Chapter 185. Fundamentals of Antibiotics

1. What resistance patterns are associated with major nosocomial pathogens?

2. When should empiric antibiotic therapy be initiated? When is it permissible to delay empiric antibiotic therapy, if ever?

3. What impact do pharmacokinetics and pharmacodynamics have on antibiotic choice?

4. In what circumstances is combination antibiotic therapy appropriate?

5. What are the major adverse effects of antibiotics?

In this chapter, we will use the terms antibacterial agents and antibiotics interchangeably. The term antibiotic has sometimes been defined as a substance produced by microbes that inhibits the growth of other microbes, especially bacteria. More broadly, it is used to characterize any agent with antibacterial properties, whether found in nature or synthesized artificially. Antibiotics are prescribed to approximately one-third of all hospitalized patients, and account for greater than 10% of hospital pharmacy expenditures. Up to one-half of antibiotic orders may be unnecessary, poorly chosen, or dosed incorrectly. Indiscriminant use of broad-spectrum agents is also believed to be a key contributor to emerging worldwide antimicrobial resistance. The estimated additional hospital costs associated with drug-resistant hospital-acquired bacterial infections in the United States is estimated to several billions of dollars. Hospitalists should be familiar with the currently available antibiotics, their penetration into various tissues, their major adverse effects, and their spectrum of activity relative to local patterns of antibacterial resistance. Antimicrobial prescribing should also take into account issues of cost and potential for the emergence of resistance.

Major categories, subdivisions, and some individual antibacterial agents are summarized in Table 185-1 according to their mode of action, and side effects and drug-drug interactions are listed in Table 185-2. Antibacterial spectrum often overlaps between different antibiotic groups. However, few antibiotics are active against multidrug- or highly resistant bacteria. Thus, the clinician may sometimes have several choices of potentially appropriate antibiotics, but at other times may only have only one, or even none.

When choosing antibacterial therapy in the hospital setting, it is crucial to know local patterns of antibiotic susceptibility and resistance. The phenotypes of bacterial organisms with respect to resistance may be important due to their abundance, virulence, or the near absence of antibiotics active against them. Most isolates of pneumococci are still sensitive to amoxicillin and especially to ceftriaxone. However, because pneumococcal infections are very common, even a small percentage of resistant isolates results in a large absolute number of infections by them. Vancomycin-resistant enterococci (VRE) are not highly virulent, but they are important in the hospital setting because of their tendency to attack debilitated and immunosuppressed patients, and because of the limited options for therapy. Methicillin-resistant Staphylococcus aureus (MRSA), by contrast, is virulent and increasingly common in both the hospital and community settings. The susceptibility patterns of nosocomial strains of MRSA differ from those of strains of community-acquired Staphylococcus aureus (cMRSA). Hospital-acquired MRSA is usually susceptible only to vancomycin, teicoplanin (available in Europe, but not the United States), and newer agents such as linezolid, tigecycline, and daptomycin. In addition to these drugs, cMRSA is also often susceptible to trimethoprim-sulfamethoxazole, fluoroquinolones, and clindamycin.

Gram-negative rods encountered in the hospital may have characteristic susceptibility phenotypes. Resistance to ampicillin and trimethoprim-sulfamethoxazole (TMP-SMX) is rising among Escherichia coli, limiting the role of these antibiotics as empiric therapy for urinary tract infections (UTIs). In addition, all enteric Gram-negative rods, and especially Escherichia coli and Klebsiella pneumoniae, may produce various types of beta-lactamases. Extended-spectrum beta-lactamases (ESBLs) are of particular significance. They confer resistance to all penicillins and commonly to all cephalosporins, and they are often accompanied by resistance to TMP-SMX, aminoglycosides, and fluoroquinolones. Carbapenems are at present the best choice for infections caused by ESBL-producing Gram-negative organisms.

Enterobacter cloacae and Enterobacter aerogenes commonly exhibit multidrug resistance due to the production of broad-spectrum AmpC beta-lactamase. AmpC confers resistance to second- and third-generation cephalosporins, monobactams, and many beta-lactam/beta-lactamase inhibitor combinations. However, Enterobacter species are often susceptible to TMP-SMX or fluoroquinolones, as well as carbapenems and fourth-generation cephalosporins, such as cefepime, which are considered the last line of antibiotics active against Gram-negative infections, and should be withheld from use as much as possible.

Pseudomonas aeruginosa and Acinetobacter baumannii often cause nosocomial infections in patients with multiple comorbidities, prolonged hospitalizations, prior antibiotic exposure, and intensive care unit (ICU) admission. These two bacteria are almost universally multidrug-resistant. In many countries, a large proportion of isolates are susceptible only to carbapenems, polymyxins, and tigecycline (the later being active only against Acinteobacter). Isolates of Pseudomonas aeruginosa, Acinetobacter baumannii, and even Klebsiella pneumoniae resistant to all commercially available antibiotics have been described in recent literature, primarily in ICU settings. Carbapenems, antipseudomonal penicillins, and the relatively toxic polymyxins in various combinations have been the main agents used against such isolates.

Proper Diagnosis

A high degree of clinical suspicion for possible bacterial infection should always constitute the basis for initiation of antibiotic therapy. Unnecessary use of antibacterial agents can harm both the patient and the hospital environment. First, initiation of antibiotics without adequate documentation of infection can obscure or delay the correct diagnoses. Second, antibiotics may cause toxicity, including life-threatening Clostridium difficile colitis. Third, when administering an antibiotic to a patient, the clinician should consider this antibiotic inactive against future infections in the same patient for a period of time of at least three months, according to some experts. Finally, antibiotic use leads to emergence of resistance microorganisms in the hospital environment.

As with most diseases, a careful history should always be the mainstay of diagnosis. Patients should be questioned about exposure to the health care system, vaccination, living conditions, animal exposure, travel history, sexual practices, chronic diseases, immunodeficiency, recent surgery, and indwelling foreign bodies such as heart valves and joint prostheses. While the physical examination is similar to that performed in other patients, hospitalists examining patients with suspected infection should pay special attention to aspects that are often neglected, such as skin, lymph nodes, surgical wounds, decubitus ulcers, intravenous line sites, and the genital region. Helpful laboratory investigations include complete blood count, certain biochemical tests, indices of inflammation, such as erythrocyte sedimentation rate, C-reactive protein, and perhaps procalcitonin, serology, urinalysis, and pharyngeal swabs with rapid testing for group A streptococcus or influenza A. Cultures from the infected site remain the gold standard for diagnosis, definite identification of the responsible pathogen, and optimization of antibiotic therapy. Finally, clinicians should distinguish between true infection and situations mimicking infections that need only limited or no antibiotic therapy (eg, UTI versus asymptomatic bacteriuria, pneumonia versus exacerbation of chronic obstructive pulmonary disease).

Timing of Therapy

Once the diagnosis of a possible infection is made, the clinician has to decide on the appropriate timing of antibiotic administration. Early initiation of antibiotic therapy should be the rule for all severe infections. There is considerable evidence that patients with sepsis have better outcomes, including reduced mortality, when antibiotic therapy is initiated as soon as possible. A large retrospective study of 2700 ICU patients with septic shock revealed that mortality was proportional to the time to initiation of appropriate antimicrobial therapy. Timely administration of antibiotics has a positive impact on a variety of serious infections. When central nervous system infection and especially meningitis is suspected, antibiotic therapy should be administered well before the definitive documentation of infection. Thus, only a few minutes delay is acceptable between the clinical suspicion of meningitis and lumbar puncture. When longer delays are expected, antibacterial therapy should be administered immediately, despite the consequent decreased diagnostic yield of lumbar puncture. Similarly, pneumonia-associated in-hospital morbidity and mortality is significantly reduced by prompt antibacterial therapy.

Delays in antimicrobial therapy are justified only in specific clinical settings. As a rule, antibiotic therapy is withheld only in hemodynamically stable patients, when the diagnosis of bacterial infection is doubtful, or when an acute or chronic infection is suspected that is best treated after identification of the responsible pathogen. Examples of clinical scenarios in which it is appropriate to delay antibiotics include a young adult with tonsillitis while awaiting the results of heterophile antibody testing, patients with possible endocarditis, in whom therapy should be delayed for at least an hour to perform three sets of aerobic and anaerobic blood cultures, and patients with suspected chronic infections due to brucellosis, syphilis, and other pathogens that mimic a wide variety of other diseases, and require specific therapy that differs from most empiric antibacterial therapies.

Antibiotic therapy initiated prior to identification of the causative infecting agent is termed empiric, a Greek word meaning derived from experience. Empiric antibiotics are chosen based on knowledge of the microbes typically causing infection at a specific body site, antibiotic penetration into this site, local patterns of antimicrobial susceptibility, and antibacterial side effect profile. Empiric antibiotic therapy usually has a broad spectrum of activity, including resistant organisms prevalent in relevant clinical settings. For example, the emergence of cMRSA has led to the increasing use of vancomycin (considered a last-line antibiotic in other situations) or sophisticated, costly newer agents in the empiric therapy of skin and soft tissue infections. Similarly, the rise of multidrug-resistant Gram-negative organisms such as Klebsiella pneumoniae, Acinetobacter baumanii, and Pseudomonas aeruginosa has led to the use of last-line antibacterials such as meropenem or colistin as empiric therapy in certain ICUs. When culture results become available, the broad spectrum of activity of the empiric regimen should be narrowed as much as possible (step-down).

Pharmacokinetics

Antibiotic pharmacokinetics, or the effects of the body on the antibiotic, depend on drug absorption, tissue distribution, metabolism, and excretion. Antibiotic absorption is most relevant to the oral administration of antibiotics. Some antibiotics, such as linezolid or levofloxacin, are nearly entirely absorbed from the gut. By contrast, oral vancomycin achieves negligible serum levels, but this route is best when high intraluminal concentrations of the drug in the gastrointestinal tract are desired, as when treating Clostridium difficile colitis. Tissue distribution plays a significant role in the success or failure of antibiotic therapy. For example, an antibiotic with poor distribution to the lungs is unlikely to be effective for pneumonia. Some antibiotics reach useful levels in one tissue compartment, but not another; for example, tigecycline is distributed widely in the body, but serum concentrations may be inadequate to treat bacteremias and endocarditis. In general, it is harder to achieve therapeutic antibiotic concentrations in the central nervous system, bones, pleural cavity, prostate, and poorly vascularized infected foci, such as abscesses. Fluoroquinolones often penetrate tissues that other antibiotics fail to reach, including bone and prostate. Local factors may also interfere with tissue distribution. Infections on foreign bodies often fail to respond to treatment because of the difficulty in penetrating the bacterial biofilms formed on their surfaces. Rifampin may be active in staphylococcal infections forming biofilm on foreign bodies.

Drug metabolism also affects antibiotic levels. Patients with hepatic or renal insufficiency may accumulate drugs, putting them at higher risk of toxicity, and need corresponding dose adjustment. Hepatic metabolism in particular is affected by genetic characteristics (pharmacogenetics) and drug-drug interactions. For example, isoniazid is metabolized by liver acetylation. About 50% of Europeans and Americans are slow acetylators, and will have higher isoniazid levels, increasing the risk for polyneuropathy and other toxicities. In addition, many antibiotics are metabolized by hepatic cytochrome P450 enzymes, resulting in numerous drug-drug interactions, as discussed below in the section on antibiotic toxicity.

Pharmacodynamics

Antibiotic pharmacodynamics, or the effect of the antibiotic on the bacteria, are related to antibiotic concentrations in blood and tissues, time, and bacterial inhibition or killing. Based on pharmacodynamic characteristics, antimicrobial agents can be divided into those that exhibit primarily concentration-dependent bacterial killing or time-dependent bacterial killing. Concentration-dependent agents, such as aminoglycosides and fluoroquinolones, achieve more bacterial killing at higher drug concentrations. Time-dependent agents, such as beta-lactams and vancomycin, accomplish maximal killing at concentrations four times higher than the minimum inhibitory concentration (MIC), with no extra bacteriocidal effects at higher concentrations.

For concentration-dependent agents, the main parameters predictive of successful clinical outcomes are the peak concentration achieved over minimum inhibitory concentration (MIC) ratio (Cmax/MIC) and the area under the concentration-time curve to MIC ratio (AUC/MIC). Thus, one or two high daily doses are considered optimal for aminoglycosides and fluoroquinolones. For time-dependent agents such as beta-lactams, the time with achieved concentrations above the MIC (t >MIC) is the pharmacodynamic parameter most predictive of bacterial killing and clinical efficacy. Thus, in theory, beta-lactam antibiotics are optimally administered in three or four doses daily, with some evidence supporting continuous administration. However, some beta-lactams antibiotics, such as ceftriaxone and ertapenem, have half-lives long enough to allow for once-daily dosing, without compromising concentrations in blood and tissues during the 24-hour period.

Combination antibiotic therapy is indicated when (1) broader empiric coverage is desired than can be accomplished with a single agent; (2) infection is proven or suspected to involve bacteria such as Mycobacterium tuberculosis with a high potential to develop resistance against a single agent; (3) when the site of infection may interfere with the ability of certain antibiotics to reach satisfactory concentrations (eg, aminoglycosides are not used alone in pneumonia, and vancomycin is not used as monotherapy in meningitis); (4) when infections are expected to be polymicrobial, as in intra-abdominal or pelvic infections; and (5) when a drug combination results in synergy, or a decrease in the MIC of each drug when tested together against a pathogen in vitro, as seen with the use of ampicillin and gentamicin in enterococcal endocarditis. However, this in vitro increase in activity does not always translate into superior clinical results, for reasons that are not always clear. Various synergistic combinations of beta-lactams with aminoglycosides have not been shown to be better than beta-lactam monotherapy in serious infections causing sepsis, according to a large meta-analysis of 64 trials including 7586 patients. Similarly, the development of several broad-spectrum beta-lactam agents, such as antipseudomonal penicillins combined with beta-lactamase inhibitors, fourth-generation cephalosporins, and carbapenems, has led to the widespread use of beta-lactam monotherapy in febrile neutropenia, a clinical setting in which combinations of beta-lactams with aminoglycosides had once been standard.

Antibiotic combinations, especially beta-lactams with aminoglycosides, were often used to prevent the development of resistance until recently. These two drug categories have different mechanisms of action, often exhibit synergy in vitro, and provide a wide spectrum of antibacterial activity. Animal studies in the 1980s provided some evidence that this combination could forestall the emergence of resistance, but meta-analyses in patients have not shown a clear benefit. Table 185-3 shows common indications for using antibiotic combinations in everyday clinical practice, as well as commonly used antibiotic combinations according to indication.

Antibiotics toxicities may arise from impaired drug metabolism, as in renal failure or hepatic insufficiency, from direct antibiotic effects, and from drug interactions. Beta-lactams are mainly metabolized in kidneys, and thus dose adjustment is needed in renal failure. An important exception is ceftriaxone, which is primarily metabolized in the liver. Gastrointestinal discomfort and mild diarrhea are the most common side effects of beta-lactams, with severe diarrhea or pseudomembranous colitis occurring less often. Up to 10% of patients report a penicillin allergy, but the true prevalence is probably closer to 1%, as most of these patients will not have a reaction if rechallenged. Possible reactions include skin rash, drug fever, nephritis, Coombs-positive hemolytic anemia, leucopenia, and even contact dermatitis. Interstitial nephritis usually results in mild, reversible renal failure, and is mainly associated with antistaphylococcal penicillins. Seizures may occur with high doses of imipenem or penicillins. Penicillin-induced anaphylaxis occurs with approximately 1 out of 10,000 administrations. Skin testing predicts penicillin IgE-mediated allergy and anaphylaxis, but not toxic epidermal necrolysis and Stevens-Johnson syndrome. Allergic reactions are much less common with cephalosporins. Fewer than 10% of patients with penicillin allergy are also allergic to cephalosporins or carbapenems. Monobactams, such as aztreonam, do not cross react with penicillin, having a substantially different chemical structure (Figure 185-1).

Aminoglycosides are excreted by the kidneys, reaching high concentrations in the urinary tract. However, they are nephrotoxic in 10–25% of patients, leading to acute kidney injury from tubular necrosis, mesangial damage, and renal vasoconstriction. This is more common in elderly, severely ill, or dehydrated patients. The serum creatinine should be closely monitored in patients at risk for aminoglycoside-related nephrotoxicity. If possible, the concomitant use of other nephrotoxic agents such as loop diuretics, vancomycin, and amphotericin B should be avoided, as these may potentiate aminoglycoside nephrotoxicity. The nephrotoxicity of aminoglycosides is usually ameliorated by dose adjustment or drug withdrawal. Once-daily administration and short treatment duration (≤ 7 days) may reduce the risk of nephrotoxicity. Ototoxicity with sensorineural deafness is the other major adverse effect of aminoglycosides. It manifests initially with high-frequency hearing loss, which may be irreversible in some cases. It is more common with prolonged administration, as with streptomycin use for tuberculosis or brucellosis, or gentamicin use in endocarditis. Neuromuscular blockade may occur when aminoglycosides are erroneously given by rapid intravenous push, or when they are administered to ICU patients receiving other inhibitors of the neuromuscular junction. Aminoglycoside use is generally avoided during pregnancy, since these antibiotics cross the placenta. Trough levels should be measured in patients receiving aminoglycosides for longer than one week, to ensure that drug accumulation is not occurring. Peak levels are usually not measured, except to ensure therapeutic levels in serious infections such as endocarditis.

Glycopeptides such as vancomycin commonly exhibit mild nephrotoxicity, which may become worse if dose adjustments are not performed based on renal clearance, similar to aminoglycosides. Serum levels of vancomycin are helpful in preventing toxicity in patients with preexisting renal failure, unstable renal function, and the elderly. There is no contraindication to coadministration with other nephrotoxic agents such as aminoglycosides, but this should be performed with great caution and only when necessary. Red person syndrome is a mild anaphylactoid reaction that causes flushing of the upper trunk and head of the patient. It is most common with the first dose of vancomycin, diminishes with subsequent doses, and can be prevented by slow administration over two hours. Some patients require premedication with antihistamines to minimize symptoms. True allergic skin rash may also occur occasionally with glycopeptide use. Ototoxicity may occur in patients with preexisting renal impairment and prolonged exposure.

Macrolides generally have mild side effects. Gastrointestinal upset due to increased motility is common. This is especially evident with erythromycin, which has been used to treat patients with gastroparesis and pseudoobstruction syndromes. Macrolides may also cause mild hepatotoxicity and QT interval prolongation. However, the most important macrolide adverse effect is drug interactions. Macrolides inhibit the CYP3A4 enzyme of the cytochrome P450 of the liver, leading to decreased metabolism and potentially toxic levels of several substances, including statins, warfarin, digoxin, carbamazepine, valproic acid, benzodiazepines, ergotamine, theophylline, vinca alkaloids, and cyclosporine. Polymorphic ventricular tachycardia (torsades de pointes), often preceded by QT prolongation, may occur when a macrolide is administered with other potent P450 inhibitors, such as cisapride, terfenadine, or astemizole.

Gastrointestinal upset is common with tetracyclines. Esophageal irritation and ulceration may occur with oral administration; this may be minimized by drinking generous amounts of water with pills and avoiding ingestion at bedtime. Phototoxicity is very common. Hepatotoxicity may occur with large intravenous doses. Use of tetracycline, and to a lesser degree doxycycline, is contraindicated in patients with preexisting renal failure, because tetracyclines may worsen renal insufficiency. Teteracyclines should be avoided in pregnancy, as they may cause fetal skeletal deformities and dental discoloration, and maternal fulminant hepatic failure. Drug interactions with digoxin, warfarin, theophylline, and statins are similar to macrolides, but usually milder. Carbamazepine and phenytoin induce hepatic metabolism of tetracyclines, reducing their activity. Antacids, cimetidine, and milk decrease tetracycline absorption.

Diarrhea is the most common side effect of fluoroquinolones. C difficile colitis is relatively common with respiratory fluoroquinolones such as levofloxacin and moxifloxacin, but occurs less often with older fluoroquinolones such as ciprofloxacin. All fluoroquinolones may cause QT prolongation, although this is minimal with ciprofloxacin and other older fluoroquinolones. Other important, occasional side effects include phototoxicity, tendon rupture—especially involving the Achilles tendon—seizures, and confusion. Tendinopathy and central nervous system side effects are more common in the elderly. Fluoroquinolone use should generally be avoided in patients younger than 18 years old because of their epileptogenic potential and adverse effects on developing joints. Fluoroquinolones may be teratogenic, and are contraindicated in pregnancy. Ciprofloxacin inhibits theophylline metabolism; this interaction does not occur with respiratory fluoroquinolones. Fluoroquinolones require dose adjustment in severe renal failure.

Lincosamides, such as clindamycin, most commonly cause gastrointestinal distress. Clindamycin was once considered the major cause of C difficile colitis, but its role in the epidemiology of C difficile has declined due to a rise in C difficile infections related to beta-lactams and fluoroquinolones. Sulfonamides may cause nephrotoxicity from crystalluria, especially with the use of high doses in the treatment of conditions such as nocardiosis and toxoplasmosis. Patients receiving high-dose sulfonamides should be adequately hydrated, and dose adjustments should be made for patients with renal failure. Sulfa antibiotics may trigger hemolysis in patients with G6PD deficiency, and rarely lead to toxic epidermal necrolysis/Stevens-Johnson syndrome. Sulfonamides decrease the metabolism of phenytoin, warfarin, and digoxin. In combination with trimethoprim, they may act as antimetabolites and cause marrow aplasia.

Patients tolerating oral medications or nasogastric feeding may benefit from oral antibiotic administration at lower cost. For example, oral levofloxacin and metronidazole approach 100% bioavailability, and oral fluconazole is 90% bioavailable. The greater ease of administration and avoidance of adverse events associated with parenteral therapy, such as bacterial and fungal line infections or post-infusion phlebitis, may result in decreased hospital length of stay. Patients may also benefit from improved mobility. The ease of administration should not, however, lead to indiscriminate use of these and newer agents. For example, overuse of linezolid, an expensive drug active against most strains of VRE and MRSA, may promote resistance in these Gram-positive organisms.

In recent decades, the prevalence of nosocomial antibacterial resistance has been rising, with increased in-hospital morbidity and mortality. Acute care hospitals should take steps to reduce the inappropriate use of antibiotics. This may require developing antibiotic guidelines, educational strategies, and computer support. Specific practices are necessary to combat antibacterial resistance. Hospitals must have surveillance systems, involving physicians, infection control practitioners, nurses, and the microbiology laboratory, to identify increases in the incidence of infections caused by resistant pathogens. This surveillance also includes monitoring of antibiotic consumption, and correlating usage with patterns of antibiotic resistance. When specific antibiotics appear to be linked to the emergence of resistance, local antibiotic policies may be needed to limit use. In most cases, antibiotic use is itself the main factor promoting resistance. Antibiotic use may also be reduced by infection control measures, such as hand washing, isolation precautions, and immunization.

The principle of “hit hard, hit fast” or the prompt use of high and appropriately divided antibiotic doses should be applied to most hospital infections. Depending on antibiotic mode of action and pharmacodynamic characteristics, low antibiotic concentrations favor the development of bacterial resistance, mainly by means of stepwise mutations. Thus, for example, development of resistance to fluoroquinolones is linked to low areas under the concentration-time curve to MIC ratio (AUC/MIC), while beta-lactam resistance is linked to low time with achieved concentrations above the MIC (t > MIC). The risk of emergence of resistance is highest in organisms with MICs close to susceptibility breakpoint.

Prolonged, intense antibiotic pressure eventually selects for resistant organisms, which may heavily colonize both the affected patient and the hospital environment in general, including other hospital patients not receiving antibiotics. Thus, antibiotic therapy should be as brief as possible in order not to compromise overall efficacy, and de-escalation or narrowing the spectrum of antibiotic therapy, should be performed as soon as antimicrobial sensitivity data are available. Other suggested measures to limit antibiotic resistance, such as periodic cycling or “crop rotation” of antibiotic classes, or antibiotic combination therapy, have inadequate evidence to support their use at this time.