The data reported here suggest that, although there is no increased risk of overall malignancies in patients with RA compared with the general population, there may be a defined pattern of risk for site-specific malignancies. Based on observed versus expected cases, there was considerable variation in the calculated SIRs among the individual studies for site-specific malignancies. Nevertheless, the random effects meta-analysis demonstrated an overall pattern that was generally consistent with the risk trends reported in the individual studies. This pattern included a clear increase in risks of lung cancer and lymphoma, both Hodgkin and non-Hodgkin, and a potential decrease in risks of colorectal cancer and breast cancer.
The increased risk of lymphoma is especially notable since this malignancy was associated with the highest relative risk, especially for Hodgkin lymphoma, which was more than three-fold higher than in the general population. Although one study suggested a decrease in lymphoma risk (an approximate 50% reduction in non-Hodgkin lymphoma with no reported cases of Hodgkin disease) [6
], these results were ascribed to the rarity of these malignancies and the small population that was followed (n = 862), although other studies with similarly small populations (that is, less than or equal to 800 patients) seemed to follow the trends seen in our meta-analysis [10
There have been a number of hypothesized explanations for the differences in the risk of certain malignancies in patients with RA compared with patients without the disease. Possible mechanisms for an increased risk of lymphoma in RA patients include the fact that RA results in persistent immunologic stimulation (which may lead to clonal selection and predispose CD5+
B cells to malignant transformation), decreases the number and function of T-suppressor lymphocytes (including those directed against the pro-oncogenic Epstein-Barr virus), and decreases natural killer cell activity in the synovial fluid, tissue, blood, and lymph [24
]. Inflammation is believed to play a key role in the risk of lymphoma; epidemiologic studies have suggested that, among patients with RA, higher inflammatory activity is a major risk determinant of lymphoma [25
]. Meanwhile, the role of RA treatment remains somewhat uncertain; large cohort studies have not confirmed any treatment-related effects; however, it is premature to make conclusions about the risk associated with anti-TNFs with the currently available data [26
]. It has been suggested that a minority of RA patients (those with the worst disease) carry much of the increased risk of lymphoma because of their disease rather than their treatment [27
The observed association between RA and lung cancer may result from several factors. Cigarette smoking would explain an indirect association between RA and lung cancer as smoking is an independent risk factor for both conditions. The direct causal association of RA with lung cancer may be mediated by chronic inflammation and/or the presence of interstitial lung disease. Systemic chronic inflammation has been reported to be a risk factor for lung cancer [28
]. A recent 10-year population-based observational cohort study reported that baseline serum C-reactive protein was significantly associated with lung cancer, independent of smoking [29
]. In addition, RA has been shown to affect the lungs; autopsy studies have shown some degree of interstitial lung disease in the majority of people with RA [30
] and the mortality from pulmonary disease in RA is approximately twice that of the general population [31
The explanation for the reduced risk of colorectal cancer is most likely due to the increased use of nonsteroidal anti-inflammatory drugs (NSAIDs) and cyclooxygenase-2 (COX-2)-selective inhibitors by patients with RA. These medications have consistently been associated with a decreased risk of colorectal cancer; a recent meta-analysis of all randomized controlled trials and observational studies concluded that COX-2 inhibitors and NSAIDs reduce the incidence of colonic adenomas and that NSAIDs also reduce the incidence of colorectal cancer [32
]. The hypothesis underlying this protective association is thought to be the inhibition of COX-2 and subsequently prostaglandin production [33
The strength of this review is its reliance on real-world clinical data obtained from observational studies rather than randomized placebo-controlled trials that reflect a selected cohort of patients. It provides precise estimates of the malignancy risk in RA patients. However, there are some limitations in the individual studies, as well as with the systematic review and meta-analysis, which should be considered when interpreting the data.
The primary limitation is the heterogeneity among studies in terms of the data sources, populations examined, and study designs. The diversity of study methodologies in some cases may have resulted in bias. Such an example is the study by Mariette and colleagues [15
], which was not strictly cohort-derived; it involved identification of new lymphoma cases based on consultation between rheumatology and oncology departments. Uncertainty regarding the size of the RA population evaluated may account for their very high reported relative risk of 7.4 for Hodgkin lymphoma. Sources of selection bias may include the use of hospitalization records for identification of populations.
Other limitations include the possibility of misclassification and the wide variation in follow-up. There may have been misclassification of the inclusion of patients into the RA populations, and there may have been uncertainty surrounding the diagnostic accuracy of the malignancies. Several of the studies were dependent on database analyses and relied on diagnostic codes, whereas another used patient self-report followed by medical record validation. Follow-up times ranged from 1 year to as long as 17 years, and it is possible that in some cases the variability observed in the SIRs may result from these differences. However, these individual study limitations may be compensated for, in part, within the context of performing such a meta-analysis as presented here.
The analysis presented does not attempt to determine causality of risk or adjust for other risk factors that may contribute to the observed increases or decreases in risks, as these data were not readily available in the individual studies. This is especially relevant with respect to severity of disease as well as RA treatment. In relation to treatment effects, nearly all patients in these studies have received treatment for their RA, and it is becoming increasingly likely that treatment of RA is initiated early in the disease process. Consequently, it is difficult to separate the underlying risk associated exclusively with the disease from some of the potential treatment effects, especially when many patients may be taking multiple medications for RA as well as for comorbid conditions. Nevertheless, the consistent findings among the studies included in this meta-analysis where patients were taking diverse medications are consistent with the recent suggestion that it is the underlying inflammation rather than treatment that contributes to the risk [25