Immune competence is of importance for the occurrence and outcome of malignancies, as exemplified by the effects of immune checkpoint inhibitors in the treatment of malignancies.¹ An increased risk for malignancies has been one of the main concerns since the introduction of biological disease-modifying anti-rheumatic drugs (bDMARDs) for the treatment of chronic inflammatory arthritis. Most treatment guidelines have therefore issued caution against using bDMARDs (tumour necrosis factor inhibitors (TNFi) in particular) in patients with a history of cancer within 5–10 years. So far, most (though not all) studies of cancer incidence following treatment with TNFi and other bDMARDs, and of recurrence of pre-treatment cancers following treatment with TNFi, have been reassuring.^2–9 The 2015 ACR recommendations for treatment of rheumatoid arthritis (RA) recommend that patients with a history of previous solid organ malignancy should be treated as patients without this condition,¹⁰ though acknowledging the low level of evidence, whereas previous recommendations suggested rituximab.¹¹ Similarly, there is no consensus regarding the time period from cancer diagnosis until the safe initiation of a bDMARD. Thus, scientific evidence supporting clinical decision-making in this context is scarce.

The aim of the present study was to assess the relative use of different bDMARDs in patients with RA and history of cancer. We used real-life data from the DANBIO (Denmark), ROB-FIN (Finland), NOR-DMARD (Norway) and ARTIS (Sweden) bDMARD registers. We identified patients with RA who initiated any bDMARD between year 2010–2017, regardless of type or number of prior bDMARDs. We identified patients with a clinical rheumatologist-assigned diagnosis of RA regardless of fulfilment of exact classification criteria. We identified the subgroup of patients with prior malignancy 10-year prior to starting the bDMARD in question through linkage to national cancer registers. Any malignancy (invasive or in situ) apart from benign tumours was defined as malignancy. Patients could contribute more than one treatment course. Both non-melanoma and melanoma skin cancer were included. The frequency of RA patients with a history of malignancy (according to the definition above) in each bDMARD group, as well as basic demographic and disease characteristics (age, gender, number of prior bDMARDs, years from cancer diagnosis until start of the bDMARD) was assessed across the different bDMARD groups. Switches from bio-original to biosimilar were regarded as one treatment.

A total of 42 638 RA patients initiating a bDMARD treatment were included (table 1). Initiators of non-TNFi biologics were generally older than TNFi-initiators, with the highest age at start for rituximab, especially in Sweden and Finland (table 1). Overall, among the bDMARD initiators in Denmark, Finland, Norway and Sweden, 344/11 230=3%, 288/4766=6%, 56/1876=3% and 1703/24 766=6.9%, respectively, had prior cancer. Whereas there was little variation across individual TNFi inhibitors ranging from 1% to 6%, the proportion of patients with a history of cancer at treatment start was higher among patients on non-anti-TNF bDMARDs, especially for rituximab (8–17%). The median time (years) since the cancer diagnosis ranged from 2 to 7 years, with a tendency towards a shorter time for rituximab (table 1).

As expected, we noted that the proportion of patients starting a bDMARD during the period 2010–2017 with a prior malignancy was low. Among these initiators, however, there was a clear preference for non-TNFi, in particular rituximab. The latter could in part be explained by differences in age at treatment start, as patients on rituximab tend to be older compared with patients on other bDMARDs. However, the small differences in median age among patients with history of cancer across the bDMARD groups under study supports the hypothesis that there is a preference for rituximab by clinicians for treatment of patients with history of cancer. Rituximab is being used for several haematological malignancies, which might at least partly explain the preference, although the underlying evidence for this preference in other types of cancers remains incomplete. Another interesting observation was that the proportion of female patients with a history of cancer was somewhat higher compared with patients with no history of cancer in the TNFi groups, but not in the non-TNFi group. A possible explanation for this could be the different choice of bDMARD in different types of cancer. Finally, there is heterogeneity not only in treatment channelling, but also due to different prescription patterns across countries.

Our results underscore both the reluctance to use bDMARDs and especially TNFi in RA patients with a history of malignancy, which implies a risk for undertreatment of some patients, and the need for more data on the benefit–risk ratio in this treatment context.