Three Genetic Polymorphisms of Homocysteine-Metabolizing Enzymes and Risk of Coronary Heart Disease: Appraisal of a Recent Meta-Analysis

W e read with great interest the recent article entitled “Three Genetic Polymorphisms of Homocysteine-Metabolizing Enzymes and Risk of Coronary Heart Disease: A Meta-Analysis Based on 23 Case-Control Studies” by Chen et al. (2011). To date, several studies have focused on the role of homocysteine-metabolizing enzymes polymorphisms in coronary heart disease (CHD) risk, however, the results of those studies remain inconclusive and ambiguous. Chen et al. (2011) conducted a meta-analysis to examine the associations between three genetic polymorphisms of genes encoding homocysteine-metabolizing enzymes (methionine synthase [MTR] A2756G, methionine synthase reductase [MTRR] A66G, and N⁵, N¹⁰-methylenetetrahydrofolate reductase [MTHFR] A1298C) and susceptibility to CHD. The results of this meta-analysis suggest that MTR A2756G polymorphism is associated with an increased risk of CHD for Europeans, whereas, for MTRR A66G and MTHFR A1298C, no obvious associations were found. It is an interesting study. Nonetheless, several issues related to the article seem worthy of comment.

First, although the meta-analysis stated that the observed frequencies of genotypes conformed to Hardy–Weinberg equilibrium (HWE) in control groups, one study of MTHFR A1298C (Abu-Amero et al., 2003) was out of HWE. Several results of HWE test in their meta-analysis were wrong. As shown in Table 1, another four studies violated HWE for genotypic distribution of the control population.

Second, in the methods section, they stated that “statistical heterogeneity was measured using the Q-test, and p<0.05 was considered statistically significant.” p≤0.10 rather than <0.05 usually indicates the heterogeneity between studies (Kavvoura and Ioannidis, 2008). In addition, when there was lack of heterogeneity (p>0.10) among the included studies, the fixed-effects model should be used. Some erroneous applications of model were found in the results section of the associations between MTRR A66G and MTHFR A1298C polymorphisms and risk of CHD. As described in the Table 4 by Chen et al. (2011), the subgroup analysis for MTRR A66G polymorphism in hospital-based case-control population should use fixed-effects model since the p-value based on Q-test is much greater than 0.10. Similarly, in the Table 5 (Chen et al., 2011), fixed-effects models should be used for the comparison CC versus AA in total population and all subgroup analyses. Additionally, they also should select the fixed-effects models when conducting the comparison CC versus CA+AA in total population and all subgroup analyses except for larger group subdivided by sample size.

Third, stratified analysis by ethnicity should be more carefully considered. The meta-analysis by Chen et al. (2011) contained four studies of Indian populations which should be cautiously treated when performing subgroup analysis by ethnicity (Ranjith et al., 2003; Angeline et al., 2007; Vinukonda et al., 2009; Vijaya Lakshmi et al., 2011). It has been shown (Saha et al., 2005) that Indian populations exhibit a high degree of genetic admixture and a greater genetic proximity with other world populations. Studies have indicated that the south Indian population is genetically closer to Asian populations (Bamshad et al., 2001), whereas the north Indian population is closer to Caucasian populations (Khan et al., 2007). When a study contains a mixed population, it should be divided into the different ethnicities if data are available. Therefore, it might be more reasonable to treat the ethnicities of the studies by Vinukonda et al. (2009) and Angeline et al. (2007) as Asian, whereas the populations from the studies by Ranjith et al. (2003) and Vijaya Lakshmi et al. (2011) should be considered as mixed populations rather than Caucasian due to indefinite geographical location.