (HSH). They are not as easily absorbed as sugar, so they do not raise blood glucose levels as much as conventional sugars. Therefore, sugar alcohols are often used in food products that are labeled as sugar free, such as chewing gum, lozenges, hard candy, and sugar-free ice cream. However, if consumed in large quantities, they will raise blood glucose and can also cause bloating and diarrhea.

Fish Oils and Other Oils

Omega-3 fatty acids in high doses have been shown to lower plasma triglycerides and VLDL cholesterol. They may also reduce platelet aggregation. In the Lyon Diet Heart Study in nondiabetic patients, a high intake of α-linolenic acid was beneficial in secondary prevention of coronary heart disease. This diet, which is rich in vegetables and fruits, also supplies a high intake of natural antioxidants. There is limited clinical information on the use of these oils in patients with diabetes.

Agents for the Treatment of Hyperglycemia

The drugs for treating type 2 diabetes (Table 17–13), other than insulin, fall into several categories. (1) Drugs that act on the sulfonylurea receptor complex of the β cell: sulfonylureas remain the most widely prescribed drugs for treating hyperglycemia. The meglitinide analog repaglinide and the d-phenylalanine derivative nateglinide also bind the sulfonylurea receptor and stimulate insulin secretion. (2) Drugs that principally lower glucose levels by their actions on liver, skeletal muscle, or adipose tissue: metformin works primarily in the liver, and the peroxisome proliferator-activated receptor agonists (PPARs) rosiglitazone and pioglitazone appear to have their main effects on skeletal muscle and adipose tissue. (3) Drugs that principally affect absorption of glucose: the α-glucosidase inhibitors acarbose and miglitol are currently available drugs in this class. (4) Drugs that mimic incretin effects or prolong incretin action: the GLP-1 receptor agonists and the DPP-4 inhibitors fall into this category. (5) Other drugs: these include pramlintide, which lowers glucose by suppressing glucagon and slowing gastric emptying.

Drugs that Act on the Sulfonylurea Receptor Complex

Sulfonylureas

The sulfonylureas contain a sulfonic acid-urea nucleus that can be modified by chemical substitutions to produce agents that have similar qualitative actions but differ widely in potency. They bind the ATP-sensitive potassium channels (KATP) on the surface of pancreatic β cells, resulting in closure of the channel and depolarization of the β cell. This depolarized state permits calcium to enter the cell and actively promote insulin release (Figure 17–5).

In β cells, the ATP-sensitive channels contain four copies each of two subunits, the regulatory subunit SUR1, which binds ATP, ADP, and sulfonylureas, and the potassium channel subunit Kir6.2. KATP channels composed of the same subunits are found in α cells, GLP-1 secreting intestinal l-cells, and the brain. The SUR1/Kir6.2 complexes are opened by diazoxide and closed by sulfonylureas at low concentrations (IC50 about 1 nM for glyburide). Inactivating mutations in SUR1 or Kir6.2 cause persistent depolarization of the β cells and have been identified in patients with hyperinsulinemic hypoglycemia of infancy (see Chapter 18). Activating mutations in SUR1 or Kir6.2 prevent depolarization of the β cells and have been identified in patients with neonatal diabetes (see section on Neonatal Diabetes earlier).

KATP channels with different subunit combinations are found in other tissues. The combination of SUR2A/Kir6.2 is found in cardiac and skeletal muscle, and SUR2B/Kir6.1 in vascular smooth muscle. Channel configurations containing SUR2 subunits are insensitive to diazoxide but sensitive to other potassium channel agonists such as pinacidil and cromakalim. Certain channel closers have much higher affinity for SUR1-containing channels than SUR2-containing channels (the sulfonylureas tolbutamide and gliclazide and the meglitinides nateglinide and mitiglinide), while others have similar affinities for both of them (glyburide, glimepiride, and repeglinide). It remains uncertain whether the different affinities of these drugs for the two classes of receptors have clinical relevance.

Sulfonylureas are not indicated for use in type 1 diabetes patients since these drugs require functioning pancreatic β cells to produce their effect on blood glucose. These drugs are used in patients with type 2 diabetes, in whom acute administration improves the early phase of insulin release that is refractory to acute glucose stimulation. Sulfonylureas are metabolized by the liver and apart from acetohexamide, whose metabolite is more active than the parent compound, the metabolites of all the other sulfonylureas are weakly active or inactive. The metabolites are excreted by the kidney and, in the case of the second-generation sulfonylureas, partly excreted in the bile. Sulfonylureas are generally contraindicated in patients with severe liver or kidney impairment. Idiosyncratic reactions are rare, with skin rashes or hematologic toxicity (leukopenia, thrombocytopenia) occurring in less than 0.1% of users.

First-generation sulfonylureas (tolbutamide, tolazamide, acetohexamide, and chlorpropamide)—Tolbutamide is supplied in tablets of 500 mg. It is rapidly oxidized in the liver to an inactive form. Because its duration of effect is short (6-10 hours), it is usually administered in divided doses (eg, 500 mg before each meal and at bedtime). The usual daily dose is 1.5 to 3 g; some patients, however, require only 250 to 500 mg daily. Acute toxic reactions such as skin rashes are rare. Because of its short duration of action, which is independent of renal function, tolbutamide is probably the safest agent to use in elderly patients, in whom hypoglycemia would be a particularly serious risk. Prolonged hypoglycemia has been reported rarely, mainly in patients receiving certain drugs (eg, warfarin, phenylbutazone, or sulfonamides) that compete with sulfonylureas for hepatic oxidation, resulting in maintenance of high levels of unmetabolized active sulfonylureas in the circulation.

Tolazamide, acetohexamide, and chlorpropamide are rarely used. Chlorpropamide has a prolonged biologic effect, and severe hypoglycemia can occur especially in the elderly, because their renal clearance declines with aging. Its other side effects include alcohol-induced flushing and hyponatremia due to its effect on vasopressin secretion and action. Tolazamide is comparable to chlorpropamide in potency but is devoid of disulfiram-like or water-retaining effects. Its duration of action may last up to 20 hours, with maximal hypoglycemic effect occurring between the 4th and 14th hours.

Second-generation sulfonylureas: glyburide, glipizide, gliclazide, and glimepiride—these agents have similar chemical structures, with cyclic carbon rings at each end of the sulfonylurea nucleus; this causes them to be highly potent (100- to 200-fold more so than tolbutamide). The drugs should be used with caution in patients with cardiovascular disease as well as in elderly patients, in whom hypoglycemia would be especially dangerous.

• Glyburide (glibenclamide)—Glyburide is supplied in tablets containing 1.25, 2.5, and 5 mg. The usual starting dose is 2.5 mg/d, and the average maintenance dose is 5 to 10 mg/d given as a single morning dose. If patients are going to respond to glyburide, they generally do so at doses of 10 mg/d or less, given once daily. If they fail to respond to 10 mg/d, it is uncommon for an increase in dosage to result in improved glycemic control. Maintenance doses higher than 20 mg/d are not recommended and may even worsen hyperglycemia. Glyburide is metabolized in the liver into products with such low hypoglycemic activity that they are considered clinically unimportant unless renal excretion is compromised. Although assays specific for the unmetabolized compound suggest a plasma half-life of only 1 to 2 hours, the biologic effects of glyburide clearly persist for 24 hours after a single morning dose in diabetic patients. Glyburide is unique among sulfonylureas in that it not only binds to the pancreatic β cell membrane sulfonylurea receptor but also becomes sequestered within the β cell. This may also contribute to its prolonged biologic effect despite its relatively short circulating half-life.
• A formulation of micronized glyburide, which apparently increases its bioavailability, is available in easy to divide tablet sizes of 1.5, 3, and 6 mg.
• Glyburide has few adverse effects other than its potential for causing hypoglycemia. It should not be used in patients with liver failure and renal failure because of the risk of hypoglycemia. Elderly patients are at particular risk of hypoglycemia even with relatively small daily doses. For this reason, drugs with a shorter half-life (eg, tolbutamide or possibly glipizide) are preferable in the treatment of type 2 diabetes in the elderly patient. Glyburide does not cause water retention, as chlorpropamide does, and even slightly enhances free water clearance.
• Glipizide (glydiazinamide)—Glipizide is supplied in tablets containing 5 and 10 mg. For maximum effect in reducing postprandial hyperglycemia, this agent should be ingested 30 minutes before breakfast, because rapid absorption is delayed when the drug is taken with food. The recommended starting dose is 5 mg daily but 2.5 mg dose may be sufficient in elderly patients with early diabetes. The dose can gradually be increased by 2.5 or 5 mg increments. Although as much as 15 mg can be given as a single daily dose before breakfast, most patients do better with divided dosing, taking 5 mg before each meal or taking 10 mg before breakfast and before dinner. The maximum recommended dose is 40 mg/d, although doses above 20 mg probably provide little additional benefit in poor responders. Because of its lower potency and shorter half-life, it is preferable to glyburide in elderly patients. At least 90% of glipizide is metabolized in the liver to inactive products, and only a small fraction is excreted unchanged in the urine. Glipizide therapy is contraindicated in patients with liver failure.
• Glipizide is also available as a slow-release preparation (Glucotrol-XL, 5 and 10 mg tablets). The medication is enclosed in a nonabsorbable shell that contains an osmotic compartment that expands slowly, thereby slowly pumping out the glipizide in a sustained manner. It provides extended release during transit through the gastrointestinal tract, with greater effectiveness in lowering of prebreakfast hyperglycemia than the shorter duration, immediate-release standard glipizide tablets. However, this formulation appears to have sacrificed glipizide's reduced propensity for severe hypoglycemia compared with longer acting glyburide without showing any demonstrable therapeutic advantages over glyburide.
• Gliclazide (not available in the United States)—This drug is another intermediate duration sulfonylurea with a duration of action of about 12 hours. It is available as 80-mg tablets. The recommended starting dose is 40 to 80 mg/d with a maximum dose of 320 mg. Doses of 160 mg and above are given as divided doses before breakfast and dinner. The drug is metabolized by the liver, and the metabolites and conjugates have no hypoglycemic effect. An extended-release preparation is also available.
• Glimepiride—This sulfonylurea is supplied in tablets containing 1, 2, and 4 mg. It has a long duration of effect with a half-life of 5 hours, allowing once-daily administration. Glimepiride achieves blood glucose lowering with the lowest dose of any sulfonylurea compound. A single daily dose of 1 mg/d has been shown to be effective, and the maximal recommended dose is 8 mg. It is completely metabolized by the liver to relatively inactive metabolic products.

Meglitinide Analogs

Repaglinide is supplied as 0.5, 1, and 2 mg tablets. Its structure is similar to that of glyburide but lacks the sulfonic acid-urea moiety. It also acts by binding to the sulfonylurea receptor and closing the ATP-sensitive potassium channel. It is rapidly absorbed from the intestine and then undergoes complete metabolism in the liver to inactive biliary products, giving it a plasma half-life of less than 1 hour. The drug therefore causes a brief but rapid pulse of insulin. The starting dose is 0.5 mg three times a day 15 minutes before each meal. The dose can be titrated to a maximum daily dose of 16 mg. Like the sulfonylureas, repaglinide can be used in combination with metformin. Hypoglycemia is the main side effect. In clinical trials, when the drug was compared with glyburide, a long-acting sulfonylurea, there was a trend toward less hypoglycemia. Like the sulfonylureas, it causes weight gain. Metabolism is by the cytochrome P4503A4 isoenzyme. Other drugs that induce or inhibit this isoenzyme may increase or inhibit the metabolism of repaglinide. The drug may be useful in patients with renal impairment or in the elderly.

Mitiglinide is a benzylsuccinic acid derivative that is very similar to repaglinide in its clinical effects. It binds to the sulfonylurea receptor causing a brief pulse of insulin. It is given as a 5 or 10 mg dose just before a meal and reduces the postprandial rise in blood glucose. It has been approved for use in Japan.

δ-Phenylalanine Derivative

Nateglinide is supplied in tablets of 60 and 120 mg. This drug binds the sulfonylurea receptor and closes the ATP-sensitive potassium channel. The drug is rapidly absorbed from the intestine, reaching peak plasma levels within 1 hour. It is metabolized in the liver and has a plasma half-life of about 1.5 hours. Like repaglinide, it causes a brief rapid pulse of insulin, and when given before a meal, it reduces the postprandial rise in blood glucose. The 60-mg dose is used in patients with mild elevations in HbA1c. For most patients, the recommended starting and maintenance dosage is 120 mg three times a day before meals. Like the other insulin secretagogues, its main side effects are hypoglycemia and weight gain.

Drugs that Act on Insulin Target Tissues

Metformin

Metformin (1,1-dimethylbiguanide hydrochloride) is used, either alone or in conjunction with other oral agents or insulin, in the treatment of patients with type 2 diabetes.

Metformin acts primarily through AMPK (Figure 17–8), which it activates by uncoupling mitochondrial oxidative phosphorylation and increasing cellular AMP levels. Metformin's therapeutic effects primarily derive from its effects on the liver, where increased AMPK activity reduces hepatic gluconeogenesis and lipogenesis. Metformin is a substrate for organic cation transporter 1, which is abundantly expressed in hepatocytes and in the gut.

Metformin has a half-life of 1.5 to 3 hours, does not bind to plasma proteins, and is not metabolized in humans, being excreted unchanged by the kidneys.

Metformin is the first-line therapy for patients with type 2 diabetes. The current recommendation is to start this drug at diagnosis. A side benefit of metformin therapy is its tendency to improve both fasting and postprandial hyperglycemia and hypertriglyceridemia in obese patients with diabetes without the weight gain associated with insulin or sulfonylurea therapy. Metformin is ineffective in patients with type 1 diabetes. Patients with renal failure (calculated GFR less than 50 mL/minute) should not get this drug because failure to excrete this drug may produce high blood and tissue levels of metformin that would stimulate lactic acid overproduction. Likewise, patients with liver failure or abusers of ethanol should not receive this drug; lactic acid production from the gut and other tissues, which rises during metformin therapy, can result in lactic acidosis when defective hepatocytes cannot remove the lactate or when alcohol-induced reduction of nucleotides interferes with lactate clearance. Finally, metformin is relatively contraindicated in patients with cardiorespiratory insufficiency, because they have a propensity to develop hypoxia that would aggravate the lactic acid production already occurring from metformin therapy. The age cutoff for prescribing metformin has not been defined and remains a function of the overall health of the patient. In general, there is concern that after the age of 65 to 70 years the potential for progressive impairment of renal function or development of a cardiac event while taking metformin raises the risk enough to outweigh the benefits of prescribing the drug for patients with type 2 diabetes.

Metformin is dispensed as 500, 850, and 1000 mg tablets. A 500 and 750 mg extended-release preparation is also available. Eighty-five percent of the maximal glucose-lowering effect is achieved by a daily dose of 1500 mg, and there is little benefit from giving more than 2000 mg daily. It is important to begin with a low dose and increase the dosage very gradually in divided doses—taken with meals—to reduce minor gastrointestinal upsets. A common schedule would be one 500 mg tablet three times a day with meals or one 850 mg or 1000 mg tablet twice daily at breakfast and dinner. The maximum recommended dose is 850 mg three times a day. Up to 2000 mg of the extended-release preparation can be given once a day.

The epithelial cells of the gut efficiently absorb metformin, where it accumulates at high concentrations and causes toxicity due to uncoupling of mitochondrial oxidative phosphorylation and increased lactate production. The most frequent side effects of metformin are gastrointestinal symptoms (anorexia, nausea, vomiting, abdominal discomfort, diarrhea), which occur in up to 20% of patients. These effects are dose-related, tend to occur at onset of therapy, and often are transient. However, in 3% to 5% of patients, therapy may have to be discontinued because of persistent diarrheal discomfort. In a retrospective analysis, it has been reported that patients who switched from immediate-release metformin to a comparable dose of extended-release metformin experienced fewer gastrointestinal side effects.

Absorption of vitamin B12 appears to be reduced during chronic metformin therapy but usually the vitamin B12 levels remain in the normal range. Screening of serum vitamin B12 levels should be considered if the patient develops a macrocytic anemia or if the patient develops peripheral neuropathy symptoms. Hypoglycemia does not occur with therapeutic doses of metformin, which permits its description as a euglycemic or antihyperglycemic drug rather than an oral hypoglycemic agent. Dermatologic or hematologic toxicity is rare.

Lactic acidosis has been reported as a side effect but is uncommon with metformin in contrast to phenformin. Metformin both increases lactate production by uncoupling mitochondrial oxidative phosphorylation, especially in the gut, and reduces lactate removal by the liver by blocking gluconeogenesis. At therapeutic doses of metformin, serum lactate levels rise only minimally if at all, since other organs such as the kidney can remove the slight excess. However, if tissue hypoxia occurs, the metformin-treated patient is at higher risk for lactic acidosis due to compromised lactate removal. Similarly, when kidney function deteriorates, affecting not only lactate removal by the kidney but also metformin excretion, plasma levels of metformin rise far above the therapeutic range and can increase lactate production and block hepatic uptake sufficiently to provoke lactic acidosis even in the absence of other causes of increased lactic acid production. Acute renal failure can occur rarely following radiocontrast administration. Metformin therapy should therefore be temporarily halted on the day of radiocontrast administration and restarted a day or two later after confirmation that renal function is normal. Almost all reported cases of lactic acidosis have involved subjects with associated risk factors that should have contraindicated its use (kidney, liver, or cardiorespiratory insufficiency, alcoholism, advanced age).

Peroxisome Proliferator–Activated Receptor Agonists

Thiazolidinediones are insulin sensitizers exerting their antidiabetic effects through the activation of PPARγ (see discussion on PPAR nuclear receptors in insulin action and insulin resistance earlier). Observed effects of thiazolidinediones include increased glucose transporter expression (GLUT 1 and GLUT 4); decreased free fatty acid levels; decreased hepatic glucose output; increased adiponectin and decreased resistin release from adipocytes; and increased differentiation of preadipocytes into adipocytes. They have also been demonstrated to decrease levels of plasminogen activator inhibitor type 1, matrix metalloproteinase 9, C-reactive protein, and interleukin-6. Like the biguanides, this class of drugs does not cause hypoglycemia. Troglitazone, the first drug in this class to go into widespread clinical use, was withdrawn from clinical use because of drug-associated fatal liver failure.

Two other drugs in the same class are available for clinical use: rosiglitazone and pioglitazone. Both are effective as monotherapy and in combination with sulfonylureas, metformin, or insulin. When used as monotherapy, these drugs lower HbA1c by about 1 or 2 percentage points. When used in combination with insulin, they can result in a 30% to 50% reduction in insulin dosage, and some patients can come off insulin completely. The combination of a thiazolidinedione and metformin has the advantage of not causing hypoglycemia. Patients inadequately managed on sulfonylureas can do well on a combination of sulfonylurea and rosiglitazone or pioglitazone. About 25% of patients in clinical trials fail to respond to these drugs, presumably because they are significantly insulinopenic.

In addition to glucose-lowering, the thiazolidinediones have effects on lipids and other cardiovascular risk factors. Rosiglitazone therapy is associated with increases in total cholesterol, LDL cholesterol (15%), and HDL cholesterol (10%). There is a reduction in free fatty acids of about 8% to 15%. The changes in triglycerides were generally not different from placebo. Pioglitazone in clinical trials lowered triglycerides (9%) and increased HDL cholesterol (15%) but did not cause a consistent change in total cholesterol and LDL cholesterol levels. A prospective randomized comparison of the metabolic effects of pioglitazone and rosiglitazone in patients who had previously taken troglitazone (now withdrawn from clinical use) showed similar effects on HbA1c and weight gain. Pioglitazone-treated subjects, however, had lower total cholesterol, LDL cholesterol, and triglyceride levels when compared with rosiglitazone. A study in patients with type 2 diabetes who were not on lipid-lowering therapy has recently confirmed this difference in lipid-lowering effects of the two thiazolidinediones. Small prospective studies have also shown that treatment with these drugs leads to improvement of biochemical and histological features of nonalcoholic fatty liver disease. The thiazolidinediones also may limit vascular smooth muscle proliferation after injury and there are reports that troglitazone and pioglitazone reduce neointimal proliferation after coronary stent placement. Also, in one double-blind, placebo-controlled study, rosiglitazone was shown to be associated with a decrease in the ratio of urinary albumin to creatinine.

The dosage of rosiglitazone is 4 to 8 mg daily and of pioglitazone 15 to 45 mg daily, and the drugs do not have to be taken with food. Rosiglitazone is primarily metabolized by the CYP 2C8 isoenzyme and pioglitazone is metabolized by CYP 2C8 and CYP 3A4.

Safety concerns and some troublesome side effects have emerged about this class of drugs that limit their use.

A meta-analysis of 42 randomized clinical trials with rosiglitazone suggested that this drug increases the risk of angina pectoris or myocardial infarction. Although subsequent studies have cast doubt on this finding, the FDA has required the manufacturer to include a boxed warning about the potential risk of heart attacks on the drug label. A meta-analysis of clinical trials with pioglitazone did not show a similar finding.

Edema occurs in about 3% to 4% of patients receiving monotherapy with rosiglitazone or pioglitazone. The edema occurs more frequently (10% to 15%) in patients receiving concomitant insulin therapy and may result in congestive heart failure. The drugs are contraindicated in diabetic individuals with New York Heart Association class III and IV cardiac status. Thiazolidinediones have also rarely been reported as being associated with new onset or worsening macular edema which resolved or improved once the drug was discontinued.

In experimental animals, rosiglitazone stimulates bone marrow adipogenesis at the expense of osteoblastogenesis resulting in a decrease in bone mineral density. An increase in fracture risk in women (but not men) has been reported with both rosiglitazone and pioglitazone. The fracture risk is in the range of 1.9 per 100 patient-years with the thiazolidinedione. In at least one study of rosiglitazone, the fracture risk was increased in premenopausal as well as postmenopausal women. This increased risk for fractures is perhaps the most important reason for limiting the use of this class of drugs.

Other side effects include anemia, which occurs in 4% of patients treated with these drugs; it may be due to a dilutional effect of increased plasma volume rather than a reduction in red cell mass. Weight gain occurs, especially when the drug is combined with a sulfonylurea or insulin. Some of the weight gain is fluid retention, but there is also an increase in total fat mass.

Troglitazone, the first drug in this class, was withdrawn from clinical use because of drug-associated fatal liver failure. The two currently available agents, rosiglitazone and pioglitazone, have thus far not caused hepatotoxicity. The FDA has, however, recommended that patients should not initiate drug therapy if there is clinical evidence of active liver disease or pretreatment elevation of the alanine aminotransferase (ALT) level that is 2.5 times greater than the upper limit of normal. Obviously, caution should be used in initiation of therapy in patients with even mild ALT elevations. Liver biochemical tests should be performed prior to initiation of treatment and periodically thereafter.

Drugs that Affect Glucose Absorption

Alpha-Glucosidase Inhibitors

Drugs of this family are competitive inhibitors of intestinal brush border α glucosidases. Two of these drugs, acarbose and miglitol, are available for clinical use in the United States. Voglibose, another α glucosidase inhibitor, is available in Japan, Korea, and India. Acarbose and miglitol are potent inhibitors of glucoamylase, α-amylase, and sucrase. They are less effective on isomaltase and are ineffective on trehalase or lactase. Acarbose binds 1000 times more avidly to the intestinal disaccharidases than do products of carbohydrate digestion or sucrose. A fundamental difference exists between acarbose and miglitol in their absorption. Acarbose has the molecular mass and structural features of a tetrasaccharide, and very little (∼2%) crosses the microvillar membrane. Miglitol, however, is structurally similar to glucose and is absorbable. Both drugs delay the absorption of carbohydrates and reduce postprandial glycemic excursion.

• Acarbose is available as 50 and 100 mg tablets. The recommended starting dose is 50 mg twice daily, gradually increasing to 100 mg three times daily. For maximal benefit on postprandial hyperglycemia, acarbose should be given with the first mouthful of food ingested. In diabetic patients it reduces postprandial hyperglycemia by 30% to 50%, and its overall effect is to lower the HbA1c by 0.5% to 1%. The principal adverse effect, seen in 20% to 30% of patients, is flatulence. This is caused by undigested carbohydrate reaching the lower bowel, where gases are produced by bacterial flora. In 3% of cases, troublesome diarrhea occurs. This gastrointestinal discomfort tends to discourage excessive carbohydrate consumption and promotes improved compliance of patients with type 2 diabetes with their diet prescriptions.
• When acarbose is given alone, there is no risk of hypoglycemia. However, if combined with insulin or sulfonylureas, it may increase risk of hypoglycemia from these agents. A slight rise in hepatic aminotransferases has been noted in clinical trials (5% vs 2% in placebo controls, and particularly with doses >300 mg/d). This generally returns to normal on stopping this drug. In the UKPDS, approximately 2000 patients on diet, sulfonylurea, metformin, or insulin therapy were randomized to acarbose or placebo therapy. By 3 years, 60% of the patients had discontinued the drug, mostly because of gastrointestinal symptoms. In the 40% of patients who remained on the drug, acarbose was associated with a 0.5% lowering of HbA1c compared with placebo.
• Miglitol is similar to acarbose in terms of its clinical effects. It is indicated for use in diet- or sulfonylurea-treated patients with type 2 diabetes. Therapy is initiated at the lowest effective dosage of 25 mg three times a day. The usual maintenance dose is 50 mg three times a day, although some patients may benefit from increasing the dose to 100 mg three times a day. Gastrointestinal side effects occur as with acarbose. The drug is not metabolized and is excreted unchanged by the kidney. Theoretically, absorbable α-glucosidase inhibitors could induce a deficiency of one or more of α-glucosidases involved in cellular glycogen metabolism and biosynthesis of glycoproteins. This does not occur in practice because—unlike the intestinal mucosa, which is exposed to a high concentration of the drug—circulating plasma levels are 200-fold to 1000-fold lower than those needed to inhibit intracellular α glucosidases. Miglitol should not be used in renal failure because its clearance is impaired in this setting.

Incretins

The gut makes several incretins, gut hormones that amplify postprandial insulin secretion, including glucagon-like peptide-1 (GLP-1, Figure 17–9) and glucose-dependent insulinotropic polypeptide (GIP). Therapeutic drugs in this class include GLP-1 receptor agonists and dipeptidyl peptidase 4 (DPP-4) inhibitors, which increase levels of both GLP-1 and GIP.

When GLP-1 is infused in patients with type 2 diabetes, it stimulates insulin secretion and lowers glucose levels. GLP-1, unlike the sulfonylureas, has only a modest insulin stimulatory effect under normoglycemic conditions. This means that GLP-1 administration has a lower risk of causing hypoglycemia than the sulfonylureas. GLP-1, in addition to its insulin stimulatory effect, also has a number of other pancreatic and extrapancreatic effects (Table 17–4).

GLP-1 Receptor Agonists

GLP-1 is rapidly proteolyzed by DPP-4 and by other enzymes such as endopeptidase 24.11. It is also cleared rapidly by the kidney. As a result, the half-life of GLP-1 is only 1 to 2 minutes. The native peptide, therefore, cannot be used therapeutically. Instead, the approach taken has been to develop metabolically stable analogs or derivatives of GLP-1 that are not subject to the same enzymatic degradation or renal clearance. Two GLP-1 receptor agonists, exenatide and liraglutide, are currently available for clinical use.

• Exenatide or exendin 4 is a GLP-1 receptor agonist isolated from the saliva of the Gila monster (a venomous lizard) that is resistant to DPP-4 action and is cleared from the plasma by glomerular filtration. When given to patients with type 2 diabetes by subcutaneous injection twice a day, this compound lowers blood glucose and HbA1c levels. Exenatide appears to have the same effects as GLP-1 on glucagon suppression and gastric emptying. In clinical trials, adding exenatide to the therapeutic regimen of patients with type 2 diabetes who are already taking metformin or a sulfonylurea (or both) further lowered the HbA1c value by 0.4% to 0.6% over a 30-week period. These patients also lost 3 to 6 lb in weight. In an open-label extension study up to 80 weeks, the HbA1c reduction was sustained, and there was further weight loss (to a total loss of ∼10 lb).
• The main side effect was nausea, affecting over 40% of patients. The nausea was dose-dependent and declined with time. The risk of hypoglycemia was higher in subjects on sulfonylureas. The delay in gastric emptying may affect the absorption of some other drugs; therefore, antibiotics and oral contraceptives should be taken 1 hour before exenatide doses. Low-titer antibodies against exenatide develop in over one-third (38%) of patients, but the clinical effects are not attenuated. High-titer antibodies develop in a subset of patients (∼6%), and in about half of these cases, an attenuation of glycemic response has been seen.
• The FDA has received 30 postmarketing reports of acute pancreatitis in patients taking exenatide.The pancreatitis was severe (hemorrhagic or necrotizing) in six instances and two of these patients died. Many of the patients had other risk factors for pancreatitis, but the possibility remains that the drug was causally responsible for some cases. Patients taking exenatide should be advised to seek immediate medical care if they experience unexplained, persistent, severe abdominal pain. The drug should be discontinued if pancreatitis is suspected. The FDA also reported 16 cases of renal impairment and 62 cases of acute renal failure in patients taking exenatide. Some of these patients had preexisting kidney disease and others had one or more risk factors for kidney disease. A number reported nausea, vomiting, and diarrhea and it is possible that these side effects cause volume depletion and contributed to the development of renal failure.
• Exenatide is dispensed as two fixed-dose pens (5 and 10 μg). It is injected 60 minutes before breakfast and before dinner. Patients should be prescribed the 5-μg pen for the first month and then, if tolerated, the dose should be increased to 10 μg twice daily. The drug is not recommended in patients with glomerular filtration rate less than 30 mL/min.
• Liraglutide is a soluble fatty acid acylated GLP-1 analog—with replacement of lysine with arginine at position 34 and the attachment of a C16 acyl chain to a lysine at position 26. The fatty-acyl GLP-1 retains affinity for GLP-1 receptors, but the addition of the C16 acyl chain allows for noncovalent binding to albumin, both hindering DPP-4 access to the molecule and contributing to a prolonged half-life and duration of action. The half-life is approximately 12 hours allowing the drug to be injected once a day.
• In clinical trials lasting 26 and 52 weeks, adding liraglutide to the therapeutic regimen (metformin, sulfonylurea, thiazolidinedione) of patients with type 2 diabetes further lowered the HbA1c value. Depending on the dose and design of the study, the HbA1c decline was in the range of 0.6% to 1.5%. The patients had sustained weight loss of 1 to 6 lb.
• Like exenatide, the most frequent side effect is nausea and vomiting affecting approximately 10% of subjects. There was also an increase in incidence of diarrhea. About 2% to 5% of subjects withdrew from the studies because of the GI symptoms. Pancreatitis may also occur—in the clinical trials there were seven cases of pancreatitis in the liraglutide-treated group with one case in the comparator group (2.2 vs 0.6 cases per 1000 patient-years). Liraglutide stimulates C-cell neoplasia and causes medullary thyroid carcinoma in rats. Human C-cells express very few GLP-1 receptors and the relevance to human therapy is unclear but because of the animal data, the drug should not be used in patients with personal or family history of medullary thyroid carcinoma or multiple endocrine neoplasia syndrome type 2.
• The dosing is initiated at 0.6 mg daily increased after 1 week to 1.2 mg daily. An additional increase in dose to 1.8 mg is recommended if needed for optimal glycemic control. Titration should of course be based on tolerability. There is limited experience using the drug in renal failure but no dose adjustment is recommended.

DPP-4 Inhibitors

An alternative to the use of GLP-1 receptor agonists involves inhibition of the enzyme DPP-4 with prolongation of the action of endogenously released GLP-1 and GIP. Two oral DPP-4 inhibitors, sitagliptin and saxagliptin, are available for the treatment of type 2 diabetes in the United States. An additional DPP-4 inhibitor, Vildagliptin, is available in Europe.

• Sitagliptin in clinical trials was shown to be effective in lowering glucose when used alone and in combination with metformin and pioglitazone. In various clinical trials, improvements in HbA1c ranged from 0.5% to 1.4%. The usual dose of sitagliptin is 100 mg once daily, but the dose is reduced to 50 mg daily if the calculated creatinine clearance is 30 to 50 mL/min and to 25 mg for clearances less than 30 mL/min. Unlike exenatide, sitagliptin does not cause nausea or vomiting. It also does not result in weight loss. The main adverse effect appears to be a predisposition to nasopharyngitis or upper respiratory tract infection. A small increase in neutrophil count of approximately 200 cells/mL has also occurred. Since its FDA approval and clinical use, there have been reports of serious allergic reactions to sitagliptin, including anaphylaxis, angioedema, and exfoliative skin conditions, including Stevens-Johnson syndrome. There have also been reports of pancreatitis (88 cases including two cases of hemorrhagic or necrotizing pancreatitis). The frequency of these events is unclear. DPP-4 is a pleiotropic enzyme that inactivates a variety of peptide hormones, neuropeptides, and chemokines and DPP-4 inhibitors have been shown to prolong the action of neuropeptide Y and substance P. Whether its inhibition over a long-term period will have negative consequences is not known.
• Saxagliptin when added to the therapeutic regimen (metformin, sulfonylurea, thiazolidinedione) of patients with type 2 diabetes further lowered the HbA1c value by about 0.7% to 0.9%. The dose is 2.5 or 5 mg once a day. The 2.5 mg dose should be used in patients with calculated creatinine clearance less than 50 mL/minute. It lowers HbA1c by about 0.6% when added to metformin, glyburide, or thiazolidine in various 24-week clinical trials. The drug is weight neutral. The main adverse reactions were upper respiratory tract infection, naspharyngitis, headache, and urinary tract infection. There is also small reversible dose-dependent reduction in absolute lymphocyte count which remains within normal limits. Hypersensitivity reactions such as urticaria and facial edema occurred in 1.5% of patients on the drug compared to 0.4% with placebo. The metabolism of saxagliptin is through CYP 3A4/5; so strong inhibitors or inducers of CYP 3A4/5 will affect the pharmacokinetics of saxagliptin and its active metabolite.
• Vildagliptin, like the other DPP-4 inhibitors, lowers HbA1c by about 0.5% to 1% when added to the therapeutic regimen of patients with type 2 diabetes. The dose is 50 mg once a day or twice a day. The adverse reactions include upper respiratory tract infections, nasopharyngitis, dizziness, and headache. Rare cases of hepatic dysfunction including hepatitis have been reported. Liver function testing is recommended, quarterly the first year, and periodically thereafter.

Others

Pramlintide is a synthetic analog of islet amyloid polypeptide (IAPP) that when given subcutaneously (1) delays gastric emptying, (2) suppresses glucagon secretion, and (3) decreases appetite. It is approved for use both in type 1 and insulin-treated type 2 patients. In 6-month clinical studies with type 1 and insulin-treated type 2 patients, those on the drug had approximately a 0.4% reduction in HbA1c and lost about 1.7 kg compared to placebo. The HbA1c reduction was sustained for 2 years, but some of the weight was regained. The drug is given immediately before the meal by injection. Hypoglycemia is the most concerning adverse event, and it is recommended that the short-acting or premixed insulin doses be reduced by 50% when the drug is started. Since the drug slows gastric emptying, recovery from hypoglycemia can be a problem because of delay in absorption of the fast-acting carbohydrate. Nausea was the other main side effect, affecting 30% to 50% of subjects. It tended to improve with time. In patients with type 1 diabetes, pramlintide is initiated at the dose of 15 μg before each meal and titrated by 15 μg increments to a maintenance dose of 30 or 60 μg before each meal. In patients with type 2 diabetes, the initiation dose is 60 μg premeals increased to 120 μg in 3 to 7 days if no significant nausea occurs.

A combination of pramlintide and recombinant human leptin (metreleptin) is currently being evaluated in clinical trials for weight loss.

Drug Combinations

A number of drug combinations, glyburide-metformin (Glucovance), glipizide-metformin (Metaglip), repaglinide–metformin (Prandi-Met), rosiglitazone-metformin (Avandamet), pioglitazone-metformin (ACTOplus Met), sitagliptin-metformin (Janumet), rosiglitazone-glimepiride (Avandaryl), and pioglitazone-glimepiride (Duetact) are available. These combinations, however, limit the clinician's ability to optimally adjust dosage of the individual drugs and for that reason are of questionable merit.

Insulin

Insulin is indicated for individuals with type 1 diabetes as well as for those with type 2 diabetes whose hyperglycemia does not respond to diet therapy and other diabetes drugs.

Human insulin is now produced by recombinant DNA techniques (biosynthetic human insulin). Animal insulins are no longer available in the United States. With the availability of highly purified human insulin preparations, immunogenicity has been markedly reduced, thereby decreasing the incidence of therapeutic complications such as insulin allergy, immune insulin resistance, and localized lipoatrophy at the injection site. The insulins are also quite stable and refrigeration while in use is not necessary. During travel, reserve supplies of insulin can be readily transported without significant loss of potency, provided they are protected from extremes of heat or cold.

Insulins in Europe and United States are available only in a concentration of 100 U/mL (U100). Low dose (0.3 mL) disposable insulin syringes allow for accurate dosing of as low as 1 or 2 U. If extremely low doses of insulin are needed, diluted insulin can be prepared using diluent available from the insulin manufacturer. For use in rare cases of severe insulin resistance in which large quantities of insulin are required, a U500 (500 U/mL) regular human insulin is available from Eli Lilly.

Insulin preparations differ with regard to their time of onset and duration of biologic action (Figure 17–11; Table 17–14). The short-acting preparations are regular insulin and the rapidly acting insulin analogs (Table 17–15). They are dispensed as clear solutions at neutral pH and contain small amounts of zinc to improve their stability and shelf life. The long-acting preparations are neutral protamine hagedorn (NPH) insulin and the long–acting insulin analogs. NPH insulin is dispensed as a turbid suspension at neutral pH with protamine in phosphate buffer. The long-acting insulin analogs are dispensed as clear solutions; insulin glargine is at acidic pH and insulin detemir is at neutral pH. The rapidly acting insulin analogs and the long-acting insulins are designed for subcutaneous administration, while regular insulin can also be given intravenously. While insulin aspart has been approved for intravenous use (eg, in hyperglycemic emergencies), there is no advantage in using insulin aspart over regular insulin by this route.

It is important to recognize that values given for time of onset of action, peak effect, and duration of action are only approximate and that there is great variability in these parameters from patient to patient and even in a given patient depending on the size of the dose, the site of injection, the degree of exercise, the avidity of circulating anti-insulin antibodies, and other less well-defined variables.

Short Acting Insulin Preparations

Regular Insulin

Regular insulin is a short-acting, soluble crystalline zinc insulin whose hypoglycemic effect appears within 30 minutes after subcutaneous injection, peaks at about 2 hours, and lasts for about 5 to 7 hours when usual quantities (ie, 5-15 U) are administered. Intravenous infusions of regular insulin are particularly useful in the treatment of diabetic ketoacidosis and during the perioperative management of insulin-requiring diabetics. Regular insulin is indicated when the subcutaneous insulin requirement is changing rapidly, such as after surgery or during acute infections—although the rapidly acting insulin analogs may be preferable in these situations.

For very insulin-resistant subjects who would otherwise require large volumes of insulin solution, a U500 preparation of human regular insulin is available. Because a U500 syringe is not available, a U100-insulin syringe or tuberculin syringe is used to measure doses. The physician should carefully note dosages in both units and volume to avoid overdosage.

Rapidly Acting Insulin Analogs

Insulin lispro (Humalog) is an insulin analog produced by recombinant technology, wherein two amino acids near the carboxyl terminal of the B chain have been reversed in position; proline at position B28 has been moved to B29 and lysine has been moved from B29 to B28. Insulin aspart (Novolog) is a single substitution of proline by aspartic acid at position B28. Insulin glulisine (Apidra) differs from human insulin in that the amino acid asparagine at position B3 is replaced by lysine and the lysine in position B29 by glutamic acid. These changes reduce the tendency to form hexamers in these three analogs, in contrast to human insulin. When injected subcutaneously, the analogs quickly dissociate into monomers and are absorbed very rapidly, reaching peak serum values in as soon as 1 hour—in contrast to regular human insulin, whose hexamers require considerably more time to dissociate and become absorbed. The amino acid changes in these analogs do not interfere with their binding to the insulin receptor, with the circulating half-life, or with their immunogenicity, which are all identical to those of human regular insulin.

Clinical trials have demonstrated that the optimal times of preprandial subcutaneous injection of comparable doses of the rapid-acting insulin analogs and of regular human insulin are 20 minutes and 60 minutes before the meal, respectively. Although this more rapid onset of action has been welcomed as a great convenience by patients with diabetes who object to waiting as long as 60 minutes after injecting regular human insulin before they can begin their meal, patients must be taught to ingest adequate absorbable carbohydrate early in the meal to avoid hypoglycemia during the meal. Another desirable feature of rapidly acting insulin analogs is that their duration of action remains at about 4 hours irrespective of dosage. This contrasts with regular insulin, whose duration of action is prolonged when larger doses are used.

The rapidly acting analogs are also commonly used in pumps. In a double-blind crossover study comparing insulin lispro with regular insulin in insulin pumps, persons using insulin lispro had lower HbA1c values and improved postprandial glucose control with the same frequency of hypoglycemia. The concern remains that in the event of pump failure, users of the rapidly acting insulin analogs will have more rapid onset of hyperglycemia and ketosis.

The structural differences between insulin lispro and human insulin may be sufficient to prevent insulin lispro from binding to human insulin antibodies in some patients, and there have been case reports of successful use of insulin lispro in those rare patients who have a generalized allergy to human insulin or who have severe antibody-mediated insulin resistance.

Long Acting Insulin Preparations

• Neutral Protamine Hagedorn (NPH), or isophane, insulin is an intermediate-acting insulin in which the onset of action is delayed by combining two parts of soluble crystalline zinc insulin with one part protamine zinc insulin. The mixture has equivalent concentrations of protamine and insulin, so that neither is in excess (isophane). Its onset of action is delayed by 2 to 4 hours, and its peak response is generally reached in about 8 to 10 hours. Because its duration of action is often less than 24 hours (with a range of 10-20 hours), most patients require at least two injections daily to maintain a sustained insulin effect.
• Flocculation of suspended particles may occasionally frost the sides of a bottle of NPH insulin or clump within bottles from which multiple small doses are withdrawn over a prolonged period. This instability is a rare phenomenon and might occur less frequently if NPH human insulin were refrigerated when not in use and if bottles were discarded after 1 month of use. Patients should be vigilant for early signs of frosting or clumping of the NPH insulin, because it indicates a pronounced loss of potency. Several cases of diabetic ketoacidosis have been reported in patients with type 1 diabetes who had been inadvertently injecting this denatured insulin.
• Insulin glargine is an insulin analog in which the asparagine at position 21 of the A chain of the human insulin molecule is replaced by glycine and two arginines are added to the carboxyl terminal of the B chain. The arginines raise the isoelectric point of the molecule close to neutral, making it more soluble in an acidic environment. In contrast, human insulin has an isoelectric point of pH 5.4. Insulin glargine is a clear insulin that, when injected into the neutral pH environment of the subcutaneous tissue, forms microprecipitates that slowly release the insulin into the circulation. It lasts for about 24 hours without any pronounced peaks and is given once a day to provide basal coverage. This insulin cannot be mixed with the other insulins because of its acidic pH. When this insulin was given as a single injection at bedtime to type 1 diabetes patients, fasting hyperglycemia was better controlled when compared with bedtime NPH insulin. The clinical trials also suggest that there may be less nocturnal hypoglycemia with this insulin when compared with NPH insulin.
• In one clinical trial involving patients with type 2 diabetes, insulin glargine was associated with a slightly more rapid progression of retinopathy when compared with NPH insulin. The frequency was 7.5% with the analog and 2.7% with the NPH. This observation however, was not confirmed in a 5-year open label prospective study of 1024 subjects randomized to NPH or insulin glargine. There was no increase in the risk of progression of retinopathy with the analog insulin.
• In in vitro studies, insulin glargine has a sixfold greater affinity for the IGF-1 receptor compared with the human insulin. There has also been a report that insulin glargine has increased mitogenicity compared with human insulin in a human osteosarcoma cell line. Circulating levels of insulin glargine, however, are low, and the clinical significance of these observations is not yet clear. An observational study from Germany of 127,031 patients who had exposure to regular insulin, short-acting insulin analogs, and insulin glargine reported a strong correlation between increased insulin dose and cancer risk. Moreover insulin glargine, dose for dose, appeared to carry a higher risk than regular insulin. Additional studies are needed to confirm or refute this observation. Because of lack of safety data, use of insulin glargine during pregnancy is not recommended.
• Insulin detemir is an insulin analog in which the tyrosine at position 30 of the B chain has been removed and a 14-C fatty acid chain (tetradecanoic acid) is attached to the lysine at position 29 by acylation. The fatty acid chain makes the molecule more lipophilic than native insulin, and the addition of zinc stabilizes the molecule and leads to formation of hexamers. After injection, self-association at the injection site and albumin binding in the circulation via the fatty acid side chain lead to slower distribution to peripheral target tissues and prolonged duration of action. The affinity of insulin detemir is four- to fivefold lower than that of human soluble insulin and, therefore, the U100 formulation of insulin detemir has a concentration of 2400 nmol/mL compared to 600 nmol/mL for NPH. The duration of action for insulin detemir is about 17 hours at therapeutically relevant doses. It is recommended that the insulin is injected once or twice a day to achieve a stable basal coverage. Apparently this insulin has been reported to have lower within-subject pharmacodynamic variability compared to NPH insulin and insulin glargine. In vitro studies do not suggest any clinically relevant albumin binding interactions between insulin detemir and fatty acids or protein-bound drugs. Because there is a vast excess (∼400,000) of albumin-binding sites available in plasma per insulin detemir molecule, it is unlikely that hypoalbuminemic disease states will affect the ratio of bound to free insulin detemir.

Insulin Mixtures

Because intermediate-acting insulins require several hours to reach adequate therapeutic levels, their use in patients with type 1 diabetes requires supplements of regular insulin or insulin lispro or insulin aspart preprandially. For convenience, regular or NPH insulin may be mixed together in the same syringe and injected subcutaneously in split dosage before breakfast and supper. It is recommended that the regular insulin be withdrawn first, then the NPH insulin. No attempt should be made to mix the insulins in the syringe, and the injection is preferably given immediately after the syringe is loaded. Stable premixed insulins (70% NPH and 30% regular) are available as a convenience to patients who have difficulty mixing insulin because of visual problems or insufficient manual dexterity.

With increasing use of rapid-acting insulin analogs as a preprandial insulin, it has become evident that combination with an intermediate-acting or long-acting insulin is essential to maintain postabsorptive glycemic control. It has been demonstrated that insulin lispro can be acutely mixed with NPH insulin without affecting its rapid absorption. Premixed preparations of insulin lispro and NPH insulin however are unstable because of exchange of insulin lispro with the human insulin in the protamine complex. Consequently, over time, the soluble component becomes a mixture of regular and insulin lispro at varying ratios. In an attempt to remedy this, an intermediate insulin composed of isophane complexes of protamine with insulin lispro was developed and given the name NPL (neutral protamine lispro). Premixed combinations of NPL and insulin lispro are now available for clinical use (Humalog Mix 75/25 and Humalog Mix 50/50). These mixtures have a more rapid onset of glucose-lowering activity compared with 70% NPH/30% regular human insulin mixture and can be given within 15 minutes before or after starting a meal. Similarly, a 70% insulin aspart protamine/30% insulin aspart (NovoLogMix 70/30) is now available. The main advantages of these new mixtures are that (1) they can be given within 15 minutes of starting a meal and (2) they are superior in controlling the postprandial glucose rise after a carbohydrate-rich meal. These benefits have not translated into improvements in HbA1c levels when compared with the usual 70% NPH/30% regular mixture.

The longer acting insulin analogs (insulin glargine or insulin detemir) cannot be acutely mixed with either regular insulin or the rapid-acting insulin analogs.

Methods of Insulin Administration

• Insulin syringes and needles—Disposable plastic syringes with needles attached are available in 1-mL, 0.5-mL, and 0.3-mL sizes. Their finely honed 30- to 31-gauge attached needles have greatly reduced the pain of injections. They are light, not susceptible to damage, and convenient when traveling. Moreover, their clear markings and tight plungers allow accurate measurement of insulin dosage. The low-dose syringes have become increasingly popular, because most patients take less than 30 U at one injection and also it is available with half unit markings. Two lengths of needles are available: short (8 mm) and long (12.7 mm). Long needles are preferable in obese patients to reduce the variability of insulin absorption.
• Sites for injection—Any part of the body covered by loose skin can be used as an injection site, including the abdomen, thighs, upper arms, flanks, and upper-outer quadrants of the buttocks. In general, regular insulin is absorbed more rapidly from upper regions of the body such as the deltoid area or the abdomen rather than from the thighs or buttocks. Exercise appears to facilitate insulin absorption when the injection site is adjacent to the exercising muscle. Rotation of sites continues to be recommended to avoid delayed absorption when fibrosis or lipohypertrophy occurs due to repeated use of a single site. However, considerable variability of absorption rates from different regions, particularly with exercise, may contribute to the instability of glycemic control in certain patients with type 1 diabetes if injection sites are rotated indiscriminately over different areas of the body. Consequently, diabetologists recommend limiting injection sites to a single region of the body and rotating sites within that region. It is possible that some of the stability of glycemic control achieved by infusion pumps may be related to the constancy of the region of infusion from day to day. For most patients, the abdomen is the recommended region for injection because it provides a considerable area in which to rotate sites, and there may be less variability of absorption with exercise than when the thigh or deltoid areas are used. The effect of anatomic regions appears to be much less pronounced with the analogs.
• Insulin pen injector devices—Insulin pens eliminate the need for carrying insulin vials and syringes. Cartridges of insulin lispro, insulin aspart and insulin glargine are available for reusable pens (Lilly, Novo Nordisk, and Owen Mumford). Disposable prefilled pens are also available for insulin lispro, insulin aspart, insulin glulisine, insulin detemir, insulin glargine, NPH, 70% NPH/30% regular, 75% NPL/25% insulin lispro, and 70% insulin aspart protamine/30% insulin aspart. Thirty-one-gauge needles (5, 6, and 8 mm long) for these pens make injections almost painless.
• Insulin pumps—Several small portable open loop pumps for the delivery of insulin are in the market. These devices contain an insulin reservoir and a pump programmed to deliver regular insulin subcutaneously; they do not contain a glucose sensor. With improved methods for self-monitoring of blood glucose at home (see later), these pump systems are becoming increasingly popular. In the United States, Medtronic MiniMed, Insulet, Animas, and Roche insulin infusion pumps are available for subcutaneous delivery of insulin. These pumps are small (about the size of a pager) and easy to program. They have many features, including the ability to record a number of different basal rates throughout a 24-hour period and adjust the time over which bolus doses are given. They are able also to detect pressure build-up if the catheter is kinked. The catheter connecting the insulin reservoir to the subcutaneous cannula can be disconnected so the patient can remove the pump temporarily (eg, for bathing). Ominpod (Insulet Corporation) is an insulin infusion system in which the insulin reservoir and infusion set are integrated into one unit (pod), so it does not use a catheter. The pod, placed on the skin, delivers subcutaneous basal and bolus insulin based on wirelessly transmitted instructions from a personal digital assistant.
• The great advantage of continuous subcutaneous insulin infusion (CSII) is that it allows for establishment of a basal profile tailored to the patient. The patient, therefore, is able to eat with less regard to timing because the basal insulin infusion should maintain a constant blood glucose level between meals. Also, the ability to adjust the basals makes it easier for the patient to manage glycemic excursions that occur with exercise. The pumps also have software that can assist the patient to calculate boluses based on glucose reading and carbohydrates to be consumed. They also keep track of the time elapsed since last insulin bolus and the patient is reminded of this when he or she attempts to give additional correction bolus before the effect of the previous bolus has worn off (insulin on board feature). This feature reduces the risk of overcorrecting and subsequent hypoglycemia.
• CSII therapy is appropriate for patients who are motivated, mechanically adept, educated about diabetes (diet, insulin action, treatment of hypo- and hyperglycemia), and willing to monitor their blood glucose four to six times a day. Known complications of CSII include ketoacidosis, which can occur when insulin delivery is interrupted, and skin infections. Another major disadvantage is the cost and the time demanded of physicians and staff in initiating therapy. Almost all patients use the rapid-acting insulin analogs in their pumps. Clinical trials have shown that when compared with regular insulin, subjects using rapid-acting insulin analogs in pumps had lower HbA1c values and improved postprandial glucose control with the same frequency of hypoglycemia. There does remain a concern that in the event of pump failure, the insulin analogs could result in more rapid onset of hyperglycemia and ketosis.
• Inhaled insulin—Exubera, the first inhaled insulin preparation approved by the FDA is no longer available; the manufacturer stopped marketing it because of lack of demand. In clinical trials, Exubera was as effective as subcutaneous regular insulin in controlling postprandial glucose excursions. Physicians, however, were reluctant to prescribe Exubera for a number of reasons, including a lack of long-term safety data on pulmonary function, availability of other insulin delivery systems, cost, lack of insurance coverage, and awkward dosing. The manufacturer also subsequently reported that there were six cases of lung cancer in patients who were on inhaled insulin and one case in the comparator-treated patients. All the patients who developed lung cancer had a history of prior cigarette smoking. There is currently only one pharmaceutical company that is still conducting clinical trials with inhaled insulin (Technosphere insulin, MannKind Corporation).
• Pancreas transplantation—This transplantation done at the time of renal transplantation is becoming more widely accepted. Patients undergoing simultaneous pancreas and kidney transplantation have an 85% chance of pancreatic graft survival and a 92% chance of renal graft survival after 1 year. About 1200 pancreas transplants are performed yearly in the United States. About 75% of the transplants are perfomed at the same time as the kidney transplant with both organs from the same donor. About 15% of the pancreas transplants are performed after a previously successful kidney transplant and 10% are performed without a kidney transplant. Solitary pancreatic transplantation in the absence of a need for renal transplantation should be considered only in those rare patients who fail all other insulin therapeutic approaches and who have frequent severe hypoglycemia or who have life-threatening complications related to their lack of metabolic control.
• Islet cell transplantation—This is a minimally invasive procedure, and investigators in Edmonton, Canada, reported initial insulin independence in a small number of patients with type 1 diabetes, who underwent this procedure. Using islets from multiple donors and steroid-free immunosuppression, percutaneous transhepatic portal vein transplantation of islets was achieved in more than 20 subjects. Several centers around the world have successfully replicated this experience. Unfortunately, although all of the initial cohorts achieved insulin independence posttransplantation, some for more than 2 years of follow-up, a decline in insulin secretion occurred with time, and the subjects have again required supplemental insulin. However, patients with successful transplants had complete correction of severe hypoglycemic reactions, leading to a marked improvement in overall quality of life. Islet transplant trials with different immunosuppressive regimens are currently underway. Even if long-term insulin independence is demonstrated, wide application of this procedure for the treatment of type 1 diabetes is limited by the dependence on multiple donors and the requirement for potent long-term immunotherapy.

Steps in the Management of the Diabetic Patient

History and Examination

A complete history is taken and physical examination is performed for diagnostic purposes and to rule out the presence of coexisting or complicating disease. Nutritional status should be noted, particularly if catabolic features such as progressive weight loss are present despite a normal or increased food intake. The family history should include not only the incidence but also the age at onset of diabetes in other members of the family, and it should be noted whether affected family members were obese, whether they required insulin, and whether they developed complications from their diabetes. Other factors that increase cardiovascular risk, such as a smoking history, presence of hypertension or hyperlipidemia, or oral contraceptive pill use should be documented.

A careful physical examination should include baseline height and weight, pulse rate, and blood pressure. If obesity is present, it should be characterized as to its distribution, and a waist-to-hip ratio should be recorded. All peripheral arterial pulses should be examined, noting whether bruits or other signs of atherosclerotic disease are present. Neurologic and ophthalmologic examinations should be performed, with emphasis on investigation of abnormalities that may be related to diabetes, such as neovascularization of the retina or stocking/glove sensory loss in the extremities.

Laboratory Diagnosis

See also Laboratory Findings in diabetes mellitus above. Laboratory diagnosis should include documentation of the presence of fasting hyperglycemia (plasma glucose ≥126 mg/dL [7 mmol/L]), postprandial (post-glucose tolerance test) values consistently ≥200 mg/dL (11.1 mmol/L) or HbA1c ≥6.5%. An attempt should be made to characterize the diabetes as type 1 or type 2, based on the clinical features present and on whether ketonuria accompanies the glycosuria. For the occasional patient, measurement of ICA, ICA 512, GAD, and insulin antibodies can help in distinguishing between type 1 and type 2 diabetes. Many newly diagnosed patients with type 1 diabetes still have significant endogenous insulin production, and C-peptide levels may not reliably distinguish between type 1 and type 2 diabetes. Other baseline laboratory measurements that should be made part of the record include hemoglobin A1c, lipid profile, serum creatinine and electrolytes, complete blood count, electrocardiogram, and urine albumin measurement (type 2 patient).

Patient Education and Self-Management Training

Since diabetes is a lifelong disorder, education of the patient and the family is probably the most important obligation of the clinician who provides initial care. The best persons to manage a disease that is affected so markedly by daily fluctuations in environmental stress, exercise, diet, and infections are the patients themselves and their families. It must be remembered that education is necessary not only for patients with newly diagnosed diabetes and their families, but also for patients with diabetes of any duration who may never have been properly educated about their disorder or who may not be aware of advances in diabetes management. The teaching curriculum should include explanations by the physician or nurse of the nature of diabetes and its potential acute and chronic hazards and how they can be recognized early and prevented or treated. Self-monitoring of blood glucose should be emphasized, especially in insulin-requiring diabetic patients, and instructions must be given on proper testing and recording of data. Patients must also be helped to accept the fact that they have diabetes; until this difficult adjustment is made, efforts to cope with the disorder are likely to be futile. Counseling should be directed at avoidance of extremes such as compulsive rigidity or self-destructive neglect.

Patients on insulin should have an understanding of the actions of basal and bolus insulins. They should be taught how to determine whether the basal dose is appropriate and how to adjust the rapidly acting insulin dose for the carbohydrate content of a meal. Patients and their families or friends should also be taught to recognize signs and symptoms of hypoglycemia and how to institute appropriate therapy for hypoglycemic reactions. Strenuous exercise can precipitate hypoglycemia, and patients should know how much to reduce their insulin dosage in anticipation of strenuous activity or to take supplemental carbohydrate. Injection of insulin into a site farthest away from the muscles most involved in exercise may help ameliorate exercise-induced hypoglycemia, since insulin injected in the proximity of exercising muscle may be more rapidly mobilized. Exercise training also increases the effectiveness of insulin and insulin doses should be adjusted accordingly. Because infections, particularly pyogenic ones with fever and toxemia, provoke a marked increase in insulin requirements, patients must be taught how to appropriately administer supplemental rapid-acting insulin as needed to correct hyperglycemia during infections.

Type 2 diabetes on noninsulin medications should be informed about the time of onset, peak action, duration of action, and any adverse effects of pharmacologic agents being used. They should also learn to inquire about possible drug interactions whenever any new medications are added to their regimens.

The targets for blood glucose control should be elevated appropriately in elderly patients since they have the greatest risk if subjected to hypoglycemia and the least long-term benefit from more rigid glycemic control. Patients should be provided advice on personal hygiene, including detailed instructions on foot and dental care. All infections (especially pyogenic ones) provoke the release of high levels of insulin antagonists such as catecholamines or glucagon and thus bring about a marked increase in insulin requirements. Patients who are oral agents may decompensate and temporarily require insulin. All patients receiving therapy that can cause hypoglycemia should wear a MedicAlert bracelet or necklace that clearly states that insulin or an oral sulfonylurea drug is being taken. Patients should be told about community agencies, such as American Diabetes Association chapters, that can serve as a continuing source of instruction.

Finally, vigorous efforts should be made to persuade new diabetics who smoke to give up the habit, since large vessel peripheral vascular disease and debilitating retinopathy are less common in nonsmoking diabetic patients.

Specific Therapy

With the publication of data from the DCCT and the UKPDS, there has been a shift in the guidelines regarding acceptable levels of control. The ADA recommends that for patients with either type 1 and type 2 diabetes, the goal is to achieve preprandial blood glucose values of 80 to 120 mg/dL and an average bedtime glucose of 100 to 140 mg/dL and HbA1c of <7% (nondiabetic range: 4%-6%). Obviously, these goals should be modified by taking into account the patient's ability to carry out the treatment regimen, the risk of severe hypoglycemia, and other patient factors that may reduce the benefit of such tight control.

Type 1 Diabetes

Patients with type 1 diabetes require replacement therapy with exogenous insulin. This should be instituted under conditions of an individualized diabetic diet with multiple feedings and normal daily activities so that an appropriate dosage regimen can be developed.

At the onset of type 1 diabetes, many patients recover some pancreatic β cell function and may temporarily need only low doses of exogenous insulin to supplement their own endogenous insulin secretion. This is known as the honeymoon period. Within 8 weeks to 2 years, however, most of these patients show either absent or negligible pancreatic β cell function. At this point, these patients should be switched to a more flexible insulin regimen with a combination of rapid-acting insulin analogs or regular insulin together with intermediate-acting or long-acting insulin. At a minimum, the patient should be on a three-injection regimen and frequently may need four or more injections. Twice-daily split-dose insulin mixtures cannot maintain near-normalization of blood glucose without hypoglycemia (particularly at night) and are not recommended. Self-monitoring of blood glucose levels is required for determining the optimal adjustment of insulin dosage and the modulation of food intake and exercise in type 1 diabetes.

Certain caveats should be kept in mind regarding insulin treatment. Considerable variations in absorption and bioavailability exist, even when the same dose is injected in the same region on different days in the same individual. Such variation often can be minimized by injecting smaller quantities of insulin at each injection and consequently using multiple injections. Furthermore, a given insulin dosage may demonstrate considerable variability in pharmacokinetics in different individuals, either because of insulin antibodies that bind insulin with different avidity or for other as yet unknown reasons. A properly educated patient should be taught to adjust insulin dosage by observing the pattern of recorded self-monitored blood glucose levels and correlating it with the approximate duration of action and the time to peak effect after injection of the various insulin preparations (Table 17–14).

A combination of rapid-acting insulin analogs and long-acting insulins (insulin glargine or insulin detemir) allows for more physiologic insulin replacement. In clinical studies, combinations of rapid-acting insulin analogs (insulin lispro or insulin aspart) with meals together with intermediate-acting (NPH) or longer acting insulin (insulin glargine) for basal coverage have now been shown to have improved HbA1c values with less hypoglycemia when compared with a regimen of regular insulin with meals and NPH at night. Table 17–16 illustrates some regimens that might be appropriate for a 70-kg person with type 1 diabetes eating meals of standard carbohydrate intake and moderate to low fat content.

Multiple injections of NPH insulin can be mixed in the same syringe as the insulin lispro, insulin aspart, and insulin glulisine. Insulin glargine is usually given once in the evening to provide 24-hour coverage. This insulin should not be mixed with any of the other insulins and must be given as a separate injection. There are occasional patients in whom insulin glargine does not seem to last for 24 hours, and in such cases it needs to be given twice a day.

Continuous subcutaneous insulin infusion by portable battery-operated open loop devices currently provides the most flexible approach, allowing the setting of different basal rates throughout the 24 hours and permitting patients to delay or skip meals and vary meal size and composition (see Methods of Insulin Administration, earlier). The dosage is usually based on providing 50% of the estimated insulin dose as basal and the remainder as intermittent boluses prior to meals. For example, a 70-kg man requiring 35 U of insulin per day may require a basal rate of 0.7 U/h throughout the 24 hours with the exception of 3 am to 8 am, when 0.8 U/h might be appropriate (for the dawn phenomenon). The meal bolus would depend on the carbohydrate content of the meal and the premeal blood glucose value. One unit per 15 g of carbohydrate plus 1 U for 50 mg/dL of blood glucose above a target value (eg, 120 mg/dL) is a common starting point. Further adjustments to basal and bolus dosages would depend on the results of blood glucose monitoring. Most patients use the rapid-acting insulin analogs in the pumps.

One of the more difficult therapeutic problems in managing patients with type 1 diabetes is determining the proper adjustment of insulin dose when the early morning blood glucose level is high before breakfast (Table 17–17). Prebreakfast hyperglycemia is sometimes due to the Somogyi effect, in which nocturnal hypoglycemia evokes a surge of counterregulatory hormones to produce high blood glucose levels by 7 am. However, a more common cause of prebreakfast hyperglycemia is the waning of the evening or bedtime insulin and/or the dawn phenomenon. That is, reduced tissue sensitivity to insulin between 5 am and 8 am (dawn), due to spikes of growth hormone released hours before, at onset of sleep. Table 17–17 shows that diagnosis of the cause of prebreakfast hyperglycemia can be facilitated by self-monitoring of blood glucose at 3 am in addition to the usual bedtime and 7 am measurements. This is required for only a few nights, and when a particular pattern emerges from monitoring blood glucose levels overnight, appropriate therapeutic measures can be taken. Prebreakfast hyperglycemia due to the Somogyi effect can be treated by reducing the dose of either intermediate- or long-acting insulin analog at bedtime. For hyperglycemia due to waning of overnight basal insulin and/or dawn phenomenon, an increase in the evening dose of the basal insulin or shifting it from dinnertime to bedtime (or both) can be effective. A bedtime dose either of insulin glargine or insulin detemir provides more sustained overnight insulin levels than human NPH and may be effective in managing refractory prebreakfast hyperglycemia. If this fails, insulin pump therapy may be required with a higher basal insulin infusion rate (eg, from 0.8 U/h to 0.9 U/h from 6 am until breakfast).

Type 2 Diabetes

Therapeutic recommendations are based on the relative contributions of β cell insufficiency and insulin insensitivity in individual patients. The possibility that the individual patient has a specific etiological cause for their diabetes should always be considered, especially when the patient does not have a family history of type 2 diabetes or does not have any evidence of central obesity or insulin resistance. Such patients should be evaluated for other types of diabetes such as LADA or MODY. Patients with LADA should be prescribed insulin when their disease is diagnosed and treated like patients with type 1 diabetes. It is also important to note that many type 2 patients have a progressive loss of β cell function and will require additional therapeutic interventions with time.

• Weight reduction—One of the primary modes of therapy in the obese patient with type 2 diabetes is weight reduction. Normalization of glycemia can be achieved by reducing adipose stores, with consequent restoration of tissue sensitivity to insulin. A combination of caloric restriction, increased exercise, modification of behavior, and consistent reinforcement of good eating habits is required if a weight reduction program is to be successful. Knowledge of the symptoms of diabetes and an understanding of the risks and complications of diabetes often increase the patient's motivation for weight reduction. Even so, significant weight loss is seldom achieved and is even more difficult to maintain in the morbidly obese patient. Weight control is variable in moderately obese patients, depending on the enthusiasm of the therapist and the motivation of the patient.
• For selected patients, medical or surgical options for weight loss should be considered (also see Chapter 20). Sibutramine is a combined serotonin-norepinephrine reuptake inhibitor that is associated with modest suppression of appetite. It results in about 3 to 5 kg weight loss over a 6 to 12 month period. Its side effects include dry mouth, anorexia, insomnia, dizziness, and hypertension. It is also not recommended in patients with cardiac disease or stroke. In clinical trials in obese type 2 patients, the patients who lost weight with sibutramine also had reduction in HbA1c levels
• Orlistat is a reversible inhibitor of gastric and pancreatic lipases and prevents the hydrolysis of dietary triglycerides. These triglycerides are then excreted in the feces. In a 1-year study in obese patients with type 2 diabetes, those taking orlistat had lost more weight, had lower HbA1c values, and had improved lipid profiles. The main adverse reactions were gastrointestinal, with oily spotting, oily stool, flatus, and fecal urgency and frequency. Malabsorption of fat-soluble vitamins also occurs. Patients should take a multivitamin tablet containing fat-soluble vitamins at least 2 hours before or 2 hours after the administration of orlistat.
• Bariatric surgery (Roux-en-Y or gastric banding) typically result in substantial weight loss, and long-term followup has shown that many patients with diabetes who undergo these procedures maintain the weight reduction.
• Nonobese patients with type 2 diabetes frequently have increased visceral adiposity—the so-called metabolically obese normal-weight patient. There is less emphasis on weight loss but exercise remains an important aspect of treatment.
• Antihyperglycemic agents—The current recommendation is to start metformin therapy at diagnosis and not wait to see if the patient can achieve target glycemic control with weight management and exercise. Metformin is advantageous because, apart from lowering glucose without the risk of hypoglycemia, it also lowers triglycerides and promotes some modest weight loss. The drug, however, cannot be used in patients with renal failure, and some patients have gastrointestinal side effects at even the lowest doses. Under these circumstances the choice of the initial agent depends on a number of factors, including comorbid conditions, adverse reactions to the medications, ability of the patient to monitor for hypoglycemia, drug cost, and patient and physician preferences. Sulfonylureas have been available for many years, and their use in combination with metformin is well established. They do, however, have the propensity to cause hypoglycemia and weight gain. Thiazolidinediones improve peripheral insulin resistance and lower glucose without causing hypoglycemia. They also have been reported to improve nonalcoholic fatty liver disease. In addition, they have beneficial effects on the lipid profile and some other cardiovascular risk factors. They decrease microalbuminuria, and reduce neointimal tissue hyperplasia after coronary artery stent placement. Thiazolidinediones, however, can cause fluid retention and are contraindicated in patients with heart failure. They also very commonly increase weight, which patients find distressing, affecting adherence. The drugs are associated with increased fracture risk in women and this adverse effect significantly limits their use. Pioglitazone is the preferred choice because of concerns regarding rosiglitazone and the risk of ischemic heart disease. Both drugs are contraindicated in patients with active liver disease and in patients with liver enzyme levels ≥2.5 times the upper limit of normal. The α-glucosidase inhibitors have modest glucose-lowering effects and have gastrointestinal side effects. They have a lower risk of hypoglycemia than the sulfonylureas and promote weight loss. The GLP-1 receptor agonists (exenatide and liraglutide) have a lower risk of hypoglycemia than the sulfonylureas, and they promote weight loss. However, they need to be given by injection, cause nausea, may cause pancreatitis, and are contraindicated in patients with gastroparesis. The DPP-4 inhibitors (sitagliptin and saxagliptin) also have a low risk of hypoglycemia, and they do not cause nausea or vomiting. They can also be used in patients with kidney impairment. There are, however, reports of serious allergic reactions, including anaphylaxis, angioedema, and Stevens-Johnson syndrome. There is concern that sitagliptin, like the GLP-1 receptor agonists, may cause pancreatitis.
• When patients are not well controlled on their initial therapy (usually metformin) then a second agent should be added. In those patients where the problem is hyperglycemia after a carbohydrate-rich meal (such as dinner), a short-acting secretagogue before meals may suffice to put the glucose levels into the target range. Patients with severe insulin resistance or nonalcoholic fatty liver disease or microalbuminuria may be candidates for pioglitazone. Subjects who are very concerned about weight gain may benefit from a trial of a GLP-1-receptor agonist or DPP-4 inhibitor. If two agents are inadequate, then a third agent is added, although data on efficacy with such combined therapy are limited. When the combination of oral agents (and injectable GLP-1 receptor agonists) fail to achieve target glycemic control in patients with type 2 diabetes or if there are contraindications to their use, then insulin treatment should be instituted. Various insulin regimens may be effective. One proposed regimen is to continue the oral combination therapy and then simply add a bedtime dose of NPH or long-acting insulin analog (insulin glargine or insulin detemir) to reduce excessive nocturnal hepatic glucose output and improve fasting glucose levels. If the patient does not achieve target glucose levels during the day, then daytime insulin treatment can be initiated. A convenient insulin regimen under these circumstances is a split dose of 70/30 NPH/regular mixture (or Humalog Mix 75/25 or NovoLogMix 70/30) before breakfast and before dinner. If this regimen fails to achieve satisfactory glycemic goals or is associated with unacceptable frequency of hypoglycemic episodes, then a more intensive regimen of multiple insulin injections can be instituted as in patients with type 1 diabetes. Metformin principally reduces hepatic glucose output, and it is reasonable to continue with this drug when insulin therapy is instituted. The thiazolidinediones, which improve peripheral insulin sensitivity, can be used together with insulin, but this combination is associated with more weight gain and peripheral edema. The sulfonylureas also have been shown to be of continued benefit. There is limited information on the benefits of continuing the GLP1-receptor agonists or the DPP-4 inhibitors once insulin therapy is initiated. Weight-reducing interventions should continue even after initiation of insulin therapy and may allow for simplification of the therapeutic regimen in the future.

Immunopathology of Insulin Therapy

At least five molecular classes of insulin antibodies are produced during the course of insulin therapy: IgA, IgD, IgE, IgG, and IgM. Human insulin is much less antigenic than the older formulations of animal (especially beef) insulins, but because of its hexameric presentation at therapeutic injection doses, it is also treated as a foreign substance by the immune system and results in detectable—albeit low—titers of insulin antibodies in most patients.

Insulin Allergy

Insulin allergy, a hypersensitivity reaction of the immediate type, is a rare condition in which local or systemic urticaria occurs immediately after insulin injection. This reaction is due to histamine release from tissue mast cells sensitized by adherence of IgE antibodies to their surface. In severe cases, anaphylaxis can occur. The appearance of a subcutaneous nodule at the site of insulin injection, occurring several hours after the injection and lasting for up to 24 hours, has been attributed to an IgG-mediated complement-binding Arthus reaction. Because sensitivity was often due to noninsulin protein contaminants, the highly purified insulins have markedly reduced the incidence of insulin allergy, especially of the local variety. Antihistamines, corticosteroids, and even desensitization may be required, especially for systemic hypersensitivity in an insulin-dependent patient. A protocol for allergy testing and insulin desensitization is available from the Lilly Company. A trial of insulin analogs should also be considered. There is a case report of successful use of insulin lispro in the face of generalized allergy to human insulin.

Immune Insulin Resistance

All patients who receive insulin (including insulin analogs) develop a low titer of circulating IgG antibodies, and this neutralizes the rapid action of insulin to a small extent. With the animal insulins, a high titer of circulating antibodies is sometimes developed, resulting in extremely high insulin requirements, often to more than 200 U/d. This is now very rarely seen with the switch to the highly purified human insulins and has not been reported with use of the analogs.

Lipodystrophy at Injection Sites

Rarely, a disfiguring atrophy of subcutaneous fatty tissue occurs at the site of insulin injection. Although the cause of this complication is obscure, it seems to represent a form of immune reaction, particularly because it occurs predominantly in females and is associated with lymphocyte infiltration in the lipoatrophic area. This complication has become even less common because of the development of highly purified insulin preparations of neutral pH. Injection of highly purified preparations of insulin directly into the atrophic area often results in restoration of normal contours.

Lipohypertrophy, on the other hand, is not a consequence of immune responses; rather, it seems to be due to the pharmacologic effects of depositing insulin in the same location repeatedly. It is prevented by rotation of injection sites. There is a case report of a male patient who had intractable lipohypertrophy (fatty infiltration of injection site) with human insulin but no longer had the problem when he switched to insulin lispro.

Acute Complications of Diabetes Mellitus

Hypoglycemia

Hypoglycemic reactions (see later and Chapter 18) are the most common complications that occur in patients with diabetes who are treated with insulin. Hypoglycemia may result from delay in taking a meal, from unusual physical exertion without supplemental calories, or from an increase in insulin dose. In addition, it can occur in any patient taking oral agents that stimulate pancreatic β cells (eg, sulfonylureas, meglitinide, d-phenylalanine analogs), particularly if the patient is elderly, has renal or liver disease, or is taking certain other medications that alter metabolism of the sulfonylureas (eg, phenylbutazone, sulfonamides, or warfarin). It occurs more frequently with the use of long-acting sulfonylureas.

The signs and symptoms of hypoglycemia may be divided into those resulting from stimulation of the autonomic nervous system and those arising from neuroglycopenia (insufficient glucose for normal central nervous system function). When the blood glucose falls to around 54 mg/dL, the patient starts to experience both sympathetic (tachycardia, palpitations, sweating, tremulousness) and parasympathetic (nausea, hunger) nervous system symptoms. If these autonomic symptoms are ignored and the glucose levels fall further (to around 50 mg/dL), then neuroglycopenic symptoms appear, including irritability, confusion, blurred vision, tiredness, headache, and difficulty speaking. A further decline in glucose (below 30 mg/dL) can then lead to loss of consciousness or even a seizure.

With repeated episodes of hypoglycemia, there is adaptation and autonomic symptoms do not occur until the blood glucose levels are much lower and so the first symptoms are often due to neuroglycopenia. This condition, which is referred to as hypoglycemic unawareness, results from failure of the sympathetic nervous system to respond to hypoglycemia (Table 17–18). This adaptation of the central nervous system to recurrent hypoglycemic episodes is due to upregulation of the GLUT 1 transporters at the blood-brain barrier and increased glucose transport into the brain despite subnormal levels of plasma glucose. It has been shown that hypoglycemic unawareness can be reversed by keeping glucose levels high for a period of several weeks. Except for sweating, most of the sympathetic symptoms of hypoglycemia are blunted in patients receiving β-blocking agents for angina pectoris or hypertension. Though not absolutely contraindicated, these drugs must be used with caution in insulin-requiring diabetics, and β1-selective blocking agents are preferred.

Hypoglycemia in insulin-treated patients with diabetes occurs as a consequence of three factors: behavioral issues, impaired counterregulatory systems, and complications of diabetes.

Behavioral issues include injecting too much insulin for the amount of carbohydrates ingested. Drinking alcohol in excess, especially on an empty stomach, can also cause hypoglycemia. In patients with type 1 diabetes, hypoglycemia can occur during or even several hours after exercise, and so glucose levels need to be monitored and food and insulin adjusted. Some patients do not like their glucose levels to be high, and they treat every high glucose level aggressively. These individuals who stack their insulin, that is, give another dose of insulin before the first injection has had its full action, can develop hypoglycemia.

Counterregulatory issues resulting in hypoglycemia include impaired glucagon response and impaired sympatho-adrenal responses (Table 17–19). Patients with diabetes of greater than 5 years duration lose their glucagon response to hypoglycemia. As a result, they are at a significant disadvantage in protecting themselves against falling glucose levels. Once the glucagon response is lost, their sympatho-adrenal responses take on added importance. Unfortunately, aging, autonomic neuropathy, or hypoglycemic unawareness due to repeated low glucose levels further blunts the sympatho-adrenal responses. Occasionally, Addison disease develops in persons with type 1 diabetes mellitus; when this happens, insulin requirements fall significantly, and unless insulin dose is reduced, recurrent hypoglycemia will develop.

Complications of diabetes that increase the risk for hypoglycemia include autonomic neuropathy, gastroparesis, and renal failure. The sympathetic nervous system is an important system alerting the individual that the glucose level is falling by causing symptoms of tachycardia, palpitations, sweating, and tremulousness. Failure of the sympatho-adrenal responses increases the risk of hypoglycemia. In patients with gastroparesis, insulin given before a meal promotes maximal glucose uptake into cells before the food is absorbed, causing the glucose levels to fall. Finally, in renal failure, hypoglycemia can occur presumably because of decreased insulin clearance as well as loss of the renal contribution to gluconeogenesis in the postabsorptive state.

Treatment

To treat insulin-induced hypoglycemia, the diabetic patient should carry glucose tablets or juice at all times. For most episodes, ingestion of 15 g of carbohydrate is sufficient to reverse the hypoglycemia. The patient should be instructed to check the blood glucose in 15 minutes and treat again if the glucose level is still low.

A parenteral glucagon emergency kit (1 mg) should be provided to every patient with diabetes who is on insulin therapy. Family or friends should be instructed how to inject it subcutaneously or intramuscularly into the buttock, arm or thigh in the event that the patient is unconscious or refuses food. The drug can occasionally cause vomiting and the unconscious patient should be turned on his or her side to protect the airway. Glucagon mobilizes glycogen from the liver raising the blood glucose by about 36 mg/dL in about 15 minutes. After the patient recovers consciousness additional oral carbohydrate should be given. Glucagon is contraindicated in sulfonylurea–induced hypoglycemia where it paradoxically causes insulin release. People with diabetes on hypoglycemic drug therapy should also wear a MedicAlert bracelet or necklace or carry a card in his or her wallet.

Medical personnel treating severe hypoglycemia can give 50 mL of 50% glucose solution by rapid intravenous infusion. If intravenous therapy is not available, 1 mg of glucagon can be injected intramuscularly. If the patient is stuporous and glucagon is not available, small amounts of honey or maple syrup or glucose gel (15 g) can be inserted within the buccal pouch, although, in general, oral feeding is contraindicated in unconscious patients. Rectal administration of maple syrup or honey (30 mL per 500 mL of warm water) has been effective.

Most patients who arrive at emergency departments in hypoglycemic coma appear to recover fully; however, profound hypoglycemia or delays in therapy can result in permanent neurologic deficit or even death. Furthermore, repeated episodes of hypoglycemia may have a cumulative adverse effect on intellectual functioning. The physician should carefully review with the patient the events leading up to the hypoglycemic episode. Associated use of other medications, as well as alcohol or narcotics, should be noted. Careful attention should be paid to diet, exercise pattern, insulin or sulfonylurea dosage, and general compliance with the prescribed diabetes treatment regimen. Any factors thought to have contributed to the development of the episode should be identified and recommendations made in order to prevent recurrences of this potentially disastrous complication of diabetes therapy.

If the patient is hypoglycemic from use of a long-acting oral hypoglycemic agent (eg, chlorpropamide or glyburide) or from high doses of a long-acting insulin, admission to hospital for treatment with continuous intravenous glucose and careful monitoring of blood glucose is indicated.

Coma

Coma is a medical emergency calling for immediate evaluation to determine its cause so that proper therapy can be started. Patients with diabetes may be comatose because of hypoglycemia, diabetic ketoacidosis, hyperglycemic hyperosmolar coma, or lactic acidosis. When evaluating a comatose diabetic patient, these must be considered in addition to the myriad causes included in the differential diagnosis of coma (eg, cerebrovascular accidents, head trauma, intoxication with alcohol, or other drugs).

After emergency measures have been instituted (airway protection; laboratory tests; intravenous dextrose unless fingerstix blood glucose shows hyperglycemia), a careful history (from family, friends, or paramedics), physical examination, and laboratory evaluation are required to resolve the differential diagnosis. Patients in deep coma from a hyperosmolar nonketotic state or from hypoglycemia are generally flaccid and have quiet breathing—in contrast to patients with acidosis, whose respirations are rapid and deep if the pH of arterial blood has dropped to 7.1 or below. When hypoglycemia is a cause of the coma, the state of hydration is usually normal. Although the clinical laboratory remains the final arbiter in confirming the diagnosis, a rapid estimate of blood glucose and ketones can be obtained by the use of bedside glucose and ketone meters (see Laboratory Findings in diabetes mellitus, above). Table 17–20 is a summary of some laboratory abnormalities found in diabetic patients with coma attributable to diabetes or its treatment.

Diabetic Ketoacidosis

This acute complication of diabetes mellitus may be the first manifestation of previously undiagnosed type 1 diabetes or may result from increased insulin requirements in type 1 diabetes patients during the course of infection, trauma, myocardial infarction, or surgery. The National Data Group reports an annual incidence of five to eight episodes of diabetic ketoacidosis per 1000 diabetic patients. In all cases, precipitating factors such as infection should be searched for and treated appropriately. Poor compliance, either for psychological reasons or because of inadequate patient education, is probably the most common cause of diabetic ketoacidosis, particularly when episodes are recurrent. In adolescents with type 1 diabetes, recurrent episodes of severe ketoacidosis often indicate the need for counseling to alter this behavior.

Diabetic ketoacidosis has been found to be one of the more common serious complications of insulin pump therapy, occurring in approximately 1 per 80 patient-months of treatment.

Patients with type 2 diabetes may also develop ketoacidosis under severe stress such as sepsis, trauma, or major surgery.

Pathogenesis

Acute insulin deficiency results in rapid mobilization of energy from stores in muscle and fat depots, leading to an increased flux of amino acids to the liver for conversion to glucose and of fatty acids for conversion to ketones (acetoacetate, β-hydroxybutyrate, and acetone). In addition to this increased availability of precursor, there is a direct effect of the low insulin-glucagon ratio on the liver that promotes increased production of ketones as well as of glucose. In response to both the acute insulin deficiency and the metabolic stress of ketosis, the levels of insulin-antagonistic hormones (corticosteroids, catecholamines, glucagon, and GH) are consistently elevated. Furthermore, in the absence of insulin, peripheral utilization of glucose and ketones is reduced. The combination of increased production and decreased utilization leads to an accumulation of these substances in blood, with plasma glucose levels reaching 500 mg/dL (27.8 mmol/L) or more and plasma ketones reaching levels of 8 to 15 mmol/L or more. β-Hydroxybutyrate is the predominant ketone and its ratio to acetoacetate increases from 1:1 to as much as 5:1.

The hyperglycemia causes osmotic diuresis leading to depletion of intravascular volume. As this progresses, impaired renal blood flow reduces the kidney's ability to excrete glucose, and hyperosmolality worsens. Severe hyperosmolality (>330 mOsm/kg) correlates closely with central nervous system depression and coma.

In a similar manner, impaired renal excretion of hydrogen ions aggravates the metabolic acidosis that occurs as a result of the accumulation of the ketoacids, β-hydroxybutyrate, and acetoacetate. The accumulation of ketones may cause vomiting, which exacerbates the intravascular volume depletion. In addition, prolonged acidosis can compromise cardiac output and reduce vascular tone. The result may be severe cardiovascular collapse with generation of lactic acid, which then adds to the already existent metabolic acidosis.

Clinical Features

Symptoms and Signs

The appearance of diabetic ketoacidosis is usually preceded by a day or more of polyuria and polydipsia associated with marked fatigue, nausea, and vomiting. Eventually, mental stupor ensues and can progress to frank coma. On physical examination, evidence of dehydration in a stuporous patient with rapid and deep respirations and the fruity breath odor of acetone strongly suggest the diagnosis. Postural hypotension with tachycardia indicates profound dehydration and salt depletion. Abdominal pain and even tenderness may be present in the absence of abdominal disease, and mild hypothermia is usually present.

Laboratory Findings

Typically, the patient with moderately severe diabetic ketoacidosis has a plasma glucose of 350 to 900 mg/dL (19.4-50 mmol/L), serum ketones are positive at a dilution of 1:8 or greater, hyperkalemia of 5 to 8 mEq/L, slight hyponatremia of approximately 130 mEq/L, hyperphosphatemia of 6 to 7 mg/dL, and an elevated blood urea nitrogen and creatinine. Acidosis may be severe (pH ranging from 6.9 to 7.2 with a bicarbonate concentration ranging from 5 to 15 mEq/L); pCO2 is low (15-20 mm Hg) secondary to hyperventilation.

The fluid depletion is typically about 100 mL/kg. The hyperkalemia occurs despite total body potassium depletion, because of the shift of potassium from the intracellular to extracellular spaces in systemic acidosis. The average total body potassium deficit resulting from osmotic diuresis, acidosis, and gastrointestinal losses is about 3 to 5 mEq/kg body weight. Similarly despite the elevated serum phosphate, total body phosphate is generally depleted. Serum sodium is generally reduced, due to loss of sodium ions by polyuria and vomiting (7-10 mEq/kg), and because severe hyperglycemia shifts intracellular water into the interstitial compartment (for every 100 mg/dL of plasma glucose above normal, serum sodium decreases by 1.6 mEq/L). Serum osmolality can be directly measured by standard tests of freezing point depression or can be estimated by calculating the molarity of sodium, chloride, and glucose in the serum. A convenient formula for estimating effective serum osmolality is:

The effective serum osmolality in humans is generally between 280 and 300 mOsm/kg. These calculated estimates are usually 10 to 20 mOsm/kg lower than values recorded by standard cryoscopic techniques. Central nervous depression or coma occurs when the effective serum osmolality exceeds 320 to 330 mOsm/L

Blood urea nitrogen and serum creatinine are invariably elevated because of dehydration. Urea exerts an effect on freezing point depression as measured in the laboratory, but it is freely permeable across cell membranes and therefore not included in calculations of effective serum osmolality. Serum creatinine may also be falsely elevated due to interference from acetoacetate with some automated creatinine assays. However, most laboratories can correct for these interfering chromogens by using a more specific method, if asked to do so.

The nitroprusside reagents (Acetest and Ketostix) used for the bedside assessment of ketoacidemia and ketoaciduria measure only acetoacetate and its by-product, acetone. The sensitivity of these reagents for acetone, however, is quite poor, requiring over 10 mmol/L, which is seldom reached in the plasma of ketoacidotic subjects—although this detectable concentration is readily achieved in urine. Thus, in the plasma of ketotic patients, only acetoacetate is measured by these reagents. The more prevalent β-hydroxybutyrate has no ketone group and is therefore not detected by the conventional nitroprusside tests. This takes on special importance in the presence of circulatory collapse during diabetic ketoacidosis, wherein an increase in lactic acid can shift the redox state to increase β-hydroxybutyrate at the expense of the readily detectable acetoacetate. Bedside diagnostic reagents would then be unreliable, suggesting no ketonemia in cases where β-hydroxybutyrate is a major factor in producing the acidosis. Under these circumstances, β-hydroxybutyrate can be directly measured at the bedside using the Precision Xtra meter (Abbott Diagnostics). Many clinical laboratories now offer β-hydroxybutyrate measurement.

In about 90% of cases, serum amylase is elevated. However, this often represents salivary as well as pancreatic amylase and correlates poorly with symptoms of pancreatitis, such as pain and vomiting. Therefore, in patients with diabetic ketoacidosis, an elevated serum amylase does not justify a diagnosis of acute pancreatitis; serum lipase may be useful if the diagnosis of pancreatitis is being seriously considered.

Treatment

Patients with mild DKA are alert and have pH between 7.25 and 7.30; those with moderate DKA have pH between 7.0 and 7.24 and are alert or slightly drowsy; and those with severe DKA are stuporous and have pH <7.0. Those with mild DKA can be treated in the emergency room, but those with moderate or severe DKA require admission to the intensive care unit or step-down unit.

The therapeutic goals are to restore plasma volume and tissue perfusion; reduce blood glucose and osmolality toward normal; correct acidosis; replenish electrolyte losses; and identify and treat precipitating factors. Gastric intubation is recommended in the comatose patient to prevent vomiting and aspiration that may occur as a result of gastric atony, a common complication of diabetic ketoacidosis. In patients with preexisting cardiac or renal failure or those in severe cardiovascular collapse, a central venous pressure catheter or a Swan-Ganz catheter should be inserted to evaluate the degree of hypovolemia and to monitor subsequent fluid administration.

Plasma glucose should be recorded hourly and electrolytes and pH at least every 2 to 3 hours during the initial treatment period. A bedside glucose meter should be used to titrate the insulin therapy.

• 1. Fluid replacement. In most adult patients, the fluid deficit is 4 to 5 L. Once the diagnosis of diabetic ketoacidosis is established in the emergency department, administration of at least 2 L of isotonic saline (0.9% saline solution) in an adult patient in the first 2 to 3 hours is necessary to help restore plasma volume and stabilize blood pressure while acutely reducing the hyperosmolar state. In addition, by improving renal plasma flow, fluid replacement also restores the renal capacity to excrete hydrogen ions, thereby ameliorating the acidosis as well. After the first 2 L of fluid have been given, the fluid should be changed to 0.45% saline solution given at a rate of 300 to 400 mL/h, because water loss exceeds sodium loss in uncontrolled diabetes with osmotic diuresis. Failure to give sufficient volume replacement (at least 3-4 L in 8 hours) to restore normal perfusion is one of the most serious therapeutic shortcomings affecting satisfactory recovery. In the same way, excessive fluid replacement (>5 L in 8 hours) may contribute to acute respiratory distress syndrome or cerebral edema. When blood glucose falls to approximately 250 mg/dL, the fluids should be changed to a 5% glucose solution to maintain plasma glucose in the range of 250 to 300 mg/dL. This prevents the development of hypoglycemia and also reduces the likelihood of cerebral edema, which may result from a too rapid decline of blood glucose.
• 2. Insulin. Immediately after the initiation of fluid replacement, a rapid bolus of 0.3 U of regular insulin per kilogram of body weight should be given intravenously to prime the tissue insulin receptors. This inhibits both gluconeogenesis and ketogenesis while promoting utilization of glucose and keto acids. Following the initial bolus, an insulin infusion is initiated at a rate of 0.1 U/kg/h. When a continuous infusion of insulin is used, 25 U of regular human insulin should be placed in 250 mL of isotonic saline and the first 50 mL of solution flushed through to saturate the tubing before connecting it to the intravenous line. The insulin infusion should be piggy-backed into the fluid line so that the rate of fluid replacement can be changed without altering the insulin delivery rate. If the plasma glucose level fails to fall at least 10% in the first hour, a repeat loading dose is recommended. Rarely, a patient with insulin resistance is encountered; this requires doubling the insulin dose every 2 to 4 hours if severe hyperglycemia does not improve after the first two doses of insulin and fluid replacement. The insulin dose should be adjusted with the goal of lowering the glucose concentration by about 50 to 70 mg/dL/h. If clinical circumstances prevent use of insulin infusion, then the insulin can be given intramuscularly. An initial 0.15 U/kg of regular insulin is given intravenously, and at the same time, the same size dose is given intramuscularly. Subsequently, regular insulin is given intramuscularly hourly at a dose of 0.1 U/kg until the blood glucose falls to around 250 mg/dL, when the insulin can be given subcutaneously. Insulin therapy, either as a continuous infusion or as injections given every 1 to 2 hours, should be continued until arterial pH has normalized.
• 3. Potassium. Total body potassium loss from polyuria and vomiting may be as high as 200 mEq. However, because of shifts of potassium from cells into the extracellular space as a consequence of acidosis, serum potassium is usually normal to slightly elevated prior to institution of treatment. As the acidosis is corrected, potassium flows back into the cells, and hypokalemia can develop if potassium replacement is not instituted. If the patient is not uremic and has an adequate urine output, potassium chloride in doses of 10 to 30 mEq/h should be infused during the second and third hours after beginning therapy. Replacement should be started sooner, if the initial serum potassium is inappropriately normal or low, and should be delayed, if serum potassium fails to respond to initial therapy and remains above 5 mEq/L, as in cases of renal insufficiency. Cooperative patients with only mild ketoacidosis may receive part or all of their potassium replacement orally.
• An electrocardiogram can be of help in monitoring the patient's potassium status: high peaked T waves are a sign of hyperkalemia, and flattened T waves with U waves are a sign of hypokalemia.
• Foods high in potassium content should be prescribed when the patient has recovered sufficiently to take food orally. Tomato juice has 14 mEq of potassium per 240 mL, and a medium-sized banana has about 10 mEq.
• 4. Sodium bicarbonate. The use of sodium bicarbonate in management of diabetic ketoacidosis has been questioned because clinical benefit was not demonstrated in one prospective randomized trial and because of the following potentially harmful consequences: (1) development of hypokalemia from rapid shift of potassium into cells if the acidosis is overcorrected, (2) tissue anoxia from reduced dissociation of oxygen from hemoglobin when acidosis is rapidly reversed (leftward shift of the oxygen dissociation curve), and (3) cerebral acidosis resulting from lowering of cerebrospinal fluid pH. It must be emphasized, however, that these considerations are less important when severe acidosis exists. It is therefore recommended that bicarbonate be administered to diabetic patients in ketoacidosis if the arterial blood pH is 7.0 or less with careful monitoring to prevent overcorrection.
• One to two ampules of sodium bicarbonate (one ampule contains 44 mEq/50 mL) should be added to 1 L of 0.45% saline. (Note: Addition of sodium bicarbonate to 0.9% saline would produce a markedly hypertonic solution that could aggravate the hyperosmolar state already present.) This should be administered rapidly (over the first hour). It can be repeated until the arterial pH reaches 7.1, but it should not be given if the pH is 7.1 or greater, because additional bicarbonate increases the risk of rebound metabolic alkalosis as ketones are metabolized. Alkalosis shifts potassium from serum into cells, which can precipitate a fatal cardiac arrhythmia. As noted earlier, serious consideration should be given to placement of a central venous catheter when administering fluids to severely ill patients with cardiovascular compromise.
• 5. Phosphate. Phosphate replacement is seldom required in treating diabetic ketoacidosis. However, if severe hypophosphatemia of less than 1 mg/dL (<0.35 mmol/L) develops during insulin therapy, a small amount of phosphate can be replaced per hour as the potassium salt. Correction of hypophosphatemia helps restore the buffering capacity of the plasma, thereby facilitating renal excretion of hydrogen. It also corrects the impaired oxygen dissociation from hemoglobin by regenerating 2,3-diphosphoglycerate. However, three randomized studies in which phosphate was replaced in only half of a group of patients with diabetic ketoacidosis did not show any apparent clinical benefit from phosphate administration. Moreover, attempts to use potassium phosphate as the sole means of replacing potassium have led to a number of reported cases of severe hypocalcemia with tetany. To minimize the risk of inducing tetany from too rapid replacement of phosphate, the average deficit of 40 to 50 mmol of phosphate should be replaced intravenously at a rate no greater than 3 to 4 mmol/h in a 60- to 70-kg person. A stock solution (Abbott) provides a mixture of 1.12 g KH2PO4 and 1.18 g K2HPO4 in a 5-mL single-dose vial (this equals 22 mmol of potassium and 15 mmol of phosphate). One-half of this vial (2.5 mL) should be added to 1 L of either 0.45% saline or 5% dextrose in water. Two liters of this solution, infused at a rate of 400 mL/h, corrects the phosphate deficit at the optimal rate of 3 mmol/h and provides 4.4 mEq of potassium per hour. Additional potassium should be administered as potassium chloride to provide a total of 10 to 30 mEq of potassium per hour, as noted earlier. If the serum phosphate remains below 2.5 mg/dL after this infusion, a repeat 5-hour infusion can be given.
• 6. Hyperchloremic acidosis during therapy. Because of the considerable loss of keto acids in the urine during the initial phase of therapy, substrate for subsequent regeneration of bicarbonate is lost, and correction of the total bicarbonate deficit is hampered. A portion of the bicarbonate deficit is replaced with chloride ions infused in large amounts as saline to correct the dehydration. In most patients, as the ketoacidosis clears during insulin replacement, a hyperchloremic, low-bicarbonate pattern emerges with a normal anion gap. This is a relatively benign condition that reverses itself over the subsequent 12 to 24 hours once intravenous saline is no longer being administered.

Transition to Subcutaneous Insulin Regimen

Once the diabetic ketoacidosis is controlled and the patient is awake and able to eat, subcutaneous insulin therapy can be initiated. Initially, the patient may still have significant tissue insulin resistance and may require a total daily insulin dose of 0.6 to 0.7 U/kg. Half of the total daily dose can be given as a long-acting basal insulin and the other half as short-acting insulin premeals. The patient should get injection of the basal insulin and a dose of the rapid-acting insulin analog with the first meal and the insulin infusion discontinued an hour later. The overlap of the subcutaneous insulin action and insulin infusion is necessary to prevent relapse of diabetic ketoacidiosis. The increased tissue insulin resistance is only present for a few days at most and as the patient improves the doses of both basal and bolus insulins should be reduced to avoid hypoglycemia. In fact, a patient with new diagnosis of type 1 diabetes who still has significant β-cell function may not require any basal insulin and only very low doses of rapid-acting insulin analogs before meals.

Complications and Prognosis

Low-dose insulin infusion and fluid and electrolyte replacement combined with careful monitoring of patients' clinical and laboratory responses to therapy have dramatically reduced the mortality rates of diabetic ketoacidosis to less than 5%. However, this complication remains a significant risk in the aged who have mortality rates over 20% and in patients in profound coma in whom treatment has been delayed. Acute myocardial infarction and infarction of the bowel following prolonged hypotension worsen the outlook. Prior kidney dysfunction worsens prognosis because the kidney plays a key role in compensating for pH and electrolyte abnormalities. Symptomatic cerebral edema occurs primarily in the pediatric population. Risk factors for development include severe baseline acidosis, rapid correction of hyperglycemia, and excess volume administration in the first 4 hours. Onset of headache or deterioration in mental status during treatment should lead to consideration of this complication. Intravenous mannitol at a dosage of 1 to 2 g/kg given over 15 minutes is the mainstay of therapy. Excess crystalloid infusion can precipitate pulmonary edema. Acute respiratory distress syndrome is a rare complication of treatment for DKA.

Disposition

After recovery and stabilization, patients should receive intensive detailed instructions about how to avoid this potentially disastrous complication of diabetes mellitus. They should be taught to recognize the early symptoms and signs of ketoacidosis.

Urine ketones or capillary blood β-hydroxybutyrate should be measured in patients with signs of infection or in those using an insulin pump when capillary blood glucose is unexpectedly and persistently high. When heavy ketonuria and glycosuria persist on several successive examinations, supplemental regular insulin should be administered, and liquid foods such as lightly salted tomato juice and broth should be ingested to replenish fluids and electrolytes. Patients should be instructed to contact the physician if ketonuria persists and, especially, if vomiting develops, or if appropriate adjustment of the infusion rate on an insulin pump does not correct the hyperglycemia and ketonuria. Table 17–21 summarizes the guidelines for patients regarding ketone testing and what to do with the results. In adolescents, recurrent episodes of severe diabetic ketoacidosis often indicate poor compliance with the insulin regimen, and these patients should receive intensive family counseling.

Hyperglycemic, Hyperosmolar State

This form of hyperglycemic coma is characterized by severe hyperglycemia, hyperosmolality, and dehydration in the absence of significant ketosis. It occurs in patients with mild or occult diabetes and patients are typically middle-aged or elderly. Lethargy and confusion develop as serum osmolality exceeds 300 mOsm/kg, and coma can occur if osmolality exceeds 330 mOsm/kg. Underlying renal insufficiency or congestive heart failure is common, and the presence of either worsens the prognosis. A precipitating event such as pneumonia, cerebrovascular accident, myocardial infarction, burns, or recent operation can often be identified. Certain drugs, such as phenytoin, diazoxide, glucocorticoids, and thiazide diuretics, have been implicated in its development, as have procedures associated with glucose loading such as peritoneal dialysis.

Pathogenesis

A partial or relative insulin deficiency may initiate the syndrome by reducing glucose utilization by muscle, fat, and liver, while promoting hyperglucagonemia and increasing hepatic glucose output. The result is hyperglycemia that leads to glycosuria and osmotic diuresis with obligatory water loss. The presence of even small amounts of insulin is believed to prevent the development of ketosis by inhibiting lipolysis in the adipose stores. Therefore, even though a low insulin-glucagon ratio promotes ketogenesis in the liver, the limited availability of precursor free fatty acids from the periphery restricts the rate at which ketones are formed. If a patient is unable to maintain adequate fluid intake because of an associated acute or chronic illness or has suffered excessive fluid loss (eg, from burns or therapy with diuretics), marked dehydration results. As plasma volume contracts, renal insufficiency develops; this, then, limits renal glucose excretion and contributes markedly to the rise in serum glucose and osmolality. As serum osmolality exceeds 320 to 330 mOsm/kg, water is drawn out of cerebral neurons, resulting in mental obtundation and coma.

Clinical Features

Symptoms and Signs

The onset of the hyperglycemic, hyperosmolar, nonketotic state may be insidious, preceded for days or weeks by symptoms of weakness, polyuria, and polydipsia. A history of reduced fluid intake is common, whether due to inappropriate absence of thirst, gastrointestinal upset, or, in the case of elderly or bedridden patients, lack of access to water. A history of ingestion of large quantities of glucose-containing fluids, such as soft drinks or orange juice, can occasionally be obtained; these patients are usually less hyperosmolar than those in whom fluid intake was restricted. The absence of toxic features of ketoacidosis may retard recognition of the syndrome and thus delay institution of therapy until dehydration is profound. Because of this delay in diagnosis, the hyperglycemia, hyperosmolality, and dehydration in hyperglycemic, hyperosmolar, nonketotic coma is often more severe than in diabetic ketoacidosis.

Physical examination reveals the presence of profound dehydration (orthostatic fall in blood pressure and rise in pulse, supine tachycardia, or even frank shock, dry mucous membranes, decreased skin turgor). The patient may be lethargic, confused, or comatose. Kussmaul respirations are absent unless the precipitating event for the hyperosmolar state has also led to the development of metabolic acidosis (eg, sepsis or myocardial infarction with shock).

Laboratory Findings

Severe hyperglycemia is present, with blood glucose values ranging from 800 to as high as 2400 mg/dL (44.4-133.2 mmol/L). In mild cases, where dehydration is less severe, dilutional hyponatremia as well as urinary sodium losses may reduce serum sodium to about 120 to 125 mEq/L—this protects, to some extent, against extreme hyperosmolality. Once dehydration progresses further, however, serum sodium can exceed 140 mEq/L, producing serum osmolalities of 330 to 440 mOsm/kg (normal, 280 to 295 mOsm/kg; see the section Diabetic Ketoacidosis for a convenient method for estimating serum osmolality). Ketosis is usually absent or mild; however, a small degree of ketonuria may be present if the patient has not been eating because of illness. Acidosis is not a part of the hyperglycemic, hyperosmolar state, but it may be present (often lactic acidosis) because of other acute underlying conditions (eg, sepsis, acute renal failure, myocardial infarction). Prerenal azotemia is the rule with blood urea nitrogen frequently over 100 mg/dL.

The physician must initiate a careful search for the event that precipitated the episode of hyperglycemic, hyperosmolar state if it is not obvious after the initial history and physical examination. Chest x-rays and cultures of blood, urine, and other body fluids should be obtained to look for occult sources of sepsis. Cardiac enzymes and serial electrocardiograms can be ordered to look for evidence of silent myocardial infarction.

Treatment

There are some differences in fluid, insulin, and electrolyte replacement in this disorder, as compared with diabetic ketoacidosis. However, in common with the treatment of ketoacidotic patients, careful monitoring of the patient's clinical and laboratory response to therapy is essential.

• 1. Fluid replacement. The fluid deficit may be as much as 100 to 200 mL/kg or about 9 L. If circulatory collapse is present, fluid therapy should be initiated with isotonic saline. In all other cases, initial replacement with hypotonic (usually 0.45%) saline is preferable, because these patients are hyperosmolar with considerable loss of body water and excess solute in the vascular compartment. As much as 4 to 6 L of fluid may be required in the first 8 to 10 hours. Careful monitoring of fluid quantity and type, urine output, blood pressure, and pulse is essential. Placement of a central venous pressure catheter should be strongly considered to guide replacement of fluid, especially if the patient is elderly or has underlying renal or cardiac disease. Because insulin therapy decreases plasma glucose and therefore serum osmolality, a change to isotonic saline may be necessary at some time during treatment. Once blood glucose reaches 250 mg/dL, 5% dextrose in 0.45% or 0.9% saline solution should be substituted for the sugar-free fluids. When consciousness returns, oral fluids should be encouraged.
• 2. Electrolyte replacement. Hyperkalemia is less marked, and much less potassium is lost in the urine during the osmotic diuresis of hyperglycemic, hyperosmolar, nonketotic coma than in diabetic ketoacidosis. There is, therefore, less severe total potassium depletion, and less potassium replacement is needed to restore potassium stores to normal. However, because the initial serum potassium usually is not elevated and because it declines rapidly as insulin therapy allows glucose and potassium to enter cells, it is recommended that potassium replacement be initiated earlier than in ketotic patients: 10 mEq of potassium chloride can be added to the initial liter of fluid administered if the initial serum potassium is not elevated and if the patient is making urine. When serum phosphate falls below 1 mg/dL during insulin therapy, phosphate replacement can be given intravenously with the same precautions as those outlined for ketoacidotic patients (see earlier). If the patient is awake and cooperative, part or all of the potassium and phosphate replacement can be given orally.
• 3. Insulin therapy. In general, less insulin is required to reduce the hyperglycemia of nonketotic patients than is the case for patients in diabetic ketoacidosis. In fact, fluid replacement alone can decrease glucose levels considerably. An initial insulin bolus of 0.15 U/kg is followed by an insulin infusion rate of 0.1 U/kg/h titrated to lower blood glucose by 50 to 70 mg/dL/h. Once the patient has stabilized and the blood glucose falls to around 250 mg/dL, insulin can be given subcutaneously.

Complications and Prognosis

The severe dehydration and low output state may predispose the patient to complications such as myocardial infarction, stroke, pulmonary embolism, mesenteric vein thrombosis, and disseminated intravascular coagulation. Fluid resuscitation remains the primary approach to the prevention of these complications. Low-dose heparin prophylaxis is reasonable but benefits of routine anticoagulation remain doubtful. Rhabdomyolysis is a recognized complication of the hyperosmolar state, and it should be looked for and treated.

The overall mortality rate of hyperglycemic, hyperosmolar, nonketotic coma is over 10 times that of diabetic ketoacidosis, chiefly because of its higher incidence in older patients, who may have compromised cardiovascular systems or associated major illnesses. When patients are matched for age, the prognoses of these two forms of hyperosmolar coma are reasonably comparable.

Disposition

After the patient is stabilized, the appropriate form of long-term management of the diabetes must be determined. Insulin treatment should be continued for a few weeks, but the patients usually recover sufficient endogenous insulin secretion to make a trial of diet or diet plus oral agents worthwhile. When the episode occurs in a patient who has known diabetes, then education of the patient and caregivers should be instituted. They should be taught how to recognize situations (gastrointestinal upset, infection) that predispose to recurrence of hyperglycemic, hyperosmolar state as well as detailed information on how to prevent the escalating dehydration (small sips of sugar-free liquids, increase in usual hypoglycemic therapy, or early contact with the physician) that culminates in hyperosmolar coma.

Lactic Acidosis

When severely ill diabetic patients present with profound acidosis and an anion gap over 15 mEq/L but relatively low or undetectable levels of keto acids in plasma, the presence of excessive plasma lactate (>5 mmol/L) should be considered, especially if other causes of acidosis such as uremia are not present.

Pathogenesis

Lactic acid is the end product of the anaerobic metabolism of glucose. Normally, the principal sources of this acid are the erythrocytes (which lack the enzymes for aerobic oxidation), skeletal muscle, skin, and brain. The chief pathway for removal of lactic acid is by hepatic (and to some degree renal) uptake for conversion first to pyruvate and eventually back to glucose, a process that requires oxygen. Lactic acidosis occurs when excess lactic acid accumulates in the blood. This can be the result of overproduction (tissue hypoxia), deficient removal (hepatic failure), or both (circulatory collapse). Lactic acidosis is not uncommon in any severely ill patient suffering from cardiac decompensation, respiratory or hepatic failure, septicemia, or infarction of the bowel or extremities. Type A lactic acidosis is associated with tissue hypoxia from hypovolemia or endotoxic shock and need not be associated with hyperglycemia. Type B lactic acidosis is defined as that which occurs in the absence of clinical evidence for tissue hypoxia and is associated with diabetes per se or with biguanide therapy.

With the discontinuance of phenformin therapy in the United States, lactic acidosis in patients with diabetes mellitus has become uncommon, but it still must be considered in the acidotic diabetic patient if the patient is seriously ill, and especially if the patient is receiving metformin therapy as well. Most cases of metformin-associated lactic acidosis occur in patients in whom there were contraindications to the use of metformin, in particular renal failure.

Clinical Features

Symptoms and Signs

The main clinical features of lactic acidosis are marked hyperventilation and mental confusion, which may progress to stupor or coma. When lactic acidosis is secondary to tissue hypoxia or vascular collapse, the clinical presentation is variable, reflecting that of the prevailing catastrophic illness. In the rare instance of idiopathic or spontaneous lactic acidosis, the onset is rapid (usually over a few hours), the cardiopulmonary status is stable, and mentation may be relatively normal.

Laboratory Findings

Plasma glucose can be low, normal, or high in diabetic patients with lactic acidosis, but usually it is moderately elevated. Plasma bicarbonate and arterial pH are quite low. An anion gap is present (calculated by subtracting the sum of the plasma bicarbonate and chloride from the plasma sodium; normal is 12-15 mEq/L). Ketones are usually absent from plasma, but small amounts may be present in urine if the patient has not been eating recently. Other causes of anion gap metabolic acidosis should be excluded—for example, uremia, diabetic or alcoholic ketoacidosis, and salicylate, methanol, ethylene glycol, or paraldehyde intoxication. In the absence of azotemia, hyperphosphatemia may be a clue to the presence of lactic acidosis.

The diagnosis is confirmed by demonstrating, in a sample of blood that is promptly chilled and separated, a plasma lactate concentration of 6 mmol/L or higher (normal is ∼1 mmol/L). Failure to rapidly chill the sample and separate the plasma can lead to falsely high plasma lactate values as a result of continued glycolysis by the red blood cells. Frozen plasma remains stable for subsequent assay.

Treatment

The cornerstone of therapy is aggressive treatment of the precipitating cause. An adequate airway and good oxygenation should be ensured. If hypotension is present, fluids and, if appropriate, pressor agents must be given to restore tissue perfusion. Appropriate cultures and empiric antibiotic coverage should be instituted in any seriously ill patient with lactic acidosis in whom the cause is not immediately apparent. Alkalinization with intravenous sodium bicarbonate to keep the pH above 7.2 has been recommended in the emergency treatment of severe lactic acidosis. However, there is no evidence that the mortality rate is favorably affected by administering bicarbonate, and the matter is at present controversial, particularly because of the hazards associated with bicarbonate therapy. Hemodialysis may be useful in those cases associated with metformin toxicity. Dichloroacetate, an anion that facilitates pyruvate removal by activating pyruvate dehydrogenase, reverses certain types of lactic acidosis in animals, but in a prospective controlled clinical trial involving 252 patients with lactic acidosis, dichloroacetate failed to alter either hemodynamics or survival.

Chronic Complications of Diabetes Mellitus

(Table 17–22)

In most patients with diabetes, a number of pathologic changes occur at variable intervals during the course of the disease. These changes involve the vascular system for the most part; however, they also occur in the nerves, the skin, and the lens. In addition to these complications, patients with diabetes have an increased incidence of certain types of infections and may handle their infections less well than the general population.

Classifications of Diabetic Vascular Disease

Diabetic vascular disease is conveniently divided into two main categories: microvascular disease and macrovascular disease.

Microvascular Disease

Disease of the smallest blood vessels, the capillary and the precapillary arterioles, is manifested mainly by thickening of the capillary basement membrane. Microvascular disease involving the retina leads to diabetic retinopathy, and disease involving the kidney causes diabetic nephropathy. Small vessel disease may also involve the heart, and cardiomegaly with heart failure has been described in diabetic patients with patent coronary arteries.

Macrovascular Disease

Large vessel disease in diabetes is essentially an accelerated form of atherosclerosis. It accounts for the increased incidence of myocardial infarction, stroke, and peripheral gangrene in diabetic patients. Just as in the case of atherosclerosis in the general population, the exact cause of accelerated atherosclerosis in the diabetic population remains unclear. Abnormalities in vessel walls, platelets and other components of the clotting system, red blood cells, and lipid metabolism have all been postulated to play a role. In addition, there is evidence that coexistent risk factors such as cigarette smoking and hypertension may be important in determining the course of the disease.

Prevalence of Chronic Complications by Type of Diabetes

Although all of the known complications of diabetes can be found in both types of the disease, some are more common in one type than in the other. In type 1 diabetes, end-stage renal disease develops in up to 40% of patients, compared with less than 20% of patients with type 2 diabetes. Although blindness occurs in both types, it occurs more commonly as a result of severe proliferative retinopathy, vitreous hemorrhages, and retinal detachment in type 1 disease, whereas macular edema and ischemia are the usual cause in type 2. Similarly, although diabetic neuropathy is common in both type 1 and type 2 diabetes, severe autonomic neuropathy with gastroparesis, diabetic diarrhea, resting tachycardia, and postural hypotension is much more common in type 1.

In patients with type 1 diabetes, complications from end-stage renal disease are a major cause of death, whereas patients with type 2 diabetes are more likely to have macrovascular diseases leading to myocardial infarction and stroke as the main causes of death. Cigarette use adds significantly to the risk of both microvascular and macrovascular complications in diabetic patients.

Molecular Mechanisms by Which Hyperglycemia Causes Microvascular and Macrovascular Damage

Epidemiological data and prospective intervention studies such as the DCCT have confirmed the central role of glucose in the development of chronic diabetic complications.

Four molecular mechanisms of glucose-induced damage have been proposed: (1) increased polyol pathway flux; (2) increased intracellular advanced glycation end product (AGE) formation; (3) activation of protein kinase C; and (4) increased hexosamine pathway flux.

Increase flux through the polyol pathway consumes NADPH. Because this cofactor is needed to regenerate reduced glutathione, NADPH depletion is predicted to exacerbate intracellular oxidative stress and cause cellular injury. Inhibitors of aldose reductase, the first enzyme in the polyol pathway have been shown to improve diabetic neuropathy.

Intracellular autoxidation of glucose results in production of intracellular dicarbonyls (glyoxal, 3-deoxyglucosone, methylglyoxal), also referred to as AGE precursors. These precursors damage target tissues by modifying intracellular and extracellular proteins and matrix components. Intracellular protein modifications may alter cellular functions. Modifications of extracellular matrix proteins result in abnormal interactions with other matrix proteins and integrins. The modified plasma proteins bind to receptors on endothelial cells, mesangial cells, and macrophages causing expression of cytokines and growth factors including interleukin 1, IGF-1, TNF-α, transforming growth factor-β (TGF-β), macrophage colony stimulating factor, granulocyte-macrophage stimulating factor, platelet-derived growth factor, thrombomodulin, tissue factor, vascular cell adhesion molecule 1 (VCAM 1), and vascular endothelial growth factor (VEGF). Induction of VEGF has been implicated in the vascular hyperpermeability associated with diabetes.

Protein kinase C isoforms β and Δ are activated by diacylglycerol whose levels are increased by elevated intracellular glucose. Activation of these isoforms leads to alterations in expression of endothelial nitric oxide synthase, endothelin 1, VEGF, TGF-β, PAI-1, and activation of nuclear factor κB and NADPH oxidases. Inhibitors of the protein kinase β isoform improve retinopathy and nephropathy in experimental models.

Hyperglycemia increases hexosamine pathway flux by providing more fructose-6-phosphate for the rate-limiting enzyme of the pathway glutamine: fructose-6-phosphate amidotransferase. Activity of this pathway leads to increased donation of N-acetylglucosamine moieties to serine and threonine moieties of complication–promoting factors such as PAI-1 or TGF-β.

It has been proposed that all four of these pathways are associated with overproduction of superoxide by mitochondria. High ambient glucose leads to increased substrate flux through glycolysis and the tricarboxylic acid cycle. This leads to increased potential difference across the inner mitochondrial membrane and generation of superoxide by the electron transport chain. The increased production of superoxide reduces glyceraldehyde-3-phosphate dehydrogenase (GAPDH) activity, which in turn leads to upstream increase in intracellular glucose and accumulation of glycolytic intermediates such as glyceraldehyde 3-phosphate and fructose 6-phosphate. The increased intracellular glucose leads to increased flux through the polyol pathway and also is the primary initiating event in the formation of both intracellular and extracellular AGEs. The glycolytic intermediates are important initiators of the hexosamine pathway (fructose 6-phosphate) or the protein kinase C pathway (glyceraldehyde-3-phosphate).

Genetic Factors in Susceptibility to Development of Chronic Complications of Diabetes

Although no genetic susceptibility genes have been identified as yet, two unrelated observations indicate that approximately 40% of people with diabetes may be unusually susceptible to the ravages of hyperglycemia or other metabolic sequelae of an inadequate insulin effect.

1. In one retrospective study of 164 juvenile-onset diabetic patients with a median age at onset of 9 years, 40% were incapacitated or dead from end-stage renal disease with proliferative retinopathy after a 25-year follow-up, whereas the remaining subjects were either mildly affected (40%) or had no clinically detected microvascular disease (20%). This study was completed long before the availability of glycemic self–monitoring methodology, so it is unlikely that any of these patients were near optimal glycemic control.

2. Data from renal transplantation indicate that only about 40% of normal kidneys developed evidence of moderate to severe diabetic nephropathy within 6 to 14 years of being transplanted into diabetic subjects with end-stage renal failure, whereas as many as 60% were only minimally affected. These observations support the hypothesis that although approximately 60% of people suffer only minimal consequences from hyperglycemia and other metabolic hazards of insulin insufficiency, 40% or so suffer severe, potentially catastrophic microvascular complications if the disease is poorly controlled. The genetic mechanisms for this increased susceptibility are as yet unknown but could be related to one or more of the molecular mechanisms outlined above. Identification of the genetic mechanism(s) would be very helpful in justifying more intensive insulin therapy in the group susceptible to complications in an effort to achieve near–normalization of blood glucose. The remaining 60% of less susceptible individuals might then be spared the inconveniences of strict glycemic control as well as the risks of hypoglycemia inherent in present methods of intensive insulin therapy.

Specific Chronic Complications of Diabetes Mellitus

(Table 17–22)

Ophthalmologic Complications

Diabetic Retinopathy

For early detection of diabetic retinopathy, adolescent or adult patients who have had type 1 diabetes for more than 5 years and all patients with type 2 diabetes should be referred to an ophthalmologist for examination and follow-up. In patients with type 1 diabetes, after 10 to 15 years, 25% to 50% of patients show some signs of retinopathy. This prevalence increases to 75% to 95% after 15 years and approaches 100% after 30 years of diabetes. In patients with type 2 diabetes, 60% have nonproliferative retinopathy after 16 years. When hypertension is present in a patient with diabetes, it should be treated vigorously because hypertension is associated with an increased incidence and accelerated progression of diabetic retinopathy.

Pathogenesis and Clinical Features

Two main categories of diabetic retinopathy exist: nonproliferative and proliferative. Diabetic macular edema can occur at any stage

Nonproliferative (background) retinopathy represents the earliest stage of retinal involvement by diabetes and is characterized by such changes as microaneurysms, dot hemorrhages, exudates, and retinal edema. During this stage, the retinal capillaries leak proteins, lipids, or red cells into the retina. When this process occurs in the macula (clinically significant macular edema), the area of greatest concentration of visual cells, there is interference with visual acuity; this is the most common cause of visual impairment in type 2 diabetes.

Proliferative retinopathy involves the growth of new capillaries and fibrous tissue within the retina and into the vitreous chamber. It is a consequence of small vessel occlusion, which causes retinal hypoxia; this in turn stimulates new vessel growth. New vessel formation may occur at the optic disc or elsewhere on the retina. Proliferative retinopathy can occur in both types of diabetes but is more common in type 1, developing about 7 to 10 years after onset of symptoms, with a prevalence of 25% after 15 years' duration. Prior to proliferation of new capillaries, a preproliferative phase often occurs in which arteriolar ischemia is manifested as cotton-wool spots (small infarcted areas of retina). Vision is usually normal until vitreous hemorrhage or retinal detachment occurs. Proliferative retinopathy is a leading cause of blindness in the United States, particularly because it increases the risk of retinal detachment.

Treatment

Extensive scatter xenon or argon photocoagulation and focal treatment of new vessels reduce severe visual loss in those cases in which proliferative retinopathy is associated with recent vitreous hemorrhages or in which extensive new vessels are located on or near the optic disk. Macular edema, which is more common than proliferative retinopathy in patients with type 2 diabetes (up to 20% prevalence), has a guarded prognosis, but it has also responded to scatter therapy with improvement in visual acuity if detected early. Injection of bevacizumab (Avastin), an antivascular endothelial growth factor (anti-VEGF), into the eye has been shown to stop the growth of the new blood vessels in diabetic eye disease. Avoiding tobacco use and correction of associated hypertension are important therapeutic measures in the management of diabetic retinopathy. There is no contraindication to using aspirin in patients with proliferative retinopathy.

Cataracts

Two types of cataracts occur in diabetic patients: subcapsular and senile. Subcapsular cataract occurs predominantly in patients with type 1 diabetes, may come on fairly rapidly, and has a significant correlation with the hyperglycemia of uncontrolled diabetes. This type of cataract has a flocculent or snowflake appearance and develops just below the lens capsule.

Senile cataract represents a sclerotic change of the lens nucleus. It is by far the most common type of cataract found in either diabetic or nondiabetic adults and tends to occur at a younger age in diabetic patients, particularly when glycemic control is poor.

Two separate abnormalities found in diabetic patients, both of which are related to elevated blood glucose levels, may contribute to the formation of cataracts: (1) glycosylation of the lens protein, and (2) an excess of sorbitol, which is formed from the increased quantities of glucose found in the insulin-independent lens. Accumulation of sorbitol leads to osmotic changes in the lens that ultimately result in fibrosis and cataract formation.

Glaucoma

Glaucoma occurs in approximately 6% of persons with diabetes. It is generally responsive to the usual therapy for open-angle disease. Closed-angle glaucoma can result from neovascularization of the iris in diabetic persons, but this is relatively uncommon except after cataract extraction when accelerated new vessel growth involving the angle of the iris obstructs outflow.

Renal Complications

Diabetic Nephropathy

Pathogenesis and Clinical Findings

About 4000 cases of end-stage renal disease due to diabetic nephropathy occur annually among diabetic patients in the United States. This represents about one-third of all patients being treated for renal failure. The cumulative incidence of nephropathy differs between the two major types of diabetes. Patients with type 1 diabetes, who have not received intensive insulin therapy and have had only fair to poor glycemic control, have a 30% to 40% chance of having nephropathy after 20 years—in contrast to the much lower frequency in patients with type 2 diabetes, who are not receiving intensive therapy, in whom only about 15% to 20% develop clinical renal disease. However, because so many more individuals are affected with type 2 diabetes, end-stage renal disease is much more prevalent in people with type 2 diabetes in the United States and especially throughout the rest of the world. There is no question that improved glycemic control and more effective therapeutic measures to correct hypertension can reduce the incidence of end-stage renal disease in both types of diabetes in the future.

Diabetic nephropathy is initially manifested by proteinuria; subsequently, as kidney function declines, urea and creatinine accumulate in the blood. Thickening of capillary basement membranes and of the mesangium of renal glomeruli produces varying degrees of glomerulosclerosis and renal insufficiency. Diffuse glomerulosclerosis is more common than nodular intercapillary glomerulosclerosis (Kimmelstiel-Wilson lesions); both produce heavy proteinuria.

Sensitive radioimmunoassay methods have permitted detection of microgram concentrations of urinary albumin. Conventional 24-hour urine collections, in addition to being inconvenient for patients, also show wide variability of albumin excretion, since several factors such as sustained erect posture, dietary protein, and exercise tend to increase albumin excretion rates. For these reasons, an albumin-creatinine ratio in an early morning spot urine collected upon awakening is preferable. In the early morning spot urine, a ratio of albumin (μg/L) to creatinine (mg/L) of <30 μg/mg creatinine is normal, and a ratio of 30 to 300 μg/mg creatinine suggests abnormal microalbuminuria. At least two of early morning spot urine collections over a 3- to 6-month period should be abnormal before a diagnosis of microalbuminuria is justified.

Subsequent renal failure can be predicted by persistent urinary albumin excretion rates exceeding 30 μg/min. Glycemic control as well as a low-protein diet (0.8 g/kg/d) may reduce both the hyperfiltration and the elevated microalbuminuria in patients in the early stages of diabetes and those with incipient diabetic nephropathy. Increased microalbuminuria correlates with increased levels of blood pressure and antihypertensive therapy decreases microalbuminuria. Evidence from some studies—but not the UKPDS—supports a specific role for ACE inhibitors in reducing intraglomerular pressure in addition to lowering systemic blood pressure. An ACE inhibitor (captopril, 50 mg twice daily) in normotensive diabetics impedes progression to proteinuria and prevents the increase in albumin excretion rate. Since microalbuminuria has been shown to correlate with elevated nocturnal systolic blood pressure, it is possible that normotensive diabetic patients with microalbuminuria have slightly elevated systolic blood pressure during sleep, which is lowered during antihypertensive therapy. This action may contribute to the reported efficacy of ACE inhibitors in reducing microalbuminuria in normotensive patients.

If treatment is inadequate, then the disease progresses with proteinuria of varying severity, occasionally leading to nephrotic syndrome with hypoalbuminemia, edema, and an increase in circulating LDL cholesterol as well as progressive azotemia. In contrast to all other renal disorders, the proteinuria associated with diabetic nephropathy does not diminish with progressive renal failure (patients continue to excrete 10 to 11 g daily as creatinine clearance diminishes). As renal failure progresses, there is an elevation in the renal threshold at which glycosuria appears.

Patients with diabetic nephropathy should be evaluated and followed by a nephrologist.

Treatment

Hemodialysis has been of limited success in the treatment of renal failure due to diabetic nephropathy, primarily because of progression of large vessel disease with resultant death and disability from stroke and myocardial infarction. Growing experience with chronic ambulatory peritoneal dialysis suggests that it may be a more convenient method of providing adequate dialysis with a lower incidence of complications.

Renal transplantation, especially from related donors, is often successful. For patients with compatible donors and no contraindications (such as severe cardiovascular disease), it is the treatment of choice.

Necrotizing Papillitis

This unusual complication of pyelonephritis occurs primarily in diabetic patients. It is characterized by fever, flank pain, pyuria, and sloughing of renal papillae in the urine. It is treated by intravenous administration of appropriate antibiotics.

Renal Decompensation After Radiographic Dyes

The use of radiographic contrast agents in diabetic patients with reduced creatinine clearance has been associated with the development of acute renal failure. Contrast-induced nephropathy is defined as an increase in serum creatinine of at least 0.5 mg/dL or 25% compared with baseline after exposure to intravenous contrast. The increase in serum creatinine occurs shortly after the procedure and peaks 3 to 5 days later before improving over the next 1 to 3 weeks. Although frequently transient, it can cause permanent impairment of renal function. Diabetic patients with normal renal function do not appear to be at increased risk for contrast nephropathy. If a contrast study is considered essential, patients with a serum creatinine of 1.5 to 2.5 mg/dL should be adequately hydrated before the procedure. Hydration with saline has been the cornerstone of contrast nephropathy prevention: intravenous saline 1 mL/kg/h is started 12 hours before the procedure and continued for 12 hours afterward. Some recent studies suggest that sodium bicarbonate infusions are more effective, and an alternate option is to infuse sodium bicarbonate: 150 mL of sodium bicarbonate (1 mEq/mL) is added to 1 L of 5% dextrose and infused at 3.5 mL/kg/h for 1 hour before the procedure; 1.2 mg/kg/h during the procedure and for 6 hours afterward. N-acetylcysteine has also been shown in some trials to decrease the incidence of contrast nephropathy. One regimen consists of using oral N-acetylcysteine 600 mg twice a day starting the day before the procedure for a total of four doses. A combination of N-acetylcysteine and intravenous saline or intravenous sodium bicarbonate may be more beneficial. Radiographic contrast material should not be given to a patient with a serum creatinine greater than 3 mg/dL unless the potential benefit outweighs the high risk of acute renal failure.

Neurologic Complications (Diabetic Neuropathy)

Peripheral and autonomic neuropathy are the two most common complications of both types of diabetes. Up to 50% of patients with type 2 diabetes are affected. The pathogenesis of both types of neuropathy is poorly understood. Some lesions, such as the acute cranial nerve palsies and diabetic amyotrophy, have been attributed to ischemic infarction of the involved peripheral nerve. The much more common symmetric sensory and motor peripheral neuropathies and autonomic neuropathy are felt to be due to metabolic complications.

Unfortunately, there is no consistently effective treatment for any of the neuropathies. However, several long-term clinical trials have definitively shown that normalization of blood glucose levels can prevent development and progression of this devastating complication.

Peripheral Neuropathy

Distal Symmetric Polyneuropathy

This is the most common form of diabetic peripheral neuropathy in which loss of function appears in a stocking-glove pattern and is due to an axonal neuropathic process. Longer nerves are especially vulnerable—hence the impact on the foot. Both motor and sensory nerve conduction is delayed in the peripheral nerves, and ankle jerks may be absent.

Sensory involvement usually occurs first and is generally bilateral, symmetric, and associated with dulled perception of vibration, and temperature. Pain, when present, can range from mild discomfort to severe incapacitating symptoms (see later). The sensory deficit may eventually be of sufficient degree to prevent patients from feeling pain. Patients with symptoms of sensory neuropathy should be examined with a 5.07 Semmes-Weinstein filament. Those who cannot feel the filament must be considered at risk for unperceived neuropathic injury.

The denervation of the small muscles of the foot results in clawing of the toes and displacement of the submetatarsal fat pads anteriorly. These changes, together with the joint and connective tissue changes, alter the biomechanics of the foot, increasing plantar pressures. This combination of decreased pain threshold, abnormally high foot pressures, and repetitive stress (eg, walking) can lead to calluses and ulcerations in the high-pressure areas such as over the metatarsal heads. Peripheral neuropathy also predisposes to the development of Charcot arthropathy. Other predisposing factors for this condition include autonomic neuropathy and trauma. The acute Charcot foot presents with pain and swelling, and if the condition is untreated, it leads to a rocker bottom deformity and ulceration. The early radiologic changes are of joint subluxation and periarticular fractures. As the process progresses, there is frank osteoclastic destruction leading to deranged and unstable joints particularly in the midfoot.

Not surprisingly, the key issue for healing of neuropathic ulcers in a foot with good vascular status is mechanical unloading. Additionally, any infection should be treated with debridement and appropriate antibiotics. Healing times of 8 to 10 weeks are typical. In the occasional patient, where healing appears refractory, platelet-derived growth factor (becaplermin) should be considered for local application. A postmarket epidemiologic study showed increased cancer deaths in patients who had used three or more tubes of becaplermin on their leg or feet ulcers, and there is now a boxed warning in the drug label. Once ulcers are healed, therapeutic footwear is key to preventing recurrences. Custom-molded shoes are reserved for patients with significant foot deformities. Other patients with neuropathy may require accommodative insoles that distribute the load over as wide an area as possible. Patients with foot deformities and loss of protective threshold should seek regular care from a podiatrist. They should be educated about appropriate footwear, and those patients with loss of protective threshold should be instructed to inspect their feet daily for reddened areas, blisters, abrasions, or lacerations.

Isolated Peripheral Neuropathy

Involvement of the distribution of only one nerve (mononeuropathy) or of several nerves (mononeuropathy multiplex) is characterized by sudden onset with subsequent recovery of all or most of the function. This neuropathology has been attributed to vascular ischemia or traumatic damage. Cranial and femoral nerves are commonly involved, and motor abnormalities predominate. The patient with cranial nerve involvement usually presents with diplopia. Clinical examination reveals signs of single third, fourth, or sixth nerve weakness, and the pupil is spared. A full recovery of function occurs in 6 to 12 weeks. Diabetic amyotrophy presents with onset of severe pain in the front of the thigh. Within a few days or weeks of the onset of pain, the patient develops weakness and wasting of the quadriceps. Usually as the weakness appears, the pain tends to improve. Management includes analgesia and improved diabetic control. The symptoms improve over 6 to 18 months.

Painful Diabetic Neuropathy

Hypersensitivity to light touch and occasionally severe burning pain, particularly at night, can become physically and emotionally disabling. Amitriptyline, 25 to 75 mg at bedtime, has been recommended for pain associated with diabetic neuropathy. Dramatic relief has often resulted within 48 to 72 hours. This rapid response is in contrast to the 2 or 3 weeks required for an antidepressant effect. Patients often attribute benefit to having a full night's sleep after amitriptyline, compared to many previously sleepless nights occasioned by neuropathic pain. Mild to moderate morning drowsiness is a side effect that generally improves with time or can be lessened by giving the medication several hours before bedtime. This drug should not be continued if improvement has not occurred after 5 days of therapy. If amitriptyline's anticholinergic effects are too troublesome, then nortriptyline can be used. Desipramine in doses of 25 to 150 mg/d seems to have the same efficacy as amitriptyline. Tricyclic antidepressants in combination with phenothiazine or fluphenazine have been shown in two studies to be efficacious in painful neuropathy, with benefits unrelated to relief of depression. Gabapentin (900-1800 mg/d in three divided doses) has also been shown to be effective in the treatment of painful neuropathy and should be tried if the tricyclic drugs prove ineffective. Pregabalin, a congener of gabapentin, has been shown in an 8-week study to be more effective than placebo in treating painful diabetic peripheral neuropathy. However, this drug was not compared with an active control. Also, because of its abuse potential, it has been categorized as a Schedule V controlled substance. Duloxetine, a serotonin and norepinephrine reuptake inhibitor, has been approved for the treatment of painful diabetic neuropathy. In clinical trials, this drug reduced the pain sensitivity score by 40% to 50%. Capsaicin, a topical irritant, has been found to be effective in reducing local nerve pain; it is dispensed as a cream (Zostrix 0.025%, Zostrix-HP 0.075%) to be rubbed into the skin over the painful region two to four times daily. Gloves should be used for application, because hand contamination could result in discomfort if the cream comes in contact with eyes or sensitive areas such as the genitalia.

Diabetic neuropathic cachexia is a syndrome characterized by a symmetric peripheral neuropathy associated with profound weight loss (up to 60% of total body weight) and painful dysesthesias affecting the proximal lower limbs, the hands, or the lower trunk. Treatment is usually with insulin and analgesics. The prognosis is usually good, and patients typically recover their baseline weight with resolution of the painful sensory symptoms within 1 year.

Autonomic Neuropathy

Neuropathy of the autonomic nervous system is common in patients with diabetes of long duration and can be a very disconcerting clinical problem. It can affect many diverse visceral functions including blood pressure and pulse, gastrointestinal activity, bladder function, and erectile function. Treatment is directed specifically at each abnormality.

Involvement of the gastrointestinal system may be manifested by nausea, vomiting, and postprandial fullness (from gastric atony); symptoms of reflux or dysphagia (from esophageal involvement); constipation and recurrent diarrhea, especially at night (from involvement of the small bowel and colon); and fecal incontinence (from anal sphincter dysfunction). Gallbladder function is altered, and this enhances stone formation.

Gastroparesis should be considered in type 1 diabetic patients in whom unexpected fluctuations and variability in their blood glucose levels develops after meals. Radiographic studies of the stomach and radioisotopic examination of gastric emptying after liquid and solid meals are of diagnostic value in these patients. Metoclopramide has been of some help in treating diabetic gastroparesis. It is a dopamine antagonist that has central antiemetic effects as well as a cholinergic action to facilitate gastric emptying. It is given in a dose of 10 mg orally three or four times a day, 30 minutes before meals and at bedtime. Drowsiness, restlessness, fatigue, and lassitude are common adverse effects. Tardive dyskinesia and extrapyramidal effects can occur especially when the drug is used for longer than 3 months and the FDA has cautioned against its chronic use. Erythromycin appears to bind to motilin receptors in the stomach and has been found to improve gastric emptying in doses of 250 mg three times daily over the short term, but its effectiveness seems to diminish over time. Tegaserod (Zelnorm), the partial 5-HT4 agonist, has been withdrawn from the US market by the FDA, so it is no longer used for this indication. In selected patients, injections of botulinum toxin into the pylorus can reduce sphincter resistance and enhance gastric emptying. Gastric electrical stimulation has been reported to improve symptoms and quality of life indices in patients with gastroparesis refractory to pharmacologic therapy.

Diarrhea associated with autonomic neuropathy has occasionally responded to broad-spectrum antibiotic therapy, although it often undergoes spontaneous remission. Refractory diabetic diarrhea is often associated with impaired sphincter control and fecal incontinence. Therapy with loperamide, 4 to 8 mg daily, or diphenoxylate with atropine, two tablets up to four times a day, may provide relief. In more severe cases, tincture of paregoric or codeine (60 mg tablets) may be required to reduce the frequency of diarrhea and improve the consistency of the stools. Clonidine has been reported to lessen diabetic diarrhea; however, its usefulness is limited by its tendency to cause orthostatic hypotension in these patients who already have autonomic neuropathy. Constipation usually responds to stimulant laxatives such as senna. Metamucil and other bulk-providing agents may relieve either the diarrhea or the constipation phases, which often alternate.

Inability to completely empty the bladder can sometimes occur. Bethanechol in doses of 10 to 50 mg three times a day has occasionally improved emptying of the atonic urinary bladder. Catheter decompression of the distended bladder has been reported to improve its function, and considerable benefit has been reported after surgical severing of the internal vesicle sphincter.

Use of Jobst fitted stockings, tilting the head of the bed, and arising slowly from the supine position are useful in minimizing symptoms of orthostatic hypotension. Some patients may require the addition of a mineralocorticoid such as fludrocortisone acetate (0.1-0.2 mg twice daily). Fludrocortisone therapy, however, can result in supine hypertension and hypokalemia. Midodrine (10 mg three times a day), an alpha adrenergic agonist, can be tried if Jobst stockings and sleeping upright prove ineffective in providing symptomatic relief.

Erectile dysfunction due to neuropathy differs from the psychogenic variety in that the latter may be intermittent (erections occur under special circumstances), whereas diabetic erectile dysfunction is usually persistent. To distinguish neuropathic or psychogenic erectile dysfunction from the erectile dysfunction caused by aortoiliac occlusive disease or vasculopathy, papaverine is injected into the corpus cavernosum of the penis. If the blood supply is competent, a penile erection occurs (Chapter 12). Urinary incontinence, with large volumes of residual urine, and retrograde ejaculation can also result from pelvic neuropathy.

There are medical, mechanical, and surgical approaches available for treatment of erectile dysfunction. Penile erection depends on relaxation of the smooth muscle in the arteries of the corpus cavernosum, and this is mediated by nitric oxide-induced cyclic 3′,5′-guanosine monophosphate (cGMP) formation. cGMP–specific phosphodiesterase type 5 (PDE5) inhibitors impair the breakdown of cGMP and improve the ability to attain and maintain an erection. Sildenafil (Viagra), vardenafil (Levitra), and tadalafil (Cialis) have been shown in placebo-controlled clinical trials to improve erections in response to sexual stimulation. The recommended dose of sildenafil for most patients is one 50 mg tablet taken approximately 1 hour before sexual activity. The peak effect is at 1.5 to 2 hours, with some effect persisting for 4 hours. Patients with diabetes mellitus using sildenafil reported 50% to 60% improvement in erectile function. The maximum recommended dose is 100 mg. The recommended dose of both vardenafil and tadalafil is 10 mg. The doses may be increased to 20 mg or decreased to 5 mg based on efficacy and side effects. Tadalafil has been shown to improve erectile function for up to 36 hours after dosing. In clinical trials, only a few adverse effects have been reported—transient mild headache, flushing, dyspepsia, and in some altered color vision. Priapism can occur with these drugs and patients should be advised to seek immediate medical attention if an erection persists for longer than 4 hours. The PDE5 inhibitors potentiate the hypotensive effects of nitrates, and their use is contraindicated in patients who are concurrently using organic nitrates in any form. Caution is advised for men who have suffered a heart attack, stroke, or life-threatening arrhythmia within the previous 6 months; men who have resting hypotension or hypertension, and men who have a history of cardiac failure or have unstable angina. Rarely, a decrease in vision or permanent visual loss has been reported after PDE5 inhibitor use.

Intracorporeal injection of vasoactive drugs causes penile engorgement and erection. Drugs most commonly used include papaverine alone, papaverine with phentolamine, and alprostadil (prostaglandin E1). Alprostadil injections are relatively painless, but careful instruction is essential to prevent local trauma, priapism, and fibrosis. Intraurethral pellets of alprostadil avoid the problem of injection of the drug.

External vacuum therapy (Erec-Aid System) is a nonsurgical treatment consisting of a suction chamber operated by a hand pump that creates a vacuum around the penis. This draws blood into the penis to produce an erection that is maintained by a specially designed tension ring inserted around the base of the penis and which can be kept in place for up to 20 to 30 minutes. Although this method is generally effective, its cumbersome nature limits its appeal.

In view of the recent development of nonsurgical approaches to therapy of erectile dysfunction, resort to surgical implants of penile prostheses is becoming less common.

Cardiovascular Complications

Heart Disease

Microangiopathy occurs in the heart and may explain the existence of congestive cardiomyopathies found in diabetic patients without demonstrable coronary artery disease. Much more commonly, however, heart failure in patients with diabetes is a consequence of coronary atherosclerosis. Myocardial infarction is three to five times more common in diabetic patients than in age-matched controls and is the leading cause of death in patients with type 2 diabetes. Cardiovascular disease risk is increased in patients with type 1 diabetes as well, although the absolute risk is lower than in patients with type 2 diabetes. Women with diabetes lose the protection against myocardial infarction that is usually present during the childbearing years. The increased risk of atherosclerosis in people with diabetes may reflect the combination of hyperlipidemia, abnormalities of platelet adhesiveness, coagulation factors, hypertension, and oxidative stress and inflammation.

Large intervention studies of risk factor reduction in diabetes are lacking, but it is reasonable to assume that reducing these risk factors would have beneficial effects. Lowering LDL cholesterol reduces first events in patients without known coronary disease and secondary events in patients with known coronary disease. These intervention studies included some patients with diabetes, and the benefits of lowering LDL cholesterol were apparent in this group. The National Cholesterol Education Program clinical practice guidelines have designated diabetes as a coronary risk equivalent and have recommended that patients with diabetes should have an LDL cholesterol goal of <100 mg/dL. Lowering LDL cholesterol to 70 mg/dL may have additional benefit and is a reasonable target for most patients with type 2 diabetes who have multiple risk factors for cardiovascular disease.

The ADA also recommends lowering blood pressure to 130/80 mm Hg or less. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) randomized 33,357 subjects (age 55 years or older) with hypertension and at least one other coronary artery disease risk factor to receive treatment with chlorthalidone, amlodipine, or lisinopril. Chlorthalidone appeared to be superior to amlodipine and lisinopril in lowering blood pressure, in reducing the incidence of cardiovascular events, in tolerability, and in cost. The study included 12,063 individuals with type 2 diabetes. The Heart Outcomes Prevention Evaluation (HOPE) study randomized 9297 high-risk patients who had evidence of vascular disease or diabetes plus one other cardiovascular risk factor to receive ramipril or placebo for a mean period of 5 years. Treatment with ramipril resulted in a 25% reduction of the risk of myocardial infarction, stroke, or death from cardiovascular disease. The mean difference in blood pressure between the placebo and ramipril groups was 2.2 mm Hg systolic and 1.4 mm Hg diastolic, and the reduction in cardiovascular event rate remained significant after adjustment for this small difference in blood pressure. The mechanism underlying this protective effect of ramipril is unknown. Patients with type 2 diabetes who already have cardiovascular disease or microalbuminuria should therefore be considered for treatment with an ACE inhibitor. More clinical studies are needed to address the question of whether patients with type 2 diabetes, who do not have cardiovascular disease or microalbuminuria, would specifically benefit from ACE inhibitor treatment.

Aspirin (81-325 mg daily) has been shown to effectively inhibit thromboxane synthesis by platelets and reduce the risk of diabetic atherothrombosis without increasing risks of either vitreous or gastrointestinal hemorrhage. Use of low-dose enteric-coated aspirin is recommended in adults with diabetes and evident macrovascular disease, in those with increased cardiovascular risk factors, or in those patients older than 30 years. Contraindications for aspirin therapy include age less than 21 years (because of risk of Reye syndrome), aspirin allergy, bleeding tendency (eg, anticoagulant therapy), recent gastrointestinal bleeding, or active hepatic disease. The Early Treatment Diabetic Retinopathy Study (ETDRS) showed that aspirin does not influence the course of proliferative retinopathy. There was no statistically significant difference in the severity of vitreous/preretinal hemorrhages or their rate of resolution between the aspirin and placebo groups. Thus, it appears that there is no contraindication to aspirin use to achieve cardiovascular benefit in diabetic patients who have proliferative retinopathy.

Peripheral Vascular Disease

Atherosclerosis is markedly accelerated in the larger arteries. It is often diffuse, with localized enhancement in certain areas of turbulent blood flow, such as at the bifurcation of the aorta or other large vessels. Clinical manifestations of peripheral vascular disease include ischemia of the lower extremities, impotence, and intestinal angina.

The incidence of gangrene of the feet in people with diabetes is 30 times that in age-matched controls. The factors responsible for the development of this condition, in addition to peripheral vascular disease, are small vessel disease, peripheral neuropathy with loss of both pain sensation and neurogenic inflammatory responses, and secondary infection. In two-thirds of patients with ischemic gangrene, pedal pulses are not palpable. In the remaining one-third who has palpable pulses, reduced blood flow through these vessels can be demonstrated by plethysmographic or Doppler ultrasound examination. Prevention of foot injury is imperative. Agents that reduce peripheral blood flow such as tobacco and propranolol should be avoided. Control of other risk factors such as hypertension is essential. Cholesterol-lowering agents are useful as adjunctive therapy when early ischemic signs are detected and when dyslipidemia is present. Patients should be advised to seek immediate medical care if a diabetic foot ulcer develops. Improvement in peripheral blood flow with endarterectomy and bypass operations is possible in certain patients.

Skin and Mucous Membrane Complications

Chronic pyogenic infections of the skin may occur, especially in poorly controlled diabetic patients. Eruptive xanthomas can result from hypertriglyceridemia, associated with poor glycemic control. Candidal infection can produce erythema and edema of intertriginous areas below the breasts, in the axillas, and between the fingers. It causes vulvovaginitis in most chronically uncontrolled diabetic women with persistent glucosuria and is a frequent cause of pruritus. While antifungal creams containing miconazole or clotrimazole offer immediate relief of vulvovaginitis, recurrence is frequent unless glucosuria is reduced.

Necrobiosis lipoidica diabeticorum are oval or irregularly shaped plaques with reddish demarcated borders and a glistening yellowish surface usually located over the anterior surfaces of the legs or the dorsal surfaces of the ankles. Rarely, lesions can occur on the hands, fingers, forearms, face, and scalp. The necrobiosis lesion usually starts out as an oval violaceous patch that slowly expands. The advancing border is red and the central area turns yellow-brown. The thinning of the dermis in the center of the lesion leads to the shiny surface and prominent telangiectasia. It also allows the subcutaneous fat to become more visible hence the yellowish appearance. Pathologically, the lesions show degeneration of collagen, granulomatous inflammation of subcutaneous tissues and blood vessels, capillary basement membrane thickening, and obliteration of vessel lumina. The condition is associated with type 1 diabetes, although it can occur in patients with type 2 diabetes and also in people without diabetes. It occurs in about 0.3% of patients with diabetes, usually in patients in their 30s and 40s, and is about three times more common in women than in men. In some studies an association with microalbuminuria and retinopathy has been reported. Improving glycemic control may help the condition. The first-line treatment includes topical and subcutaneous corticosteroids. Second-line treatments include systemic steroids, cyclosporine, ticlodipine, nicotinamide, clofazimine, fumarate esters, intralesional etanercept, and topical psoralen with ultraviolet A. Pulsed dye lasers can improve the appearance of telangiectasias. Flare-ups are frequent. No treatment is completely effective.

Shin spots are uncommon in adult diabetics. They are brownish, rounded, painless atrophic lesions of the skin in the pretibial area.

Bone and Joint Complications

Bone and joint complications are generally attributed to metabolic or vascular sequelae of long-standing diabetes.

Diabetic cheiroarthropathy is a syndrome of chronic progressive stiffness of the hand secondary to contracture and tightening of the skin over the joints. It is characterized by inability to flatten the palms against a flat surface (prayer sign). It is believed to be due to glycosylation of collagen and perhaps other proteins in connective tissue. It is associated with poor glycemic control and with longer duration of diabetes.

Dupuytren contractures consist of nodular thickening and contracture of the palmar fascia of the hand, producing flexure contractures of the fingers. It can occur in the absence of diabetes, but is more common in people with diabetes. The etiology is unclear but there may be an inflammatory component. Glucocorticoid injection into discrete nodules can sometimes help but the standard treatment is surgical fasciectomy.

Carpal tunnel syndrome occurs when the median nerve is compressed within the carpal tunnel. It is more common in people with diabetes, especially those who also have diabetic cheiroarthropathy. It is presumably due to glycosylation of collagen and other proteins in the connective tissue.

Patients with adhesive capsulitis of the shoulder (frozen shoulder) complain primarily of stiffness and loss of range of motion. They may also have shoulder pain. It occurs more frequently in people with long-standing diabetes—the incidence being two to four times higher than in the general population. The patients may also have diabetic cheiroarthropathy or Dupuytren contractures. In most cases it is a limited condition that responds to physical therapy. Patients should be warned that recovery may take 6 to 18 months. For a few patients surgery may be necessary. Adhesive capsulitis of the hip has also been described.

Data on bone mineral density and fracture risk in people with diabetes are contradictory. Bone mineral density has been reported to be low, normal, or high in type 2 diabetes patients. Type 2 diabetes patients do appear to be at increased risk for nonvertebral factures. Women with type 1 diabetes have an increased risk of fracture when compared to women without diabetes. It is likely that other factors such as duration of diabetes and diabetes complications such as neuropathy and renal disease affect both the bone mineral density and fracture risk.

Diffuse idiopathic skeletal hyperostosis (DISH) is a skeletal disease characterized by ossification of the anterior longitudinal ligament of the spine and various extraspinal ligaments. It causes stiffness and decreased range of spinal motion. The peripheral joints most commonly affected are the metacarpophalangeal joints, elbows, and shoulders. Diabetes, obesity, hypertension, and dyslipidemia are risk factors for this condition.

Hyperuricemia and gout are disorders associated with the metabolic syndrome. Thus, it is not surprising that type 2 diabetes patients are at increased risk for acute gout as well as chronic tophaceous gout.

Bursitis, particularly of the shoulders and hips, occurs more frequently than expected in patients with diabetes.

Infection

There are also several unusual infections that occur almost exclusively in diabetic patients (eg, emphysematous cholecystitis, mucormycosis, malignant otitis externa, and necrotizing papillitis). As noted earlier, atherosclerosis with peripheral vascular disease is very common in the diabetic population, and the resultant ischemia undoubtedly plays a role in the frequent and severe lower extremity infections seen in these patients.

Management of Diabetes in the Hospitalized Patient

A number of studies have observed that hospitalized patients with diabetes and those with new-onset hyperglycemia (ie, those without a preadmission diagnosis of diabetes) have higher inpatient morbidity and mortality. The morbidity and mortality in diabetics is twice that of nondiabetics. Those with new-onset hyperglycemia have even a higher mortality—almost eightfold greater in one study. These observations have led to increased interest in improving glucose control in the hospitalized diabetic patient.

Most patients with diabetes are hospitalized for reasons other than their diabetes. Up to 10% to 15% of all hospitalized patients have diabetes. Audits suggest that as many as a 30% of hospitalized diabetic patients have inappropriate management of their diabetes such as being given metformin where contraindicated; failure to act on high blood glucose levels; omission of diabetes medication; no record of diabetes complications; inappropriate insulin management; or blood glucose monitoring. Use of outpatient oral therapies or insulin regimens in the hospital is challenging. Patients are not eating according to their normal schedule and are often fasting for procedures. It is, therefore, usual to use insulin, subcutaneous or intravenous, in the hospitalized patient. Insulin is safe to use in patients with cardiac, renal, and liver disease and its dosing can be adjusted to match changing inpatient needs.

Surgery represents a stressful situation during which most of the insulin antagonists (catecholamines, GH, corticosteroids) are mobilized. In the patient with diabetes, this can lead to a worsening of hyperglycemia and perhaps even ketoacidosis. The aim of medical management of people with diabetes during the perioperative period is to minimize these stress-induced changes. Recommendations for management depend both on the patient's usual diabetic regimen and on the type of surgery (major or minor) to be done.

For people with diabetes controlled with diet alone, no special precautions must be taken unless diabetic control is markedly disturbed by the procedure. If this occurs, small amounts of short-acting insulin, as needed, will correct the hyperglycemia. Patients on oral agents should not take them on the day of surgery. If there is significant hyperglycemia, small amounts of short-acting insulin are given as needed. If this approach does not provide adequate control, an insulin infusion should be started in the manner indicated below. The oral agents can be restarted once the patient is eating normally after the operation. It is important to order a postoperative serum creatinine level to ensure normal renal function prior to restarting metformin therapy.

Patients taking insulin represent the only serious challenge to management of diabetes when surgery is necessary. However, with careful attention to changes in the clinical or laboratory picture, glucose control can be managed successfully. The protocol used to control the glucose depends on the kind of diabetes (type 1 or type 2); whether it is minor surgery (<2 hour and patient eating afterward) or major surgery (longer than 2 hours, invasion of body cavity, not eating afterward); and the preoperative insulin regimen (basal bolus or premixed insulin twice a day or premeal bolus only or regular before meals and NPH at bedtime). Type 1 patients have to be on some insulin to prevent the development of diabetic ketoacidosis. Many type 2 patients on insulin are fine without insulin for a few hours. Ideally patients with diabetes should undergo surgery early in the morning. Table 17–23 summarizes the approach for these various kinds of patients. One insulin infusion method adds 10 U of regular insulin to 1 L of 5% dextrose in half-normal saline, and this is infused intravenously at a rate of 100 to 180 mL/h. This gives the patient 1 to 1.8 U of insulin per hour, which, except in the most severe cases, generally keeps the blood glucose within the range of 100 to 250 mg/dL (5.5-13.9 mmol/L). The infusion may be continued for several days if necessary. Plasma glucose or blood glucose should be determined every 2 to 4 hours to be sure metabolic control is adequate. If it is not, adjustments in the ratio of insulin to dextrose in the intravenous solution can be made.

An alternative method, which is gaining increasing popularity, consists of separate infusions of insulin and glucose delivered by pumps to permit independent adjustments of each infusion rate, depending on hourly variation of blood glucose values. There are a number of different algorithms available for insulin infusions. Table 17–24 provides guidelines for management with an insulin drip and an algorithm designed to achieve glycemic control in the range of 120 to 180 mg/dL blood glucose.

After surgery, when the patient has resumed an adequate oral intake, subcutaneous administration of insulin can be resumed, and intravenous administration of insulin and dextrose can be stopped half an hour after the first subcutaneous dose. Insulin needs may vary in the first several days after surgery because of continuing postoperative stresses and because of variable caloric intake. In this situation, multiple doses of regular insulin, guided by blood glucose determinations, can keep the patient in acceptable metabolic control.

In the intensive care units, glucose levels are controlled most frequently using insulin infusions. Patients on total parental nutrition (TPN) can have insulin added to the bag. Standard TPN contains 25% dextrose, so 50 mL/h infusion delivers 12.5 g of dextrose per hour.

On the general surgical and medical wards, most patients are managed on subcutaneous insulin regimens. Limited cross–sectional and prospective studies suggest that the best glucose control is achieved on a combination of basal and bolus regimen with 50% of daily insulin needs provided by intermediate or long-acting insulins. Standardized order sets prompt medical personnel to write more physiological insulin orders, reduce errors, and include algorithms for recognition and treatment of hypoglycemia. Table 17–25 is an example of one such order set for a subcutaneous insulin regimen.

Targets for Glucose Control in the Hospitalized Patient

A number of studies have observed that hospitalized patients with diabetes and those with new-onset hyperglycemia (ie, those without a preadmission diagnosis of diabetes) have higher inpatient morbidity and mortality. Also, poor glucose control in diabetics at admission is associated with increased size of myocardial infarction and in some studies increased nosocomial infections after surgery. These observations have led to the question of whether lowering glucose levels close to normal in the hospital improves outcomes.

A prospective trial in surgical ICU patients (Leuven 1 study) reported that aggressive treatment of hyperglycemia (blood glucose <110 mg/dL) reduced mortality (35 deaths vs 63, p <0.04) and morbidity. Only a small number of persons in this study (204 of 1548) had a diagnosis of diabetes preoperatively, and so this study suggests that controlling hyperglycemia per se (independent of a diagnosis of diabetes) was beneficial. The benefits, however, were principally seen in patients who were in the ICU for longer than 5 days, and it was unclear whether the benefits would also apply to most surgical patients who stay in the ICU for only 1 to 2 days.

The same investigators performed a similar prospective trial among 1200 medical ICU patients (Leuven 2 study) and reported that aggressive treatment of hyperglycemia reduced morbidity (reduction in acquired renal injury and increased early weaning from mechanical ventilation) but not mortality. Again, as in the surgical ICU study, only a small number of persons (16.9%) had a diagnosis of diabetes at admission.

The findings of the Leuven studies however have not been confirmed in other prospective studies. Two other ICU-based studies (Glucontrol and VISEP) that attempted to confirm the findings were unable to do so. These two studies, however, were stopped prematurely for the following reasons: an interim analysis (falsely) suggested increased mortality in the test group in the Glucontrol study, and hypoglycemic events increased sevenfold in the intensive treatment group in the VISEP study. A large multicenter, multinational study (NICE-SUGAR) recruited 6104 surgical and medical ICU patients with hyperglycemia (20% had diabetes) and randomized them to tight control (81-108 mg/dL) or less tight control (<180 mg/dL). The tight group achieved blood glucose levels of 115 ± 18 and the conventional group 144 ± 23. There were more deaths (829 vs 751) in the tight glucose control group compared to the less tight glucose control group (p = 0.02). The intensively treated group also had more cases of severe hypoglycemia (206 vs 15 cases), and the excess deaths in the intensive group were due to cardiovascular events.

A study on tight intraoperative glycemic control during cardiac surgery also failed to show any benefit; if anything, the intensively treated group had more events. The United Kingdom Glucose Insulin in Stroke Trial (GIST-UK) failed to show beneficial effect of tight glycemic control in stroke patients; however, the investigators acknowledged that, because of slow recruitment, the study was underpowered.