Statistics from Altmetric.com
We read with interest the report by Yimet al of a failure to observe an association between polymorphisms of the microsomal epoxide hydrolase (mEPHX) gene and chronic obstructive pulmonary disease (COPD).1 This contrasts with the findings of earlier studies.2
There is debate in the literature on the place of association studies in the investigation of late onset complex disorders.3Failure to replicate an initial report of a positive association is common4-6 and it is important that the reasons for this are established.
The authors correctly state that their failure to replicate the results of earlier studies may be a reflection of the marked racial differences in the frequency of the mEPHX gene within their population. However, their study also lacks power. Given, for example, their reported frequency of 75% for the wild type homozygous exon 4, a sample size of 80 subjects would only be able to detect a difference of 22% (e.g. 75% versus 53%) between the case and control groups (two tailed p value = 0.05, power = 0.8).
Phenotypic heterogeneity is a problem in the genetic dissection of complex traits and hampers comparisons between studies. The authors are rigorous in their spirometric criteria used to define cases. However, their COPD group also includes never smokers and those with minimal pack-year histories. It is not restricted to adult onset disease, potentially containing chronic asthmatics. This phenotypically heterogeneous group could reduce the likelihood of demonstrating an association. The calculation of phenotypic “scores” is one solution to the clinical diversity which the label “COPD” describes.7
Finally, the importance of age and sex matching cannot be understated, not only because controls may develop disease, but also to minimise the effects that will occur to the gene pool as the population ages. We would therefore urge caution before abandoning a role for this candidate gene in this population. Further studies in extended populations with rigorously matched controls are clearly needed.
authors' reply We thank Dr Ruse and colleagues for their interest and comment on our study.1-1 They mentioned three points: (1) sample size, (2) the possibility of including asthmatic patients in the COPD groups, and (3) failure of age and sex matching between the disease and control groups.
We agree with them that our sample size was not large enough to detect small differences between the two groups (COPD 83, control 76). The strict criteria used in our study to select patients with disease or healthy smokers made our sample size smaller.
They suggested the possibility that we may have included asthmatic patients in the COPD groups because of the minimal smoking history in some patients. It is well known that there are risk factors for developing COPD other than smoking history such as environmental tobacco smoking (passive smoking), ambient air pollution, and occupation. It is therefore possible for non-smokers to develop COPD. Although we vigorously excluded patients with minimal asthmatic features in order to select a phenotypically homogeneous group, it is true that some patients with chronic asthma cannot be differentiated from patients with COPD by any method.
Gene frequencies do not vary according to sex in the general population and the lack of sex matching in our study may not influence the result. When we excluded six women from the COPD group the result was the same. Although we adjusted for the effect of age by stratification, it is clear that an age matched control group would have been better. The first and only study which suggested the role of genotypes of microsomal epoxide hydrolase (mEPHX) in the pathogenesis of COPD also lacked age and sex matching because the control group was anonymous.1-2
We agree with Dr Ruse and colleagues that further large scale rigorously matched case control studies are needed to clarify the role of this candidate gene in the pathogenesis of COPD.
Jae-Joon Yim et al 2-1 reported that genetic polymorphisms in microsomal epoxide hydrolase (mEPHX), glutathione-S transferase (GST) M1, and GST T1 genes are not associated with the development of chronic obstructive pulmonary disease (COPD) in Koreans. However, we strongly propose the possibility that the frequency for the mutant type of mEPHX exon 3 polymorphism (codon 113) was overestimated in their study as we have recently found a haplotype with a novel polymorphism (codon 119, accession #AB035519) within the antisense primer they used, and thus half of the individuals heterozygous at codon 113 could be misclassified as homozygous mutant using the same protocol in Japanese subjects.2-2 In fact, the allele frequency for mEPHX exon 3 polymorphism reported by Yim et al 2-1 was not in Hardy-Weinberg equilibrium, suggesting the existence of some problems.
We have investigated the association between mEPHX gene and susceptibility to COPD in Japanese subjects and identified a novel single nucleotide polymorphism at codon 119 (AAG to AAA) 20 bp downstream of the codon 113 polymorphism (estimated allele frequency 0.29). This is a silent substitution and is unlikely to have any biological significance by itself. However, the variant type of this polymorphism (AAA) showed strong linkage disequilibrium with the wild type at codon 113. Since the novel polymorphism at codon 119 existed within the antisense primer used for codon 113 polymorphism, in individuals with 113:wild and 119:variant in one allele and 113:variant and 119:wild in another, the latter allele with the higher homology to the antisense primer was preferentially amplified as if it was an homologous variant for codon 113. In the Japanese population about half of the wild allele for codon 113 showed variant at codon 119 and almost all the variant alleles for codon 113 showed wild at codon 119.2-2 As a consequence, about half the individuals heterozygous at codon 113 were misclassified as homozygous variants and the allele frequency was not in Hardy-Weinberg's equilibrium using the primer set used by Yim et al 2-1and Smith and Harrison.2-3 The miscalculated allele frequency was the same as that reported by Yim et al.2-1 The true genotype at mEPHX codon 113 could be determined by direct sequencing of the PCR products amplified with an antisense primer designed outside the original one. Since previous reports2-3 2-4 in Caucasians using the same protocol reported quite a low frequency of the homozygous mutant at codon 113, and the allele frequency was in Hardy-Weinberg's equilibrium, the novel polymorphism at codon 119 is unlikely to exist in Caucasians. Thus, the novel polymorphism at codon 119 or the haplotype at codons 113 and 119 are probably specific for Asians and we strongly suggest that Yim et al should carry out direct sequence analyses for codon 113 using an antisense primer outside the codon 119 to determine the true genotype and to re-evaluate the relationship with COPD in the Asian population.
authors' reply We appreciate the comment by Dr Yoshikawa and colleagues on the possibility of overestimating the frequency of homozygous mutant genotype of microsomal epoxide hydrolase (mEPHX) exon 3 and agree that further explanation is needed for the fact that in our study3-1 the allele frequencies of mEPHX in exon 4 are in Hardy-Weinberg equilibrium but those of exon 3 are not. The suggestion by Yoshikawa et al 3-2 that patients with a heterozygous genotype of mEPHX exon 3 can be misclassified as a homozygous mutant due to polymorphism at codon 119 may be a good explanation for this observation, and we plan to sequence the PCR product of exon 3 amplified with an antisense primer outside the original one we used. We expect this to reveal the prevalence of a single nucleotide polymorphism at codon 119 of exon 3.
There is one fact which is overlooked by Dr Yoshikawa and colleagues. In our opinion there is no reason to assume that the prevalence of a single nucleotide mutation at codon 119 of mEPHX is different between patients with COPD and healthy smokers and, if the prevalence of mutation at codon 119 of mEPHX exon 3 is similar in the two groups, the real distributions of genotypes of mEPHX exon 3 are also similar.
As mentioned above, the sequencing of the PCR product of exon 3 amplified with an antisense primer outside the original one will clarify this confusion and further research on the functional significance of a single nucleotide polymorphism at codon 119 of mEPHX exon 3 will provide us with a more complete understanding of this polymorphism.