Objective To study the risk of a second malignant neoplasm (SMN) and mortality in patients with rheumatoid arthritis (RA) with a history of a primary cancer diagnosis and treated with biological disease-modifying antirheumatic drugs (bDMARD).
Methods Among patients with RA (n=15 286) registered in the DANBIO Register during 2000–2011, 1678 had a primary cancer according to the Danish Cancer Registry. HRs for SMN and death were calculated.
Results During follow-up there were 279 patients with RA contributing person-years to the bDMARDs use before their primary cancer diagnosis, 220 to the only after, 92 to the both before and after, while 1203 patients with RA contributed to the non-use strata. Ever use of bDMARDs was associated with a HR of 1.11 (95% CI 0.74 to 1.67) for developing a SMN compared with non-use (cancer site adjusted). The HR for death associated with bDMARD use before the primary cancer diagnosis was increased 1.53 (95% CI 1.13 to 2.09). After further adjustment for extent of the primary cancer, the HR for death was 1.20 (95% CI 0.88 to 1.63) for bDMARDs use before cancer, 1.36 (95% CI 0.78 to 2.39) for bDMARD use only after cancer and 1.22 (95% CI 0.70 to 2.13) for use both before and after the cancer.
Conclusions Among patients with RA with a history of cancer, treatment with bDMARDs was not associated with increased risk of SMN. No clear conclusion can be drawn regarding mortality in bDMARD-treated patients with RA.
- DMARDs (biologic)
Statistics from Altmetric.com
The safety of treatment with biological disease-modifying antirheumatic drugs (bDMARD) has been carefully studied for the past 15 years; however, it is still largely unknown whether this treatment is safe in patients with arthritis with a history of cancer. The role of tumour necrosis factor-alpha (TNF-α) in carcinogenesis is complex and includes both a tumour destructive and promoting capacity.1 A meta-analysis of selected randomised clinical trials (RCT) of TNF-α inhibitor (TNF-I) treatment with infliximab or adalimumab in patients with rheumatoid arthritis (RA) reported a three times increased risk of cancer during short-term follow-up.2 However, more recent meta-analyses of RCTs and observational studies have reported no overall increased risk of primary cancer in patients with RA treated with TNF-I,3 4 but increased risks of specific cancer types have been reported or are still being discussed.5–7
Second malignant neoplasms (SMN) are an increasing challenge as the survival after first primary cancer has improved substantially for most cancer types. The concern for SMNs or cancer recurrences in patients with a history of a first primary cancer diagnosis has led to some reluctance in treating this subset of patients with RA with TNF-I. In clinical practice, rituximab is often the bDMARD of preference in patients with RA with a history of cancer, since rituximab (RTX) is used in the treatment of lymphomas and hence not considered tumour promoting.
Only one study has investigated whether TNF-I therapy and other bDMARDs influence the risk of SMN in patients with a prior cancer; they found that patients from the UK with RA and prior malignancy selected to receive either TNF-I or RTX did not have an increased risk of a future incident malignancy.8
The only study investigating mortality in patients receiving TNF-I after their first cancer is a Swedish study of patients with RA with breast cancer where no difference in all-cause mortality was reported among TNF-I treated versus non-treated.9
Thus, after more than 15 years of widespread clinical use of TNF-I, we still lack knowledge about the treatment-associated risk of SMN and mortality in RA patients with a history of a primary cancer. Observational studies with long-term follow-up, detailed drug history and independent and complete registration of incident SMN and deaths are needed. The aim of the present study was to investigate risk of any SMN and risk of death among bDMARD-treated patients with RA with a cancer by linking the national Danish DANBIO Registry with the Danish Cancer Registry.
Materials and methods
Since 2000, Danish rheumatologists have registered rheumatologic patients treated with bDMARDs in the DANBIO Registry. Variables in DANBIO include diagnosis, start and stop dates for treatment with bDMARDs and disease activity measures.10 All departments of rheumatology in Denmark report to DANBIO and patients are principally followed lifelong. From 2000 to 2006, reporting to DANBIO was voluntarily, but became compulsory in 2006 where online registration was implemented (www.danbio-online.dk). Coverage of bDMARD-treated patients is >90%. From 2000, patients with arthritis treated with bDMARDs have also been registered in DANBIO, and from 2006 rheumatology departments across Denmark have been committed to register all newly diagnosed or newly referred patients with RA regardless of biological or non-biological treatment. DANBIO has been approved by the National Board of Health as a national quality registry. Treatment with bDMARDs is tax paid and free of charge for the individual treated patients.
The Civil Registration System
Since 1968, the Civil Registration System (CRS) has assigned a unique personal identification number to all Danish residents (CPR number), which enables linkage of information for a person throughout various registers.11 The CRS contains updated information on vital status including date of death and emigration.
The Danish Cancer Register
All cancer cases in Denmark have been registered in the Danish Cancer Register (DCR) since 1943. The registration was based on paper notification forms from various sources until 2004.12 From 2004, the DCR is based on electronic reporting from other health registries.13 Tumours are coded according to the International Classification of Diseases 7 or 10. The completeness of DCR has been assessed for specific types of cancer and found to be high.14–17 In the DCR, an SMN is defined as a tumour arising in a different organ or in different parts of an organ as specified by the 4-digit topography code as well as a tumour arising in the same organ with a different 4-digit morphology code. Extent of the cancer disease has been reported as localised, regional spread, distant metastases or unknown through 2003. Since 2004, the tumour, node, metastases (TNM) classification has been used. To be able to use comparable data before and after 2004, the TNM codes were converted into the four categories for extent of disease.
Cohort identification and follow-up
A total of 15,286 patients with RA registered in DANBIO between January 2000 and December 2011 were linked to the CRS (information on dates of death or emigration) and the DCR (identification of all cancer cases through 2011 except non-melanoma skin cancer). We excluded 70 patients with an SMN diagnosis before or at the same day as entry in DANBIO. In this paper, we study the following exposure groups: patients with a first primary cancer and non-use with bDMARDs (1203 contributing patients where 1176 remained never exposed during follow-up); patients with a first primary cancer and ever use of bDMARDs (502 contributing patients), who were further stratified to: patients with bDMARDs use before their first cancer (279 contributing patients where 190 remained only before users during follow-up); patients with bDMARDs use only after their first cancer (220 contributing patients); and patients with bDMARDs use both before and after first cancer (92 contributing patients). For all patients, follow-up for SMN and death started at the date of first cancer diagnosis or at entry in DANBIO, whichever occurred last. End of follow-up was date of diagnosis with an SMN, death, emigration or 31December 2011, whichever occurred first.
Hazard Ratio (HR)s and 95% CIs for any type of SMN as well as mortality were estimated using Cox proportional hazard models stratified according to treatment and adjusted for age (underlying time scale), sex, calendar time (2000–2004, 2005–2009, 2010–2011) and cancer site. All variables were time dependent except sex. The bDMARDs included TNF-I inhibitors (infliximab, etanercept, adalimumab, golimumab, certolizumab), rituximab and other (abatacept, tocilizumab). Additional stratified analyses for SMN risk and mortality were performed by type of bDMARDs (TNF-I and rituximab) after primary cancer. Effect modification analyses for use of bDMARD before or after cancer and mortality were carried out by sex, age at cancer diagnosis and time since primary cancer. All tests were likelihood ratio tests.
For the subset of patients with cancer for whom extent of disease was known (n=1326), the mortality analyses were carried out with and without adjustment for this variable. We included three categories in the adjustment for extent of disease: localised, regional spread and distant metastasis.
A total of 1678 patients with RA diagnosed with a first primary cancer were identified. Type of first primary cancer and RA characteristics at the time of entry in DANBIO are presented in table 1.
Table 2 shows the risk of an SMN among patients with RA with previous cancer comparing use of bDMARDs with non-use.
No increased risk of a SMN was observed in patients with previous cancer and ever use of bDMARDs compared with non-use, HR 1.11 (95% CI 0.74 to 1.67) (table 2). Similar results were seen when ever use of bDMARDs was stratified according to timing of use (before and/or after cancer) (table 2) and when use of bDMARDs before cancer and use of bDMARDs after cancer were stratified on sex and age at first primary cancer diagnosis (online supplementary table S1). The HR of SMN was 1.25 (95% CI 0.71 to 2.18) among patients initiating bDMARDs more than 5 years after the first cancer diagnosis compared with non-users, while the HR was 0.92 (95% CI 0.40 to 2.10) among patients initiating bDMARDs less than 5 years after the first cancer diagnosis.
Supplementary file 1
During follow-up, 342 patients died. The overall mortality after a first primary cancer was increased for ever use of bDMARDs, HR 1.25 (95% CI 0.99 to 1.57) (table 3).
For 1326 patients with a first primary cancer diagnosis who had information on extent of the cancer disease, the HR for death for ever bDMARD use compared with non-use was 1.35 (95% CI 1.04 to 1.76), which decreased slightly after adjustment for extent of cancer, HR=1.23 (95% CI 0.94 to 1.60) (table 3). In a model that did not adjust for extent of disease, the HR was 1.53 (95% CI 1.13 to 2.09) if treatment with bDMARDs was given before the first primary cancer; but after adjustment for extent of disease, the HR was attenuated to 1.20 (95% CI 0.88 to 1.63). If treated with bDMARDs both before and after first cancer, the HR increased from 0.99 (95% CI 0.57 to 1.73) to 1.22 (95% CI 0.70 to 2.13) after adjustment for extent of disease. A similar tendency was seen, if bDMARDs was started only after the first cancer with HRs of 1.19 (95% CI 0.69 to 2.04) and 1.36 (95% CI 0.78 to 2.39), respectively, before and after adjustment for extent of disease. The HR for death adjusted for extent of cancer disease was 1.90 (95% CI 0.89 to 4.09) if treatment was started ≥5 years after first primary cancer and 1.13 (95% CI 0.70 to 1.82) if treatment was started <5 years after first primary cancer. Results for stratifications on sex and age at cancer are shown in online supplementary table S2.
We found no increase in the risk of an SMN in RA patients with a history of cancer who had received bDMARDs compared with non-treated. This was seen independently of type and timing of bDMARDs. In relation to mortality, adjustment for extent of cancer disease affected the HRs for death in different directions depending on whether use of bDMARDs was initiated before and/or after the cancer diagnosis. The adjusted risks of dying were non-significantly elevated for all strata defined by the timing and type of bDMARDs.
Second malignancies have become an increasingly important concern in oncology during the last two decades, as they now comprise the sixth most common group of malignancies after skin, colorectal, lung, breast and prostate cancers.18 The aetiology of SMN includes shared environmental risk factors, genetic predisposition as well as radiation and chemotherapy used to treat the primary cancer and for patients with RA long-standing inflammatory disease activity may contribute as well.
In our study, we found no increased risk of an SMN among bDMARD-treated patients with RA compared with non-treated. The study from the British Society for Rheumatology Biologics Register reported an age and gender-adjusted HR of 0.55 (95% CI 0.35 to 0.86) for the TNF-I cohort and 0.43 (95% CI 0.10 to 1.80) for the RTX cohort in comparison with the synthetic DMARDs cohort.8 They concluded that patients with RA and prior malignancy selected to receive either TNF-I or RTX in the UK do not have an increased risk of future incident malignancy. This is in line with our results stratified on TNF-I and RTX. In a previous study, we found no human papillomavirus-related cancers after bDMARD treatment in 327 women with a history of precancerous lesions of the uterine cervix although an increased occurrence could not be ruled out due to low number of events.19
We found an increased mortality for ever use of bDMARDs compared with non-use among patients with RA with a history of cancer that became non-significant after adjustment for extent of first primary cancer disease. A Swedish study showed a relative risk of dying following cancer of 1.1 (95% CI 0.8 to 1.6) associated with exposure to TNF-I.20 This Swedish study reported that cancers occurring in patients with RA who were undergoing or had a history of treatment with biological agents were not characterised by any markedly differences in stage at presentation compared with non-users.20 However, the Swedish study did not investigate the timing of exposure to TNF-I relative to the timing of the diagnosis of the first primary cancer as we did. In patients with RA treated with bDMARDs before first cancer, the HR for death was significantly increased; however, after adjustment for extent of disease the HR was attenuated and no longer significant. This indicates that these patients had more severe cancer disease than those not treated. The underlying explanation could be that bDMARDs induce more aggressive tumours or that extent of disease influences the choice of RA treatment after the cancer diagnosis. The observation of a stronger association after adjustment for extent of disease among patients treated both before and after first cancer speaks against the first mentioned explanation. In addition, prior or active cancer disease is a relative contraindication for TNF-I treatment in Denmark, and the decision on whether treatment with bDMARDs should be initiated after a primary cancer is taken for each individual case in the local rheumatology department. A tendency to use bDMARD treatment in less severe cancer cases is supported by the observation that the risk of death was higher after adjustment for extent of disease among patients treated after the cancer diagnosis. However, numbers were small, so consequences of adjustments must be interpreted with caution. Further studies with longer follow-up and/or inter-registries collaboration are necessary to answer this important clinical question.
The strengths of our study include high coverage of biologically treated patients (>90%) as well as a high validity of the RA diagnosis with detailed information on diagnostic features and disease manifestations in the DANBIO register. Also, registration in DANBIO was independent of registration in the DCR. The numbers of SMNs and deaths were small and limited the scope of the study and the possibilities to perform more detailed analyses, for example, looking into the influence from RA disease activity. High disease activity may be associated with an increased risk of SMN and is known to be associated with increased mortality,21 so a higher disease activity associated with bDMARD use may have confounded our results. However, a time-dependent cumulative disease activity measure would be difficult to build into the analyses of our available data. More insights into the issue of whether bDMARDs affect cancer prognosis could have been provided by including data on recurrences, but unfortunately, these are not recorded in the DCR. In addition, it would have been highly relevant to study cancer-specific mortality in addition to all-cause mortality, but cause of death was not collected under the present study protocol. Extended future studies should aim at taking the factors mentioned above into consideration.
Using the national Danish registers, we found no indication that use of DMARDs in patients with RA with a history of cancer increased the risk of SMN. No firm conclusion could be reached regarding whether treatment with bDMARDs influences mortality in these patients.
We thank all rheumatology departments in Denmark for reporting to the DANBIO Registry.
Handling editor Tore K Kvien
Contributors LD: contributed to study idea, conception and design, literature search, data collection, the analysis and interpretation of data, drafting the manuscript and approving the final version. RC: contributed to the interpretation of data, references, revising the manuscript and approving the final version. IMJH: contributed to study conception and design, interpretation of data, revising the manuscript and approving the final version. LEK and MLH: contributed to interpretation of data, revising the manuscript and approving the final version. LM: contributed to study design, literature search, data collection, the analysis and interpretation of data, revising the manuscript and approving the final version.
Funding This study was supported by grants from the Danish Rheumatism Association, the Danish Cancer Society and the Danish Council for Independent Research.
Competing interests LD: have received speaking fees outside the submitted work from UCB, MSD and Janssen pharmaceutica; IMJ: Roche Pharmaceuticals; LEK: Pfizer, AbbVie, Amgen, UCB, Celgene, BMS, Biogen, Forward Pharma, MSD, Novartis, Eli Lilly and Janssen Pharmaceuticals; MLH: AbbVie, BMS, MSD, Pfizer, Orion, Novartis, Biogen, Eli Lilly, Celltrion.
Ethics approval The project is entirely register based, and the patients have not been contacted. No approval by the ethical committees was needed according to Danish law. Approval by the Danish Data Protection Agency was obtained (HGH-2016-099, I-Suite no: 04979).
Provenance and peer review Not commissioned; externally peer reviewed.
Presented at An abstract of this study was presented at the 2017 American College of Rheumatology Annual Meeting.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.