Current issues of ACP Journal Club are published in Annals of Internal Medicine


Meta-analysis: Long-term, low-dose methotrexate for rheumatoid arthritis or psoriatic disease increases the risk for liver toxicity

ACP J Club. 1991 Sept-Oct;115:61. doi:10.7326/ACPJC-1991-115-2-061

Source Citation

Whiting-O'Keefe QE, Fye KH, Sack KD. Methotrexate and histologic hepatic abnormalities: a meta-analysis. Am J Med. 1991 Jun;90:711-6. [PubMed ID: 1828327]



To assess the risk for hepatic fibrosis and cirrhosis in patients with rheumatoid or psoriatic arthritis who were receiving low-dose methotrexate therapy.

Data sources

English language articles that reported liver biopsy data on patients who were treated with low-dose methotrexate for psoriasis, psoriatic arthritis, or rheumatoid arthritis were identified through a computerized search.

Study selection

Studies had to have used a histologic description mappable to the Roenigk classification (grade I [normal] to grade IV [cirrhosis with fibrosis and nodular regeneration]) and to have given comparable information on dose and duration of methotrexate use. 15 studies were identified.

Data extraction

2 investigators independently collected data from the articles on dosage and duration of treatment with methotrexate, hepatic histologic grades, and amount of alcohol consumed by patients (≥ or < 100 g/wk). Results were analyzed by study, weighted for numbers of patients.

Main results

Of 636 patients, 334 had rheumatoid arthritis and 299 had psoriasis or psoriatic arthritis. Mean duration of therapy was 211 weeks, and mean total cumulative dose of methotrexate was 2536 mg. The maximum weekly dose was 30 mg; the mean dose was 11.3 mg/wk. 178 patients (28%, 95% CI 24% to 32%) progressed ≥ 1 histologic grade on liver biopsy; advanced liver disease (histologic grades IIIB or IV) occurred in 32 patients (5%). 68 patients with light alcohol consumtion (26%) progressed compared with 11 patients with heavy alcohol consumption (73%) {CI for the 47% difference 24% to 70%}* (P < 0.001). The incidence of advanced changes was greater in patients who were heavy drinkers (18%) than those who were light drinkers (5%) (P < 0.001). Fewer patients with rheumatoid arthritis (24%) than patients with psoriatic disease (33%) progressed {CI for the 9% difference 2% to 16%}* (P = 0.02). Patients with rheumatoid arthritis were less likely to have advanced disease than those with psoriatic disease (3% vs 8% {CI for the 5% difference 1.5% to 8.5%}* P = 0.005). Using linear regression anaylsis, the mean cumulative dose was associated with histologic progression (P = 0.01), but duration, dose/wk, or the maximum dose/wk was not, and none of these was associated with progression to histologic grades IIIB or IV. The mean risk for progressing 1 histologic grade/1000 mg methotrexate consumed was 7% (CI 2% to 11%).


Patients who are receiving long-term, low-dose methotrexate for rheumatoid arthritis or psoriatic disease are at risk for progressive changes evident on hepatic biopsy. The risk increases with cumulative dose of methotrexate and for heavy drinkers.

Source of funding: Not stated.

Address for reprint: Dr. Q.E. Whiting O'Keefe, Medaware, Inc., 2202 Edgewood Road, Redwood City, CA 94062, USA.

*Numbers calculated from data in article.


Low-dose methotrexate is being used with increasing frequency as treatment for rheumatoid and psoriatic arthritis. With prolonged use, methotrexate clearly causes clinically apparent liver disease. Histologic abnormalities are not reliably preceded by clinical symptoms or liver enzyme elevations. Consequently, most clinicians will monitor, using routine liver biopsy, patients who are receiving such therapy. This study provides information on the results of such monitoring.

The study suffers, however, from the common problems of meta-analyses. It could not account for the studies that were not included because they were missed by the investigators' literature search or selection criteria. Also, because untreated patients were not available, it is uncertain how much of the observed liver disease was caused by the methotrexate. In particular, it is unclear how much of the progression in patients who drank alcohol would have occurred even without the methotrexate. Nevertheless, it is unlikely that the dramatically high rates of progressive liver disease that were seen could have occurred spontaneously.

The study's most important limitations are clinical. First, the clinical importance of these biopsy abnormalities is uncertain. What proportion would have healed on their own without further progression and without clinical symptoms? The investigators felt it necessary to stop the drug for only 10% of the patients with progressive liver disease apparent on biopsy. Second, this paper still provides no data on the interpretability, utility, and recommended frequency of monitoring these patients with liver biopsy. Although the authors provide a recommendation, it does not appear to be based on their data.

It is clear from the study that biopsy abnormalities are much more frequent than otherwise would have been suspected in patients who are receiving low-dose methotrexate. The frequency of clinically important hepatotoxic changes still remains to be determined, as well as the usefulness of liver biopsy in predicting these changes.

Brian L. Strom, MD
University of PennsylvaniaPhiladelphia, Pennsylvania, USA