Re: “Association Between the CYP4F2 Gene rs1558139 and rs2108622 Polymorphisms and Hypertension: A Meta-Analysis” by Geng et al. (Genet Test Mol Biomarkers 2019;23:342-347;DOI: 10.1089/gtmb.2018.0202)

Dear Editor,

Geng et al. (2019) recently published a meta-analysis on association of two CYP4F2 polymorphisms and hypertension, and included three and six studies for rs1558139 and rs2108622, respectively. As their main results “the dominant model for rs1558139, and both the homozygous model and the recessive model of rs2108622 were statistically significant giving values of 0.83 (0.71-0.96), 0.83 (0.71-0.98), and 1.24 (1.07-1.44), respectively.” However, there are some issues about their included studies, genetic methods meta-analysis, and results interpretation, which should be noticed.

First of all, there are some errors in the text. Genotype of Fan et al. (2011) study for rs2108622 in Table 1 is incorrect, and it should be presented as shown in Table 1. Also, author referred to Luo et al. (2015) study as “a meta-analysis of the rs2108622 polymorphisms and hypertension,” whereas in contrast they indicated that “no meta-analysis for rs1558139, rs2108622 polymorphisms, and hypertension has been reported.”

Missing studies are second issue in this meta-analysis. With regard to the study by Geng et al. (2019) “a manual search and a review of the references were performed to identify additional articles.” It seems that authors missed two articles associated with rs2108622: (1) author indicated Zhang et al. (2010) study as a related article, but missed this study in the meta-analysis. (2) Li et al. (2013) study was referenced in previous meta-analysis, whereas it had been missed from review of the references. The genotyping data in these case-control studies are in Hardy-Weinberg equilibrium and they need to be included. The genotype of these studies is presented in Table 1.

Third comment is related to meta-analysis and interpretation. First, the authors indicated that “A homozygote model, a recessive model, and a dominant model were used to analyze the eligible data”; however, they presented figure 2 based on allelic model. Also, meta-analysis was based on both alleles, which made both risk and protective effects in different genetic models and requires some modifications. Authors referenced C allele for rs15581139 (CC vs. TT) and G allele for rs2108622 (GG vs. AA) based on homozygote model, whereas T and A allele were selected in dominant (CT+TT vs. CC and GA+AA vs. GG) and recessive models (TT vs. CT+CC and AA vs. GG+GA), which is misleading and makes problems in interpretation. In this study, we carried out all meta-analysis of rs15581139 and rs2108622 based on C and G alleles to make consistent results (dominant model CC+CT vs. TT and GG+GA vs. AA; recessive model CC vs. CT+TT and GG vs. GA+AA). For rs15581139 significant risk effects in both of recessive and heterozygote models were seen, whereas for rs2108622, significant protective effects on dominant and homozygote models were seen (Table 2). In another meta-analysis we also included data related to two missed articles (Table 3). Also, we carried out subgroup analysis based on ethnicity. Interestingly, the result for East Asian population for rs2108622 was not significant (Table 3). Also, it seems to be better to attribute the results of rs15581139 variant to East Asian population.

Finally, based on different genetic models, it is clear that C allele of rs15581139 increases the risk of hypertension in East Asian populations, whereas G allele of rs2108622 decreases the risk of hypertension in overall population. Also, AA genotype increased risk of hypertension in more than two other genotypes (GG or AG).