Incidence of melanoma and other malignancies among rheumatoid arthritis patients treated with methotrexate

Abstract

Objective

To determine cancer risk in a cohort of 459 rheumatoid arthritis (RA) patients treated with methotrexate in community practice.

Methods

All RA patients who started methotrexate prior to June 1986 and were attending 1 of 6 rheumatologists were studied. Demographic data were matched to the State Cancer Registry to identify all malignancies (except nonmelanoma skin cancer) for 1983–1998, and to the National Death Index to identify all deaths to the end of 1999. Followup started on the date when methotrexate was started and ended either on the last confirmed date on which the patient was seen by the rheumatologist or at death. Standardized incidence ratios (SIRs) were calculated using state population cancer rates stratified by sex, age (in 5-year groups), and calendar year.

Results

There were 4,145 person-years of followup (average 9.3 years). Eighty-seven malignancies were identified (14 before, 64 during, and 9 after the followup period). There was an estimated 50% excess risk of malignancy among methotrexate-exposed RA patients relative to the general population (SIR 1.5, 95% confidence interval [95% CI] 1.2–1.9), with a 3-fold increase in melanoma (SIR 3.0, 95% CI 1.2–6.2), a 5-fold increase in non-Hodgkin's lymphoma (SIR 5.1, 95% CI 2.2–10.0), and an almost 3-fold increase in lung cancer (SIR 2.9, 95% CI 1.6–4.8).

Conclusion

Compared with the general population, methotrexate-treated RA patients have an increased incidence of melanoma, non-Hodgkin's lymphoma, and lung cancer. There may be a role for regular skin cancer screening for all RA patients, particularly those receiving immunosuppressive therapy.

INTRODUCTION

Rheumatoid arthritis (RA) is a chronic, inflammatory disease of unknown etiology that affects approximately 1% of the adult population worldwide (1). In Australia, 2.4% of the population report having RA (2). If left untreated, RA results in joint deformity and destruction leading to significant disability and decreased quality of life. It may also shorten life expectancy with increased deaths due to cardiovascular disease, infection, and cancer (3). Population-based and hospital-based RA cohort studies have variably reported an increased incidence of Hodgkin's and non-Hodgkin's lymphoma, leukemia, myeloma, and lung cancer and a reduced incidence of colorectal cancer in persons with RA compared with the general population (4).

An association between methotrexate (MTX), a commonly prescribed disease-modifying antirheumatic drug (DMARD) used to treat RA, and cancer has also been suggested (4). There have been numerous case reports of patients with RA treated with MTX developing cancer, particularly lymphoma, and in some cases the tumors have regressed or disappeared when MTX was discontinued (5). These reports have prompted concern that MTX itself is oncogenic in patients with RA (6); however, studies to date have been inconclusive (4).

Previous studies of cancer incidence in MTX-treated patients with RA have been conducted in Europe and the US. The incidence and types of malignancies that occur in Australia may differ from those reported elsewhere due to differences in population and environmental factors. For example, the incidence of melanoma and other types of skin cancer in the general population in Australia is higher than elsewhere, attributed to the outdoor lifestyle and the reduced ozone layer in this region. It is not known whether this increased risk is accentuated in patients with RA treated or not treated with MTX.

We have previously studied the long-term outcome of a cohort of 459 patients with RA treated with MTX in community practice in Australia (7, 8). The objective of the current study was to determine the incidence of cancer in this cohort compared with the general population and compared with published studies that have determined the incidence of malignancy in MTX-treated RA cohorts in other countries.

PATIENTS AND METHODS

Patients.

We previously assembled a cohort of all patients attending the community-based private practices of 6 rheumatologists in Melbourne, Australia, who met the American College of Rheumatology criteria for classic or definite RA (9) and who had begun to receive MTX therapy prior to June 1986 (7, 8). Patients' medical records were reviewed and data were extracted and recorded on standardized data collection forms. Information up to at least April 1, 1995 or within 1 year of death was sought. If this information was unavailable from the medical record of the attending rheumatologist, data were sought from the patient's general practitioner, and/or subsequent rheumatologist (if known), and/or hospital records. These data have been reported previously (7, 8).

Identification of malignancy.

The demographic details of all patients were matched to the Victorian State Cancer Registry. This registry records details of all malignancies occurring in Victoria, apart from nonmelanocytic skin cancers. Notification of malignancy in the registry is mandatory by law and virtual complete ascertainment is achieved by notification from pathology laboratories and hospital medical record departments and by screening of death certificates. The International Classification of Diseases, Ninth Revision is used to code site of malignancy (10), and the International Classification of Diseases for Oncology morphology rubrics to code histologic type (11, 12). At the time of the study, the registry was complete up until the end of 1998. The date of diagnosis of malignancy as recorded in the Victorian State Cancer Registry was used in the analysis.

Identification of deaths.

The Australian Institute of Health and Welfare also matched the demographic details of all patients to the National Death Index to identify all deaths up until the end of 1999.

The study was approved by the Ethics Committees of Monash University and The Cancer Council of Victoria.

Statistical analysis.

For all patients, followup started on the date they first started taking MTX and ended either on the last confirmed date seen by the rheumatologist or at death. The Victorian State Cancer Registry became population based in 1982. Due to the absence of population-based cancer registry data prior to 1983, we were unable to exploit the observation time of RA patients prior to exposure to MTX.

Standardized incidence ratios (SIRs) for all malignancies and for selected cancers were calculated using the Victorian population cancer rates stratified by sex, age (in 5-year groups), and calendar year. The SIR compares the malignancy incidence observed in the RA cohort with that expected if the cohort developed malignancy at the same rates as persons in the population of Victoria, Australia. An SIR value >1 indicates an elevated malignancy incidence in the RA cohort relative to the general population.

Crude incidence curves for development of malignancy from the time of initiation of MTX were produced using the Kaplan-Meier method. A Cox regression analysis was performed including sex, age, positive rheumatoid factor, and ever use of azathioprine and cyclophosphamide. All analyses were performed using Stata for Windows, version 6 (StataCorp, College Station, TX).

RESULTS

We followed 458 RA patients (309 women, 149 men) for a total of 4,145 person-years (an average of 9.3 years) and identified 87 malignancies. Of these, 64 were observed during the followup period, 14 were diagnosed before cohort entry (i.e., before initiation of MTX treatment), and 9 were diagnosed after exit from followup.

The cumulative incidence curve for the development of any malignancy from time of initiation of MTX is displayed in Figure 1. The sites or types of malignancy are displayed in Table 1.

Figure 1. Kaplan-Meier survival estimate for malignancy incidence (all malignancies) from time of initiation of methotrexate (proportion and 95% confidence interval).

Table 1. Malignancies occurring in a rheumatoid arthritis cohort of 458 patients prior to methotrexate (MTX) initiation, during followup, and after exit from followup*

Prior to MTX (n = 14)		During followup (n = 64)		After exit from followup (n = 9)
Malignancy	No.	Malignancy	No.	Malignancy	No.
Colorectal	2	Lung	14	Melanoma	2
Bladder	2	Non-Hodgkin's lymphoma	8	Prostate	2
Breast	2	Melanoma	7	Non-Hodgkin's lymphoma	2
Urethra	1	Colorectal	6	Thyroid	1
Glottis	1	Bladder	4	Bladder	1
Cervix	1	Breast	4	Colorectal	1
Melanoma	1	Cervix	3
Kaposi's sarcoma	1	Unknown primary	2
Liver	1	Uterus	2
Lung	1	Brain	2
Uterus	1	Prostate	2
		Pancreas	2
		Liver	1
		Hodgkin's lymphoma	1
		Myeloma	1
		Parotid gland	1
		Sarcoma	1
		Ureter	1
		Ampulla of Vater	1
		Pleura	1

There was an estimated 50% excess risk of malignancy among RA patients exposed to MTX relative to the general population (SIR 1.5, 95% confidence interval [95% CI] 1.2–1.9) (Table 2). The risk of non-Hodgkin's lymphoma was more than 5 times higher in RA patients than in the general population (SIR 5.1, 95% CI 2.2–10.0). Other cancers for which there was evidence of increased risk were lung cancer (SIR 2.9, 95% CI 1.6–4.8) and melanoma (SIR 3.0, 95% CI 1.2–6.2).

Table 2. Standardized incidence ratios (SIRs) for malignancy overall and for specific malignancies in a methotrexate-treated rheumatoid arthritis cohort of 458 patients*

	Observed cancers, no.	Expected cancers, no.	SIR	95% CI
* 95% CI = 95% confidence interval.
Overall	64	42.6	1.5	1.2–1.9
Lung	14	4.9	2.9	1.6–4.8
Non-Hodgkin's lymphoma	8	1.6	5.1	2.2–10.0
Melanoma	7	2.3	3.0	1.2–6.2
Colorectal	6	7.4	0.8	0.3–1.8
Bladder	4	1.9	2.2	0.6–5.5
Breast	4	6.1	0.7	0.2–1.7
Liver	1	0.3	3.7	0.1–20.8
Hodgkin's lymphoma	1	0.1	8.9	0.2–49.8

The results of the Cox regression analysis are presented in Table 3. There was an increased risk of malignancy for MTX-treated patients over age 70 years (n = 68) compared with the general population (hazard ratio 2.94, 95% CI 1.38–6.29) and an increased risk for those who had ever used cyclophosphamide (n = 36; hazard ratio 2.57, 95% CI 1.38–5.69). Use of azathioprine (n = 166) appeared to have a protective effect (hazard ratio 0.51, 95% CI 0.27–0.96).

Table 3. Relative risk of malignancy in a methotrexate-treated rheumatoid arthritis cohort adjusting for sex, age, positive rheumatoid factor, and ever used azathioprine or cyclophosphamide*

	No. (%)	Person-years	Hazard ratio	95% CI
* 95% CI = 95% confidence interval. † Baseline category.
Sex
Female	309 (67)	2,747	1.0
Male	149 (33)	1,398	1.18	0.69–2.03
Age, years
<40	70 (15)	705	0.28	0.06–1.26
40–49	78 (17)	747	1.29	0.55–3.02
50–59†	129 (28)	1,240	1.0
60–69	113 (25)	959	1.98	0.99–3.93
≥70	68 (15)	495	2.94	1.38–6.29
Positive rheumatoid factor
Yes	280 (61)	2,559	1.0
No	98 (21)	889	0.57	0.26–1.22
Unknown	80 (17)	698	1.43	0.77–2.66
Azathioprine
No	292 (64)	2,525	1.0
Yes	166 (36)	1,620	0.51	0.27–0.96
Cyclophosphamide
No	422 (92)	3,814	1.0
Yes	36 (8)	332	2.57	1.38–5.69

DISCUSSION

In this Australian cohort of patients with RA exposed to MTX, we found that relative to the general population, there was an overall increased incidence of malignancy and, specifically, a 3-fold increased risk of melanoma, 5-fold increased risk of non-Hodgkin's lymphoma, and a 3-fold increased risk of lung cancer. This study is, to our knowledge, the first to report an increased risk of melanoma in patients with RA treated with MTX compared with the general population, although several case reports of melanoma occurring in patients with RA treated with MTX have been published (13–15). A recent study that pooled administrative data from 2 US states and 1 Canadian province reported a 2.3-fold increased risk (95% CI 1.55–3.22) of melanoma in patients with RA treated with tumor necrosis factor inhibitors compared with the general population (16). However, 55% of their cohort had received MTX previously and 39% were receiving MTX at the time of initiation of treatment with a biologic agent.

There may be a greater risk of melanoma in Australia where the general population risk of melanoma is also higher than in other parts of the world (17). Environmental factors such as increased exposure to ultraviolet (UV) radiation may play a role in accentuating the risk of melanoma in patients with RA exposed to MTX in our setting. In support of these findings, we have recently observed that the self-reported prevalence of melanoma in RA patients starting biologic therapy and enrolled in the population-based Australian Rheumatology Association Database (ARAD) national registry was greater than expected (18) relative to the Australian general population (17). All patients had been exposed to MTX therapy as well as various other DMARDs. However, self-report may be unreliable, and these data need to be verified by record linkage of the ARAD to the Australian National Cancer Statistics Clearing House, currently underway.

While several studies have suggested an increased risk of nonmelanocytic skin cancer in RA (19–21), we were unable to ascertain the risk of nonmelanocytic skin cancer in our RA cohort because the Victorian State Cancer Registry does not collect these data. An increased risk of melanoma and nonmelanocytic skin cancers in MTX-treated patients with RA may be analogous to the now well-established increased risk of skin cancer (22), first noted in Australia in the 1970s, in recipients of solid organ transplantation (23). The risk has been particularly noted for squamous cell and basal cell carcinoma, while an increased risk of melanoma has been reported in some studies (24–27) but not all studies (22). In Australia, a 2-fold increased risk of melanoma has been observed among kidney transplant recipients relative to the nontransplant population (28). Both immunosuppressive therapy and solar UV radiation are believed to be important risk factors for the development of these malignancies, which may explain the higher rates in Australia compared with those reported elsewhere (28).

It has been postulated that immunosuppression may confer specific effects on UV radiation–related carcinogenesis. For example, in a population-based study in Sweden involving 5,356 patients, there was an increased risk of nonmelanocytic skin cancers, which was higher for sun-exposed areas of the skin and the lip and increased over time (22). Similarly, a single-site study in Norway reported a 65-fold increased risk of skin squamous cell carcinoma (SCC) and a 20-fold increased risk of lip SCC, with the risk related to the degree of immunosuppression (26). An increased duration of exposure to immunosuppressive therapy in patients with RA has also been associated with an increased risk of malignancy (29, 30).

The 5-fold increased incidence of non-Hodgkin's lymphoma in our RA cohort relative to the general population is consistent with the findings of other studies that have examined the overall cancer risk for patients with RA (19, 20, 31–33). While multiple case reports have described spontaneous remission of lymphomas in patients with RA upon withdrawal of MTX therapy (5), studies in the US (34) and France (35) have found that there is insufficient epidemiologic evidence to establish a causal role for MTX treatment in the development of lymphomas in RA. Although Baecklund et al found a strong association between inflammatory activity and increased lymphoma risk, there was no association between increased lymphoma risk and any specific drug therapy (36). They postulated that the observed increased risk may be explained by the fact that patients with more severe disease would have higher inflammatory activity and would therefore be more likely to be treated with second-line DMARDs, although few patients during their study period (1965–1983) had been treated with immunosuppressive therapy. A recent study examining the effect of MTX and anti–tumor necrosis factor therapy on lymphoma risk was similarly unable to establish a causal relationship between RA treatments and development of lymphoma (34). Those authors also postulated that the observed increased lymphoma incidence may reflect channeling bias, that is, channeling of high-risk patients to DMARD treatment.

The 3-fold increased incidence of lung cancer in our RA cohort relative to the general population is also in keeping with previous observations of an increased risk of lung cancer in patients with RA (20, 31–33). Although smoking history was not available for our cohort, a relationship between smoking and the development of lung cancer in patients with RA has been reported (37). Kauppi et al (38) reported an increased risk of lung cancer for Finnish men and suggested that smoking was responsible because Finnish men are more likely to smoke than Finnish women. A recent study comparing 23 RA patients with lung cancer with 6,570 patients with lung cancer alone observed few significant differences between the groups, and concluded that RA does not influence lung cancer stage or shorten survival time for patients with lung cancer (39). Given the paucity of studies investigating the link between RA and lung cancer, it is difficult to draw conclusions about the increased incidence of lung cancer that we observed and the role, if any, of MTX.

We observed a 2.5-fold increased cancer risk for MTX-treated patients with RA exposed to cyclophosphamide, but contrary to expectation, no increased risk with exposure to azathioprine. The latter result is likely to be spurious because it is at odds with other studies including the Rheumatoid Arthritis Azathioprine Registry study, which demonstrated a trend toward an overall increased risk of cancer with azathioprine use (relative risk 1.4) and an increased risk of hematopoietic malignancies (32). Previous studies have also found azathioprine use to be associated with an increased risk of malignancy in patients undergoing immunosuppression for non–transplant-related indications (40). With the increasing use of biologic therapy in the treatment of RA, it will be important to determine whether these agents also augment cancer risk for MTX-treated patients. This is of particular concern in Australia, where prescription of biologic therapy under the government-subsidized scheme is reserved for patients with severe disease who have failed to respond to multiple DMARDs including MTX, and concomitant MTX therapy is a requirement for treatment with some biologic agents (infliximab and anakinra).

Our study has several important limitations. First, because the study was not population based, there is the potential for selection bias. We ensured that all RA patients attending 1 of 6 community-based rheumatology practices who had started MTX therapy prior to June 1986 were included, so it is likely that our patient population is representative of patients with RA seen in community-based rheumatology practices in Melbourne, Australia. In contrast, because cohorts of patients with RA ascertained from hospital clinics are known to have more comorbidities than those ascertained from the community, the true population-based risk of malignancy may be even higher. In support of this, a Swedish study that included 3 cohorts of patients observed an increased risk of nonmelanocytic skin cancer in a prevalent cohort of RA patients admitted to the hospital between 1990 and 2003 relative to the general population, but no increased risk in an incident cohort diagnosed with RA between 1995 and 2003 (41). Notably, there was also an increased risk in the third cohort treated with tumor necrosis factor α antagonists between 1999 and 2003. Second, at the time of the study, the cancer registry was complete up until the end of 1998, and cancers in our study population identified after this time would not have been included in the estimate of cancer risk. Third, without a control group of RA patients who have not been exposed to MTX, it is not possible to determine how much of the increased risk of malignancy we observed can be attributed to MTX.

In summary, our study confirms that in an Australian setting, there is an increased risk of malignancy for patients with RA exposed to MTX relative to the general population, in particular an increased risk of melanoma, non-Hodgkin's lymphoma, and lung cancer. An increased risk of melanoma has not been reported previously but is in keeping with the self-reported prevalence of melanoma in a population-based cohort of patients with RA exposed to MTX enrolled in ARAD and the well-established increased risk of both melanoma and nonmelanocytic skin cancers for transplant recipients in Australia. Further investigation is needed to determine whether this risk is unique to Australia and what role MTX, immunosuppression per se, and/or environmental factors such as exposure to UV radiation play in its development. Our findings, taken together with other studies investigating the risk of skin cancer in patients with RA (21), may support a role for regular skin cancer screening for all patients with RA, particularly those receiving immunosuppressive therapy.

AUTHOR CONTRIBUTIONS

Dr. Buchbinder had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis

Study design. Buchbinder, Wluka, Hall.

Acquisition of data. Buchbinder, Wluka, Giles, Harkness, Lewis, Littlejohn, Miller.

Analysis and interpretation of data. Buchbinder, Barber, Wluka, Hall, Jolley.

Manuscript preparation. Buchbinder, Barber, Wluka, Giles, Hall, Littlejohn, Jolley.

Statistical analysis. Heuzenroeder, Jolley.

Literature review. Buchbinder, Barber.

Patient provision. Ryan.