While many studies have investigated the risk of cancer following exposure to tumour necrosis factor (TNF) inhibitors,1–3 data are scarce with respect to people with cancer prior to treatment.2 ,4 To our knowledge, no studies have looked at cancer risk following exposure to anti-TNF in patients with a history of premalignant conditions, such as cervical dysplasia and Barrett's oesophagus. Guidelines from the British Society for Rheumatology (BSR) advise caution when using anti-TNF in subjects with premalignant conditions.5 Given the uncertainties regarding the use of anti-TNF in people with premalignant conditions, we investigated the outcome of women with rheumatoid arthritis (RA) with a previous history of carcinoma in situ (CIS) of the cervix registered with the BSR Biologics Register.6

The BSR Biologics Register, established in 2001, is following two cohorts of patients: one cohort of patients with RA starting biologic therapy and a second biologic-naive cohort with active RA receiving non-biologic disease modifying antirheumatic therapy (nbDMARD).6 Only women with CIS cervix prior to registration were included in this analysis, identified by flagging with the UK cancer registries using the International Classification of Diseases (ICD) codes 233.1 (ICD 9) and D06 (ICD 10). Cancer registration is mandatory in the UK and completeness is >99%.7 The ICD codes for CIS cervix comprised cervical intraepithelial neoplasia (CIN III), with or without mention of severe dysplasia. Mild to moderate dysplasia (CIN I or CIN II) was not included. Separate to this, rheumatologists were asked to provide details of all prior cancers in patients at the time of recruitment to the study. Participants were followed until first female genital cancer, death, last received follow-up or 31 March 2011, whichever came first. Subjects in the nbDMARD cohort were censored on starting biologic therapy. Incident female genital cancers, defined as any new or recurrent cancer (including CIS) of the female genital organs (ICD 10 C51–58, D06–07), were identified through (1) flagging with the national cancer and death registries, (2) 6-monthly patient diaries for 3 years and (3) 6-montly physician questionnaires for 3 years and annually thereafter.

In all, 238 subjects had previous CIS cervix among 11 738 women: nbDMARD-only 48/2654 (1.8%) and anti-TNF 190/9084 (2.1%). Recording of prior CIS cervix by the rheumatologist was low in both groups (table 1) which may reflect either a choice to not include a premalignant condition on the questionnaire or being unaware of the diagnosis. Overall, 73 (38%) anti-TNF treated patients started therapy within 10 years of CIS cervix, including 27 (14%) within 5 years. Two incident female genital cancers (metastatic squamous cell cancer of the vulva and metastatic cervical cancer—both fatal) were reported in the nbDMARD cohort (table 1). In each case, the CIS cervix occurred 13 years previously. No female cancers were reported during 893 person-years follow-up in the anti-TNF cohort. A strength of the study was linkage to the cancer registry, limiting reporting bias between the cohorts. However, despite using a large national cohort of women with RA, this study lacked power to detect a clinically important difference in the risk of incident cancer between the cohorts. In summary, there were no new or recurrent female genital cancers among women with pre-existing CIS cervix selected for treatment with anti-TNF in the BSRBR.