An abundance of population-specific monomorphic SNPs may or may not be meaningful: a commentary on differences in allele frequencies of familial hypercholesterolemia SNPs in the Malaysian population

In previous issue of the Journal of Human Genetics, Alex et al.¹ report the differences in allele frequencies of familial hypercholesterolemia (FH) single-nucleotide polymorphisms (SNPs) in the Malaysian population. As one of the most common inherited disorders in humans, autosomal dominant hypercholesterolemia (ADH) (MIM #143890) or its more common name, FH is caused predominantly by mutations in LDLR in the majority of patients. Additionally, less strong but important associations were also identified for ApoE and PCSK9. Either heterozygous or homozygous mutations in these genes undoubtedly led to hypercholesterolemia ranging from mild-to-severe phenotypes. However, these findings underplay several previous studies highlighting the importance of SNPs as genetic modifying factors in ADH, which affect cholesterol regulation differently across populations (reviewed in Fahed and Nemer²).

The article by Alex et al. aimed to identify ADH variants in self-identified multiethnic Malaysians including 140 clinically diagnosed FH patients and 111 healthy controls by evaluating 310 previously reported ADH-associated SNPs mainly on the three ADH-causing genes and a few genes involved in the cholesterol metabolic pathways. These SNPs under interrogation comprised either causative SNPs or merely linked to ADH SNPs (as a signpost), whereas multiple-nucleotide polymorphisms, as well as structural variations, were excluded from this study. Furthermore, the genotyping platform was a custom-designed microarray in which the designed probes were based on common SNPs from three databases, namely BHF, dbSNP and SNPedia.

The data presented by the authors were interesting, that is, in both subject groups 44.1% (137/310) monomorphic (mono-allelic) SNPs were found. Their report is very similar in terms of number on other genes in African Americans or European Americans³ and the Danish population⁴. However, the high proportion of invariable SNPs observed could be caused either by a rare haplotype presented in the Human Genome Project or possibly by genotyping errors, frequently found in the SNP database.⁵ Recently, 8.32% of biallelic coding SNPs in dbSNP were classified as artifacts, called ‘single nucleotide differences’ (SNDs), which result from highly similar (paralogous) genes.⁶ LDLR, which has several paralogs including VLDLR, LRP, LRP1B, LRP8 and LRP1 (www.genecards.org), is the only paralogous gene reported in Alex et al’s study. Given that the paralogs of LDLR potentially introduce SNDs, but only small number of monomorphic SNPs (28 out of 137) were reported, when comparing with other reported non-paralogous genes, APOB (73 SNPs) and PCSK9 (36 SNPs).

The authors, however, claimed that 23 monomorphic SNPs in Malaysians are truly unique as these SNPs in other populations published elsewhere were reported as polymorphic. For example, the authors reported rs61318752, which is a SNP in LDLR, as a monomorphic. Considering that some Malaysians are of Chinese descent, on the contrary, this rs61318752 SNP was previously reported for Han Chinese with MAF of 11.4%. Even though population-specific monomorphic sites could be explained by either low mutation rate or positive natural selection, before making such a conclusion, it is imperative to thoroughly consider the study populations, patient recruitment, SNP selection criteria, reliability of genotyping methods applied and publicly available SNP databases used for comparison.

The research population studied by Alex et al. was 251 Malaysians consisting of Malay (61.8%), Indian (22.7%) and Chinese (15.5%). Such samples are of Southeast Asian ancestry in which the data could provide a better understanding of genetic factors influencing ADH by the race than those reported with European ancestry (reviewed in Fahed and Nemer²). However, using larger samples that are ethnically homogeneous would be more appropriate as the allele frequency estimated from small groups of controls deviates more from the true frequency than those from the larger ones.⁷ Alex et al. recruited the ADH patients on the basis of the commonly used Dutch Lipid Clinic Network criteria.⁸ The SNP selection criteria were comprehensive and also based on effects of SNPs to the disease information collected and curated in SNPedia (http://www.SNPedia.com). Nonetheless, the variant causality in the analysis can be frequently misclassified by either inclusion of non-causal variants or exclusion of causal variants⁹. Genotyping results by Alex et al. were validated by direct sequencing of randomly selected samples. The authors compared their results with multiple SNP databases including dbSNP, HAPMAP, mutDB and BHF to assess the uniqueness of the SNPs uncovered across worldwide populations.¹⁰ It would be better if a recent database of Pan-Asian population, PanSNPdb¹¹, was also taken into account.

The presence of monomorphic SNPs was rarely mentioned and analyzed in previous reports. This is possibly due to publication bias; positive findings were published, whereas negative ones were ignored. Alex et al. presented the first evaluation of population differentiation in allele frequencies for ADH-associated variants in Malaysians (as indicated by high Fst levels of 0.1285–0.3552), contributing to ADH molecular research. The authors should not prematurely draw the conclusion that these SNPs could be the targets of positive selection using Fst values.¹² However, further studies of the disease in Malaysians and possibly other Southeast Asian populations could utilize these data to select informative variants for either disease-associated studies or molecular screening for preventive medicine in the future.