Lumican and muscarinic acetylcholine receptor 1 gene polymorphisms associated with high myopia

Sir,

Lin et al¹ report that polymorphisms in the lumican gene are associated with high myopia. They studied four SNPs in 182 highly myopic cases and 78 emmetropic controls, recruited from a sample of 3000 Taiwanese-born medical students aged 16–25 years old (Table 1a). They also published a second study addressing the same question,² testing five SNPs—including four of those above—in 201 high myopes and 86 emmetropic controls, also recruited from 3000 Taiwanese-born volunteers aged 16–25 years old. For the four SNPs common to both studies, the results were similar.

Table 1a HWE test P-values originally reported by Lin et al³

In neither article is the other study referred to, which suggests that the two investigations are unlikely to be an original study followed by a replication study. Thus, it seems possible that the same SNPs were genotyped twice (albeit using different methods) in a common set of subjects and reported as being independent results. If true, this would contravene established scientific practice (because it produces a false impression of the weight of evidence supporting a gene–disease association).

In another recent publication, Lin et al³ genotyped four SNPs in the CHRM1 gene in their Taiwanese case–control subjects. Genotype frequencies were hugely different (P<10⁻⁷) between cases and controls for two SNPs (rs544978 and rs544269) suggesting strong association.³

Lin et al tested for departure from Hardy–Weinberg equilibrium (HWE), as this can be indicative of genotyping error⁴ (Table 1a). We repeated these tests (R genetics HWE.test function) using the data in Table 2 of their article.³ Our results (Table 1b) showed that both myopia-associated SNPs were not in HWE in controls and that Lin et al's HWE test results were erroneous. Although departure from HWE can occur in case–control samples if the disease–polymorphism association is strong,⁴ such strong association was not detected at the CHRM1 locus in a high-myopia GWA in Japanese subjects.⁵ The possibility of genotyping error weakens the evidence of an association between CHRM1 polymorphisms and high myopia. We also noticed inconsistencies between the data reported by Lin³ and the HapMap database, for example in the alleles of rs544978 and the allele frequencies of rs2186410 (Table 2a and b).

Table 1b HWE test P-values re-calculated from the data of Lin et al³

Table 2a Characteristics of CHRM1 polymorphisms reported by Lin et al³

Table 2b Characteristics of CHRM1 polymorphisms in the HapMap database