Statistics from Altmetric.com
In a case control study on the associations between functional genetic polymorphisms in biotransformation enzymes and Crohn’s disease, we found a strong association between the Tyr113His (348T>C) polymorphism in exon 3 of the microsomal epoxide hydrolase (EPHX1) gene and Crohn’s disease.1 The three referees all agreed that the study was interesting and should be published so that other groups can attempt to replicate the results in independent study cohorts. This was done recently by Cuthbert and colleagues (Gut 2004;53:1386) who investigated 344 controls and 307 patients with Crohn’s disease, and who were unable to reproduce our results. In addition, they reported that our data for the EPHX1 exon 3 polymorphism in the control group were not in Hardy-Weinberg equilibrium (HWE), as also noticed previously by Györffy and colleagues.2 Our data on EPHX1 exon 3 genotyping were obtained by restricted fragment length polymorphism (RFLP) analyses by applying the method described by Lancaster and colleagues.3
However, recently it was reported that a silent substitution polymorphism (G to A) at codon 119 of the EPHX1 gene may exist, which may flaw the polymerase chain reaction (PCR) RFLP method applied by us, as the presence of this polymorphism may disturb proper binding of the reverse primer, covering the 119 G>A area, resulting in over-classification of His113 alleles.4,5 Therefore, we developed a dual colour allele specific discrimination assay for genotyping the polymorphism at codon 113 of the EPHX1 gene. EPHX1 genotypes were detected with the iCycler iQ Multicolour Real Time Detection System (Bio-Rad Laboratories, Veenendaal, the Netherlands) using molecular beacons. PCR was performed with the forward primer 5′-CAA CTC CAA CTA CCT GAA G-3′ and the reverse primer 5′-TGA CAT ACA TCC CTC TCT G-3′ in the presence of the FAM labelled wild-type beacon (5′-CGC GAT GAT TCT CAA CAG ATA CCC TCA CTT CAA TCG CG-3′) and the HEX labelled mutant beacon (5′-CGC GAT ATT CTC AAC AGA CAC CCT CAC TTC AAT CGC G-3′). The 25 μl reaction mixture contained 200 ng of genomic DNA, 10 mM Tris/HCl (pH 9.0), 50 mM KCl, 0.1% Triton X-100, 4 mM MgCl2, 0.25 mM dNTPs, 50 ng of each primer, 200 nM of each beacon, and 2.5 U Taq-DNA-polymerase. The PCR conditions were three minutes at 95°C, then 40 cycles of 30 seconds at 95°C, 30 seconds at 59°C, and 30 seconds at 72°C. Fluorescent signals were measured at 59°C. Genotypes were assigned using the iCycler iQ Optical System, software version 3.1. At each PCR run (in 96 well plates) sterile H2O instead of genomic DNA was added in several wells as a negative control for amplification. As the PCR-RFLP analyses were performed in the first half of 1999, only some of the samples were still available (125 of 149 controls and 149 of 151 cases) and these were re-evaluated by the iCycler method.
Genotype distribution of the EPHX1 Tyr113His polymorphism in patients with Crohn’s disease and controls was now in HWE (χ2 = 2.47, p = 0.12 and χ2 = 0.82, p = 0.37, respectively) and genotype distribution was not significantly different between cases and controls (χ2 = 3.5, p = 0.17). The Tyr allele frequencies of 0.70 and 0.68 obtained for cases and controls, respectively, were very similar to the corresponding values of 0.71 and 0.70, as reported by Cuthbert et al.
Thus in answer to the question as posed by Cuthbert et al: “Genetic association between EPHX1 and Crohn’s disease: population stratification, genotyping error, or random chance?”, we can conclude that a genotyping error was responsible for our earlier published association between the EPHX1 Tyr113His polymorphism and Crohn’s disease.1 We now clearly state that our revised data do not support an association between this EPHX1 polymorphism and Crohn’s disease. Similar genotyping errors may also be present in several other studies on the EPHX1 exon 3 polymorphism in association with a variety of diseases, as many studies were based on methods using a reverse primer covering the “119 silent mutation area” of the EPHX1 gene.3,6–9 This may also have consequences for interpretation of results in the cited papers. However, a rapid literature search by Pubmed revealed more than 100 papers on EPHX1 polymorphisms over the past 10 years, suggesting that many more papers may deal with genotyping problems, as outlined above.
In addition, Cuthbert et al also reported that another polymorphism tested in our study,1 the CYP1A1 exon 7 Ile/Val polymorphism, was not in HWE in the control group. This is correct but this deviation from HWE may be attributed to random chance, due to the rarity of the Val allele in our population, which makes the χ2 test inappropriate under such conditions. For instance, genotype distribution is in accordance with HWE when only two individuals less would have been classified as Val/Val homozygotes.
We thank Cuthbert et al and Györffy and colleagues2 for their interest in our work. In addition, we conclude that (interpretation of) data in many other published studies on the EPHX1 Tyr113His (exon 3) polymorphism should be critically re-evaluated.
Conflict of interest: None declared.