Drug-induced liver injury due to nitrofurantoin: similar clinical features, but different HLA risk alleles in an independent cohort

We read with great interest the recent article on the subject of nitrofurantoin-related drug-induced liver injury (DILI) in the DILIN cohort from Chalasani and colleagues. Independently of the DILIN project, projects entitled DILIGEN and iDILIC, involving both retrospective and prospective recruitment of DILI cases outside the USA for study of genetic susceptibility, took place from 2004 to 2014. A total of 26 nitrofurantoin DILI cases from these studies passed expert adjudication and were classified as possible, probable or highly probable based on Roussel Uclaf Causality Assessment Method (RUCAM) causality scores of 3 or higher. These cases were included in a recent genome-wide association study (GWAS) on DILI. The DILIGEN/iDILIC cases were recruited from the United Kingdom (n=21), Netherlands (n=3), Australia (n=1) and South America (n=1). All had white European ethnicity. Interestingly, the clinical features of this cohort were very similar to those described in the recent report with 85% of cases female, a mean age of 62 years and the majority of cases showing hepatocellular injury (73%). For RUCAM, 70% had a score of 6 or above which represents probable or highly probable causality. When we divided the cohort on the basis of nitrofurantoin exposure time, 31% had received the drug for <7 days, 31% for 8 to 365 days and 38% for >365 days. This longer exposure group is smaller in our cohort than in the recent DILIN report.

As previously described in detail, HLA genotype can be imputed from GWAS data with a high level of reliability. Imputed HLA genotypes had already been obtained using this approach for all 26 nitrofurantoin cases as part of the larger GWAS. As summarised in Table 1, which shows HLA alleles showing differences between cases and controls with uncorrected p values <0.05, certain HLA alleles were more common among the nitrofurantoin DILI cases compared with population controls. The strongest association was with HLA-A*33:01, which we reported previously to be a risk factor for DILI due to several drugs, particularly terbinafine. The haplotype HLA-DRB1*07:01–HLA-DQA1*02:01–HLA-DQB1*02:02, previously reported by others to be a risk factor for some forms of DILI (e.g. due to lapatinib and asparaginase),^, also appears more common in our cohort but none of our nitrofurantoin DILI cases had exposure to other drugs for which this haplotype association had been reported. None of the cases in our cohort were positive for DRB1*11:04, in contrast with the findings of Chalasani et al.. It is possible that we did not detect this association in our cohort due to a lower incidence of cases with DILI after prolonged nitrofurantoin exposure. However, we did not see a significant increase in frequency of HLA-A*01:01 in our cohort either (OR 1.29 (95% CI 0.66-2.54), p= 0.45), even though this was detected more commonly in patients with short nitrofurantoin exposure in the DILIN cohort.

Despite the fact that it is possible to obtain interesting data on genetic risk factors for DILI, especially HLA, using small case cohorts, we believe some caution is needed. In particular, use of replication cohorts is generally recommended in genetic studies. Assembly of a larger cohort of nitrofurantoin DILI cases would be useful both in resolving the issue of at risk HLA genotypes and in identifying non-HLA genetic risk factors for this serious and relatively common form of DILI.