It follows from the preceding discussion that extrapolation across different eth-nogeographic groups for a given trait and a given drug may not be permissible. On the other hand, observations on ethnic specificity may be useful in other ways, e.g., to improve the diagnosis and clinical care of patients or to provide a fuller understanding of a given trait. For example, African, Mediterranean, and Oriental males affected by G6PD deficiency may be more susceptible to hemolysis induced by exposure to more than 200 therapeutics, but the potential to cause clinically significant hemolysis of drugs and other agents may differ from that among Caucasians. Thus, several agents (e.g., trinitrotoluene, quinidine, nitro-furazone, chloramphenicol) are capable of inducing hemolytic reactions of greater severity and of longer duration among Caucasians than among G6PD-deficient African-Americans.31
Another example concerns ethnic differences in response to alcohol. Screening of various populations indicates that an alcohol dehydrogenase (ALDH2) deficiency occurs at varying frequencies (8-45%) in populations of Mongoloid origin but is not found in Caucasian or Negroid populations. Facial flushing, an acute vasomotor dilation in response to ethanol, has attracted attention by its association with variant forms of ALDH2. Among Japanese, homozygotes and most heterozygotes for the atypical (''Asian'') ALDH2 are flushers, while those homozygous for the usual ALDH2 are nonflushers. Nearly 86% of Japanese subjects who always experienced facial flushing have inactive ALDH2, whereas infrequent flushing or an absence of flushing is associated with active ALDH2.64 As a consequence of the aversive vascular effects of ethanol, Japanese men and women with the ''Asian'' form of ALDH2 drink significantly less alcohol than those with the ''Caucasian'' form, and are more highly protected from alcoholism.65 Thus, ethnic variations have shown that flushing may act as a deterrent to ethanol abuse, but may also serve as a useful biomarker of the ''Asian'' ALDH2 phenotype that is easily perceived by patients and physicians.
A patient who exhibits an unexpected clinical response to a drug whose disposition and metabolism are dependent on a known pharmacogenetic trait characterized by ethnic specificities poses another situation of therapeutic interest. It should raise the question of whether the ethnicity of the patient suggests a basis for the response. Consider the CYP2D6* polymorphism as the trait of interest. As already mentioned, the disposition of many commonly used therapeutic agents is subject to control by this polymorphism, and differences in response to medicines between the extreme CYP2D6* phenotypes, the poor and ultrarapid phenotypes, can be quite dramatic (see Chapter 2). Since the frequency of CYP2D6* poor metabolizers is significantly higher among Africans and Caucasians compared to
Asians, an unexpected response to any given drug might be expected to occur more frequently among the former populations. Thus, the failure of poor meta-bolizers to experience analgesia from codeine66,68 and to be protected against dependence on the oral opiates would be more likely to occur among Africans and Caucasians than among Asian patients.69 Since poor metabolizers are more likely to experience interactions with other drugs70 and to experience the neu-rotoxic effects of amphetamine analogs such as 3,4-methylenedioxymetham-phetamine (MDMA, also referred to as ''Ecstasy''),71 these unexpected responses would also be expected to occur more frequently among Africans and Caucasians than among Asians. Similar considerations may apply to the analysis of the basis for an unexpected response to a given drug in connection with ethnic differences in the ultrarapid metabolizer phenotype. Accordingly, the unexpected failure to respond to nortriptyline,41 or of an unexpected exaggerated response of CYP2D6* ultrarapid metabolizers to codeine,66 would be expected to occur more frequently among African or Saudi Arabian patients than among Asian patients. Consequently, when a patient experiences an unexpected response to a given drug, the physician should consider whether the ethnic or geographic origin of the patient suggests a basis for the response.
The use of probe substrates alone or in various combinations to assess the contribution is noted in Appendix B, but caution should be observed in the selection of probe drugs for phenotyping in different ethnic populations. Wen-nerholm and colleagues evaluated four different CYP2D6 probes (codeine, debrisoquine, dextromethorphan, and metoprolol) in whites and in black Tan-zanian subjects carrying the African-specific CYP2D6*17 and *29 alleles.68 The data showed that CYP2D6*17 has altered substrate specificity in vivo compared with the common CYP2D6 variants in white subjects, and that the CYP2D6*29 variant contributes to slower metabolism of some CYP2D6 substrates in black Tanzanians.
Further study of genetically variable phenotypes exhibiting ethnic specificity may also provide new information about the structures of the genes and regulatory pathways that may be responsible for unexpected or unusual drug responses. Thus, ethnic specificity due to inactive or low activity enzymes, or to other defective protein variants, may be explained alternatively as a deletion of an
entire gene, , as truncated genes, , and as missense mutations of the coding region.76,77 Consider, for example, differences at the CYP2D locus between Chinese and Caucasians.78 Up to 50% of CYP2D6 alleles are accounted for by a variant in which a mutation at residue 34 replaces serine with proline. This substitution results in an unstable enzyme with lower mean CYP2D6 activity in Chinese persons compared to that in Caucasians, which explains the relatively slower metabolism of drug substrates for CYP2D6 and the lower doses of anti-depressant and neuroleptic drugs that are used among Chinese. Examples involving interethnic variation in which phenotypes are due to high activity enzymes
may be explained as regulatory variants, as kinetic variants, , and as duplicated genes.50,82
Another excellent example to illustrate how ethnic differences in dietary habits may provide a starting point for pharmacogenetic investigation of evolutionary issues is provided by the analysis of lactase nonpersistence (lactose intolerance). Lactase nonpersistence is a genetically unusual autosomal recessive trait that is determined by a polymorphic gene (LCT) as described in Appendix A. LCT is located on chromosome 2q21.83,84 For many years efforts to find LCT variants responsible for lactase persistence proved fruitless; but it was suggested that a c/s-acting element contributed to this phenotype.85,86 In 2002 a breakthrough occurred when a C/T single nucleotide polymorphism (SNP) located in the 13th intron of a completely different gene, MCM6, was found to be strongly associated with the trait in Finnish families.84 Sequence analysis of the complete 47-kb region of interest and association analysis revealed that a SNP DNA variant, C/T-13910, approximately 14 kb upstream of but separate from the LCT locus, was completely associated with lactase nonpersistence. A second variant, G/A-22018, 8 kb telomeric to C/T-13910 was also associated with the trait.84 Tissue-based and /n v/fro functional assays met the requirement that the variant had a c/s-acting effect on LCT promoter activity, and the occurrence of the variant C/T-13910 in distantly related populations indicated that it was very old. However, follow-up studies showed that while this SNP was a good predictor of lactase persistence in Europeans, it was a poor predictor in Africans even though they harbored what appeared to be lactase nonpersistent alleles.87 Pursuit of the possibility that African populations must harbor one or more additional variants that confer a similar lactase-persistence phenotype resulted in a second breakthrough by Tishkoff and colleagues.88 They sampled patterns of variation in 43 African ethnic populations, including dairying and nondairying groups, and discovered a new SNP, G/C-14010, that was also located in intron 13 of MCM6, and was significantly associated with lactase persistence. And like the C/T-13910 variant, it seemed to affect LCT promoter activity. Thus, Africans and Finns show similar patterns of lactase persistence but composed of different genetic variants.
In commenting on the discoveries of Enattah et al.84 and Tishkoff et al.,88 Wooding draws attention to the extraordinary evolutionary significance that attaches to these findings.89 Taken together, they tell us that divergent human populations have been under similar pressures involving milk, the main source of dietary carbohydrate, and have converged on the same solution of prolonging lactase phlorizin hydrolase (LPH) expression into adulthood. Not only are they a testament to the powerful evolutionary influence culture can exert on our genes, but they show it has done so at least twice in different regions of the world.
The discovery of racial differences in phenylthiocarbamide (PTC) taste sensitivity in the 1930s and 1940s first alerted biologists to population variations in the human drug response (see Figure 1.3). Since the demonstration of raciogeo-graphic variation in polymorphisms of G6PD and isoniazid (NAT2) acetylation in the 1950s, allele-specific differences in gene expression between ethnic populations has been an important part of pharmacogenetic analysis. Originally, allele-specific differences in levels of gene expression were primarily associated with epigenetic phenomena during development, with genomic X-inactivation and genomic imprinting being notable examples (see Chapter 6). More recent studies have shown that differences in allele-specific gene expression among nonimprinted autosomal genes are also relatively common.90,91 Such differences have also been shown to be heritable,92 and can be mapped as quantitative traits.90,93
Extensive pharmacogenetic studies of numerous metabolic and receptor traits, including G6PD deficiency and isoniazid acetylation (NAT2), have shown that nonsynonymous, allele-specific, coding sequence polymorphisms capable of modifying protein structure and function account for the ethnic variation in human drug response that has been identified. However, the proportion of the total variation in human genetic diversity for which such variations account is unclear. In contrast to coding sequence polymorphisms where the consequences of non-synonymous polymorphisms can be identified at the level of the protein pheno-type, the genetic cause or causes of quantitative variation are more difficult to define. More than 30 years ago, shortly after the discovery of the genetic code and before the invention of recombinant DNA technologies, Mary-Claire King and Allan Wilson performed a study to determine the genetic distance between humans and chimpanzees from comparisons of human and chimpanzee proteins. Of particular interest to this discussion were the conclusions that a relatively small number of genetic changes in gene expression systems, which they ascribed to redundancies in the genetic code or to differences in nontranscribed regions of the genome, were more likely to account for the major organismal differences between humans and chimpanzees than changes in amino acid sequences. Today, there is growing evidence at experimental and population levels that genetic variations in nontranscribed regions of the genome, as in regulatory sequences of the 5' and 3' untranslated regions, may underlie a substantial fraction of complex human disorders including ethnic variations of human drug responses.90,93 Other recent studies point to the possibility that synonymous polymorphisms, in contrast to nonsynonymous polymorphisms, may account, in part, for quantitative variations in protein phenotypes94-96 (see pp. 35 and 363).
Spielman et al.97 recently extended the genetic analysis of population differences from qualitative phenotypes to the quantitative expression level of genes. They determined the proportion of gene expression phenotypes that differed significantly between populations and the extent to which the phenotypic differences were attributable to specific genetic polymorphisms. Measurements of gene expression were made with the Affymetrix Genome Focus Array on annotated genes expressed in transformed lymphoblastoid cell lines from 142 persons of three populations taken from the International HapMap Project.98 Among 1097 gene expression phenotypes, 35 genes whose mean expression differed by 2-fold or more were identified from a total of 4197 genes expressed in the lymphoblastoid cell lines. Among the 1097 expression phenotypes, about 25% of those tested, there were marked differences in allele frequencies between populations. For the phenotypes with the strongest evidence of cis determinants, most of the variation was due to allele frequency differences at cis-linked regulators. Overall, Spielman et al. found that at least 25% of the gene expression phenotypes differed significantly between major population groups, and that specific genetic variations in allele frequency accounted for the difference in the most significant instances among the phenotypes that were cis regulated. In 11 phenotypes studied in detail, expression phenotypes were also largely attributable to frequency differences at the DNA level. They note that variants in coding regions of candidate genes do not account for a large proportion of disease sus-ceptibility,91 and they speculate, but do not demonstrate, that quantitative variations in gene expression are responsible instead.
In his review of regulatory polymorphisms underlying complex disease traits, Knight shows that allele-specific effects on gene expression are relatively common, are typically of modest magnitude, and are context specific.91 To date, most research has focused on the modulation of expression of regulatory polymorphisms by the process of transcriptional initiation. The functional characterization of such polymorphisms is problematic because of the many potential confounders of commonly used functional assays, and because many of the studies on allele-specific transcription factor binding and promoter analysis demonstrate no definitive mechanism beyond relative allelic differences in levels of transcription. Ways in which regulatory polymorphisms may act to alter gene expression are shown by modulation of transcriptional regulation of the Duffy binding protein, by modulation of alternative splicing at CTLA4 encoding cy-totoxic T lymphocyte antigen, and by modulation of translational efficiency by the serine protease factor, F12 (XII). To advance further our understanding of regulatory polymorphisms on gene expression, the role of DNA sequence polymorphisms will need to be considered more broadly in modulating gene expression.
Was this article helpful?