Reply:^†‡§

We appreciate the comments of Leiro et al. on the role of the combined glutathione-S-transferase M1 (GSTM1) and glutathione-S-transferase T1 (GSTT1) null genotypes in determining the susceptibility to developing idiosyncratic drug-induced liver injury (DILI) as a general mechanism that occurs regardless of the type of drug involved.1 These enzymes are predominantly expressed in the liver, and apart from being involved in drug metabolism, they modulate the byproducts of oxidative stress, an important common pathway in liver injury.2 Thus, it is not surprising that this geneticvariation may increase the risk of hepatotoxicity with several drugs independently of the final pathway leading to damage (immunological versus metabolic). Actually, no differences were found in the glutathione-S-transferase genotypes when the cases were compared according to the presence (n = 39, 25%) or absence of any of the classic hypersensitivity features.

The reasons for performing an analysis of deletions for both genes were based on the fact that single nucleotide polymorphisms leading to functional changes in GSTM1 and GSTT1 enzymes had not been identified in Caucasian subjects.3 Indeed, among the 154 DILI patients, the frequencies of the GSTM1 null and GSTT1 null genotypes were 56% and 29%, respectively. These figures were higher than those observed in the control group but not significantly different. A detailed distribution of the number of active genotypes of GSTT1 and GSTM1 in the main pharmacological classes of drugs is shown in Table 1. Interestingly, other studies have focused on antituberculosis drugs, using a case-control methodology, and they have investigated whether each genotype or the combined GSTM1 and GSTT1 null genotype could influence the susceptibility to developing hepatotoxicity. They have shown conflicting results, probably because of the small size of the cohort of affected patients, which lacks statistical power.4

Putative genetic differences among control subjects if healthy individuals, instead of individuals matched for exposure to the drug, were used as controls would cause a bias, if any, toward the null hypothesis. Therefore, the design of our study fully supports the findings of a prominent effect of the double-null GSTM1 and GSTT1 genotype on DILI susceptibility. Nevertheless, the choice of controls for DILI genetic studies is challenging, and the ideal controls would be matched for demographics and equivalent exposure to the implicated drug. This statement is particularly relevant in the case of tuberculosis because the disease may itself be associated with the at-risk genotypes. A prospective study of all patients started on antituberculosis therapy to collect cases of hepatotoxicityis ongoing. Clearly, collaborative efforts will be necessary to identify bona fide cases5 and to promote an understanding of the complex mechanisms involved in DILI.

Funding: Grant by EC 07/90910