An increase of liver enzymes is the most common adverse event observed in clinical practice in arthritis patients under long-term MTX therapy; however, the exact mechanism of such disorder remains unclear. Combining histological and immunological approaches, the present study indicated that the direct role of long-term low-dose MTX in the genesis of a MTX-specific liver change with dystrophic nuclei in hepatocytes was very rarely observed. In contrast, autoimmune lesions were commonly found.
Previous results have suggested a poor correlation between the level of increase of liver enzymes and histological findings on liver biopsies in arthritis patients under long-term MTX therapy [15–17, 24, 25]. In postmortem liver biopsies performed on 188 subjects with RA on long-term low-dose MTX therapy, Ruderman and colleagues reported only two cases of hepatic fibrosis, which were not directly related to MTX but related to either chronic alcoholism or chronic viral hepatitis . Richard and colleagues performed systematic liver biopsies prior to the introduction of long-course low-dose MTX and after 1 year of therapy. Eleven cases of hepatic fibrosis were observed prior to the introduction of MTX, with no sign of worsening at 1 year . More recently, no correlation was found between liver MTX concentrations and incidence of liver toxicity . Based on these results, liver biopsy is no longer performed on a systematic basis after exposure to usually 2 g MTX.
In the present study, histological findings with dystrophic nuclei consistent with a direct MTX toxicity  were found in only two patients. These changes are not found with nonsteroidal anti-inflammatory drug therapy, the drugs most commonly found in association with MTX. It should be noticed that NASH-like lesions were not associated with fibrosis and dystrophic nuclei. As such, they were not classified as MTX-related because of the absence of the last two parameters. These isolated liver changes could result from a large number of causes, including systemic inflammation itself, metabolic abnormalities, namely diabetes, and alcohol intake. In the other changes, the heterogeneity of the histological lesions (that is, mainly AIH-like and NASH-like) and the fact that none of these appeared as MTX dose-dependent suggest a poor link between these liver lesions and MTX administration. This conclusion is in contrast with previous but older studies that concluded a need to perform sequential liver biopsies to detect direct MTX liver toxicity [18, 29]. The current recommendations do not reach the same conclusion . In our study, neither NASH-like lesions nor the other histological lesions appeared MTX dose-dependent, consistent with previous data [30, 31]. These results are in favor of a weak direct liver toxicity of long-term MTX in arthritis patients. We reached the same conclusion on safety in patients with arthritis and hepatitis C treated with MTX or etanercept [32, 33].
Interestingly, a link between the histological liver lesions and the underlying rheumatic disease was observed. The association between autoimmune rheumatic disease and hepatic lesions has been previously reported. In our study, AIH-like lesions were strongly associated with RA and with the presence of the shared epitope. An interesting and new observation is that the presence of the antibodies commonly found in AIH (anti-SMA and anti-LKM1 antibodies ) was mostly negative in these RA-associated AIH-like lesions. The reason for this observation is unclear but could be linked to the rather modest intensity of the lesions. On the other hand, anti-CCP antibodies, also described in type 1 autoimmune hepatitis, were observed in almost 90% of patients with AIH-like lesions.
Overall, the results of the present study lead one to reconsider the role of autoimmunity in the induction of hepatic disorders in arthritis patients under long-term low-dose MTX therapy. Indeed, in these patients, elevated liver enzymes were rarely the consequence of direct MTX toxic effects. The presence of an underlying latent liver disease, possibly autoimmune in nature, was commonly found - especially in RA patients. In this particular context, exact causality assessment in drug-related liver injury remains a difficult challenge, even using the new criteria . To assess the real influence of autoimmune mechanisms in the genesis of hepatic disorders in arthritis patients on MTX treatment, it would be necessary to perform a prospective study with liver biopsy both before the introduction of MTX and during treatment in control patients without elevated liver enzymes. Today, such an approach is difficult - even impossible - to follow from an ethical point of view.