Chapter 48: The Thalassemias: Disorders of Globin Synthesis

In 1925, Cooley and Lee¹ first described a form of severe anemia that occurred early in life and was associated with splenomegaly and bone changes. In 1932, George H. Whipple and William L. Bradford² published a comprehensive account of the pathologic findings in this disease. Whipple coined the phrase thalassic anemia^3,4 and condensed it to thalassemia, from θαλασσα (“the sea”), because early patients were all of Mediterranean background. The true genetic character of the disorder became fully appreciated after 1940. The disease described by Cooley and Lee is the homozygous state of an autosomal gene for which the heterozygous state is associated with much milder hematologic changes. The severe homozygous condition became known as thalassemia major. The heterozygous states, thalassemia trait, were designated according to their severity as thalassemia minor or minima.^3,5,6,7 Later, the term thalassemia intermedia was used to describe disorders that were milder than the major form but more severe than the traits.

Thalassemia is not a single disease but a group of disorders, each resulting from an inherited abnormality of globin production.⁷ The conditions form part of the spectrum of diseases known collectively as the hemoglobinopathies, which can be classified broadly into two types. The first subdivision consists of conditions, such as sickle cell anemia, that result from an inherited structural alteration in one of the globin chains. Although such abnormal hemoglobins may be synthesized less efficiently or broken down more rapidly than normal adult hemoglobin, the associated clinical abnormalities result from the physical properties of the abnormal hemoglobin (Chap. 49). The second major subdivision of the hemoglobinopathies, the thalassemias, consists of inherited defects in the rate of synthesis of one or more of the globin chains. The result is imbalanced globin chain production, ineffective erythropoiesis, hemolysis, and a variable degree of anemia.

Several monographs describe the historical aspects of thalassemia in greater detail.^5,7

DIFFERENT FORMS OF THALASSEMIA

Thalassemia can be defined as a condition in which a reduced rate of synthesis of one or more of the globin chains leads to imbalanced globin-chain synthesis, defective hemoglobin production, and damage to the red cells or their precursors from the effects of the globin subunits that are produced in relative excess.^7,8 Table 48–1 summarizes the main varieties of thalassemia that have been defined with certainty.

The β-thalassemias are divided into two main varieties. In one form, β⁰-thalassemia, there is no β-chain production. In the other form, β⁺-thalassemia, there is a partial deficiency of β-chain production. The hallmark of the common forms of β-thalassemia is an elevated level of hemoglobin A₂ in heterozygotes. In a less-common class of β-thalassemias, heterozygotes have normal hemoglobin A₂ levels. Other rare forms include varieties of β-thalassemia intermedia that are inherited in a dominant fashion, that is, heterozygotes are severely affected, and there is a variety in which the genetic determinants are not linked to the β-globin gene cluster.^7,9,10

The δβ-thalassemias are heterogeneous. In some cases, no δ or β chains are synthesized. Originally, these disorders were classified according to the structure of the hemoglobin F produced, that is, ^Gγ^Aγ(δβ)⁰- and ^Gγ(δβ)⁰-thalassemia. This classification is illogical. The conditions are best described by the globin chains that are defectively synthesized, that is, simply (δβ)⁺-, (δβ)⁰-, and (^Aγδβ)⁰-thalassemia.^7,10 In the (δβ)⁺-thalassemias, an abnormal hemoglobin is produced that has normal α chains combined with non-α chains consisting of the N-terminal residues of the δ chain fused to the C-terminal residues of the β chain. These fusion variants, called the Lepore hemoglobins, show structural heterogeneity.

The δ-thalassemias^7,10 are characterized by reduced output of δ chains and hence reduced hemoglobin A₂ levels in heterozygotes and an absence of hemoglobin A₂ in homozygotes. They are of no clinical significance except that, when inherited with β-thalassemia trait, the level of hemoglobin A₂ is reduced to the normal range.

A disorder characterized by defective ε-, γ-, δ-, and β-chain synthesis has been defined at the clinical and molecular level.^7,10 The homozygous state for this condition, εγδβ-thalassemia, presumably is not compatible with fetal survival. It has been observed only in heterozygotes.

Hereditary persistence of fetal hemoglobin (HPFH) is a heterogeneous condition characterized by persistent fetal hemoglobin.^7,9,10 It is classified into deletion and nondeletion forms. The deletion forms of HPFH can be classified, like δβ-thalassemia, as (δβ)⁰ HPFH and then subdivided according to the particular population in which this occurs and its associated molecular defect. In effect, the deletion forms of HPFH are very similar to β-thalassemia except for more efficient γ-chain synthesis and, therefore, less chain imbalance and a milder phenotype. The homozygous state is associated with mild thalassemic changes. In fact, the β-thalassemias and deletion forms of HPFH form a clinical continuum. The nondeletion forms of HPFH also are heterogeneous. In some cases, they are associated with mutations that involve the β-globin gene cluster and in which there is β-chain synthesis cis to the HPFH determinant. These conditions are subdivided into ^Gγβ⁺ HPFH and ^Aγβ⁺ HPFH. Again, they often are subclassified according to the population in which they occur, for example, Greek HPFH, British HPFH, and so on. Finally, a heterogeneous group of HPFH determinants is associated with very low levels of persistent fetal hemoglobin, the genetic loci of which, at least in some cases, are not linked to the β-globin gene cluster.

Because α chains are present in both fetal and adult hemoglobins, a deficiency of α-chain production affects hemoglobin synthesis in fetal and in adult life. A reduced rate of α-chain synthesis in fetal life results in an excess of γ chains, which form γ₄ tetramers, or hemoglobin Bart’s. In adult life, a deficiency of α chains results in an excess of β chains, which form β₄ tetramers, or hemoglobin H. Because there are two α-globin genes per haploid genome, the genetics of α-thalassemia is more complicated than that of β-thalassemia. There are two main groups of α-thalassemia determinants.^7,10 First, in the α⁰-thalassemias (formerly called α-thalassemia 1), no α chains are produced from an affected chromosome; that is, both linked α-globin genes are inactivated. Second, in the α⁺-thalassemias (formerly called α-thalassemia 2), the output of one of the linked pair of α-globin genes is defective. The α⁺-thalassemias are subdivided into deletion and nondeletion types. Both the α⁰-thalassemias and deletion and nondeletion forms of α⁺-thalassemia are extremely heterogeneous at the molecular level. There are two major clinical phenotypes of α-thalassemia: the hemoglobin Bart’s hydrops syndrome, which usually reflects the homozygous state for α⁰-thalassemia, and hemoglobin H disease, which usually results from the compound heterozygous state for α⁰- and α⁺-thalassemia.

Because the structural hemoglobin variants and the thalassemias occur at a high frequency in some populations, the two types of genetic defect can be found in the same individual. The different genetic varieties of thalassemia and their combinations with the genes for abnormal hemoglobins produce a series of disorders known collectively as the thalassemia syndromes.⁷

The β-thalassemias are distributed widely in Mediterranean populations, the Middle East, parts of India and Pakistan, and throughout Southeast Asia (Fig. 48–1).^7,11,12 The disease is common in Tajikistan, Turkmenistan, Kyrgyzstan, and the People’s Republic of China. Because of the extensive migration from areas of high gene frequency such as the Mediterranean region (e.g., Italy, Greece), Africa, and Asia to the Americas, the α- and β-thalassemia genes and clinical disease are relatively common, especially in North, but also South, America. The β-thalassemias are rare in Africa, except for isolated pockets in West Africa, notably Liberia, and in parts of North Africa. However, β-thalassemia occurs sporadically in all racial groups and has been observed in the homozygous state in persons of pure Anglo-Saxon heritage. Thus, a patient’s racial background does not preclude the diagnosis.

The δβ-thalassemias have been observed sporadically in many racial groups, although no high-frequency populations have been defined. Similarly, the hemoglobin Lepore syndromes have been found in many populations, but, with the possible exceptions of central Italy, Western Europe, and parts of Spain and Portugal, these disorders have not been found to occur at a high frequency in any particular region.

The α-thalassemias occur widely throughout Africa, the Mediterranean countries, the Middle East, and Southeast Asia (Fig. 48–2).^7,11,12 The α⁰-thalassemias are found most commonly in Mediterranean and Oriental populations, but are extremely rare in African and Middle Eastern populations. However, the deletion forms of α⁺-thalassemia occur at a high frequency throughout West Africa, the Mediterranean, the Middle East, and Southeast Asia. In United States, approximately 30 percent of Americans of African descent carry the gene α⁺-thalassemia. Up to 80 percent of the population of some parts of Papua New Guinea are carriers for the deletion form of α⁺-thalassemia. How common the nondeletion forms of α⁺-thalassemia are in any particular populations is uncertain, but they have been reported quite frequently in some of the Mediterranean island populations and in the Middle Eastern and Southeast Asian populations. Because the hemoglobin Bart’s hydrops syndrome and hemoglobin H disease require the action of an α⁰-thalassemia determinant, these disorders are found at a high frequency only in Southeast Asia and in parts of the Mediterranean region. The α-chain termination mutants, such as hemoglobin Constant Spring, seem to be particularly common in Southeast Asia. Approximately 4 percent of the population in Thailand are carriers.

In 1949, J.B.S. Haldane¹³ suggested that thalassemia had reached its high frequency in tropical regions because heterozygotes are protected against malaria.¹³ Although many population studies have tested this hypothesis, elucidation of some of the extremely complex population genetics underlying polymorphic systems such as the thalassemias has been possible only with the advent of recombinant DNA technology.

In each of the high-frequency areas for the β-thalassemias, a few common mutations and varying numbers of rare mutations are seen (see Fig. 48–1). Furthermore, in each of these regions the pattern of mutations is different, usually found in the context of different haplotypes in the associated β-globin gene cluster.^11,14,15 Similar observations have been made in the α-thalassemias (see Fig. 48–2).^7,11 These studies suggest the thalassemias arose independently in different populations and then achieved their high frequency by selection. Although some movement of the thalassemia genes may have resulted from drift, independent mutation and selection undoubtedly provide the overall basis for their world distribution. Early studies in Sardinia, showing that β-thalassemia is less common in the mountainous regions where malarial transmission is low, supported Haldane’s suggestion that β-thalassemia reached its high frequency because of protection against malarial infections.¹⁶ For many years these data remained the only convincing evidence for a protective effect. However, later studies using malaria endemicity data and globin-gene mapping showed a clear altitude-related effect on the frequency of α-thalassemia in Papua New Guinea. In addition, a sharp cline (a gradual change of species phenotype over a geographical area) in the frequency of α-thalassemia has been found in the region stretching south from Papua New Guinea through the island populations of Melanesia to New Caledonia. This is mirrored by a similar gradient in the distribution of malaria.¹⁷ The effect of drift and founder effect in these island populations has been largely excluded by showing that other DNA polymorphisms have a random distribution through the region, with no evidence of a cline similar to that characterizing the distribution of α-thalassemia and malaria.

Firm evidence for protection of individuals with mild forms of α⁺-thalassemia against Plasmodium falciparum malaria has been provided. In a case-control study performed in Papua New Guinea, the homozygous state for α⁺-thalassemia offered approximately 60 percent protection against hospital admittance because of serious complications of malaria, notably coma or profound anemia.¹⁸ Similar levels of protection by α-thalassemia against P. falciparum malaria have been found in several different African populations.¹⁹ However, it is becoming clear that there are complex genetic epistatic interactions between protective polymorphisms of this kind. For example, although α-thalassemia and the sickle cell trait both offer strong protection against P. falciparum malaria, in those who inherit both traits, the protection is canceled out and they are fully susceptible to the disease.²⁰ Interactions of this type will have an important effect on the gene frequency of protective polymorphisms in countries in which more than one exists in the same population.

There is growing evidence that both immune and cellular mechanisms may underlie these protective effects of different red cell polymorphisms against malarial infection. Followup studies of cohorts of babies with α-thalassemia suggest that, in the first year of life, they are more prone to Plasmodium vivax and P. falciparum malaria. Because there is evidence for cross-immunization between these two species, it is possible that this effect induces early immunization that may result in babies with α-thalassemia being more resistant to P. falciparum malaria later in life.²¹ At the cellular level there is no evidence that α-thalassemia has any effect on the rates of parasite invasion and growth in red cells. However, parasitized α-thalassemic red cells are more susceptible to phagocytosis in vitro, and are less able than normal cells to form rosettes, an in vitro phenomena whereby uninfected cells bind to infected cells that is strongly associated with severity of infection, and express low levels of complement receptor 1, which is required for rosette formation.²² These highly complex immune and cellular interactions are discussed in detail in reviews.^19,23,24 Although there are less data of this kind available for the β-thalassemias, there is strong indirect evidence that their high frequency has also been maintained by protection against P. falciparum malaria.

GENETIC CONTROL AND SYNTHESIS OF HEMOGLOBIN

The structure and ontogeny of the hemoglobins are reviewed in Chaps. 7 and 49, respectively. Only those aspects with particular relevance to the thalassemia problem are discussed here.

Human adult hemoglobin is a heterogeneous mixture of proteins consisting of the major component hemoglobin A and the minor component hemoglobin A₂, which constitutes approximately 2.5 percent of the total. In intrauterine life, the main hemoglobin is hemoglobin F. The structure of these hemoglobins is similar. Each consists of two separate pairs of identical globin chains. Except for some of the embryonic hemoglobins (see below), all normal human hemoglobins have one pair of α chains. In hemoglobin A, the α chains are combined with β chains (α₂β₂), in hemoglobin A₂ with δ chains (α₂δ₂), and in hemoglobin F with γ chains (α₂γ₂).

Human hemoglobin shows further heterogeneity, particularly in fetal life, and this has important implications for understanding the thalassemias and for approaches to their prenatal diagnosis. Hemoglobin F is a mixture of molecular species with the formulas α₂γ₂^136Gly and α₂γ₂^136Ala. The γ chains containing glycine at position 136 are designated ^Gγ chains. The γ chains containing alanine are called ^Aγ chains. At birth, the ratio of molecules containing ^Gγ chains to those containing ^Aγ chains is approximately 3:1. The ratio varies widely in the trace amounts of hemoglobin F present in normal adults.

Before week 8 of intrauterine life, three embryonic hemoglobins—Gower 1 (ξ₂ε₂), Gower 2 (α₂ε₂), and Portland (ξ₂ γ₂)—are present. The ξ and ε chains are the embryonic counterparts of the adult α and β and γ and δ chains, respectively. ξ-Chain synthesis persists beyond the embryonic stage of development in some of the α-thalassemias. Persistent ε-chain production has not been found in any of the thalassemia syndromes. During fetal development, an orderly switch from ξ- to α-chain and from ε- to γ-chain production occurs, followed by β- and δ-chain production after birth.

Figure 48–3 shows the different human hemoglobins and the arrangements of the α-gene cluster on chromosome 16 and the β-gene cluster on chromosome 11.

Globin Gene Clusters

Although some individual variability exists, the α-gene cluster usually contains one functional ξ gene and two α genes, designated α₂ and α₁. It also contains four pseudogenes: ψξ₁, ψα₁, ψα₂, and θ₁.^9,10 These four pseudogenes are remarkably conserved among different species. Although it appears to be expressed early in fetal life, its function is unknown. It likely does not produce a viable globin chain. Each α gene is located in a region of homology approximately 4 kb long, interrupted by two small nonhomologous regions.^25,26,27 The homologous regions are believed to result from gene duplication, and the nonhomologous segments are believed to arise subsequently by insertion of DNA into the noncoding regions around one of the two genes. The exons of the two α-globin genes have identical sequences. The first intron in each gene is identical. The second intron of α₁ is nine bases longer and differs by three bases from that in the α₂ gene.^27,28,29 Despite their high degree of homology, the sequences of the two α-globin genes diverge in their 3′ untranslated regions 13 bases beyond the TAA stop codon. These differences provide an opportunity to assess the relative output of the genes, an important part of the analysis of the α-thalassemias.^30,31 Production of α₂ messenger RNA appears to exceed that of α₁ by a factor of 1.5 to 3. ψξ₁ and ξ₂ genes also are highly homologous. The introns are much larger than those of α-globin genes. In contrast to the latter, IVS-1 is larger than IVS-2. In each ξ gene, IVS-1 contains several copies of a simple repeated 14-bp sequence that is similar to sequences located between the two ξ genes and near the human insulin gene. The coding sequence of the first exon of ψξ₁ contains three base changes, one of which gives rise to a premature stop codon, thus making ψξ₁ an inactive pseudogene.

The regions separating and surrounding the α-like structural genes have been analyzed in detail. Of particular relevance to thalassemia is the polymorphic nature of this gene cluster.³² The cluster contains five hypervariable regions: one downstream from the α₁ gene, one between the ξ and ψξ genes, one in the first intron of both the ξ and ψξ genes, and one 5′ to the cluster. These regions consist of varying numbers of tandem repeats of nucleotide sequences. Taken together with single-base restriction fragment length polymorphisms (RFLPs), the variability of the α-globin gene cluster reaches a heterozygosity level of approximately 0.95. Thus, each parental α-globin gene cluster can be identified in the majority of persons. This heterogeneity has important implications for tracing the history of the thalassemia mutations.

Figure 48–3 shows the arrangement of the β-globin gene cluster on the short arm of chromosome 11. Each of the individual genes and their flanking regions have been sequenced.^33,34,35,36 Like the α₁ and α₂ gene pairs, the ^Gγ and ^Aγ genes share a similar sequence. In fact, the ^Gγ and ^Aγ genes on one chromosome are identical in the region 5′ to the center of the large intron yet show some divergence 3′ to that position. At the boundary between the conserved and divergent regions, a block of simple sequence may be a “hot spot” for initiation of recombination events that lead to unidirectional gene conversion.

Like the α-globin genes, the β-gene cluster contains a series of single-point RFLPs, although in this case no hypervariable regions have been identified.^37,38 The arrangement of RFLPs, or haplotypes, in the β-globin gene cluster falls into two domains. The 5′ side of the β gene, spanning approximately 32 kb from the ε gene to the 3′ end of the ψβ gene, contains three common patterns of RFLPs. The region encompassing about 18 kb to the 3′ side of the β-globin gene also contains three common patterns in different populations. Between these regions is a sequence of about 11 kb in which there is randomization of the 5′ and 3′ domains; hence, a relatively higher frequency of recombination can occur.³⁸ The β-globin gene haplotypes are similar in most populations but differ markedly in individuals of African origin. These findings suggest the haplotype arrangements were laid down very early during evolution. The findings are consistent with data obtained from mitochondrial DNA polymorphisms pointing to the early emergence of a relatively small population from Africa with subsequent divergence into other racial groups.³⁹ Again, they are extremely useful for analyzing the population genetics and history of the thalassemia mutations.

The regions flanking the coding regions of the globin genes contain a number of conserved sequences essential for their expression.^28,33 The first conserved sequence is the TATA box, which serves accurately to locate the site of transcription initiation at the CAP site, usually about 30 bases downstream. It also appears to influence the rate of transcription. In addition, two so-called upstream promoter elements are present. A second conserved sequence, the CCAAT box, is located 70 or 80 bp upstream. The third conserved sequence, the CACCC homology box, is located further 5′, approximately 80 to 100 bp from the CAP site. It can be either inverted or duplicated. These promoter sequences also are required for optimal transcription. Mutations in this region of the β-globin gene cause its defective expression and these findings provide the foundation for understating regulation of other human genes. The globin genes also have conserved sequences in their 3′ flanking regions, notably AATAAA, which is the polyadenylation signal site.

Regulation of Globin Gene Clusters

Figure 48–4 summarizes the mechanism of globin gene expression. The primary transcript is a mRNA precursor containing both intron and exon sequences. During its stay in the nucleus, it undergoes a good deal of processing that entails capping the 5′ end and polyadenylation of the 3′ end, both of which probably serve to stabilize the transcript (Chap. 10). The intervening sequences are removed from the mRNA precursor in a complex two-stage process that relies on certain critical sequences at the intron–exon junctions.

The method by which globin gene clusters are regulated is important to understanding the pathogenesis of the thalassemias. Many details remain to be determined, but studies performed over the last few years have provided at least an outline of some of the major mechanisms of globin gene regulation.^7,9,40,41,42

Most of the DNA within cells that is not involved in gene transcription is packaged into a compact form that is inaccessible to transcription factors and RNA polymerase. Transcriptional activity is characterized by a major change in the structure of the chromatin surrounding a particular gene. These alterations in chromatin structure can be identified by enhanced sensitivity to exogenous nucleases. Erythroid lineage-specific nuclease-hypersensitive sites are found at several locations in the β-globin gene cluster, which vary during different stages of development. In fetal life, these sites are associated with the promoter regions of all four globin genes. In adult erythroid cells, the sites associated with the γ genes are absent. The methylation state of the genes plays an important role in their ability to be expressed. In human and other animal tissues, the globin genes are extensively methylated in nonerythroid organs and are relatively undermethylated in hematopoietic tissues. Changes in chromatin configuration around the globin genes at different stages of development are reflected by alterations in their methylation state.

In addition to the promoter elements, several other important regulatory sequences have been identified in the globin gene clusters. For example, several enhancer sequences thought to be involved with tissue-specific expression have been identified. Their sequences are similar to the upstream activating sequences of the promoter elements. Both consist of a number of “modules,” or motifs, that contain binding sites for transcriptional activators or repressors. The enhancer sequences are thought to act by coming into spatial apposition with the promoter sequences to increase the efficiency of transcription of particular genes. It now is clear that transcriptional regulatory proteins may bind to both the promoter region of a gene and to the enhancer. Some of these transcriptional proteins, GATA-1 and NFE-2, for example, appear to be largely restricted to hematopoietic tissues.⁴⁰ These proteins may bring the promoter and the enhancer into close physical proximity, permitting transcription factors bound to the enhancer to interact with the transcriptional complex that forms near the TATA box. At least some of these hematopoietic gene transcription factors likely will be developmental-stage specific.

Another set of erythroid-specific nuclease-hypersensitive sites is located upstream from the embryonic globin genes in both the α- and β-gene clusters. These sites mark the regions of particularly important control elements. In the case of the β-globin gene cluster, the region is marked by five hypersensitive sites to DNase I treatment (HS) (an enzyme used to detect DNA-protein interaction).⁴⁰ The most 5′ site (HS5) does not show tissue specificity. HS1 through HS4, which together form the locus control region (LCR), are largely erythroid-specific. Each of the regions of the LCR contains a variety of binding sites for erythroid transcription factors. The precise function of the LCR is not known, but it is undoubtedly required to establish a transcriptionally active domain spanning the entire globin gene cluster. The α-globin gene cluster also has a major regulatory element of this kind, in this case HS40.⁴¹ This forms part of four highly conserved noncoding sequences, or multispecies conserved sequences (MCSs), called MSC-R1-R4; of these elements only MSC-R2, that is HS40, is essential for α-globin gene expression. Although deletions of this region inactivate the entire α-globin gene cluster, its action must be fundamentally different from that of the β-globin LCR because the chromatin structure of the α-gene cluster is in an open conformation in all tissues.

Some forms of thalassemia result from deletions involving these regulatory regions. In addition, the phenotypic effects of deletions of these gene clusters are strongly positional, which may reflect the relative distance of particular genes from the LCR and HS40.

Developmental Changes in Globin Gene Expression

One particularly important aspect of human globin genes is regulation of the switch from fetal to adult hemoglobin. Because many of the thalassemias and related disorders of the β-globin gene cluster are associated with persistent γ-chain synthesis, a full understanding of their pathophysiology must include an explanation for this important phenomenon, which plays a considerable role in modifying their phenotypic expression.

The complex topic of hemoglobin switching has been the subject of several extensive reviews.^7,42 β-Globin synthesis commences early during fetal life, at approximately 8 to 10 weeks’ gestation. β-Globin synthesis continues at a low level, approximately 10 percent of the total non–α-globin chain production, up to approximately 36 weeks’ gestation, after which it is considerably augmented. At the same time, γ-globin chain synthesis starts to decline so that, at birth, approximately equal amounts of γ- and β-globin chains are produced. Over the first year of life, γ-chain synthesis gradually declines. By the end of the first year, γ-chain synthesis amounts to less than 1 percent of the total non–α-globin chain output. In adults the small amount of hemoglobin F is confined to an erythrocyte population called F cells.

How this series of developmental switches is regulated is not clear. The process is not organ specific but is synchronized throughout the developing hematopoietic tissues. Although environmental factors may be involved, the bulk of experimental evidence suggests some form of “time clock” is built into the hematopoietic stem cell. At the chromosomal level, regulation appears to occur in a complex manner involving both developmental stage-specific trans-activating factors and the relative proximity of the different genes of the β-globin gene cluster to LCR. Some of the elements involved in the stage-specific regulation of human globin genes have been identified. KLF1 (erythroid Kruppel-like factor), a developmental stage–enriched protein, activates human β-globin gene expression and is involved in human γ- to β-globin gene switching.⁴³ More recently BCL11A and MYB have also been identified as being involved in this process.⁴²

Fetal hemoglobin synthesis can be reactivated at low levels in states of hematopoietic stress and at higher levels in certain hematologic malignancies, notably juvenile myeloid leukemia. However, high levels of hemoglobin F production are seen consistently in adult life only in the hemoglobinopathies.

Once cloning and sequencing of globin genes from patients with many different forms of thalassemia were possible, the wide spectrum of mutations underlying these conditions became clear. A picture of remarkable heterogeneity has emerged. For more extensive coverage of this topic, the reader is referred to several monographs and reviews.^{7,9,10,44,45,46}

β-THALASSEMIA

β-thalassemia is extremely heterogeneous at the molecular level.⁷ More than 200 different mutations have been found in association with the β-thalassemia phenotype.⁷ Broadly, they fall into deletions of the β-globin gene and nondeletional mutations that may affect the transcription, processing, or translation of β-globin messenger (Table 48–2 and Fig. 48–5). Each major population group has a different set of β-thalassemia mutations, usually consisting of two or three mutations forming the bulk and large numbers of rare mutations. Because of this distribution pattern, only approximately 20 alleles account for the majority of all β-thalassemia determinants (see Fig. 48–1).

Gene Deletions

At least 17 different deletions affecting only the β genes have been described. With one exception, the deletions are rare and appear to be isolated, single events. The 619-bp deletion at the 3′ end of the β gene is more common,⁴⁸ but even that is restricted to the Sind and Gujarati populations of Pakistan and India, where it accounts for approximately 50 percent of β-thalassemia alleles.⁴⁸ The Indian 619-bp deletion removes the 3′ end of the β gene but leaves the 5′ end intact. Many of the other deletions remove the 5′ end of the gene and leave the δ gene intact.^{49,50,51,52,53} Homozygotes for these deletions have β⁰-thalassemia. Heterozygotes for the Indian deletion have increased hemoglobin A₂ and F levels identical to those seen in heterozygotes for the other common forms of β-thalassemia. Heterozygotes for the other deletions all have unusually high hemoglobin A₂ levels.⁷ Increased δ-chain production results from increased δ-gene transcription in cis to the deletion, possibly as a result of reduced competition from the deleted 5′ β gene for transcription factors.

Other Transcriptional Mutations

Several different base substitutions involve the conserved sequences upstream from the β-globin gene.⁷ In every case, the phenotype is β⁺-thalassemia, although considerable variability exists in the clinical severity associated with different mutations of this type. Several mutations, at positions –88 and –87 relative to the mRNA CAP site, for example,^54,55 are close to the CCAAT box, whereas others lie within the TATA box homology.^56,57,58,59

Some mutations upstream from the β-globin gene are associated with even more subtle alterations in phenotype. For example, a C→T substitution at position –101, which involves one of the upstream promoter elements, is associated with “silent” β-thalassemia, that is, a completely normal (“silent”) phenotype that can be identified only by its interaction with more severe forms of β-thalassemia in compound heterozygotes.⁶⁰ A single example of an A→C substitution at the CAP site (+1) was described in an Asian Indian who, despite being homozygous for the mutation, appeared to have the phenotype of the β-thalassemia trait.⁶¹

Upstream regulatory mutations confirm the importance of the role of conserved sequences in this region as regulators of the transcription of the β-globin genes and provide the basis for some of the mildest forms of β-thalassemia, particularly those in African populations, and for some varieties of “silent” β-thalassemia.

RNA-Processing Mutations

One surprise about β-thalassemia has been the remarkable diversity of the single-base mutations that can interfere with the intranuclear processing of mRNA.

The boundaries of exons and introns are marked by invariant dinucleotides, GT at the 5′ (donor) and AG at the 3′ (receptor) sites. Single-base changes that involve either of these splice junctions totally abolish normal RNA splicing and result in the β⁰-thalassemia phenotype.^{7,62,63,64,65,66}

Highly conserved sequences involved in mRNA processing surround the invariant dinucleotides at the splice junctions. Different varieties of β-thalassemia involve single-base substitutions within the consensus sequence of the IVS-1 donor site.^{55,58,63,64,65,66,67,68,69} These mutations are particularly interesting because of the remarkable variability in their associated phenotypes. For example, substitution of the G in position 5 of IVS-1 by C or T results in severe β⁺-thalassemia.⁵⁵ On the other hand, a T→C change at position 6, found commonly in the Mediterranean region,⁷⁰ results in a very mild form of β⁺-thalassemia. The G→C change at position 5 has also been found in Melanesia and appears to be the most common cause of β-thalassemia in Papua New Guinea.⁷¹

RNA processing is affected by mutations that create new splice sites within either introns or exons. Again, these lesions are remarkably variable in their phenotypic effect, depending on the degree to which the new site is utilized compared with the normal splice site. For example, the G→A substitution at position 110 of IVS-1, which is one of the most common forms of β-thalassemia in the Mediterranean region, leads to only approximately 10 percent splicing at the normal site and hence results in a severe β⁺-thalassemia phenotype.^72,73 Similarly, a mutation that produces a new acceptor site at position 116 in IVS-1 results in little or no β-globin mRNA production and the β⁰-thalassemia phenotype.⁷⁴ Several mutations that generate new donor sites within IVS-2 of the β-globin gene have been described.^55,68

Another mechanism for abnormal splicing is activation of donor sites within exons (Fig. 48–6). For example, within exon 1 is a cryptic donor site in the region of codons 24 through 27. This site contains a GT dinucleotide. An adjacent substitution that alters the site so that it more closely resembles the consensus donor splice site results in its activation, even though the normal site is active. Several mutations in this region can activate this site so that it is utilized during RNA processing, with the production of abnormal mRNAs.^75,76,77,78 Three of the substitutions—A→G in codon 19, G→A in codon 26, and G→T in codon 27—result in reduced production of β-globin mRNA and an amino acid substitution so that the mRNA that is spliced normally is translated into protein. The abnormal hemoglobins produced are hemoglobins Malay, E, and Knossos, respectively, all of which are associated with a β-thalassemia phenotype, presumably as a result of reduced overall output of normal mRNA (Fig. 48–6). A variety of other cryptic splice mutations within introns and exons have been described.⁴⁴

Another class of processing mutations involves the polyadenylation signal site AAUAAA in the 3′ untranslated region of β-globin mRNA.^79,80,81 For example, a T→C substitution in this sequence leads to only one-tenth the normal amount of β-globin mRNA and hence the severe β⁺-thalassemia phenotype.⁷⁹

Mutations Causing Abnormal Translation of Messenger RNA

Base substitutions that change an amino acid codon into a chain termination codon, that is, nonsense mutations, prevent translation of the mRNA and result in β⁰-thalassemia. Many substitutions of this type have been described.^7,44 For example, a codon 17 mutation is common in Southeast Asia,^82,83 and a codon 39 mutation occurs at a high frequency in the Mediterranean region.^84,85

The insertion or deletion of one, two, or four nucleotides in the coding region of the β-globin gene disrupts the normal reading frame and results, upon translation of the mRNA, in the addition of anomalous amino acids until a termination codon is reached in the new reading frame. Several frameshift mutations of this type have been described.^7,44 Two mutations—the insertion of one nucleotide between codons 8 and 9 and a deletion of four nucleotides in codons 41 and 42—are common in Asian Indians.⁶³ The latter deletions are found frequently in different populations in Southeast Asia.⁸³

An unusual β⁺-thalassemia was described in a patient from the Czech Republic in whom a full-length L1 transposon was inserted into the second intron of β-globin, creating a β⁺-thalassemia phenotype by an undefined molecular mechanism.⁸⁶

Dominantly Inherited β-Thalassemia

Families in which a picture indistinguishable from moderately severe β-thalassemia has segregated in mendelian dominant fashion have been reported sporadically.^87,88 Because this condition often is characterized by the presence of inclusion bodies in the red cell precursors, it has been called inclusion body β-thalassemia. However, because all severe forms of β-thalassemia have inclusions in the red cell precursors, the term dominantly inherited β-thalassemia is preferred.^7,89 Sequence analysis has shown that these conditions are heterogeneous at the molecular level, but that many involve mutations of exon 3 of the β-globin gene. The mutations include frameshifts, premature chain termination mutations, and complex rearrangements that lead to synthesis of truncated or elongated and highly unstable β-globin gene products.^{7,89,90,91,92,93} The most common mutation of this type is a GAA→TAA change at codon 121 that leads to synthesis of a truncated β-globin chain.⁹⁴ Although an abnormal β-chain product from loci affected by mutations of this type is unusual, many of these conditions are designated as hemoglobin variants.

The reason why mutations occurring in exons 1 and 2 produce the classic form of recessive β-thalassemia whereas the bulk of the dominant thalassemias result from mutations in exon 3 has become clearer. In the former case, very little abnormal β-globin mRNA is found in the cytoplasm of the red cell precursors, whereas exon 3 mutations are associated with full-length but abnormal mRNA accumulation. The different phenotypes of these premature termination codons have been suggested to reflect a phenomenon called nonsense-mediated RNA decay, a surveillance system to prevent transport of mRNA coding for truncated peptides. Presumably this process is active in the case of exon 1 or 2 mutations, in which affected mRNAs are degraded, but is not active in the case of exon 3 mutations.^95,96,97 A complete list of the mutations that underlie the dominant β-thalassemias is given in reference 44.

Unstable β-Globin Variants

Some β-globin chain variants are highly unstable but are capable of forming a viable tetramer. The resulting unstable hemoglobins may precipitate in the red cell precursors or in the blood, giving rise to a spectrum of conditions ranging from dominantly inherited β-thalassemia to a hemolytic anemia similar to the anemia associated with other unstable hemoglobins. The first unstable hemoglobin to be described was hemoglobin Indianapolis.⁹⁸ Its structure was characterized by DNA analysis performed on stored autopsy material; however, the original description proved to be incorrect.⁹⁹

Silent β-Thalassemia

A number of extremely mild β-thalassemia alleles are either silent or almost unidentifiable in heterozygotes (see Table 48–2). Some alleles are in the region of the promoter boxes of the β-globin gene, but others involve the CAP sites or the 5′ or 3′ untranslated regions.^7,44 These alleles usually are identified by finding a form of β-thalassemia intermedia in which one parent has a typical thalassemia trait and the other parent appears to be normal but, in fact, is a carrier of one of the mild β-thalassemia alleles.

β-Thalassemia Mutations Unlinked to the B-Globin Gene Cluster

Several family studies suggest the existence of mutations that result in the β-thalassemia phenotype but do not segregate with the β-globin genes¹⁰⁰; however, their molecular basis has not been determined. Further evidence for the existence of novel mutations of this type can be found in reference 7.

Variant Forms of β-Thalassemia

In several forms of β-thalassemia, the hemoglobin A₂ level is normal in heterozygotes. Some cases result from “silent” β-thalassemia alleles, whereas others reflect the coinheritance of β- and δ-thalassemia.⁷

δβ-THALASSEMIA

The δβ-thalassemias are classified into the (δβ)⁺- and (δβ)⁰-thalassemias (Table 48–3). The (δβ)⁰-thalassemias are further divided into (δβ)⁰-thalassemia, in which both the δ- and β-globin genes are deleted, and (^Aγδβ)⁰-thalassemia, in which the ^Gγ, δ, and β genes are deleted. Because many different deletion forms of δβ-thalassemia have been described, they are further classified according to the country in which they were first identified (Table 48–3).

(δβ)⁰- and (^Aγδβ)⁰-Thalassemia

Nearly all these conditions result from deletions involving varying lengths of the β-globin gene cluster. Many different varieties have been described in different populations (see Table 48–3), although their heterozygous and homozygous phenotypes are very similar.⁷ Rare forms of these conditions result from more complex gene rearrangements. For example, one form of (^Aγδβ)⁰-thalassemia, found in Indian populations, does not result from a simple linear deletion but rather from a complex rearrangement with two deletions, one affecting the ^Aγ gene and the other the δ and β genes. The intervening region is intact but inverted.¹⁰¹ Figure 48–7 illustrates some of these conditions.

(δβ)⁺-Thalassemia

The (δβ)⁺-thalassemias usually are associated with the production of structural hemoglobin variants called Lepore.¹⁰² Hemoglobin Lepore contains normal α chains and non-α chains that consist of the first 50 to 80 amino acid residues of the δ chains and the last 60 to 90 residues of the normal C-terminal amino acid sequence of the β chains. Thus, the Lepore non-α chain is a β-fusion chain. Several different varieties of hemoglobin Lepore have been described—Washington-Boston, Baltimore, and Hollandia—in which the transition from δ to β sequences occurs at different points.⁷ The fusion chains probably arose by nonhomologous crossing over between part of the δ locus on one chromosome and part of the β locus on the complementary chromosome (Fig. 48–8). This event results from misalignment of chromosome pairing during meiosis so that a δ-chain gene pairs with a β-chain gene instead of with its homologous partner.¹⁰³ Figure 48–8 shows such a mechanism should give rise to two abnormal chromosomes: the first, the Lepore chromosome, will have no normal δ or β loci but simply a δβ fusion gene. Opposite the homologous pairs of chromosomes should be an anti-Lepore (βδ) fusion gene and normal δ and β loci. A variety of anti–Lepore-like hemoglobins have been discovered, including hemoglobins Miyada, P-Congo, Lincoln Park, and P-Nilotic.⁷ All the hemoglobin Lepore disorders are characterized by a severe form of δβ-thalassemia. The output of the γ-globin genes on the chromosome with the δβ fusion gene is not increased sufficiently to compensate for the low output of the δβ fusion product. The reduced rate of production of the δβ fusion chains of hemoglobin Lepore presumably reflects the fact that its genetic determinant has the δ gene promoter region, which is structurally different from the β-globin gene promoter and is associated with a reduced rate of transcription of its gene product.

δβ-Thalassemia-Like Disorders Resulting from Two Mutations in the β-Globin Gene Cluster

A heterogeneous group of nondeletion δβ-thalassemias has been described, most resulting from two mutations in the εγδβ-globin gene cluster (see Table 48–3). Strictly speaking, they are not all δβ-thalassemias, but they often appear in the literature under this title because their phenotypes resemble the deletion forms of (δβ)⁰-thalassemia. In the Sardinian form of δβ-thalassemia, the β-globin gene has the common Mediterranean codon 39 nonsense mutation that leads to an absence of β-globin synthesis. The relatively high expression of the ^Aγ gene in cis gives this condition the δβ-thalassemia phenotype because of a point mutation at position –196 upstream from the ^Aγ gene (see “Hereditary Persistence of Fetal Hemoglobin” below). The phenotypic picture, in which heterozygotes have 15 to 20 percent hemoglobin F and normal hemoglobin A₂ levels, is identical to that of δβ-thalassemia.¹⁰³ Another condition having the β-thalassemia phenotype, with greater than 20 percent hemoglobin F in heterozygotes, has been described in a Chinese patient in whom defective β-globin chain synthesis appears to result from an A→G change in the ATA sequence in the promoter region of the β-globin gene.¹⁰⁴ The increased γ-chain synthesis, which appears to involve both ^Gγ and ^Aγ cis to this mutation, remains unexplained. A disorder originally called δβ-thalassemia has been described in the Corfu population.^105,106 The condition results from two mutations in the β-globin gene cluster: first, a 7201-bp deletion that starts in the δ-globin gene, IVS-2, position 818 to 822, and extends upstream to a 5′ breakpoint located 1719 to 1722 bp 3′ to the ψβ-gene termination codon; and second, a G→A mutation at position 5 in the donor site consensus region of IVS-1 of the β-globin gene. The output from this chromosome consists of relatively high levels of γ chains with very low levels of β chains. The condition resembles δβ-thalassemia in the homozygous state, with almost 100 percent hemoglobin F, traces of hemoglobin A, but no hemoglobin A₂. Heterozygotes have only slightly elevated hemoglobin F levels, with a phenotype similar to “normal A₂β-thalassemia.”

εγδβ-THALASSEMIA

These rare conditions^{107,108,109,110,111,112,113} result from long deletions that begin upstream from the β-gene complex 55 kb or more 5′ to the ε gene and terminate within the cluster (see Fig. 48–7). In two cases, designated Dutch^110,111 and English,¹¹² the deletions leave the β-globin gene intact, but no β-chain production occurs even though the gene is expressed in heterologous systems.

The molecular basis for inactivation of the β-globin gene cis to these deletions was clarified by the discovery of the LCR approximately 50 kb upstream from the εγδβ-globin gene cluster (see “Genetic Control and Synthesis of Hemoglobin” above). Removal of this critical regulatory region seems to completely inactivate the downstream globin gene complex. The Hispanic form of εγδβ-thalassemia¹¹³ results from a deletion that includes most of the LCR, including four of the five DNase-1-hypersensitive sites. These lesions appear to close down the chromatin domain that usually is open in erythroid tissues and delay replication of the β-globin genes in the cell cycle. Thus, although they are rare, the lesions have been of considerable importance because analysis of the Dutch deletion first pointed to the possibility of a major control region upstream from the β-like-globin gene cluster and ultimately led to the discovery of the β-globin LCR.

HEREDITARY PERSISTENCE OF FETAL HEMOGLOBIN

This heterogeneous group of conditions produces phenotypes very similar to those of the δβ-thalassemias, except that defective β-chain production appears to be almost, but in some forms not completely, compensated by persistent γ-chain production. These conditions are best classified into deletion and nondeletion forms (Table 48–4). In the past, the conditions were classified into pancellular and heterocellular varieties, depending on the intercellular distribution of fetal hemoglobin. However, this subdivision now appears to bear little relevance to their molecular basis and probably relates more to the particular level of fetal hemoglobin and how its cellular distribution is determined.⁷

The deletion forms of HPFH are heterogeneous (see Fig. 48–7). The two African varieties result from extensive deletions of similar length (<70 kb) but with staggered ends, differing phenotypically only in the proportions of ^Gγ and ^Aγ chains produced.¹¹⁴ Another type of HPFH results from misalignment during crossing over between the ^Aγ- and β-globin genes, resulting in production of ^Aγβ fusion genes (see Fig. 48–8) that combine with α chains to form the hemoglobin variant called hemoglobin Kenya.^115,116 Hemoglobin Kenya is associated with an increased output of hemoglobin F, although at a lower level than in the deletion forms of HPFH. A theory that adequately explains the phenotypic differences between δβ-thalassemia and the deletion forms of HPFH has not been developed.⁷

The nondeletion determinants of HPFH can be classified into those that map within the β-globin gene cluster and those that segregate independently. The former are subdivided into ^Gγβ⁺ and ^Aγβ⁺ varieties, indicating persistent ^Gγ- or ^Aγ-chain synthesis in association with β-globin production directed by the β gene cis (on the same chromosome) to the HPFH determinant. Analysis of the overexpressed γ genes revealed in each case a single-base substitution in the region immediately upstream from the transcription start site.^{7,117,118,119,120} Clustering of these substitutions and lack of similar changes in normal γ genes suggest they are responsible for persistent hemoglobin F production (Fig. 48–9). This region of DNA likely is involved in binding of trans-acting proteins involved in the normal developmental repression of γ-gene expression, either by decreasing the affinity for an inhibitory factor normally present in adult life or by increasing the affinity for a factor promoting gene expression. The most common of these conditions are Greek ^Aγβ⁺ HPFH and a form of ^Gγβ⁺ HPFH, which has been found in several different African populations. If the upstream point mutations associated with persistent γ-chain production occur on the same chromosome as β-globin genes that carry β⁰-thalassemia mutations, the clinical phenotype is converted from HPFH to δβ-thalassemia, albeit with different hemoglobin A₂ levels.

In some cases, other nondeletional forms of HPFH have been related to small structural changes in the β-globin gene cluster (see Table 48–4). Although strictly speaking not a true form of HPFH, because even in homozygotes it may not be associated with increased hemoglobin F levels, the T→C polymorphism at position –158 to the ^Gγ-globin gene¹²¹ might be associated with an increased output of hemoglobin F under conditions of erythropoietic stress.

Other forms of HPFH are characterized by the persistence of low levels of fetal hemoglobin production distributed in a heterocellular manner. In all populations studied, a small proportion of individuals have an increased amount of hemoglobin F and F cells, that is, red cells that can be detected when blood films are treated with antibodies against hemoglobin F. Although this condition originally was called the Swiss form of HPFH because it was first recognized in Swiss army recruits,¹²² it is observed in every racial group. Using a variety of genetic approaches, it has become clear that a number of genes may be involved in the generation of heterocellular HPFH, including loci at Xp22.2-p22.3,6q23,8q, and 2p15^{123,124,125,126,127,128}; the latter linkage has been identified as the oncogene BCL11A. The mechanism whereby these different loci affect the level of F cells in normal individuals and increase their levels in conditions like thalassemia and sickle cell anemia remain to be determined, but their coinheritance with these conditions may have an extremely beneficial effect of their associated phenotypes.¹²⁹

δ-THALASSEMIA

Several point mutations and deletions that reduce δ-globin synthesis have been described. They are summarized in reference 7.

α-THALASSEMIA

Table 48–5 summarizes the different classes of α-thalassemia mutations. The α-globin gene haplotype can be written αα, indicating the α₁ and α₂ genes, respectively. A normal individual has the genotype αα/αα. A deletion involving one (–α) or both (– –) α genes can be further classified based on its size, written as a superscript; thus, –α^3.7 indicates a deletion of 3.7 kb including one α gene. When the sizes of the deletions are not established, a superscript describing their geographic or family origin is useful; thus, – – ^MED describes a deletion of both α genes first identified in individuals of Mediterranean origin. In thalassemia haplotypes in which both genes are intact, that is, nondeletion lesions, the nomenclature α^NDα is given, with the superscript ND indicating the gene is thalassemic. However, when the precise molecular defect is known, as in hemoglobin Constant Spring, for example, α^NDα can be replaced by the more informative α^CSα. The molecular pathology and population genetics of the α-thalassemias have been the subject of several extensive reviews.^{7,41,45,130,131}

α⁰-Thalassemia

Many deletions that involve both α genes, and therefore abolish α-chain production from the affected chromosome, have been described (Fig. 48–10).⁷ Several of the 3′ breakpoints fall within a 6- to 8-kb region at the 3′ end of the α-globin complex, suggesting this represents a breakpoint cluster region with a high level of recombination.¹³² In at least five of the deletions, the 5′ breakpoints also appear to cluster. This gives rise to a situation in which the 5′ breakpoints are located approximately the same distance apart and in the same order along a chromosome as their respective 3′ breakpoints. It is possible that such staggered deletions arise from illegitimate recombination events that delete an integral number of chromatin loops as they pass through their nuclear attachment points during replication. This mechanism has also been suggested to underlie some of the deletion forms of HPFH. One of these deletions (– –^MED) involves a more complex rearrangement that introduces a new piece of DNA bridging the two breakpoints in the α-gene cluster. This new sequence originates upstream from the α cluster and appears to have been replicated into the junction in a manner suggesting that the upstream segment of DNA also lies at the base of a replication loop. At least some of these deletions seem to have arisen by recombination events between Alu repeat sequences.

Several other mechanisms for the generation of α⁰-thalassemia have been identified. In one case of unusual genetic interest, a long (>18 kb) deletion that removes the α₁ gene and the region downstream was identified in which the α₂ gene remains intact but is completely inactivated, giving the α⁰-thalassemia phenotype. Although the inactive α₂ gene retains all its local and remote cis-regulatory elements, its expression is completely silenced and its CpG island is completely methylated as a result of transcription of antisense RNA expressed from a locus that had been juxtaposed to the α₂ gene because of the large deletion.^133,134 In some cases, this condition results from a terminal truncation of the short arm of chromosome 16 to a site 50 kb distal to the α-globin genes.¹³⁵ It is interesting that the telomeric consensus sequence (TTAGGGG)n has been added directly to the site of the break. Because this mutation is stably inherited, telomeric DNA alone appears sufficient to stabilize the broken chromosome end. This observation raises the possibility that other genetic diseases result from chromosomal truncations.

Several deletions have been identified that appear to downregulate α-globin genes by removing the α-globin LCR (HS40).^7,136,137 In each case, the α-globin genes are left intact, although in one the 3′ breakpoint is found between the ξ and ψξ genes, thus removing the ξ gene. These deletions appear to completely inactivate the α-globin gene complex, just as deletions of the β-globin LCR inactivate the entire β-gene complex. Such deletions have not been observed in the homozygous state, presumably because they would be lethal.

α⁺-Thalassemia Gene Deletions

The most common forms of α⁺-thalassemia (–α^3.7 and –α^4.2) involve deletion of one or the other of the duplicated α-globin genes (see Figs. 48–10 and 48–11).

Each α gene is located within a region of homology approximately 4 kb long, interrupted by two nonhomologous regions. The homologous regions are believed to have resulted from an ancient duplication event and to have subsequently subdivided, presumably by insertions and deletions, to give three homologous subsegments referred to as X, Y, and Z (see Fig. 48–11). The duplicated Z boxes are 3.7 kb apart, and the X boxes are 4.2 kb apart. Misalignment and reciprocal crossover between these segments at meiosis can give rise to chromosomes with either single (–α) or triplicated (ααα) α-globin genes. Such an occurrence between homologous Z boxes deletes 3.7 kb of DNA (rightward deletion). A similar crossover between the two X blocks deletes 4.2 kb of DNA (leftward deletion –α^4.2).¹³⁸ The corresponding triplicated α-gene arrangements are referred to as ααα^anti–3.7 and α^anti–4.2.^139–141 More detailed analysis of these crossover events indicates they occur more commonly in the Z box. At least three different –α^3.7 deletions have been found, depending on exactly where the crossover occurred.¹⁴² These deletions are designated –α^3.7I, –α^3.7II, and –α^3.7III, respectively. Other, rarer deletions of a single α gene have been observed.⁷

Nondeletion α-Thalassemia

Because expression of the α₂ gene is two to three times greater than expression of the α₁ gene, the finding that most of the nondeletion mutants discovered to date affect predominantly α₂ gene expression is not surprising. Presumably this is ascertainment bias because of the greater phenotypic effect of these lesions. It also is possible that defective expression of the α₂ gene has come under greater selective pressure.

Like the β-thalassemia mutations, α-thalassemia mutations⁷ can be classified according to the level of gene expression they affect (see Table 48–5). Several processing mutations have been identified. For example, a pentanucleotide deletion includes the 5′ splice site of IVS-1 of the α₂-globin gene. This mutation involves the invariant GT donor splicing sequence and thus completely inactivates the α₂ gene.¹⁴³ A second mutant of this type, found commonly in the Middle East, involves the poly-A addition signal site (AATAAA→AATAAG) and downregulates the α₂ gene by interfering with 3′ end processing.^144,145

A second group of nondeletion α-thalassemias results from mutations that interfere with translation of mRNA.⁷ Several mutations involve the initiation codon.^{146,147,148,149} In one case, for example, the initiation codon is inactivated by a T→C transition.¹⁴⁶ In another case, efficiency of initiation is reduced by a dinucleotide deletion in the consensus sequence around the start signal.¹⁴⁹ Five mutations that affect termination of translation and give rise to elongated α chains have been identified: hemoglobins Constant Spring, Icaria, Koya Dora, Seal Rock, and Pakse.⁷ Each mutation specifically changes the termination codon TAA so that an amino acid is inserted instead of the chain terminating (Fig. 48–12). This process is followed by read-through of mRNA that is not normally translated until another “in-phase” stop codon is reached. Thus, each of these variants has an elongated α chain. The “read-through” of α-globin mRNA that usually is not utilized likely reduces its stability.¹⁵⁰ Several nonsense mutations occur, for example, one in exon 3 of the α₂-globin gene.¹⁵¹ Finally, several mutations occur that cause α-thalassemia by producing highly unstable α-globin chains, including hemoglobins Quong Sze,¹⁵² Suan Doc,¹⁵³ Petah Tikvah,¹⁵⁴ and Evanston.¹⁵⁵ A complete list of nondeletion α-thalassemia alleles is given in reference 45.

Interactions of α-Thalassemia Haplotypes

Many α-thalassemia haplotypes have been described, and potentially more than 500 interactions are possible!⁷ These phenotypes result in four broad categories: (1) normal, (2) conditions characterized by mild hematologic changes but no clinical abnormality, (3) hemoglobin H disease, and (4) hemoglobin Bart’s hydrops fetalis syndrome. The heterozygous states for deletion or nondeletion forms of α⁺-thalassemia either cause extremely mild hematologic abnormalities or are completely silent. In populations where α-thalassemia is common, the homozygous state for α⁺-thalassemia (–α/–α) can produce a hematologic phenotype identical to that of the heterozygous state for α⁰-thalassemia (– –/αα), that is, mild anemia with reduced mean cell hemoglobin and mean cell volume values.

Hemoglobin H disease usually results from the compound heterozygous state for α⁰-thalassemia and either deletion or non-deletion α⁺-thalassemia. It occurs most frequently in Southeast Asia (– –^SEA/–α^3.7) and the Mediterranean region (usually – –^MED/–α^3.7).

The hemoglobin Bart’s hydrops fetalis syndrome usually results from the homozygous state for α⁰-thalassemia, most commonly – –^SEA/– –^SEA or – –^MED/– –^MED. A few infants with this syndrome who synthesized very low levels of α chains at birth have been reported. Gene-mapping studies suggest these cases result from interaction of α⁰-thalassemia with nondeletion mutations (αα^ND).

Unusual Forms of α-Thalassemia

Some unusual forms of α-thalassemia are completely unrelated to the common forms of the disease that occur in tropical populations. These conditions, which can occur in any racial groups, include α-thalassemia associated with mental retardation or leukemia. Their importance lies with the diagnostic problems they may present and, more importantly, the light that elucidation of the α-thalassemia pathology may shed on broader disease mechanisms.

Molecular Pathology of the α-Thalassemia Mental Retardation Syndrome

The first descriptions of noninherited forms of α-thalassemia associated with mental retardation suggested the lesions involving the α-globin gene locus were acquired in the paternal germ cells and that their molecular pathology might help elucidate the associated developmental changes.¹⁵⁶ Two separate syndromes of this type now are evident. In one group of patients, long deletions involve the α-globin gene cluster and remove at least 1 Mb.¹⁵⁷ This condition can arise in several ways, including unbalanced translocation involving chromosome 16, truncation of the tip of chromosome 16, and loss of the α-globin gene cluster and parts of its flanking regions by other mechanisms. These findings localize a region of approximately 1.7 Mb in band 16p13.3 proximal to the α-globin genes as being causative of mental handicap.⁴¹

The second group is characterized by defective α-globin synthesis associated with severe mental retardation and a relatively homogeneous pattern of dysmorphology.¹⁵⁸ Extensive structural studies have shown no abnormalities of the α-globin genes. These chromosomes direct the synthesis of normal amounts of α-globin in mouse erythroleukemia cells, suggesting that α-thalassemia results from deficiency of a trans-activating factor involved in regulation of the α-globin genes. This condition is encoded by a locus on the short arm of the X chromosome.¹⁵⁹ ATR-X, the gene involved, is a DNA helicase with many features of a DNA-binding protein. Many different mutations of this gene have been identified in different families with the ATR-X (α-thalassemia X-linked mental retardation) syndrome.^131,160 Studies have identified a plant homeodomain (PHD) region and an adenosine triphosphatase (ATPase)/helicase domain.¹⁶¹ Because patients with ATR-X syndrome show defective methylation of recombinant DNA arrays and related defects, this condition likely is one of a growing list of disorders that result from disordered chromatin remodeling.^162,163

α-Thalassemia and Myelodysplasia

The hematologic findings of hemoglobin H disease or mild α-thalassemia occasionally are observed in elderly patients with myeloid leukemia or the myelodysplastic syndrome. Earlier studies suggested this finding resulted from an acquired defect of α-globin synthesis in which the α-globin genes were completely inactivated in the neoplastic hemopoietic cell line.¹⁶⁴ The molecular basis for this observation now is known to reside in a variety of different mutations involving ATR-X.^41,165 The relationship of these somatic mutations of ATR-X to the neoplastic transformation remains to be determined. The molecular defect of other cases of acquired α-thalassemia, such as that seen in variable combined immunodeficiency,¹⁶⁶ also remains to be defined.

Almost all the pathophysiologic features of the thalassemias can be related to a primary imbalance of globin-chain synthesis. This phenomenon makes the thalassemias fundamentally different from all the other genetic and acquired disorders of hemoglobin production and, to a large extent, explains their extreme severity in the homozygous and compound heterozygous states (Fig. 48–13).

The anemia of β-thalassemia has three major components. First, and most important, is ineffective erythropoiesis with intramedullary destruction of a variable proportion of the developing red cell precursors. Second is hemolysis resulting from destruction of mature red cells containing α-chain inclusions. Third are the hypochromic and microcytic red cells that result from the overall reduction in hemoglobin synthesis.

Because the primary defect in β-thalassemia involves β-chain production, synthesis of hemoglobins F and A₂ should be unaffected. Fetal hemoglobin production in utero is normal. The clinical manifestations of thalassemia appear only when the neonatal switch from γ- to β-chain production occurs. However, fetal hemoglobin synthesis persists beyond the neonatal period in nearly all forms of β-thalassemia (see “Persistent Fetal Hemoglobin Production and Cellular Heterogeneity” below). β-Thalassemia heterozygotes have an elevated level of hemoglobin A₂. The elevated level appears to reflect not only a relative decrease in hemoglobin A as a result of defective β-chain synthesis but also an absolute increase in the output of δ chains both cis and trans to the mutant β-globin gene.⁷

Because α chains are shared by hemoglobins F, A, and A₂, there is no increase in hemoglobin F in the α-thalassemias. The excess γ and β chains formed as a result of defective α-chain production produce soluble homotetramers (see “Mechanisms and Consequences of Erythroid Precursor Damage and Red Cell Damage” below). Hence there is less ineffective erythropoiesis than in β-thalassemia and the major cause of anemia is hemolysis and poorly hemoglobinized red cells.

IMBALANCED GLOBIN-CHAIN SYNTHESIS

Measurements of in vitro globin-chain synthesis in the blood or marrow of patients with different types of thalassemia^167,168 and family studies that allow examination of the action of thalassemia genes in patients who also inherited α- or β-globin structural variants^7,9 provide a clear picture of the action of the thalassemia determinants. In homozygous β-thalassemia, β-globin synthesis is either absent or markedly reduced. The result is excessive production of α-globin chains. α-Globin chains are incapable of forming a viable hemoglobin tetramer, so the chains precipitate in red cell precursors. The resulting inclusion bodies can be demonstrated by both light and electron microscopy.^169,170 In the marrow, precipitation can be seen in the earliest hemoglobinized precursors and throughout the erythroid maturation pathway.¹⁷¹ These large inclusions are responsible for intramedullary destruction of red cell precursors and hence for the ineffective erythropoiesis characterizing all the β-thalassemias. A large proportion of the developing erythroblasts are destroyed within the marrow in severe cases.¹⁷² Any red cells that are released are prematurely destroyed by mechanisms that are considered below in “Mechanisms and Consequences of Erythroid Precursor and Red Cell Damage.” β-Thalassemia heterozygotes also have imbalanced globin-chain synthesis, but the magnitude of α-chain excess is much less and presumably can be resolved by the proteolytic enzymes of the red cell precursors.¹⁷³ Notwithstanding, a mild degree of ineffective erythropoiesis occurs.

Although there is marked globin-chain imbalance in the severe α-thalassemias,^7,167 the excess γ and β chains form homotetramers that do not precipitate in the red cell precursors to the same extent as excess α chains in β-thalassemia. Hence the pathophysiology of anemia is fundamentally different between the two conditions.

MECHANISMS AND CONSEQUENCES OF ERYTHROID PRECURSOR AND RED CELL DAMAGE

Damage to the red cell membrane by the globin-chain precipitation process occurs by two major routes: generation of hemichromes (Chap. 49) from excess α chains with subsequent structural damage to the red cell membrane, and similar damage mediated through the degradation products of excess α chains.^{7,174,175,176} The degradation products of free α chains—globin, heme, hemin (oxidized heme), and free iron—also play a role in damaging red cell membranes. Excess globin chains bind to different membrane proteins and alter their structure and function. Excess iron, by generating oxygen free radicals, damages several red cell membrane components (including lipids and protein) and intracellular organelles. Heme and its products can catalyze the formation of a variety of reactive oxygen species that can damage the red cell membrane. These changes are reflected in an increased rate of apoptosis of red cell precursors.¹⁷⁷ The red cells are rigid and underhydrated, leak potassium, and have increased levels of calcium and low, unstable levels of ATP. Damage to the red cells can also be mediated by the presence of rigid inclusion bodies during passage of the red cells through the spleen.

The consequences of excess non–α-chain production in the α-thalassemias are quite different. Because α chains are shared by both fetal and adult hemoglobin (Chaps. 6 and 48), defective α-chain production is manifest in both fetal and adult life. In the fetus, it leads to excess γ-chain production; in the adult, it leads to an excess of β chains. Excess γ chains form γ₄ homotetramers or hemoglobin Bart’s¹⁷⁸; excess β chains form β₄ homotetramers or hemoglobin H.¹⁷⁹ The fact that γ and β chains form homotetramers is the reason for the fundamental difference in the pathophysiology of α- and β-thalassemia. Because γ₄ and β₄ tetramers are soluble, they do not precipitate to any significant degree in the marrow, and therefore the α-thalassemias are not characterized by severe ineffective erythropoiesis. However, β₄ tetramers precipitate as red cells age, with the formation of inclusion bodies. Thus, the anemia of the more severe forms of α-thalassemia in the adult results from a shortened survival of red cells consequent to their damage in the microvasculature of the spleen as a result of the presence of the inclusions. In addition, because of the defect in hemoglobin synthesis, the cells are hypochromic and microcytic. Hemoglobin Bart’s is more stable than hemoglobin H and does not form large inclusions.

Although, as is the case in β-thalassemia, excess globin chains cause damage to the red cell membrane, the mechanisms are different in the two forms of the disease. As described in “Etiology and Pathogenesis” above, in β-thalassemia, excess α chains result in mechanical instability and oxidative damage to a variety of membrane proteins, notably protein 4.1. However, in α-thalassemia, the membranes are hyperstable, and no evidence of oxidation or dysfunction of this protein is present. Furthermore, the state of red cell hydration is different in α-thalassemia. Accumulation of excess β chains results in increased hydration. These differences in the pathophysiology of membrane damage between α- and β-thalassemia are discussed in detail in references 7 and 174,175,176.

Another factor exacerbates the tissue hypoxia of the anemia of the α-thalassemias. Both hemoglobin Bart’s and hemoglobin H show no heme–heme interaction and have almost hyperbolic oxygen dissociation curves with very high oxygen affinities. Thus, they are not able to liberate oxygen at physiologic tissue tensions; in effect, they are useless as oxygen carriers.⁷

As a consequence, infants with high levels of hemoglobin Bart’s have severe intrauterine hypoxia. This is the major basis for the clinical picture of homozygous α⁰-thalassemia, which results in the stillbirth of hydropic infants late in pregnancy or at term. Oxygen deprivation is reflected by the grossly hydropic state of the infant, presumably as a result of increased capillary permeability, and by severe erythroblastosis. Deficient fetal oxygenation probably is responsible for the enormously hypertrophied placentas and possibly for the associated developmental abnormalities that occur with the severe forms of intrauterine α-thalassemia.⁷

PERSISTENT FETAL HEMOGLOBIN PRODUCTION AND CELLULAR HETEROGENEITY

Children with severe thalassemia have an increased level of hemoglobin F that persists into childhood and later.^7,10 In the β⁰-thalassemias, hemoglobin F is the only hemoglobin produced, except for small amounts of hemoglobin A₂. Examination of the blood using staining methods specific for hemoglobin F shows that it is heterogeneously distributed among the red cells.⁷ Persistent hemoglobin F production is not a major feature of the more severe forms of α-thalassemia.

The mechanism of persistent γ-chain synthesis in the thalassemias is incompletely understood. Normal adults have small quantities of hemoglobin F that are heterogeneously distributed among the red cells. Cells with demonstrable hemoglobin F are called F cells. One important mechanism for high hemoglobin F levels in the blood of patients with β-thalassemia is cell selection.^{7,180,181,182,183} The major cause of ineffective erythropoiesis and shortened red cell survival in β-thalassemia is the deleterious effect of excess α chains on erythroid maturation in the marrow and on the survival of red cells in the blood. Therefore, red cell precursors that produce γ chains are at a selective advantage. Excess α chains combine with γ chains to produce hemoglobin F; therefore, the magnitude of α-chain precipitation is less. Differential centrifugation experiments^181,182,183 and in vivo labeling studies¹⁸⁰ have shown that populations of red cells with relatively large amounts of hemoglobin F are more efficiently produced and survive longer in the blood. The blood of patients with homozygous β-thalassemia shows remarkable cellular heterogeneity with respect to red cell survival, such as populations of cells containing predominantly hemoglobin A that are destroyed very rapidly in the spleen and elsewhere, cells with a much longer survival that contain relatively more hemoglobin F, and populations of intermediate age and hemoglobin constitution.^7,182

Although cell selection is probably the main reason for the increased levels of hemoglobin F in the red cells in β-thalassemia, other mechanisms may also be involved. In any form of “stress erythropoiesis,” that is, rapid erythroid proliferation, there is a tendency for a relative increase in γ-chain production. Furthermore, as discussed in “Hereditary Persistence of Fetal Hemoglobin” above, several genes or chromosomal locations have been defined in which polymorphisms are involved in the increased basal production of γ chains and a relative increase in the number of F cells in the blood. The interaction of these different loci appear to be responsible for high levels of hemoglobin F production in β-thalassemia and sickle cell anemia with the production of milder phenotypes.^{125,126,127,128,184} However, biosynthesis studies indicate that marrow expansion and the selective survival of F-cell precursors and their progeny are the major factors in hemoglobin F production in hemoglobin E/β-thalassemia.¹⁸³

Because a reciprocal relation exists between γ- and δ-chain synthesis, the red cells of β-thalassemia homozygotes containing large amounts of hemoglobin F have relatively low hemoglobin A₂ levels.⁷ Thus, the measured percent hemoglobin A₂ in these individuals is the average of a very heterogeneous cell population. This finding probably accounts for the extreme variability in hemoglobin A₂ levels found in homozygotes for this disorder. A further consequence of the persistence of hemoglobin F in β-thalassemia is the high oxygen affinity of the red cells.

CONSEQUENCES OF COMPENSATORY MECHANISMS FOR THE ANEMIA OF THALASSEMIA

The profound anemia of homozygous β-thalassemia and the relatively high oxygen affinity of hemoglobin F combine to cause severe tissue hypoxia. Because of the high oxygen affinity of hemoglobins Bart’s and H, a similar defect in tissue oxygenation occurs in the more severe forms of α-thalassemia. The major adaptive response to hypoxia is increased erythropoietin production. It has been found that in severely anemic children with hemoglobin E β-thalassemia, age and hemoglobin levels are independent variables in erythropoietin response and that for a given hemoglobin level there is a relatively high erythropoietin in very young children.¹⁸⁵ These observations provide an explanation for the rather unstable phenotype of many intermediate forms of β-thalassemia during early childhood. The major effect of these very high levels of erythropoietin production is expansion of the dyserythropoietic marrow. The results are deformities of the skull and face and porosity of the long bones.⁷ Extramedullary hematopoietic tumors may develop in extreme cases. Apart from the production of severe skeletal deformities, marrow expansion may cause pathologic fractures and sinus and middle ear infection as a result of ineffective drainage.

Another important effect of the enormous expansion of the marrow mass is the diversion of calories required for normal development to the ineffective red cell precursors. Thus, patients severely affected by thalassemia show poor development and wasting. The massive turnover of erythroid precursors may result in secondary hyperuricemia and gout and severe folate deficiency.

The effects of gross intrauterine hypoxia in homozygous α⁰-thalassemia have been described. In the symptomatic forms of α-thalassemia (e.g., hemoglobin H disease) that are compatible with survival into adult life, bone changes and other consequences of erythroid expansion are seen, although less commonly than in β-thalassemia.

SPLENOMEGALY: DILUTIONAL ANEMIA

Constant exposure of the spleen to red cells with inclusions consisting of precipitated globin chains gives rise to the phenomenon of “work hypertrophy.” Progressive splenomegaly occurs in both α- and β-thalassemia and may worsen the anemia.^7,10 A large spleen acts as a sump for red cells, sequestering a considerable proportion of the red cell mass. Furthermore, splenomegaly may cause plasma volume expansion, a complication that can be exacerbated by massive expansion of the erythroid marrow. The combination of pooling of the red cells in the spleen and plasma volume expansion can exacerbate the anemia in both α- and β-thalassemia.

ABNORMAL IRON METABOLISM

β-Thalassemia homozygotes that are anemic manifest increased intestinal iron absorption that is related to the degree of expansion of the red cell precursor population. Iron absorption is decreased by blood transfusion.^7,10 Increased absorption causes a steady accumulation of iron, first in the Kupffer cells of the liver and the macrophages of the spleen and later in the parenchymal cells of the liver. Most patients homozygous for β-thalassemia require regular blood transfusion; thus, transfusional siderosis adds to the iron accumulation. Iron accumulates in the endocrine glands,^7,186 particularly in the parathyroids, pituitary, pancreas, skin leading to increased pigmentation, liver, and, most important, in the myocardium.^7,187 Iron accumulation in the myocardium leads to death by involving the conducting tissues or by causing intractable cardiac failure. Other consequences of iron loading include diabetes, hypoparathyroidism, hypothyroidism, and abnormalities of hypothalamic–pituitary function leading to growth retardation and hypogonadism.^7,186 Recent work on the mechanisms of hepcidin downregulation in association with marrow hypertrophy provides a much better understanding of the mechanisms of iron loading in diseases like thalassemia and may provide new therapeutic options for the future (Chap. 43 and Ref. 188).

Accurate information is available regarding the levels of body iron, as reflected by hepatic iron, at which patients are at risk for serious complications of iron overload.^7,189 These studies, which extrapolate data obtained from patients with genetic hemochromatosis, suggest that patients with hepatic iron levels of approximately 80 μmol of iron per gram of liver, wet weight (~15 mg of iron per gram of liver, dry weight), are at increased risk for hepatic disease and endocrine organ damage. Patients with higher body iron burdens are at particular risk for cardiac disease and early death (Chap. 43).

Disordered iron metabolism is less common in the adult forms of α-thalassemia. The milder degree of anemia, fewer transfusions, and the less marked erythroid expansion of the marrow are likely explanations.

The mechanisms whereby iron, and in particular non–transferrin-bound iron mediate tissue damage, and recent evidence about the central role of hepcidin in the abnormal regulation of iron absorption in disorders like thalassemia are discussed in Chap. 42.

INFECTION

All forms of severe thalassemia appear to be associated with an increased susceptibility to bacterial infection.⁷ The reason is not known. The relatively high serum iron levels may favor bacterial growth. Another possible mechanism is blockade of the monocyte–macrophage system as a result of the increased rate of destruction of red cells. No consistent defects in white cell or immune function have been reported, and high serum iron levels as an important factor remain to be unequivocally demonstrated. The one exception is infection with Yersinia enterocolitica, a normally nonvirulent pathogen that can produce its own siderophore and hence can thrive in iron excess. Transfusion-dependent patients with thalassemia are at particular risk for blood-borne infections including hepatitis B, hepatitis C, HIV/AIDS, and, in some parts of the world, malaria.

COAGULATION DEFECTS

The increasing knowledge about the potential hypercoagulable state in some forms of thalassemia has been reviewed in detail.^{174,175,176,190} Evidence indicates that patients, particularly after splenectomy and with high platelet counts, may develop progressive pulmonary arterial disease as a result of platelet aggregation in the pulmonary circulation. Furthermore, using thalassemic red cells as a source of phospholipids, enhanced thrombin generation has been demonstrated in a prothrombinase assay. The procoagulant effect of thalassemia cells appears to result from increased expression of anionic phospholipids on the red cell surface (Chap. 33). Normally, neutral or negatively charged phospholipids are confined to the inner leaflet of the red cell membrane, an effect that is mediated by the action of aminophospholipid translocase, an enzyme sometimes known as flippase. In effect, this enzyme flips aminophospholipids that are diffused to the outer leaflet back to the inner leaflet (Chaps. 31 and 46). The current belief is that these aminophospholipids in thalassemic red cells are moved to the outer leaflet, thus providing a surface on which coagulation can be activated. Other nonspecific changes in the coagulation pathway and its antagonists have been observed in patients with different forms of thalassemia.

There is increasing evidence that, as in the case of sickle cell anemia (Chap. 49), the hemolytic component of the anemia of β-thalassemia is associated with the release of hemoglobin and arginase resulting in impaired nitric oxide availability and endothelial dysfunction with progressive pulmonary hypertension.¹⁹¹ There may be other contributions to this complication including increased coagulability and local structural damage to the lungs relating to excess iron deposition.

CLINICAL HETEROGENEITY

The pathophysiologic mechanisms described above provide the basis for the remarkably diverse clinical findings in the thalassemia syndromes.^7,192 All the manifestations of β-thalassemia can be related to excess α-chain production. Thus, any mechanism that reduces the excess of α chains should reduce the clinical severity of the disease. Several elegant “experiments of nature” have shown that this reasoning is true and, incidentally, have confirmed that globin-chain imbalance is the major factor determining the severity of the thalassemias.

Coinheritance of α-thalassemia can reduce the severity of the more severe forms of β-thalassemia.^193,194 The effect is much more marked in individuals who are homozygotes or compound heterozygotes for different forms of β⁺-thalassemia. β⁰-Thalassemia homozygotes who have inherited α-thalassemia seem to be protected little, if at all.

Severe β-thalassemia can be modified by the coinheritance of genetic determinants for enhanced production of γ chains. Several determinants may be involved. For example, inheritance of a particular RFLP haplotype in the region 5′ to the β-globin gene may be an important factor.^195,196 This particular β-globin gene haplotype is associated with a single base change, C→T, at position –158 relative to the ^Gγ-globin gene, an alteration that creates a cleavage site for the restriction enzyme XmnI.¹²¹ An excess of individuals homozygous for T (XmnI+ +) with the phenotype of thalassemia intermedia exist compared with thalassemia major in different populations.^196,197,198 Whether this polymorphism is the only factor that increases hemoglobin F production in these cases is not absolutely clear. As discussed under “Hereditary Persistence of Fetal Hemoglobin” above, it is now clear that there are loci on chromosomes 2, 6, and 8, and possibly the X chromosome, at which polymorphisms are involved in the elevation of fetal hemoglobin synthesis and that their coinheritance may significantly modify the phenotype of different forms of β-thalassemia.

Some mutations that cause β-thalassemia are associated with a mild phenotype because they result in only modest reduction of β-chain production.⁷ For example, mutations at positions –29 and –88 are associated with mild β⁺-thalassemia in Africans. Similarly, particularly mild phenotypes are commonly found with a base substitution at position 6 in IVS-1 and at position –87 in the 5′-flanking region of the β-globin gene in Mediterranean populations. The homozygous state for the IVS-1 position 6 mutation usually produces an extremely mild form of β-thalassemia. When these “mild” mutations are coinherited with more-severe β-thalassemia determinants, the compound heterozygous states are characterized by a more severe form of thalassemia intermedia. Other forms of thalassemia intermedia are associated with the homozygous state for δβ-thalassemia, the various interactions of β-thalassemia with δβ-thalassemia, and heterozygous β-thalassemia of the severe variety or in association with triplicated α-gene loci.^7,10,198 These complex interactions are the subject of several extensive reviews.^198,199,200

These mechanisms for the phenotypic variability of the β-thalassemias represent only the beginning of our understanding of the genetic diversity of these conditions. Hence, defining a series of genetic modifiers that act at different levels is useful.¹⁹² Primary modifiers represent the diversity of mutations at the β-globin gene locus. Secondary modifiers are those, such as α-thalassemia and increased hemoglobin F production, that directly modify the relative degree of the imbalanced globin chain output. However, an increasing number of tertiary modifiers, that is, genetic diversity, have an important effect on the complications of the disease. These include loci involved in iron, bone, and bilirubin metabolism and in determining resistance of susceptibility to infection. Furthermore, phenotypic diversity may reflect different degrees of adaptation to anemia and the effect of the environment. These complex issues have been reviewed¹⁹² and are illustrated in Fig. 48–14. Several extensive reviews of the pathophysiology of the intermediate forms of β-thalassemia in different populations are available.^199,200

The α-thalassemias, particularly hemoglobin H disease, show considerable clinical diversity. Some of this variability can be related to particular genotypes,^7,41 but the reasons for the heterogeneity of these disorders is not clear.

β- AND δβ-THALASSEMIAS

The most clinically severe form of β-thalassemia is thalassemia major. A milder clinical picture, characterized by a later onset and either no transfusion requirement or at least fewer transfusions than are required to treat the major form of the illnesses, is designated β-thalassemia intermedia. β-Thalassemia minor is the term used to describe the heterozygous carrier state for β-thalassemia. More extensive accounts of the clinical features of these conditions are given in two monographs.^7,9

β-THALASSEMIA MAJOR

The homozygous or compound heterozygous state for β-thalassemia, thalassemia major, produces the clinical picture first described by Cooley and Lee¹ in 1925. Affected infants are well at birth. Anemia usually develops during the first few months of life and becomes progressively more severe. The infants fail to thrive and may have feeding problems, bouts of fever, diarrhea, and other gastrointestinal symptoms. The majority of infants who develop transfusion-dependent homozygous β-thalassemia present with these symptoms within the first year of life. A later onset suggests the condition will develop into one of the intermediate forms of β-thalassemia (see “Pathophysiology” above).

The course of the disease in childhood depends almost entirely on whether the child is maintained on an adequate transfusion program.^7,9 The classic textbook picture of Cooley anemia describes the disease as it was seen before these children could be maintained with relatively normal hemoglobin levels by regular blood transfusions. If adequate transfusion is possible, children grow and develop normally and have no abnormal physical signs. Few of the complications of the disorder occur during childhood. The disease presents a problem only when the effects of iron loading resulting from ineffective erythropoiesis and from repeated blood transfusions become apparent at the end of the first decade. Children who are treated with an adequate iron chelation regimen develop normally, although some of them remain short in height.

An inadequately transfused child develops the typical features of Cooley anemia. Growth is stunted. With bossing of the skull and overgrowth of the maxillary region, the face gradually assumes a “mongoloid” appearance. These changes are associated with a characteristic radiologic appearance of the skull, long bones, and hands (Fig. 48–15). The diploe widens, with a “hair on end” or “sun ray” appearance and a lacy trabeculation of the long bones and phalanges. Gross skeletal deformities can occur. The liver and spleen are enlarged, and the pigmentation of the skin increases. Many features of a hypermetabolic state, as evidence by fever, wasting, and hyperuricemia, may develop.

The clinical course is characterized by severe anemia with frequent complications. These children are particularly prone to infection, which is a common cause of death. Spontaneous fractures occur commonly as a result of the expansion of the marrow cavities with thinning of the long bones and skull. Maxillary deformities often lead to dental problems from malocclusion. Formation of massive deposits of extramedullary hematopoietic tissue may cause neurologic complications. With the gross splenomegaly that may occur, secondary thrombocytopenia and leukopenia frequently develop, leading to a further tendency to infection and bleeding. Splenectomy is frequently performed to reduce transfusion frequency and severe thrombocytopenia; however, postsplenectomy infections are particularly common.⁷ Bleeding tendency may be seen in the absence of thrombocytopenia. Epistaxis is particularly common. These hemostatic problems are associated with poor liver function in some cases. Chronic leg ulceration may occur but is more common in thalassemia intermedia.

Children who have grown and developed normally throughout the first 10 years of life as a result of regular blood transfusion begin to develop the symptoms of iron loading as they enter puberty, particularly if they have not received adequate iron chelation.^7,9 The first indication of iron loading usually is the absence of the pubertal growth spurt and failure of the menarche. Over the succeeding years, a variety of endocrine disturbances may develop, particularly diabetes mellitus, hypogonadotrophic hypogonadism, and growth hormone deficiency. Hypothyroidism and adrenal insufficiency also occur but are less common.^7,186 Toward the end of the second decade, cardiac complications arise, and death usually occurs in the second or third decade as a result of cardiac siderosis.^187,188,189 Cardiac siderosis may cause an acute cardiac death with arrhythmia, or intractable cardiac failure. Both of these complications can be precipitated by intercurrent infection.

Even the adequately transfused child who has received chelation therapy may suffer a number of complications. Bloodborne infection, notably with hepatitis B or C,²⁰¹ HIV,²⁰² or malaria,²⁰³ is extremely common in some populations, although the frequency is decreasing with the use of widespread blood-donor screening programs. Delayed puberty and growth retardation are common and probably reflect hypogonadotrophic hypogonadism and damage to the pituitary gland.^201,204 Osteoporosis is being recognized increasingly and may, at least in part, be a reflection of hypogonadism.²⁰¹

β-THALASSEMIA INTERMEDIA

The clinical phenotype of patients designated as having thalassemia intermedia is more severe than the usual asymptomatic thalassemia trait but milder than transfusion-dependent thalassemia major.^7,199,200 The syndrome encompasses disorders with a wide spectrum of disability. At the severe end, patients present with anemia later than patients with the transfusion-dependent forms of homozygous β-thalassemia and are just able to maintain a hemoglobin level of approximately 6 g/dL without transfusion. However, their growth and development are retarded. The patients become seriously disabled, with marked skeletal deformities, arthritis, and bone pain; progressive splenomegaly; growth retardation; and chronic ulcerations above the ankles. At the other end of the spectrum, patients remain completely asymptomatic until adult life and are transfusion independent, with hemoglobin levels as high as 10 to 12 g/dL. All varieties of intermediate severity are observed. Some patients become disabled simply from the effects of hypersplenism. Intensive studies of the molecular pathology of this condition have provided some guidelines about genotype–phenotype relationships that are useful for genetic counseling (Table 48–6).

Overall, the clinical features of the intermediate forms of β-thalassemia are similar to the features of β-thalassemia major. At the severe end of the spectrum, particularly in cases of growth retardation, patients should be treated with regular transfusion. However, a number of important complications, including progressive hypersplenism, occur in patients with milder forms. Clinically significant iron loading as a result of increased absorption is seen even in patients with infrequent transfusions (Chap. 43). Iron overload results in frequent diabetes and endocrine disturbances, typically by fourth decade of life. A high incidence of pigment gallstones, skeletal deformities, bone and joint disease, leg ulcers, and thrombotic tendency, particularly after splenectomy, is observed.⁷

Hematologists should be aware that in patients heterozygous for rare forms of β-thalassemia, a phenotype of thalassemia intermedia that results in the clinical constellation of autosomal dominant thalassemia (discussed in “Pathophysiology” above) is encountered on rare occasions.

β-THALASSEMIA MINOR

The heterozygous state for β-thalassemia is usually identified during family studies of patients with more severe forms of β-thalassemia, population surveys, or, most frequently, by the chance finding of the characteristic hematologic changes during a routine study. There is an extensive literature on this condition,⁷ some of which suggests that affected individuals may have symptoms of anemia and, not infrequently, splenomegaly, while other studies suggest that the condition is completely symptomless and palpable splenomegaly does not occur. Surprisingly, none of these studies have been controlled. A controlled study reported that individuals with the β-thalassemia trait suffer from fatigue and other symptoms indistinguishable from those with mild anemias from other causes. There was no difference in the frequency of palpable splenomegaly between the thalassemic and control groups.²⁰⁵ The trait not infrequently causes a moderately severe anemia of pregnancy, in some cases requiring transfusion. Some β-thalassemia carriers have increased iron stores, although this is most often a result of inappropriate iron therapy based on a misdiagnosis. In countries where there is a relatively high frequency of genetic determinants for hemochromatosis, the possibility of their coinheritance should be borne in mind if a patient with β-thalassemia trait with an unusually high plasma iron or serum ferritin level is encountered.

α-THALASSEMIAS

Hemoglobin Bart’s Hydrops Fetalis Syndrome

This disorder is a frequent cause of stillbirth in Southeast Asia. Infants either are stillborn between 34 and 40 weeks’ gestation or are born alive but die within the first few hours.^7,206 Pallor, edema, and hepatosplenomegaly are seen. The clinical picture resembles hydrops fetalis as a result of Rh blood group incompatibility. Massive extramedullary hemopoiesis and enlargement of the placenta are noted at autopsy. A variety of congenital anomalies have been observed.

The rescue of a few infants with this syndrome by prenatal detection and exchange transfusion has been reported. These babies have grown and developed normally, although they are blood transfusion–dependent.^207,208

This condition is associated with a high incidence of maternal toxemia of pregnancy and difficulties at the time of delivery because of the massive placenta.²⁰⁶ The reason for placental hypertrophy is unknown, although severe intrauterine hypoxia is suspected because a similar phenomenon is observed in hydrops infants with Rh incompatibility.

Hemoglobin H Disease

Hemoglobin H disease was described independently in the United States and in Greece in 1956.^209,210 The clinical findings are variable. A few patients are affected almost as severely as patients with β-thalassemia major, but most patients have a much milder course.^7,211 Lifelong anemia with variable splenomegaly occurs; bone changes are unusual.

As discussed earlier in “Etiology and Pathogenesis,” a few attempts have been made to correlate the genotype with the phenotype of hemoglobin H disease. In general, as expected, patients with a nondeletion form of α-thalassemia affecting the predominant α₂ gene interacting with an α⁰-thalassemia determinant α^NDα/– –, or α^{ConstantSpring}α/– –, for example, have higher hemoglobin H levels, a greater degree of anemia, and a more severe clinical course than patients with the – –/–α genotype.^{212,213,214,215}

Milder Forms of α-Thalassemia

Because two α-globin genes exist per haploid genome, a wide spectrum of different conditions with overlapping phenotypes result from their various interactions.⁷ The carrier states for the deletion and nondeletion forms of α-thalassemia, –α/αα and α^NDα/αα, are symptomless. Similarly, the homozygous states for the deletion forms of α⁺-thalassemia, –α/–α, and the heterozygous state for α⁰-thalassemia, – –/αα, are symptomless, although they are associated with mild anemia and red cell changes. On the other hand, the homozygous states for the nondeletion forms of α-thalassemia, α^NDα/α^NDα, are associated with an extremely diverse series of phenotypes. As mentioned in “Interactions of α-Thalassemia Haplotypes” above in “Etiology and Pathogenesis,” they sometimes result in the clinical picture of hemoglobin H disease. In other patients, they are associated with only mild hypochromic anemia.⁷ The homozygous states for the chain termination mutants, notably hemoglobin Constant Spring, constitute a special case because they produce a particularly characteristic phenotype. In this case, moderate hemolytic anemia with splenomegaly are seen.^7,216,217

α-Thalassemia and Mental Retardation

The clinical phenotype of these conditions is heterogeneous. In cases associated with chromosomal deletion (tip of chromosome 16; ATR-16 [α-thalassemia chromosome 16-linked mental retardation syndrome]), the clinical defects vary with the extent of chromosomal defect; only α-thalassemia and mental retardation are constant.¹⁵⁷ To some extent this clinical variation is related to the length of the associated deletions; those which extend for 2000 kb involve the genes that are involved in tuberous sclerosis and polycystic kidney disease. In these cases the latter dominate the clinical picture, but there mental retardation and α-thalassemia are also associated.

The clinical phenotype in the second group of these disorders, which are caused by mutations of ATR-X, includes skeletal abnormalities, dysmorphic face, neonatal hypotonus, genital abnormalities, and a variety of less-constant features, in addition to mental retardation and α-thalassemia.¹⁵⁸

εγδβ-Thalassemia

The clinical picture varies with the stage of development.⁷ Neonates may be significantly anemic and require transfusions. In contrast, children and adults with this condition are asymptomatic. They have the clinical and laboratory picture of heterozygous β-thalassemia, with the exception of a normal hemoglobin A₂ level. The reason for this discrepancy of developmental differences of the clinical phenotype has not been identified. The homozygous state is assumed to be lethal.

β-THALASSEMIA MAJOR

Hemoglobin levels at presentation may range from 2 to 3 g/dL or even lower.⁷ The red cells show marked anisopoikilocytosis, with hypochromia, target cell formation, and a variable degree of basophilic stippling (Fig. 48–16). The appearance of the blood film varies, depending on whether the spleen is intact. In nonsplenectomized patients, large poikilocytes are common. After splenectomy, large, flat macrocytes and small, deformed microcytes are frequently seen. The reticulocyte count is moderately elevated, and nucleated red cells nearly always are present in the blood. These red cell forms may reach very high levels after splenectomy. The white cell and platelet counts are slightly elevated unless secondary hypersplenism occurs. Staining of the blood with methyl violet, particularly in splenectomized subjects, reveals stippling or ragged inclusion bodies in the red cells.¹⁶⁹ These inclusions can nearly always be found in the red cell precursors in the marrow. The marrow usually shows erythroid hyperplasia with morphologic abnormalities of the erythroblasts, such as striking basophilic stippling and increased iron deposition. Iron kinetic studies indicate markedly ineffective erythropoiesis, and red cell survival usually is shortened. Populations of cells with very short survival and longer-lived populations of cells are seen. The latter contain relatively more fetal hemoglobin. An increased level of fetal hemoglobin, ranging from less than 10 percent to greater than 90 percent, is characteristic of homozygous β-thalassemia. No hemoglobin A is produced in β⁰-thalassemia. The fetal hemoglobin is heterogeneously distributed among the red cells. Hemoglobin A₂ levels in homozygous β-thalassemia may be low, normal, or high. However, expressed as a proportion of hemoglobin A, the hemoglobin A₂ level almost invariably is elevated. Differential centrifugation studies indicate some heterogeneity of hemoglobin F and A₂ distribution among thalassemic red cells, but their level in whole blood gives little indication of their total rates of synthesis.

In vitro hemoglobin synthesis studies using marrow or blood show a marked degree of globin-chain imbalance. Marked excess of α-chain over β- and γ-chain production is always observed. Other aspects of the laboratory findings in this condition, including red cell survival, iron absorption, ferrokinetics, erythrokinetics, and the consequences of iron loading, were discussed earlier (see “Etiology and Pathogenesis” above).

The examination of siblings, parents, and children can be very important in confirming the diagnosis by finding the abnormalities in other family members, and the examining physician should make every effort to obtain a complete blood count in family members. With the exception of higher hemoglobin levels, the hematological changes in β-thalassemia intermedia are similar to those in β-thalassemia major (Fig. 48–17).

β-THALASSEMIA MINOR

Hemoglobin values of patients with β-thalassemia minor usually range from 9 to 11 g/dL. The most consistent finding is small, poorly hemoglobinized red cells (see Fig. 48–16), resulting in mean cell hemoglobin (MCH) values of 20 to 22 pg and mean corpuscular volume (MCV) values of 50 to 70 fL. The red cell count is usually normal or elevated and the hemoglobin and hematocrit is usually slightly below normal; however, the red cell indices are particularly useful in screening for heterozygous carriers of thalassemia in population surveys. The marrow in heterozygous β-thalassemia shows slight erythroid hyperplasia with rare red cell inclusions. Megaloblastic transformation as a result of folic acid deficiency occurs occasionally, particularly during pregnancy. A mild degree of ineffective erythropoiesis is noted, but red cell survival is normal or nearly normal. The hemoglobin A₂ level is increased to 3.5 to 7.0 percent. The level of fetal hemoglobin is elevated in approximately 50 percent of cases, usually to 1 to 3 percent and rarely to greater than 5 percent.

α-THALASSEMIAS

Hemoglobin Bart’s Hydrops Fetalis Syndrome

In infants with the hydrops fetalis syndrome, the blood film shows severe thalassemic changes with many nucleated red cells. The hemoglobin consists mainly of hemoglobin Bart’s, with approximately 10 to 20 percent hemoglobin Portland. Usually no hemoglobin A or F is present, although rare cases that seem to result from interaction of α⁰-thalassemia with a severe nondeletion form of α⁺-thalassemia show small amounts of hemoglobin A.

Hemoglobin H Disease

The blood film shows hypochromia and anisopoikilocytosis. The reticulocyte count usually is approximately 5 percent. Incubation of the red cells with brilliant cresyl blue results in ragged inclusion bodies in almost all cells. These bodies form because of precipitation of hemoglobin H in vitro as a result of redox action of the dye. After splenectomy, large, single Heinz bodies are observed in some cells (Fig. 48–18). These bodies are formed by in vitro precipitation of the unstable hemoglobin H molecule and are seen only after splenectomy. Hemoglobin H constitutes between 5 and 40 percent of the total hemoglobin. Traces of hemoglobin Bart’s may be present, and the hemoglobin A₂ level usually is slightly subnormal.

α⁰-Thalassemia and α⁺-Thalassemia Traits

The α⁰-thalassemia trait is characterized by the presence of 5 to 15 percent hemoglobin Bart’s at birth.⁷ This hemoglobin disappears during maturation and is not replaced by a similar amount of hemoglobin H. An occasional cell with hemoglobin H inclusion bodies may appear after incubation with brilliant cresyl blue. This phenomenon is often used as a diagnostic test for the α-thalassemia trait. However, the test is difficult to standardize and requires much experience to be useful. In adult life, the red cells of heterozygotes have morphologic changes of heterozygous thalassemia with low MCH and MCV values. The electrophoretic pattern is normal. Globin-synthesis studies show a deficit of α-chain production, with an α-chain–to–β-chain production ratio of approximately 0.7.

The α⁺-thalassemia trait (–α/αα) is characterized by a mild reduction in MCH and MCV values although in some cases there are normal values, 1 to 2 percent of hemoglobin Bart’s at birth in some but not all cases, and a slightly reduced α-chain–to–β-chain production ratio of approximately 0.8; thus, this genotype often is referred to as silent carrier. Extensive studies comparing the level of hemoglobin Bart’s at birth with a DNA analyses demonstrated that there is no detectable hemoglobin Bart’s in a significant number of newborns who are heterozygous for α⁺-thalassemia.^218,219 Globin gene synthetic ratios can be distinguished from normal only by studying relatively large numbers of samples and comparing the mean α–to–β ratio with that of normal control subjects. This approach is not reliable for diagnosing individual cases of the α⁺-thalassemia trait, and, unfortunately, no reliable method of diagnosis is available except for DNA analysis.

Homozygous State for Nondeletion Types of α-Thalassemia

The homozygous state for nondeletion forms of α-thalassemia involving the dominant (α₂) globin gene causes a more severe deficit of α chains than do the deletion forms of α⁺-thalassemia. In some cases, the homozygous state produces hemoglobin H disease.

The homozygous state for hemoglobin Constant Spring or other chain-termination mutations is associated with moderately severe hemolytic anemia in which, for reasons not explained, no hemoglobin H is present but small amounts of hemoglobin Bart’s persist into adult life. The homozygous states for the other nondeletion forms of α⁺-thalassemia are associated with hemoglobin H disease. In the homozygous state for hemoglobin Constant Spring, the blood picture shows mild thalassemic changes with normal-size red cells.^216,217 The hemoglobin consists of approximately 5 to 6 percent hemoglobin Constant Spring, normal hemoglobin A₂ levels, and trace amounts of hemoglobin Bart’s. The remainder is hemoglobin A.

The heterozygous state for hemoglobin Constant Spring shows no hematologic abnormality. The hemoglobin pattern is normal except for the presence of approximately 0.5 percent hemoglobin Constant Spring. The latter can be observed on alkaline starch-gel electrophoresis as a faint band migrating between hemoglobin A₂ and the origin. It is best seen on heavily loaded starch gels and is easily missed if other electrophoretic techniques are used (Fig. 48–19). In the newborn, usually 1 to 3 percent hemoglobin Bart’s is present in the cord blood.

Homozygous State for Deletion Forms of α⁺-Thalassemia

The homozygous state for deletion forms of α⁺-thalassemia is characterized by a thalassemic blood picture with 5 to 10 percent hemoglobin Bart’s at birth and hematologic findings similar to those in α⁰-thalassemia heterozygotes in adult life. In general, the –α^4.2 deletion is associated with a more severe phenotype than is the –α^3.7 deletion.⁷

The clinical and hematologic findings in homozygous β-thalassemia and hemoglobin H disease are so characteristic that the diagnosis usually is not difficult. Figure 48–20 shows a simple flowchart for laboratory investigations of a suspected case.

In early childhood, distinguishing the thalassemias from the congenital sideroblastic anemias may be difficult, but the marrow appearances in the latter are quite characteristic. Because of the high hemoglobin F levels encountered in juvenile chronic myelogenous leukemia, this disorder may superficially resemble β-thalassemia. However, the finding of primitive cells in the marrow, the absence of elevated hemoglobin A₂ levels on hemoglobin electrophoresis, the decrease in carbonic anhydrase in juvenile chronic myelogenous leukemia, and characteristic in vitro responses of myeloid progenitors in vitro to granulocyte-monocyte colony-stimulating factor (Chap. 87) readily differentiate this disorder from β-thalassemia.

LESS-COMMON FORMS OF THALASSEMIA

(δβ)⁰-Thalassemia

The homozygous state for δβ-thalassemia is clinically milder than Cooley anemia and is one form of thalassemia intermedia.^220,221,222 Only hemoglobin F is present; hemoglobins A and A₂ are not produced. Heterozygous δβ-thalassemia is hematologically similar to β-thalassemia minor.⁷ The fetal hemoglobin level is higher (range: 5 to 20 percent), and the hemoglobin A₂ value is normal or slightly reduced. As in β-thalassemia, the fetal hemoglobin is heterogeneously distributed among the red cells, thus distinguishing this disorder from HPFH (Fig. 48–21).

Heterozygosity for both β-thalassemia and δβ-thalassemia results is a condition clinically similar to but milder than Cooley anemia. The hemoglobin consists largely of hemoglobin F, with a small amount of hemoglobin A₂. This finding is seen because the associated β-thalassemia gene has usually been the β⁰ variety. δβ-Thalassemia has also been observed in individuals heterozygous for hemoglobin S or C.⁷

(δβ)⁺-Thalassemia and Hemoglobin Lepore Disorders

The hemoglobin Lepore disorders have been described in the homozygous state and in the heterozygous state, either alone or in association with β- or δβ-thalassemia, hemoglobin S, or hemoglobin C.^7,9,223 In the homozygous state, approximately 20 percent of the hemoglobin is of the Lepore type and 80 percent is fetal hemoglobin. Hemoglobins A and A₂ are absent. The clinical picture is variable. Some cases are identical to transfusion-dependent homozygous β-thalassemia; others are associated with the clinical picture of thalassemia intermedia. In the heterozygous state, the findings are similar to those of β-thalassemia minor. The hemoglobin consists of approximately 10 percent hemoglobin Lepore, with a reduced level of hemoglobin A₂ and a slight but consistent increase in fetal hemoglobin level. The Lepore hemoglobins have been found sporadically in most racial groups. In the majority of cases, chemical analysis has shown that these hemoglobins are identical to hemoglobin Lepore Washington-Boston. Hemoglobin Lepore Hollandia and Lepore Baltimore have been observed in only a few patients.^7,223

HEREDITARY PERSISTENCE OF FETAL HEMOGLOBIN

The current knowledge about the molecular pathology of HPFH was described earlier in “Etiology and Pathogenesis.” Table 48–4 summarizes the currently accepted classification and nomenclature of this complex group of conditions. The different forms of HPFH are of very little clinical importance except that they may interact with thalassemia or the structural hemoglobin variants.

(δβ)⁰ Hereditary Persistence of Fetal Hemoglobin

Homozygotes for (δβ)⁰ HPFH have 100 percent hemoglobin F. Their blood shows mild thalassemic changes, with reduced MCH and MCV values very similar to those observed in heterozygous β-thalassemia. Similarly, they have imbalanced globin chain production, with ratios in the range of those observed in β-thalassemia heterozygotes.²²⁴ Heterozygotes have approximately 20 to 30 percent hemoglobin F, slightly reduced hemoglobin A₂ values, and completely normal blood pictures. Thus, this condition appears to be an extremely well-compensated form of δβ-thalassemia in which the output of γ chains almost but not entirely compensates for the complete absence of β and δ chains. The different molecular forms of this condition show no difference in phenotype except in the proportion of ^Gγ chains. The African forms of (δβ)⁰ HPFH have been found in association with hemoglobins S and C or with β-thalassemia (Chap. 49). These compound heterozygous states are associated with little clinical disability.⁷

Nondeletion Types of Hereditary Persistence of Fetal Hemoglobin

Many nondeletion forms of HPFH associated with point mutations upstream from the γ-globin genes have been described (see Table 48–4). ^Gγ β⁺ HPFH has been found in the heterozygous and compound heterozygous states with β-globin chain variants in African populations. No associated clinical or hematologic findings have been reported. Compound heterozygotes for ^Gγ β⁺ HPFH and hemoglobins S or C produce 45 percent of the abnormal hemoglobin, approximately 30 percent hemoglobin A, and approximately 20 percent hemoglobin F containing only ^Gγ chains.^225,226

The most common form of nondeletion HPFH is ^Aγ β⁺ HPFH, which is found in Greeks.^227,228,229 In the homozygous state, no clinical or hematologic abnormalities are noted. The hemoglobin findings are characterized by approximately 25 percent fetal hemoglobin and reduced hemoglobin A₂ levels of approximately 0.8 percent.²³⁰ Heterozygotes, who also are hematologically normal, have 10 to 15 percent hemoglobin F, almost all of the ^Aγ variety. Compound heterozygotes with β-thalassemia have high hemoglobin F levels and a clinical picture that is only slightly more severe than the β-thalassemia trait.

In the British form of ^Aγ β⁺ HPFH²³¹ heterozygotes have approximately 5 to 12 percent hemoglobin F, whereas homozygotes have approximately 20 percent. No associated hematologic abnormalities are seen, although surprisingly in this form of nondeletion HPFH the hemoglobin F seems to be unevenly distributed among the red cells.

A heterogeneous group of conditions is associated with persistent production of small amounts of hemoglobin F in adult life. They are categorized under the general heading of heterocellular HPFH. Their clinical importance is that, when they are coinherited with different forms of β-thalassemia, they may lead to greater output of hemoglobin F and, hence, to a milder phenotype. This type of interaction should be suspected when one parent of a patient with β-thalassemia intermedia has an unusually high level of hemoglobin F for the β-thalassemia trait. Similarly, unaffected lateral relatives or other family members with slightly elevated hemoglobin F levels may be found.

β-THALASSEMIA ASSOCIATED WITH β-CHAIN STRUCTURAL HEMOGLOBIN VARIANTS

The most clinically important associations of β-thalassemia with β structural hemoglobin variants are sickle cell thalassemia, hemoglobin C thalassemia, and hemoglobin E thalassemia (Chap. 49). In addition, many interactions of β-thalassemia with rare structural variants have been reported.^7,9,10

Sickle cell thalassemia^7,232,233 occurs in parts of Africa and in the Mediterranean, particularly Greece and Italy. It also has been observed in the Middle East and parts of India. The clinical consequences of carrying one gene for hemoglobin S and one gene for β-thalassemia depend entirely on the type of β-thalassemia mutation. The interaction between the sickle cell gene and β⁰-thalassemia is characterized by a clinical disorder that is very similar to sickle cell anemia. Similarly, the interaction of the sickle cell gene with the more severe forms of β⁺-thalassemia associated with marked reduction in β-globin synthesis yields a similar clinical phenotype. On the other hand, the interaction of the sickle cell gene with very mild forms of β⁺-thalassemia may be quite innocuous.²³³ The latter disorder is characterized by mild anemia associated with splenomegaly and a hemoglobin composition of approximately 60 to 70 percent hemoglobin S, 25 percent hemoglobin A, and an elevated level of hemoglobin A₂. In all these interactions, one parent shows the sickle cell trait, and the other parent shows the β-thalassemia trait.

Hemoglobin C thalassemia is a mild hemolytic disorder associated with splenomegaly.^7,9,10 Again, the hemoglobin pattern varies depending on whether the thalassemia gene is the β⁺ or β⁰ type. This relatively innocuous condition has been recorded mainly in North Africa, but it also is found in West Africa. It is characterized by a mild hemolytic anemia and splenomegaly with a blood picture showing the numerous target cells characteristic of all the hemoglobin C disorders.

Hemoglobin E thalassemia, which occurs at a high frequency in the eastern half of the Indian subcontinent and throughout Southeast Asia, is one of the most important hemoglobinopathies in the world population.^{7,9,10,234,235,236,237,238,239,240} As mentioned earlier in “Etiology and Pathogenesis,” hemoglobin E is synthesized at a reduced rate and hence produces the clinical phenotype of a mild form of β-thalassemia. Hence, when hemoglobin E is inherited with β-thalassemia—and most often this is a β⁰- or severe β⁺-thalassemia mutation in Southeast Asia and India—a marked deficit of β-chain production results, with the clinical picture of severe β-thalassemia. Hemoglobin E thalassemia shows a remarkable variability in clinical expression,^{234,235,236,237,238} ranging from a mild form of thalassemia intermedia to a transfusion-dependent condition clinically indistinguishable from homozygous β-thalassemia. The reasons for this variability of expression are not understood, although some of the factors involved are identical to those that modify other forms of β-thalassemia.^239,240

In more-severe cases of hemoglobin E thalassemia, severe anemia with growth retardation, leg ulcers, bone deformity, marked tendency to infection, iron loading, and variable splenomegaly and hypersplenism are seen. Large tumor masses composed of extramedullary erythropoietic tissue may cause a variety of compression syndromes, including a clinical picture that closely mimics a cerebral tumor. Another curious picture that seems to be restricted to splenectomized patients is an obliterative occlusion of the pulmonary vasculature that is believed to result from an extremely high platelet count.²⁴¹

The clinical course and complications in transfusion-dependent patients are similar to those observed in homozygous β-thalassemia. In the milder forms, the main complications are progressive hypersplenism, organ damage as a result of progressive iron loading from an increased rate of absorption, extramedullary erythropoietic tumor masses, bone disease, and infection. The blood picture shows a typical thalassemic pattern. The hemoglobin consists of E, F, and A₂. Usually no hemoglobin A is present because the β⁰-thalassemias are particularly common in the parts of the world where hemoglobin E is found.

Newer studies emphasize the complex interactions between genetic factors,^239,240 differences in adaptation to anemia, particularly in early life (see “Pathophysiology” above), and the environment, notably proneness to malarial infection, that underlie the widely differing and unstable phenotypes of patients with hemoglobin E β-thalassemia.^238,239

β-THALASSEMIA WITH NORMAL HEMOGLOBIN A₂ LEVEL

Rare forms of β-thalassemia are seen in which heterozygotes have normal hemoglobin A₂ levels. Their main clinical importance is that they can be confused with the more severe forms of α-thalassemia in the heterozygous state and therefore may cause difficulties in genetic counseling and prenatal diagnosis. Based on hematologic studies, two main classes of “normal hemoglobin A₂ β-thalassemia”—sometimes called types 1 and 2—are seen.²⁴² Type 1 is the “silent” form of β-thalassemia. Type 2 is heterogeneous, with many cases representing the compound heterozygous state for β-thalassemia and δ-thalassemia.

“Silent” β-thalassemia^7,243 is characterized by no hematologic changes in heterozygotes. Several mild forms of β-thalassemia that underlie this phenotype are described (see Refs. 44 and 45). Although this condition can be partly identified by demonstrating a mild degree of globin-chain imbalance, with α-to-β synthesis ratios of approximately 1.5:1, it can only be diagnosed with certainty by DNA analysis. Compound heterozygotes for this condition and β⁰-thalassemia have a mild form of β-thalassemia intermedia.

Normal hemoglobin A₂ β-thalassemia type 2 in heterozygotes is indistinguishable from typical β-thalassemia with elevated hemoglobin A₂ levels.²⁴² The homozygous state has not been described. The compound heterozygous state for this gene and for β-thalassemia with raised hemoglobin A₂ levels is characterized by a clinical picture of severe transfusion-dependent β-thalassemia. Family data obtained in Italy and Sardinia suggest this condition represents the compound heterozygous state for both β-thalassemia and δ-thalassemia.^244,245 Most of the δ-thalassemias have been observed trans to β-thalassemia. However, the form of δ-thalassemia resulting from loss of an A in codon 59 occurs on the same chromosome as the hemoglobin Knossos mutation, which is associated with a mild form of β-thalassemia.²⁴⁶ This finding explains the normal level of hemoglobin A₂ associated with this condition, which is the most common form of normal hemoglobin A₂ β-thalassemia in the Mediterranean region.

Several other conditions, mentioned earlier in this chapter in “Etiology and Pathogenesis,” are associated with a phenotype that is indistinguishable from normal A₂ β-thalassemia. These conditions include the heterozygous states for the Corfu form of δβ-thalassemia and εγδβ-thalassemia.

OTHER UNUSUAL FORMS OF β-THALASSEMIA

The clinical features of the dominant β-thalassemias resemble the features of thalassemia intermedia.⁷ Moderate anemia and splenomegaly are seen, with a blood picture showing thalassemic red cell changes. The marrow shows erythroid hyperplasia with well-marked inclusion bodies in the red cell precursors. The latter may be seen in the blood after splenectomy. Hemoglobin analysis shows hemoglobins A and A₂ are present, and the hemoglobin F level is not usually elevated much higher than that seen in β-thalassemia trait. Hemoglobin A₂ levels are always raised.

Other unusual varieties of β-thalassemia include those categorized by unusually high hemoglobin F or A₂ levels. Most of these conditions result from deletions involving the β-globin gene and its promoter region. For example, the so-called Dutch²⁴⁷ form of β-thalassemia is associated with unusually high hemoglobin F levels in heterozygotes and high hemoglobin A₂ levels. Several other conditions of this type, which result from different-size deletions, have been reported (see Ref. 7).

δ⁰-THALASSEMIA

δ⁰-Thalassemia causes a complete absence of hemoglobin A₂ in homozygotes and a reduced hemoglobin A₂ level in heterozygotes.²⁴⁸ It is of no clinical significance except for its effect of reducing hemoglobin A₂ levels in β-thalassemia heterozygotes.

εγδβ-THALASSEMIA

This heterogeneous condition has been observed only in the heterozygous state in a few families.^7,108,109 It is characterized by neonatal hemolysis and, in adult life, by the hematologic picture of heterozygous β-thalassemia with normal hemoglobin A₂ levels.

α-THALASSEMIA IN ASSOCIATION WITH α- AND β-CHAIN HEMOGLOBIN VARIANTS

Several α-globin structural variants are caused by single amino acid substitutions at α-chain loci on chromosomes that carry only a single α-chain gene. Individuals who inherit variants of this type and an α⁰-thalassemia determinant have a form of hemoglobin H disease in which the hemoglobin consists of the α-chain variant hemoglobin and hemoglobin H. Well-documented examples include hemoglobin QH disease (– –/–α^Q),^249,250 hemoglobin G Philadelphia H disease (– –/–α^G),^251,252 and hemoglobin Hasharon H disease (– –/–α^Hash).²⁵³ Many examples of the coexistence of the homozygous or heterozygous states for β-chain hemoglobin variants and different α-thalassemia determinants have been reported.^7,9,10 Particularly well-characterized disorders include the various interactions of α⁰- and α⁺-thalassemia with hemoglobin E^7,234 and hemoglobin S (Chap. 49).^254,255 Carriers for these hemoglobin variants who also have the α⁰- or α⁺-thalassemia traits have thalassemic red cell indices and unusually low levels of the abnormal hemoglobin. Individuals with sickle cell anemia who have α-thalassemia show thalassemic red cell changes, more persistent splenomegaly, and lower hemoglobin F values than do patients without the thalassemia genes.

The only forms of treatment available for thalassemic children are regular blood transfusions, iron chelation therapy in an attempt to prevent iron overload, judicious use of splenectomy in cases complicated by hypersplenism, and a good standard of general pediatric care.^7,9,256 Marrow transplantation has an important role in selected cases (Chap. 23).

TRANSFUSION

Children with β-thalassemia who are maintained at a hemoglobin level of 9.5 to 14.0 g/dL grow and develop normally. They do not develop the distressing skeletal complications of thalassemia.^7,256 Maintaining a lower hemoglobin level than this range without any deleterious effects on development and with the added advantage of reducing the level of iron loading may be possible. This regimen maintains a mean pretransfusion level that does not exceed 9.5 g/dL.²⁵⁷ A transfusion program should not be started too early, and it should be initiated only when the hemoglobin level is too low to be compatible with normal development. If transfusion is started too soon, thalassemia intermedia may be missed, and the child may be transfused unnecessarily. Usually blood transfusions are given every 4 weeks on an outpatient basis. To avoid transfusion reactions, washed, filtered, or frozen red cells should be used so that the majority of the white cells and plasma-protein components are removed (Chap. 138).

IRON CHELATION

Every child who is maintained on a high-transfusion regimen ultimately develops iron overload and dies of siderosis of the myocardium. Therefore, such children must be started on a program of iron chelation within the first 2 to 3 years of life.²⁵⁶ Deferoxamine (desferrioxamine) was the first chelating agent of proven long-term value for treatment of thalassemia. It is best administered by an 8- to 12-hour overnight pump-driven infusion in the subcutaneous tissues of the anterior abdominal wall.^258,259 Chelation therapy should commence by the time the serum ferritin level reaches approximately 1000 mcg/dL. In practice, this level usually is seen after the 12th to 15th transfusion. To prevent toxicity, infants must not be overchelated when the iron burden is still low. The initial dose usually is 20 mg/kg 5 nights per week, with 100 mg of oral vitamin C (200 mg in older children and adults) on the day of infusion, after the infusion has been initiated.²⁵⁹ Some evidence and widespread opinion indicate ascorbate precipitates myocardiopathy in these patients if it is given before deferoxamine infusion is started.^260,261 In patients who are heavily iron loaded, particularly those patients with cardiac or endocrine complications, the body iron stores can be effectively lowered by continuous intravenous infusion of deferoxamine at a dose of up to 50 mg/kg body weight. The procedure usually entails insertion of an intravenous delivery system.

Extensive experience with the use of deferoxamine and its toxic effects has been reported.¹⁸⁹ No serious complications occur other than local erythema and painful subcutaneous nodules at the site of infusions and extremely rare severe allergic reactions. These reactions can be controlled, at least in part, by including 5 to 10 mg hydrocortisone in the infusion. Probably of greatest concern is neurosensory toxicity, which has been documented in up to 30 percent of cases. Toxicity causes high-frequency hearing loss that may become symptomatic.^262,263 In a few cases, the toxicity did not respond to discontinuation of the drug, and permanent hearing loss resulted. Ocular toxicity has been reported.²⁶² Symptoms include visual failure, night and color blindness, and field loss. Reversal of symptoms after discontinuation of the drug has been reported. Deferoxamine may cause bone changes and growth retardation, sometimes associated with bone pain. Body measurements characteristically show a reduced crown-pubis–to–pubis-heel ratio.²⁶⁴ These changes may be associated with radiologic abnormalities of the vertebral column. These complications can be prevented by exercising extreme care in monitoring patients receiving long-term deferoxamine therapy. Young children or individuals from whom most of the iron has been removed by chelation are at particularly high risk. Formal audiometry and ophthalmologic examinations at 6-month intervals are recommended.

Because of the practical difficulties of a nightly subcutaneous infusion of deferoxamine there has been an intensive search for effective oral chelating drugs. Two of these agents are currently available, deferiprone (Ferriprox, L1) and deferasirox. The extensive literature on these agents has been reviewed.^265,266,267 Deferiprone is administered at a dosage of 75 mg/kg in three daily doses. Unfortunately there have been limited numbers of long-term trials comparing its efficacy with deferoxamine, but overall it appears to be less effective than deferoxamine at maintaining safe body iron levels. Its administration is accompanied by a number of complications, the most important of which is neutropenia and, in some cases, agranulocytosis with some fatalities. Hence it is recommended that patients receiving this agent have a weekly white cell count. It also causes arthritis which varies in severity and between different ethnic groups. However, by virtue of its membrane-crossing capacity it has been suggested that it may be more effective in removing cardiac iron (Chap. 43). Unfortunately, to date, all the studies that suggest that it may reduce the frequency of cardiac complications in transfusion-dependent thalassemics have been retrospective and there are no long-term controlled data available. It is currently suggested that it should be used in combination with desferrioxamine, particularly for its cardiac-iron sparing effect; again, long-term prospective data are required.

The initial studies of deferasirox were promising²⁶⁶ and suggested that this agent in doses of 5 or 10 mg/kg per day, or higher in those who are heavily iron-loaded, was as effective as desferrioxamine in containing adequate hepatic iron levels. Preliminary clinical studies also showed that this agent may be effective for removing excess cardiac iron. Recent followup data have confirmed these early observations.²⁶⁷ The most frequent adverse reactions to deferasirox included gastrointestinal disturbances, transient rashes, and a nonprogressive increase in serum creatinine. It is still too early to be sure about the overall effectiveness of this agent, however, or to assess its long-term safety.

Because of the extremely well-documented data showing long-term survival of patients adequately treated with deferoxamine,^268,269,270 this agent is still recommended as a first-line choice for management of transfusion-dependent thalassemia. However, particularly in view of problems of compliance and the promising trial results of deferasirox, this drug is also being used increasingly as a first line form of treatment. Further long-term follow up data regarding its efficacy are still required however.

Careful monitoring of the degree of iron accumulation during chelation therapy is absolutely vital. The simplest approach, particularly in countries where most sophisticated technology is not available, is a regular estimation of the serum ferritin level, which should be maintained at less than 1500 mcg/L. The value of hepatic iron concentration assessment was discussed earlier in “Abnormal Iron Metabolism.” Newer noninvasive approaches to assessing body iron burden have been developed. There is now strong evidence that, with adequate calibration, the measurement and mapping of liver iron concentrations using magnetic resonance imaging (MRI) is an extremely effective approach for the regular assessment of the effectiveness of chelation therapy.²⁷¹ Similarly, there have been advances in the noninvasive estimation of myocardial iron using T2* MRI. Evidence obtained using this approach suggests that there may be a variable correlation between hepatic and cardiac iron concentrations.²⁷² Clearly functional cardiologic studies should be combined with assessment of cardiac iron levels, particularly the ejection fraction, pulmonary artery pressure, and other parameters of cardiac activity. The true value of these new approaches to assessing myocardial iron levels and function still require further study by prospective controlled trials.

Increasing evidence indicates children maintained at a high hemoglobin level do not develop hypersplenism.⁷ However, enlargement of the spleen with increased transfusion requirements occurs commonly in patients maintained at a lower hemoglobin level. Splenectomy should be performed if transfusion requirements increase dramatically or pain develops because of the size of the spleen. Because of the risk of overwhelming pneumococcal infections, splenectomy should not be performed in children younger than age 5 years. Patients should receive a pneumococcal vaccine prior to the procedure. They then should be placed on prophylactic oral penicillin after the operation. Haemophilus influenzae type B and meningococcal vaccines also are recommended.

Children with severe thalassemia are still prone to other infections. Presentation with abdominal pain, diarrhea, and vomiting should always suggest an infection with a member of the Yersinia class of bacteria. Empirical treatment should start immediately with either an aminoglycoside or a cotrimoxazole. Transfusion-transmitted virus infection is common in some populations. All chronically transfused patients should be tested annually for hepatitis C, hepatitis B, and HIV. Patients with serologic evidence of chronic active hepatitis should be considered for treatment.

As mentioned earlier in “Abnormal Iron Metabolism,” subtle endocrine deficiencies are increasingly recognized, particularly those associated with growth retardation and hypogonadism. These patients require expert endocrinologic assessment and replacement therapy when appropriate.

STEM CELL TRANSPLANTATION

By 1997, more than 1000 marrow transplants had been performed at three centers in Italy.^{273,274,275,276} Based on this experience and on later data,⁷ the prognosis evidently depended on the adequacy of iron chelation up to the time of transplantation. Hence, patients were divided into three classes: class I patients had a history of adequate iron chelation and neither liver fibrosis nor hepatomegaly; class II patients had one or two of these characteristics; and class III patients had all three characteristics. Among children in class I who had undergone transplantation early in the course of the disease, disease-free survival was assessed at 90 to 93 percent at 5 years, with a 4 percent risk of mortality related to the procedure. For class II patients, the intermediate-risk group, the survival and disease-free survival rates were 86 percent and 82 percent, respectively. For class III, the high-risk group, the survival and disease-free survival rates were 62 percent and 51 percent, respectively. Apart from the immediate complications of severe infection in the posttransplantation period, most of the problems were related to development of acute or chronic graft-versus-host disease. The overall frequency of mild to severe grades ranges from 27 to 30 percent.²⁷⁷ Modification of preparative drug regimens has reduced the frequency of drug toxicity. The occurrence of mixed chimerism may be a risk factor for graft-versus-host disease. No case of hematologic malignancy has been observed in the longest followup of patients between 15 and 20 years after transplantation. Recent experience has fully confirmed these pioneering studies²⁷⁸ and suggests that patients without matched donors could benefit from haploidentical mother-to-child transplantation. The current status of blood stem cell therapy is discussed further in Chap. 23.

GENERAL CARE

Management of thalassemia requires a high standard of general pediatric care. Infection should be treated early. If the diet is deficient in folate, supplements should be given. Supplementation probably is unnecessary in children maintained on a high-transfusion regimen. Particular attention should be paid to the ear, nose, and throat because of chronic sinus infection and middle-ear diseases resulting from bone deformity of the skull. Similarly, regular dental surveillance is essential because poorly transfused thalassemic children have a variety of deformities of the maxilla and poorly developed teeth. In the later stages of the illness, when iron loading becomes the major feature, endocrine replacement therapy may be necessary. Symptomatic treatment for metabolic bone disease and cardiac failure also may be needed.

THERAPIES OF SPECIAL TYPES OF THALASSEMIA

Hemoglobin H disease usually requires no specific therapy, although splenectomy may be of value in cases associated with severe anemia and splenomegaly.^7,9,10 Because splenectomy may be followed by a higher incidence of thromboembolic disease than occurs in splenectomized children with β-thalassemia,⁷ the spleen should be removed only in cases of extreme anemia and splenomegaly. Oxidant drugs should not be given to patients with hemoglobin H disease. The management of symptomatic sickle cell thalassemia follows the lines described for sickle cell anemia (Chap. 49).

Thalassemia intermedia presents a particularly complex therapeutic problem. Whether a child with a steady-state hemoglobin level of 6 to 7 g/dL should be transfused is difficult to determine with certainty. Probably the best compromise is to watch such children very closely during the first years of life. If they grow and develop normally and no signs of bone changes are evident, they should be maintained without transfusion. If, however, their early growth pattern is retarded or their activity is limited because of their anemia, they should be placed on a regular transfusion regimen. If hypersplenism plays a role in their anemia as the children grow older, splenectomy should be performed. Because many of these patients have significant iron loading from the gastrointestinal tract, regular estimations of serum iron and ferritin should be obtained and chelation therapy instituted when appropriate.

EXPERIMENTAL APPROACHES TO TREATMENT

Two main experimental approaches are being pursued in the search for more effective therapy of the thalassemias: (1) reactivation or augmentation of fetal hemoglobin production and (2) somatic gene therapy.

The main rationale for employing agents that have been used in attempts to increase hemoglobin F production is based on the observation that patients recovering from cytotoxic drug therapy or during other periods of erythroid expansion may reactivate hemoglobin F synthesis. In addition, the observation that butyrate analogues might have a stimulating effect on hemoglobin F production has led to a number of studies of their potential for management of thalassemia. A number of clinical trials have been performed.^{279,280,281,282} Agents that have been used include various cytotoxic drugs, erythropoietin, and several different butyrate analogues. Overall, these agents, used alone or in combination, have produced some small effects on fetal hemoglobin production, but the results of these trials have been disappointing. Some notable exceptions were seen, however, particularly several cases of homozygosity or compound heterozygosity for hemoglobin Lepore in which use of either a combination of sodium phenylbutyrate and hydroxyurea or hydroxyurea alone produced a spectacular rise in hemoglobin F production. In the case of two homozygotes for hemoglobin Lepore, the necessity for further transfusion was eliminated.²⁸³ This finding raises the intriguing possibility that certain mutations, possibly deletions of the β-globin gene cluster, are more susceptible to this type of approach. Recent progress in searching for genetic targets for modifying fetal hemoglobin synthesis has been reviewed recently.⁴²

The other experimental approach involves somatic gene therapy. Currently, the therapy is mainly directed at gene transfer into potential hematopoietic stem cells using retroviral vectors.²⁸⁴ Other approaches also are being taken, including attempts at the restoration of normal splicing in cases of splicing mutations²⁸⁵ and use of trans-splicing ribozymes to correct β-globin gene transcripts.²⁸⁶ However, studies using murine models with recombinant lentiviral vectors suggest that sustained, high-level globin gene expression may be possible, at least in this experimental system.^287,288 There continues to be slow progress toward somatic-cell gene therapy as applied to the hemoglobin disorders,²⁸⁹ with at least one apparent success and future plans for several clinical trials.

The prognosis for patients with severe forms of β-thalassemia who are adequately treated by transfusion and chelation has improved dramatically over the years. Three large studies investigated the influence of effective long-term desferrioxamine use on the development of cardiac disease.^268,269,270 In one study, patients who had maintained sustained reduction of body iron, as estimated by a serum ferritin level less than 2500 mcg/L over 12 years of followup, had an estimated cardiac disease-free survival rate of 91 percent. This finding is in contrast to patients in whom most determinations of serum ferritin level exceeded this value, in whom the estimated cardiac disease-free survival rate was less than 20 percent. In a second study, the relationship between survival and total-body iron burden was measured directly using hepatic storage iron values. Patients who had maintained hepatic iron concentrations of at least 15 mg of iron per gram of liver, dry weight, had a 32 percent probability of survival to age 25 years. No cardiac disease developed in patients who maintained hepatic iron levels below this threshold. These and other studies provide unequivocal evidence that adequate transfusion and chelation are associated with longevity and good quality of life. On the other hand, poor compliance or unavailability of chelating agents still are associated with a poor prospect of survival much beyond the second decade.

In parts of the world where the incidence of thalassemia is high, the disease places an immense economic burden on society. For example, if all the thalassemic children born in Cyprus were treated by regular blood transfusions and iron-chelating therapy, it was estimated that within 15 years the total medical budget of the island would be required to treat this single disease.²⁹⁰ Clearly, this approach was not feasible, so considerable effort was directed toward developing programs for prevention of the different forms of thalassemia.

The goal of prevention can be achieved in two ways. The first is prospective genetic counseling, that is, screening total populations while the children still are at school and warning carriers about the potential risks of marriage to another carrier. Few data are available about the value of programs of this type; a pilot study in Greece was unsuccessful.²⁹¹ Because it is believed this approach will not be successful in many populations, considerable effort has been directed toward developing prenatal diagnosis programs.

Prenatal diagnosis for prevention of thalassemia entails screening mothers at the first prenatal visit, screening the father in cases in which the mother is a thalassemia carrier, and offering the couple the possibility of prenatal diagnosis and termination of pregnancy if both mother and father are carriers of a gene for a severe form of thalassemia. Currently, these programs are devoted mainly to prenatal diagnosis of the severe transfusion-dependent forms of homozygous β⁺ or β⁰-thalassemia. Considerable experience has also been gained in prenatal diagnosis of mothers at risk for having a fetus with the hemoglobin Bart hydrops syndrome, considering the distress caused by a long and difficult pregnancy and the obstetric problems resulting from the birth of a hydropic infant with a massive placenta.

The first efforts at prenatal detection of β-thalassemia used fetal blood sampling and globin-chain synthesis analysis carried out at approximately week 18 of pregnancy. Despite the technical difficulties involved, the method was applied successfully in many countries and resulted in a reduced birth rate of infants with β-thalassemia.²⁹² The technique is associated with a low maternal morbidity rate, a fetal mortality rate of approximately 3 to 4 percent, and an error rate of 1 to 2 percent. Its main disadvantage is that it must be carried out relatively late in pregnancy. For this reason, efforts turned to first trimester prenatal diagnosis.

DNA technology has enabled diagnosis of important hemoglobin disorders in utero by fetal DNA analysis. Although analysis can be carried out on DNA derived from amniotic fluid, the approach has drawbacks because, again, it must be done relatively late in pregnancy, and often amniotic fluid cells must be grown in culture to obtain a sufficient amount of DNA.²⁹³ However, DNA can be obtained as early as week 9 of pregnancy by chorionic villus sampling. Although the safety of this technique remains to be fully evaluated and limb reduction deformities may occur when the procedure is carried out very early in pregnancy (9 or 10 weeks), chorionic villus sampling has become the major method for prenatal diagnosis of the thalassemias based on subsequent experience with the technique.^{7,293,294,295,296,297}

Remarkable advances in DNA technology have provided a variety of methods for the direct identification of mutations in fetal DNA⁷⁷ Even in families with extremely rare mutations, rapid DNA sequencing technology allows a diagnosis to be made very rapidly. The error rate using these different approaches varies, mainly depending on the experience of the particular laboratory; low rates, less than 1 percent, are reported from most centers. Potential sources of error include maternal contamination of fetal DNA and nonpaternity.

The application of this new technology has caused a major reduction in the birth rate of infants with thalassemia throughout the Mediterranean region and the Middle East, and in parts of the Indian subcontinent and Southeast Asia. Several approaches continue to be explored in an attempt to avoid the use of invasive procedures like chorion villous sampling. A variety of methods are being used to harvest fetal DNA from fetal cells in maternal blood or from maternal plasma^298,299 and there are increasing numbers of attempts at preimplantation diagnosis of thalassemias.^300,301 There is every expectation that some of these approaches will reach the clinic in the near future.³⁰²

THALASSEMIA AS A GLOBAL HEALTH PROBLEM

The remarkable advances in the diagnosis, prevention, and treatment of the thalassemias described in this chapter are only relevant to the richer countries of the world. In many developing countries in which there is a very high frequency of thalassemia, there are very limited facilities for their diagnosis and management. Because many of these countries are going through the epidemiologic transition, which involves improvements in nutrition, cleaner water supplies, and better public health services, babies with serious forms of thalassemia who previously would have died of infection or profound anemia are now surviving to present for treatment.

Approaches to the better control and management of the thalassemias in developing countries have been reviewed.^303,304 They include the development of partnerships between centers in the developed and developing countries for training workers in this field, and, once these partnerships are developed, for the further evolution of partnerships between those developing countries where there is knowledge and expertise of the field with those where no knowledge or facilities exist. Without organizations along these lines, the thalassemias will continue to cause the premature death of hundreds of thousands of infants worldwide.