This review reported that tumour necrosis factor inhibitors did not increase the risk of malignancy but they appeared to increase the risk of skin cancer, including melanoma. These findings were based on observational data so the authors urged caution in considering these results. The authors' cautious conclusions appear likely to be reliable.

To assess the risk of malignancy in patients with inflammatory arthritides treated with tumour necrosis factor inhibitors (TNFi) in clinical practice as recorded in prospective observational studies.

MEDLINE, EMBASE and The Cochrane Library were searched for relevant studies in English were published between 1998 and March 2010. The search strategy was reported in an appendix. Conference proceedings of key rheumatology conferences were searched from 2005.

Prospective observational registry studies that reported malignancy in at least 100 exposed patients with rheumatoid arthritis, psoriatic arthritis or ankylosing spondylitis who received TNFi treatment were eligible for inclusion.

Most of the included registries reported on participants with rheumatoid arthritis; one also included spondylarthritis and two included rheumatic disease more broadly. Most registries collected all malignancies but around one third recorded only specific types (most commonly lymphoma).

Two reviewers independently selected studies for inclusion. Disagreements were resolved by discussion.

One reviewer assessed the quality of included studies using the Newcastle-Ottawa scale for non-randomised studies.

Data were extracted on number of patients/patient-years, incidence rates, standardised incidence ratios and measures of relative risk (RR) for patients exposed and not exposed to TNFi.

One reviewer extracted data from the included studies. The extraction was checked by a second reviewer.

A direct random-effects meta-analysis was conducted using the most recent relative risk data on malignancy from each source to avoid repeat inclusion of cases. The degree of inconsistency between studies was assessed using the Ι² statistic.

Twenty-nine studies from 11 different registries were included in the review. Where reported, length of follow-up ranged from 198 to 29,981 patient-years. Study quality varied. Most studies scored highly on representativeness of the exposed cohort, ascertainment of exposure and demonstration that malignancy was not present at the start of the study. Only 10 of 29 studies reported on the proportion of patients lost to follow-up.

There was no evidence for a statistically significant increase in the risk of all-site malignancy (seven studies) or for a significantly increase in risk with longer exposure to TNFi (two studies).

In patients with previous malignancies there was a higher risk of a new/recurring malignancy that was not significantly increased by exposure to TNFi (two studies).

There was a significantly increased risk of non-melanoma skin cancer (1.45, 95% CI 1.15 to 1.76; four studies) and a statistically non-significant increased risk of melanoma (1.79, 95% CI 0.92 to 2.67; two studies) associated with TNFi. There was no significant increase in the risk of lymphoma (three studies).

TNFi treatments did not increase the risk of malignancy, particularly lymphoma, but they appeared to increase the risk of skin cancer, including melanoma.

This review was based on a clearly defined question. Attempts were made to identify all the relevant evidence. Studies were evaluated for methodological quality and combined statistically using established methods, though it was not always clear to which type of ratio the pooled values referred. The benefit of restricting inclusion to prospective registries was that data were available from a large number of patients who in some cases received treatment over a longer period than would be observed in clinical trials. However, as these data were observational, they were subject to confounding and several kinds of bias relating to selection and surveillance of patients. The authors of the review acknowledge these limitations and stated that their results should be considered with caution.

The authors' cautious conclusions are likely to be reliable.

Practice: The authors stated that the current approach of physicians to patients with previous malignancy was appropriate.

Research: The authors stated that given the long latency associated with many cancers, follow-up should be extended even further to assess whether TNF inhibition increased the risk of malignancy.

Partly supported by a grant from Wyeth Europa.