Patients with rheumatoid arthritis (RA) are at increased risk of lymphoma,1^–4 and anti-tumour necrosis factor (TNF) agents may further increase this risk.5 6 We conducted a retrospective cohort study of this question using the computerised data of Kaiser Permanente (KP; http://www.kaiserpermanente.org/). We defined a “severe RA cohort” of potential candidates for anti-TNF therapy as those aged 18 years and older from 1996–2002 who received at least two diagnoses of RA and were prescribed a disease-modifying antirheumatic drug (DMARD) after receiving methotrexate (n = 3982). Patients who had a previous history of anti-TNF agent were excluded. Follow-up began with the prescription of the first DMARD after methotrexate and ended with the 90th birthday, 31 December 2002, death, disenrollment, anti-TNF agent start, or diagnosis of lymphoproliferative cancer. We compared characteristics of the severe RA cohort to those of etanercept users, and we compared cancer outcomes to those of the general KP membership (n = 2 012 972).

All information was obtained through linkage. Cancer diagnoses were obtained from the KP Cancer Registry. We calculated standardised incidence rate ratios (SIR) and 95% confidence intervals using cancer rates from the general KP membership. We applied these rates to the age- and sex-specific person-time in the severe RA cohort to obtain expected numbers.

About three-quarters of the RA patients were women, and on average they were 15 years older than patients in the general population. The mean duration of follow-up was 4.4 years. About 70% of patients who were tested had a positive rheumatoid factor, and mean erythrocyte sedimentation was 41.2 mm/h. The severe RA cohort was comparable to patients using etanercept (n = 714), except that 36% of patients in the severe RA cohort had used two or more DMARDs prior to cohort entry, whereas 74% of etanercept patients had used two or more DMARDs.

Compared to the general population, patients with RA who received DMARD therapy after methotrexate were 4.7-times as likely to develop non-Hodgkin’s lymphoma and 3.8-times as likely to develop any lymphoproliferative cancer (table 1). These findings can be compared with a recent population-based, prospective cohort study from the UK,7 which followed 2105 patients with new onset inflammatory polyarthritis for an average 7.4 years. By the fifth year, 59% of patients satisfied the 1987 American College of Rheumatology (ACR) criteria for RA. Among patients who ever used methotrexate, the SIR for any lymphoma was 4.86 (95% CI, 1.78–10.57). Although our study was not planned to estimate risk among the etanercept patients, we wish to report that follow-up in the etanercept cohort was 1428 person-years, with none of these patients developing a lymphoproliferative cancer.

The key strengths of this study are the community-based setting and the availability of comprehensive data. While we did not verify the diagnoses of RA directly, our definition of severe RA has strong face validity. We believe results from this study may be useful to interpret the frequency of lymphoproliferative cancers among patients with severe RA who have received anti-TNF therapies.