Abstract
Objective. To assess the risk of total malignancy and nonmelanoma skin cancers (NMSC) in patients with rheumatoid arthritis (RA) receiving certolizumab and golimumab through a metaanalysis of data from randomized control trials (RCT).
Methods. We systematically reviewed the literature up to May 2011 in Medline databases, as well as abstracts from the 2009 and 2010 annual meetings of the European League Against Rheumatism and the American College of Rheumatology. Mantel-Haenszel method was used to determine a common odds ratio (OR). Statistical heterogeneity was assessed by chi-square Q test. We selected only RCT including more than 30 RA subjects randomly assigned to an anti-tumor necrosis factor (TNF) or a nonbiological disease-modifying antirheumatic drug (DMARD) control group.
Results. The literature search identified 793 articles; 6 (2 with certolizumab and 4 with golimumab) were selected for metaanalysis. A total of 2710 patients received at least 1 dose of certolizumab or golimumab. For anti-TNF-treated patients, 18 cancers (excluding NMSC) and 9 NMSC were observed versus 4 cases of total malignancy and 3 NMSC in control groups. Metaanalysis revealed a pooled OR of 1.06 (95% CI 0.39–2.85) for risk of total malignancy and 0.69 (95% CI 0.23–2.11) for risk of NMSC with certolizumab and golimumab versus DMARD. Heterogeneity was not significant.
Conclusion. Metaanalysis of RCT of golimumab and certolizumab did not find an increased risk of total malignancy and NMSC. These results must be confirmed with longterm extension studies and registry studies, and careful monitoring remains mandatory.
For more than 10 years, tumor necrosis factor (TNF) antagonists have shown effectiveness for rheumatoid arthritis (RA) and they are now widely used for treating active RA. In addition to adalimumab, etanercept, and infliximab, 2 new anti-TNF monoclonal antibodies, certolizumab and golimumab, have been approved for RA1,2,3,4,5,6,7,8.
In 2006, a metaanalysis of randomized controlled trials (RCT) of adalimumab and infliximab suggested that these anti-TNF monoclonal antibodies might be associated with an increased risk of malignancy9. However, data including RCT with etanercept did not find a significantly increased risk in patients with RA10. Nevertheless, doubt remains concerning the specific risk for nonmelanoma skin cancers (NMSC) in longterm extension studies11 and in a recent metaanalysis using patient-level data12.
Our objective was to assess the risk of total malignancy and NMSC in patients with RA receiving certolizumab or golimumab; we performed a systematic review and meta-analysis based on data from RCT of those 2 newly licensed anti-TNF products whose results have not been included in previous metaanalyses.
MATERIALS AND METHODS
Study selection
We performed a systematic review of the literature up to May 2011. Bibliographic references were selected from Medline databases and abstracts from 2009 and 2010 annual meetings of the European League Against Rheumatism (EULAR) and the American College of Rheumatology (ACR).
We searched Medline with PubMed using the following key terms: arthritis, rheumatoid [MeSH] OR rheumatoid arthritis [all fields] AND (neoplasm [MeSH] OR safety [MeSH]) OR (neoplasm or safety) [all fields] AND (biological therapy [MeSH] OR tumor necrosis factor-alpha [MeSH] OR antibodies, monoclonal [MeSH]) OR (monoclonal antibody OR biological response modifier OR tumor necrosis factor-alpha OR tumor necrosis factor alpha antibody OR tumor necrosis factor-alpha OR anti tumor necrosis factor) [all fields] OR certolizumab pegol [all fields] OR golimumab [all fields] OR (Drug Combination Disease (W) Modifying (W) Anti (W) Inflammatory (W) Drug OR Dmard) [all fields]. In addition, reference lists of the reports initially identified were manually searched to identify additional relevant reports.
To be included, RCT of golimumab and certolizumab had to include more than 30 RA subjects in intention-to-treat studies, randomly assigned to an anti-TNF or a nonbiological DMARD control group for at least 10 weeks. Studies with a placebo control group and treatment arms of combination biological therapies were excluded.
Total malignancy was considered as all solid (including melanoma and excluding NMSC) and hematological cancers. NMSC included basal and squamous cell skin cancers.
Statistical analysis
The pooled odds ratio (OR) for patients receiving certolizumab or golimumab versus synthetic disease-modifying antirheumatic drug (DMARD) therapy was determined using Mantel-Haenszel methods, with a continuity correction designed for sparse data (adding 0.5 to each arm of such studies). Moreover, additional sensitivity analyses were conducted, describing the stability of the results when either 1 or 2 “virtual” events were used in case no cancer event had been observed in the trials.
Analyses involved use of Revman 5.0 software developed by the Nordic Cochrane Center. OR and 95% CI are shown on forest plots. Statistical heterogeneity was assessed by the chi-square Q test, with a significance level of 0.05.
RESULTS
The literature search identified 793 articles, from which 72 were preselected on the basis of the title and abstract. After reading full texts, we selected 6 RCT with a synthetic DMARD comparison group, 2 with certolizumab6,8, and 4 with golimumab1,2,3,4 (Figure 1). A total of 2710 patients received at least 1 dose of certolizumab or golimumab, compared with 783 who received synthetic DMARD. For patients treated with anti-TNF inhibitors, 18 cancers were observed (excluding NMSC): 2 each breast, tongue, colon, and lung cancers; 1 each uterus, esophageal, liver, adrenal, papilloma, testicle, bladder, and ovarian cancers; and 2 hematological neoplasms. Only 4 cases of malignancy were observed in control groups. A metaanalysis of data from these 6 studies with patients receiving synthetic DMARD as control group revealed a pooled OR for risk of total malignancy with certolizumab and golimumab of 1.06 (95% CI 0.39–2.85) versus DMARD, without significant heterogeneity (Figure 2).
The selection process of articles.
Risk estimates of total malignancy (excluding nonmelanoma skin cancers) reported in randomized controlled trials of patients with rheumatoid arthritis treated with golimumab and certolizumab. TNF: tumor necrosis factor; DMARD: disease-modifying antirheumatic drug; M-H: Mantel-Haenszel test.
Nine cases of NMSC were observed in anti-TNF groups, compared with 3 in DMARD groups, and the pooled OR was 0.69 (95% CI 0.23–2.11; Figure 3).
Risk estimates of nonmelanoma skin cancers reported in randomized controlled trials of patients with rheumatoid arthritis treated with golimumab and certolizumab. TNF: tumor necrosis factor; DMARD: disease-modifying antirheumatic drug; M-H: Mantel-Haenszel test.
Sensitivity analyses did not significantly modify our results, as the same trend toward no significantly increased risk was observed (data not shown).
DISCUSSION
A relation between RA and carcinogenesis was suggested for many years because some registry studies found an increased risk of certain cancers (lung, lymphoma, NMSC) in patients with RA as compared with the general population, regardless of the treatment13,14.
Recently, Leombruno, et al10, in a metaanalysis, analyzed 17 RCT of adalimumab, etanercept, and infliximab compared to DMARD in RA. The OR was 1.34 (95% CI 0.75–2.39) for total malignancy. Askling, et al12, in another metaanalysis of RCT of these 3 TNF antagonists in RA (31 RCT) and other diseases (43 RCT), used a strict method listing different types of malignancy, with predefined terms that were as specific as possible irrespective of report in the trial, in order to reduce missing information. They did not find an increased risk of total malignancy, but the risk of NMSC was significantly higher with TNF antagonists (HR 2.02, 95% CI 1.11–3.95)12.
In order to complete the metaanalysis of these 3 widely used TNF antagonists, and because few RCT of these biologics had been recently published, we aimed to assess the risk of malignancy especially with 2 new monoclonal antibodies by performing a metaanalysis of data from RCT of these 2 drugs. We chose to pool total malignancy (without NMSC) as 1 group and NMSC-only as the other, because of observations of a possible increased risk of NMSC with TNF antagonist treatment11,12 and differences in prognosis of these cancers. This distinction had been made in previous metaanalyses10,12. We excluded placebo-control trials in order to compare the risk of malignancy with these 2 new TNF antagonists with the “gold standard” in RA. Data extracted from RCT enable a strict comparison with referent treatment and a comprehensive inventory of safety. However, because of their short followup (only 2 studies had data at 52 weeks and one at 48 weeks; the others were 24-week followup studies) and small sizes of different groups, RCT are limited in assessing uncommon adverse effects such as cancer. This is a significant bias for safety analysis we found in all metaanalyses of RCT.
This first metaanalysis of risk of malignancy with use of certolizumab and golimumab is reassuring, as we found no increased risk of total malignancy and NMSC. However, these results must be confirmed with longterm extension studies and registry studies, and careful monitoring remains mandatory.
- Accepted for publication November 9, 2011.
REFERENCES
In this issue
