In terms of specific cancers, an increased risk of lymphoma is well recognised in patients with RA [
77]. This risk is greatest in patients with the highest cumulative burden of disease activity [
8]. The type of lymphomas seen in patients with RA suggests an underlying immune deficiency profile prone to lymphomagenesis [
77]. Even prior to the introduction of TNFi therapies, some lymphoma events were linked to the use of csDMARDs, in particular cyclophosphamide, azathioprine and MTX, in the treatment of RA [
78]. The meta-analysis by Simon et al. found increased rates of lymphoma as expected, but additionally identified a greater risk of lung cancer (which may in part be accounted for by smoking, a shared risk factor for RA and lung cancer [
79]) and a decreased risk of colorectal cancers in patients with RA compared to the general population. A study from the UK, published in 2013, also identified an increased incidence of lymphoma and lung cancer but showed a reduction in prostate and cervical cancers in patients with RA [
80]. In an administrative claims database in South Korea, there was again an increased risk of lymphoma but lower rates of gastric cancer [
81].
A number of studies have found an increased occurrence of non-melanoma skin cancer (NMSC) in RA compared to the general population [
82‐
84]. By contrast, it is unclear whether RA is associated with malignant melanoma. There is a well-documented link between the immunocompromised host and melanoma, with evidence from patients with AIDS and post organ transplant [
85,
86]. However, data in patients with RA taking csDMARDs have been inconsistent [
80,
87]. Interestingly, a study from Australia, which has the highest background incidence of melanoma worldwide, found an increased risk of melanoma in MTX users compared to the background population [
88]. In the meta-analysis by Simon et al. [
76] the pooled standardised incidence rate for melanoma was significantly higher in RA patients, although the authors noted that few of the individual studies found a statistically significant increased risk.
Tumour necrosis factor inhibitors
As TNF is known to have anti-tumour effects, there were initial concerns that TNF inhibition could increase cancer incidence [
89]. Data on incident malignancies in the TNFi RCTs were mixed, with some earlier individual trials and meta-analyses appearing to show an increased risk of cancer [
16,
90]. However, more recent systematic reviews and meta-analyses have not been able to replicate this finding [
84,
91]. A systematic review by Askling et al. [
91], published in 2011, explored the potential cause of these different results. They noted that previous meta-analyses did not include all of the published trials, that the number of events in both treatment and control arms was low and that the definitions of cancer across individual studies may not be comparable. They attempted to address these issues in their meta-analysis. The authors were provided with patient level data from each study sponsor with more detailed information on study design, patent information and treatment details than was available in the published reports. Cancer events were identified by a search of each study sponsor’s clinical and safety database, which allowed the same definition of a cancer to be used across all studies. They found no difference between cancer rates across the three TNFi studied (INF, ETN and ADA), but did find differing rates of cancer in respective control arms.
As a result of the initial concerns, patients with a history of prior cancer were excluded from the majority of TNFi RCTs and, post-licensing, TNFi therapy was felt to be contra-indicated in patients with a previous malignancy. Therefore, information is limited on the recurrence of cancer in patients with prior malignancy treated with TNFi. Nevertheless, recent studies from administrative claims databases in the US and the Swedish ARTIS register have reported no increase in recurrent cancers in TNFi-treated versus biologic-naïve patients with similar history of prior malignancy [
92,
93]. These studies are also supported by data from the British and German registries [
94,
95]. Notably, all studies were based on small numbers of cases.
Regarding specific malignancies, the risk of lymphoma in TNFi users has been widely studied. However, as lymphoma is relatively rare, studies have struggled to identify sufficient cases of lymphoma to reach robust conclusions. Observational studies from the UK, the US and Sweden have examined the potential associated between TNFi and lymphoma in patients with RA [
4,
96‐
99]. A study from South Sweden, published in 2005, reported a large, but non-significant, relative risk of lymphoma of 4.9 (95% CI 0.9, 26.2) in patients treated with TNFi compared to csDMARDs. Two US studies using the National Data Bank for Rheumatic Diseases (NDB), found similar rates of lymphoma in TNFi users compared to other patients with RA [
4,
97]. However, the Swedish and US studies were limited by potential confounding by indication, i.e. patients with more severe disease are more likely to be prescribed TNFi. This was not adjusted for at all in the first NDB publication [
4]. The Southern Swedish and second NDB publication adjusted for the baseline Health Assessment Questionnaire score [
97], which is more a measure of disability than disease activity and therefore there may have been residual confounding. The BSRBR-RA study used propensity score matching to account for confounding by indication and other imbalances between baseline covariates between TNFi- and csDMARD-treated patients; they reported an equivocal HR for lymphoma of 1.00 (95% CI 0.56, 1.80) [
96]. The national Swedish ARTIS registry also found no increased risk of lymphoma in TNFi-exposed compared to csDMARD-exposed patients [
98]. Nevertheless, all of these studies remain potentially underpowered, either through insufficient person-years of observation or the low number of events. A recent EULAR initiative performed a joint analysis with data from 11 registers which has been published only as an abstract so far [
100]. With some 600,000 patient-years of exposure no increased risk was found. The results are reassuring not only concerning the overall incidence of lymphomas but also the spectrum of subtypes.
Data from national European registries in the UK, Germany and Denmark have consistently demonstrated no increased occurrence of solid cancers in patients treated with TNFi versus those treated with csDMARDs [
95,
101,
102]. These data are also supported by an older study from the NDB in the US [
103].
The risk of skin cancer appears to be unaltered by TNFi therapy. A meta-analysis of observational and long-term extension studies in 2012 by Le Blay et al. [
84] found no increase in NMSC in TNFi users compared to csDMARD users (pooled odds ratio 0.79; 95% CI 0.62, 1.02). Subsequent studies from the DANBIO registry in Denmark and the BSRBR-RA also found no difference in incidence of NMSC rates between TNFi and csDMARD users [
83,
102]. However, the most recent study from Sweden, the largest to date with over 2800 NMSCs, did identify a small increase in the risk of squamous cell carcinoma (SCC), but not basal cell carcinoma, in TNFi users compared to those taking csDMARDs [
82]. The studies from Denmark and the UK comprised mostly basal cell carcinomas and relatively few SCCs and so may have been underpowered to detect a small increase in SCC. The EULAR initiative mentioned above also pooled data from 11 observational registries across Europe in order to explore the risk of malignant melanoma in TNFi versus csDMARD users [
104]. This study captured data on over 130,000 patients over close to 580,000 patient-years follow-up. The overall standardised incidence ratio, adjusted for calendar year, and age and sex standardised to the reference population of each country, was higher in the ever TNFi-exposed versus biologic-naïve patients at 1.1 (95% CI 0.9, 1.4), but this did not reach statistical significance.
Other bDMARDs
A small number of studies have investigated the risk of malignancy in patients treated with other bDMARDs. Simon et al. [
105] compared rates of malignancy in the RCT data for ABT with rates seen in five observational cohorts of patients with RA taking csDMARDs. They found the standardised incidence ratios were no different in the ABT user group. Similarly, Solomon et al. [
106] investigated the risk of malignancy in a range of biologic and csDMARD users compared to MTX users and found lower rates of malignancy in TNFi users but equivalent rates (with wide confidence intervals) in patients treated with RTX and ABT. Scott et al. [
107] examined rates of recurrent NMSC in patients with RA or inflammatory bowel disease across a number of immunosuppressive therapies. Rates of recurrence were similar in ABT and RTX users compared to MTX users. The number of events in each treatment group was small and may have been underpowered to detect a true difference. Similarly, the collaborative European project described above found no increase in the risk of malignant melanoma in RTX, ABT and TCZ users compared to biologic-naïve patients [
104]; again, both the number of events and total follow-up time were small.