GSTM1 and GSTT1 genetic polymorphisms and risk of anti-tuberculosis drug-induced hepatotoxicity: an updated meta-analysis

Abstract

The results of studies investigating the associations between GSTM1 and GSTT1 polymorphisms and anti-tuberculosis drug-induced hepatotoxicity (ADIH) risk exhibit much controversy. Therefore, a meta-analysis was performed in order to examine the associations between GST variants and ADIH risk. A total of 451 relevant studies were identified through the digital medical databases Medline, Embase, and CBM published up to October 2012. Thirteen individual case-control studies were eventually recruited for GSTM1 null polymorphism (including 951 ADIH cases, 1,922 controls) and 12 studies for GSTT1 null polymorphism (847 cases, 1,811 controls). Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were appropriately calculated from fixed-effects or random-effects models. Subgroup analyses were stratified by ethnicity and different treatment combinations. The overall ORs of relevant studies that exhibited elevated ADIH risk was significantly associated with GSTM1 null genotypes (OR = 1.36, 95% CI 1.04-1.79), but for the GSTT1 polymorphism, no difference was found (OR = 0.98, 95% CI 0.82-1.18). In the subgroup analyses, the pooled results showed that GSTM1 null allele carriers had a significant association with ADIH risk in East Asians and the patients who used isoniazid (INH) + rifampicin (RMP) + pyrazinamide (PZA) + ethambutol (EMB), or + streptomycin (SM) (HRZES), but the opposite result was observed for patients using HR. Moreover, the GSTT1 null genotype evaluated the susceptibility to ADIH for tuberculosis using HRZ. This meta-analysis provides evidence that there may be an increased risk of ADIH in individuals with null genotypes of GSTM1 in the total population, especially East Asians and patients receiving HRZE or HRZES. However, polymorphisms of the GSTT1 null genotype seem to have no association with susceptibility to ADIH, except for patients receiving HRZ.

PharmGKB summary

“…A recent large-scale meta-analysis of GST variants, which included 13 and 12 case-control studies for the GSTM1 and GSTT1 null genotypes, respectively (approximately 900 cases and 1900 controls for each gene), found evidence that the null genotype of GSTM1 , but not GSTT1 , was associated with marginally increased susceptibility to ATT-DILI [114], which was consistent with the previous meta-analyses [111, 112]. …”

Genetic Variations Associated with Anti-Tuberculosis Drug-Induced Liver Injury

“…A clinical study in an Indian patient population indicated that GSTM1 and GSTT1 null genotypes are associated with high rates of AT-DILI [64]. A meta-analysis that considered 451 relevant studies before 2013 found a strong association between certain GSTM1 genotypes and an elevated risk of AT-DILI after combined treatment with isoniazid and rifampicin [65]. Association of GSTT1 null genotypes with the risk of AT-DILI only exists in some East Asian populations [65].…”

“…A meta-analysis that considered 451 relevant studies before 2013 found a strong association between certain GSTM1 genotypes and an elevated risk of AT-DILI after combined treatment with isoniazid and rifampicin [65]. Association of GSTT1 null genotypes with the risk of AT-DILI only exists in some East Asian populations [65]. Another meta-analysis-based study of 20 published association reports collectively conducted on over 5000 subjects also revealed the presence of a significant association between GSTM1 genotypes and risk of AT-DILI, but no association with GSTT1 genotypes [66].…”

The Relationship of HLA-DQ Alleles with Tuberculosis Risk: A Meta-analysis

“…A p value less than 0.1 (I 2 [ 50 %) indicated statistical significant heterogeneity existed among the studies, and random-effects model was used to calculate ORs with corresponding 95 % CIs. In contrast, p values greater than 0.1 (I 2 \ 50 %) indicated no statistical significant heterogeneity existed among studies, and the fixed-effects model was used to calculate ORs with corresponding 95 % CIs [16]. We first calculated the combined ORs using the fixed-effects model.…”