Abstract
Objective:
To characterize the incidence rate of skin cancer associated with methotrexate and hydroxychloroquine in older adults with rheumatoid arthritis (RA).
Methods:
RA patients aged ≥65 years who initiated methotrexate or hydroxychloroquine as their first disease modifying anti-rheumatic drugs (DMARDs). The primary outcome was new occurrence of any skin cancer (i.e., malignant melanoma or non-melanoma skin cancer; NMSC) based on validated algorithms (positive predictive value >83%). Secondary outcomes were malignant melanoma, NMSC, basal cell carcinoma (BCC), and squamous cell carcinoma (SCC). We estimated the incidence rates (IRs) and hazard ratios (HRs) for each outcome in the 1:1 propensity score (PS)-matched methotrexate and hydroxychloroquine groups.
Results:
We included 24,577 PS-matched pairs of methotrexate and hydroxychloroquine initiators. Compared with hydroxychloroquine (IR 25.20/1,000 person-years), methotrexate initiators (IR 26.21/1,000 person-years) had a similar risk of any skin cancer [HR 1.03 -(95%CI 0.92,1.14)] over a mean follow-up of 388 days. The HR (95%CI) associated with methotrexate was 1.39 (0.87,2.21) for malignant melanoma, 1.01(0.90,1.12) for NMSC, 1.37 (1.13,1.66) for BCC, and 0.79 (0.63,0.99) for SCC compared with hydroxychloroquine.
Conclusion:
In this large cohort of older RA patients initiating methotrexate or hydroxychloroquine as their first DMARD, we found no difference in the risk of skin cancer including malignant melanoma and NMSC. However, for specific components of NMSC, methotrexate initiators had higher risk of BCC but lower risk of SCC compared with hydroxychloroquine initiators.
Keywords: rheumatoid arthritis, skin cancer, nonmelanoma skin cancer, malignant melanoma, basal cell carcinoma, squamous cell carcinoma, cohort study
INTRODUCTION
Malignant melanoma arises from uncontrolled proliferation of melanocytes(1) and can be found in skin, central nervous system, and ocular tract with cutaneous melanoma being the most common form. In 2017, 85,686 new cases of cutaneous melanoma were reported in the U.S (incidence rate 22.7 per 100,000), marking it the sixth most common cancer in the U.S.(2) Another class of skin malignancy, nonmelanoma skin cancer (NMSC), consists of squamous cell carcinoma (SCC) and basal cell carcinoma (BCC) characterized by malignant growth of keratinocytes in epithelial layer. NMSC is the most common type of cancer world-wide, and its incidence is notably increasing partly due to advanced surveillance and aging populations.(3) The exact incidence of NMSC is hard to capture as it is not reported to cancer registers; however, the estimated total number of NMSCs treated in the United States was 5,434,193 in 2012 Medicare data—illustrating its high healthcare burden.(4)
Malignant melanoma and NMSC can occur de novo in individuals who have risk factors for skin cancer including exposure to ultraviolet radiation, fair skin, atypical nevi, or family history of skin cancer.(5) However, they can also arise secondary to immunosuppressed(6) or chronic autoimmune conditions(7) like rheumatoid arthritis (RA).(8) Skin cancers arising in RA are thought to be related to systemic inflammation and dysfunctional immune surveillance combined with adverse effects of medication.(9)
While some previous studies have shown elevated risk of skin cancer in RA patients treated with tumor necrosis factor inhibitor (TNFi),(10, 11) others have found no difference in the NMSC risk specific to RA treatments.(12, 13) Similarly, the link between conventional disease modifying anti-rheumatic drugs (DMARDs) and skin cancer is unclear. A previous study reported a greater risk of melanoma among patients with RA treated with methotrexate compared to the general population;(14) however, another study in patients with psoriasis found no association between methotrexate and the risk of cutaneous malignancy.(15) With hydroxychloroquine, a case-control study reported an increased risk for NMSC associated with high cumulative dose of hydroxychloroquine among patients with systemic lupus erythematosus and primary Sjogren’s syndrome(16). But no clear association with skin cancer risk in RA has been reported.
Recently, interest in the risk of skin cancer in RA patients related to use of DMARDs increased after a placebo-controlled randomized controlled trial of methotrexate (Cardiovascular Inflammation Reduction Trial, CIRT)(17, 18) and observational cohort studies(19, 20) reported elevated risk of NMSC in methotrexate users. However, it remains unclear how methotrexate risk compares to other commonly used conventional DMARDs. Ideally, a randomized controlled trial comparing RA drugs on the outcome of safety endpoints could provide us guidance on the risk of cancer in methotrexate versus other DMARDs. But there has been no adequately powered head-to-head comparison of conventional DMARDs on the risk of skin cancer in patients with RA.
We aimed to compare the rate of any skin cancer including malignant melanoma and NMSC after treatment initiation with methotrexate versus another commonly used non-biologic DMARD, hydroxychloroquine, among older patients with RA.
METHODS
Data Source
We conducted a new user, active comparator cohort study using Medicare fee-for-service claims data between January 1, 2006 and December 31, 2017. Medicare is a federally funded insurance program that covers legal residents in the United States who are 65 years and older or those who have certain disabilities. The database contains longitudinal information on demographics, insurance enrollment, and diagnosis or procedures claims in inpatient setting (Part A), outpatient clinics (part B), and outpatient pharmacy dispensing (Part D). This study was approved by the institutional review board of the Brigham and Women’s Hospital, Boston, Massachusetts (2011P002580–177). Personal identifiers were removed from the data to protect patient confidentiality, and therefore, the requirement for patients’ informed consent was waived. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Study Design and Population
Cohort formation began with identification of the first prescription claim for methotrexate or hydroxychloroquine through Medicare on or after January 1, 2006 (Supplemental Figure S1). Patients were required to have at least 365 days of continuous enrollment in Medicare Parts A, B, and D prior to cohort entry. Using a previously validated claims-based algorithm for RA with positive predictive value (PPV) of 87%,(21) all patients were required to have at least 1 inpatient or 2 outpatient RA diagnosis codes using International Classification of Diseases, 9th Revision (ICD-9) or 10th revision (ICD-10), separated by 7 to 365 days during the 365 days baseline period.
We excluded patients aged less than 65 years old and those with any previous dispensing for other immunosuppressants, TNFi, other biologics, or chloroquine. Patients with history of malignancy including NMSC, radiation therapy, end-stage renal disease or renal dialysis, HIV/AIDS, or psoriatic arthritis were also excluded. Additionally, we excluded those with missing age, gender, or race, no follow-up, and nursing home residents.
Study Outcomes
The primary outcome of interest was new occurrence of any skin cancer consisting of malignant melanoma or NMSC including carcinoma in situ of skin. The secondary outcomes were individual components of the primary outcome: malignant melanoma, NMSC including BCC, and SCC. Carcinoma in situ of skin was included as part of the primary outcome of any skin cancer and secondary outcome of NMSC but it was not assessed separately. Positive control outcome was prespecified as an occurrence of an ophthalmologic exam under the rationale that eye exams are routinely performed with hydroxychloroquine prescription to monitor retinal toxicity.
Outcome Validation
For our outcome ascertainment, we conducted a separate validation study of claims-based algorithms (Supplemental table S1) for cutaneous malignant melanoma and NMSC with chart review using longitudinal Medicare claims data linked to the Partners’ Research Patient Data Registry (RPDR) for the period between January 1, 2008 and September 30, 2015. The details of RPDR are described elsewhere.(22) Following algorithms were selected based on their high performance: 1) malignant melanoma was identified with 2 ICD-9 or ICD-10 diagnosis codes separated by 1 to 60 days (PPV 82%); 2) NMSC including carcinoma in situ was identified based on at least 1 ICD-9 or ICD-10 diagnosis code followed by Healthcare Common Procedure Coding System (HCPCS) codes for NMSC treatment occurring within 60 days (PPV 84%) (Supplemental Table S2).
Follow-up Time
In the primary as-treated analysis, follow-up time started on day after cohort entry and continued until the earliest of the following events: occurrence of the study outcome, end of insurance enrollment, end of study period, death, switch to a comparator drug, discontinuation of index drug, or initiation of biologics/targeted therapies. A patient could contribute to multiple events for secondary outcome analyses (i.e., a patient who had BCC could also contribute to SCC). For primary outcome, occurrence of earliest skin cancer was counted as an outcome. We allowed additional 60 days of supply to methotrexate and hydroxychloroquine prescriptions after end of day supply from the most recent prescription (“grace” period).
Covariates
We measured a total of 62 prespecified covariates (see Table 1 for the full list of covariates) during the baseline period of 365 days before the initiation of study drug. The covariates included demographics, evaluation of health status (combined comorbidity score(23), claims-based frailty index(24)), use of other drugs, RA or skin cancer-related risk factors, and healthcare utilization.
Table 1.
Selected patient characteristics in 365 days prior to cohort entry after 1:1 PS matching
Patient Characteristics | Methotrexate (N = 24,577) | Hydroxychloroquine (N = 24,577) | Abs. Std. Diff. |
---|---|---|---|
Demographic | |||
Age, years, mean (SD) | 73.80 (6.36) | 73.73 (6.45) | 0.011 |
Female | 19,502 (79.4%) | 19,528 (79.5%) | 0.003 |
Region | 0.01 | ||
Northeast | 4,190 (17.0%) | 4,189 (17.0%) | |
South | 10,739 (43.7%) | 10,628 (43.2%) | |
North central | 6,002 (24.4%) | 6,058 (24.6%) | |
West | 3,608 (14.7%) | 3,663 (14.9%) | |
Unknown | 38 (0.2%) | 39 (0.2%) | |
Race | 0.008 | ||
White | 20,656 (84.0%) | 20,663 (84.1%) | |
Black | 2,577 (10.5%) | 2,536 (10.3%) | |
Asian | 446 (1.8%) | 461 (1.9%) | |
Hispanic | 721 (2.9%) | 733 (3.0%) | |
Native American | 177 (0.7%) | 184 (0.7%) | |
Comorbidities | |||
Actinic keratosis | 1,646 (6.7%) | 1,660 (6.8%) | 0.002 |
Atopic dermatitis/contact dermatitis | 74 (0.3%) | 80 (0.3%) | 0.004 |
Psoriasis | 60 (0.2%) | 65 (0.3%) | 0.004 |
Alopecia areata | 26 (0.1%) | 30 (0.1%) | 0.005 |
Vitiligo | 38 (0.2%) | 38 (0.2%) | 0 |
Hypertension | 19,668 (80.0%) | 19,628 (79.9%) | 0.004 |
Hyperlipidemia | 17,411 (70.8%) | 17,510 (71.2%) | 0.009 |
Diabetes | 7,287 (29.6%) | 7,256 (29.5%) | 0.003 |
Combined comorbidity score, mean (SD) | 1.21 (2.08) | 1.21 (2.09) | 0.001 |
Claims-based frailty index, mean (SD) | 0.15 (0.05) | 0.15 (0.05) | 0.002 |
Medications | |||
Amiodarone | 263 (1.1%) | 271 (1.1%) | 0.003 |
First-generation macrolide | 433 (1.8%) | 418 (1.7%) | 0.005 |
Second-generation macrolide | 5,350 (21.8%) | 5,374 (21.9%) | 0.002 |
Tetracycline | 2,236 (9.1%) | 2,249 (9.2%) | 0.002 |
Antiplatelet agents | 2,251 (9.2%) | 2,234 (9.1%) | 0.002 |
Anticoagulant | 2,268 (9.2%) | 2,264 (9.2%) | 0.001 |
NSAIDs and Coxibs | 11,423 (46.5 %) | 11,424 (46.5%) | 0 |
ACE inhibitor | 7,973 (32.4%) | 7,922 (32.2%) | 0.004 |
ARBs | 5,619 (22.9%) | 5,566 (22.6%) | 0.005 |
Beta blockers | 9,739 (39.6%) | 9,672 (39.4%) | 0.006 |
Calcium channel blocker | 7,306 (29.7%) | 7,319 (29.8%) | 0.001 |
Loop diuretics | 4,761 (19.4%) | 4,733 (19.3%) | 0.003 |
Thiazides | 8,396 (34.2%) | 8,314 (33.8%) | 0.007 |
Insulin | 1,466 (6.0%) | 1,448 (5.9%) | 0.003 |
Non-insulin diabetes medications | 4,451 (18.1%) | 4,420 (18.0%) | 0.003 |
Statins | 11,899 (48.4%) | 11,867 (48.3%) | 0.003 |
Antidepressants | 7,951 (32.4%) | 7,917 (32.2%) | 0.003 |
Opioids | 15,648 (63.7%) | 15,623 (63.6%) | 0.002 |
Antipsychotics | 604 (2.5%) | 579 (2.4%) | 0.007 |
Anticonvulsants | 5,184 (21.1%) | 5,146 (20.9%) | 0.004 |
Any oral steroid use | 15,872 (64.6%) | 15,877 (64.6%) | 0 |
Recent oral steroid use* | 12,267 (49.9%) | 12,269 (49.9%) | 0 |
Cumulative prednisone-equivalent milligrams, mean (SD) | 518.98 (1,065.81) | 462.40 (826.25) | 0.059 |
Procedures/ Healthcare utilization | |||
Procedures on benign/premalignant lesions | 2,039 (8.3%) | 2,045 (8.3%) | 0.001 |
Skin biopsy | 303 (1.2%) | 313 (1.3%) | 0.004 |
Dermatologist visit | 3,256 (13.2%) | 3,262 (13.3%) | 0.001 |
Rheumatologist visit | 19,359 (78.8%) | 19,394 (78.9%) | 0.003 |
Emergency room visit | 8,260 (33.6%) | 8,329 (33.9%) | 0.006 |
Recent hospitalization* | 1,034 (4.2%) | 1,047 (4.3%) | 0.003 |
Abbreviations: Abs. Std. Diff., absolute standardized difference; ACE, angiotensin converting enzyme; ARBs, angiotensin II receptor blockers; Coxibs, selective cyclooxygenase 2 inhibitors; HCQ, hydroxychloroquine; N, number; NSAIDs, nonsteroidal anti-inflammatory drugs; NUVB, Narrow-band ultraviolet B; PDT, Photodynamic therapy; PS, propensity score; SD, standard deviation; TIA, transient ischemic attack.
Recent: 60 days prior to (including) the index date.
Statistical Analysis
We first compared the baseline characteristics of the methotrexate and the hydroxychloroquine groups. For confounding control, we used multivariable logistic regression to estimate the propensity score (PS) which is the probability of starting methotrexate versus hydroxychloroquine in this study population with model that included the 62 predefined baseline covariates.(25) Using the nearest-neighbor matching algorithm (maximum caliper width of 0.01 on the PS scale), methotrexate initiators were PS-matched to hydroxychloroquine initiators with a 1:1 ratio. After matching, we assessed covariate balance using the standardized mean differences: variables with a standardized mean difference <0.10 were determined as having acceptable imbalance.(26)
For the primary and secondary outcomes, we estimated the incidence rates (IR) and incidence rate differences (RD) with 95% confidence intervals (CI) in the PS matched groups. Cox proportional hazards regression models estimated hazard ratios (HR) comparing the risk for each outcome among methotrexate versus hydroxychloroquine initiators. The Kaplan-Meier curves were generated to assess the risk of the primary outcome in PS-matched groups over time. Prespecified subgroup analyses were conducted within strata defined by sex, race, baseline steroid use, and baseline NSAID use. We performed a sensitivity analysis with a 90-day latency period by starting the follow-up for outcomes 90 days after the cohort entry date. In addition, we calculated the number of events and incidence rates stratified by follow-up time (i.e., 0–365 days, 366–730 days, and 731+ days) to assess the risk of skin cancer related to the treatment duration. All analyses were conducted using the Aetion Evidence Platform (Aetion, Inc) and R (version 3.31; R Foundation for Statistical Computing).
RESULTS
Baseline Characteristics of Study Population
After applying the inclusion and exclusion criteria (Figure 1), our study cohort consisted of 38,842 new users of methotrexate (75.6% female; age (mean±SD) 73.59±6.27 years) and 25,291 new users of hydroxychloroquine (79.9 % female; age (mean±SD) 73.75±6.46 years). Before PS matching (Supplemental Table S3), methotrexate users had a higher proportion of recent oral steroid use (58.7% versus 48.8%), while hydroxychloroquine users had more history of skin cancer procedures (7.2% versus 8.6%) and dermatologist visit (11.0% versus 13.9%).
Figure 1.
Flowchart of study cohort selection for the methotrexate versus hydroxychloroquine comparison
Abbreviations: AIDS, acquired immunodeficiency syndrome; ESRD, end stage renal disease; HCQ, hydroxychloroquine; HIV, human immunodeficiency virus; MTX, methotrexate; N, number; NMSC, non-melanoma skin cancer; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; TNFi, tumor necrosis factor-alpha inhibitor
After 1:1 PS matching, we had 24,577 pairs of methotrexate and hydroxychloroquine initiators with the mean±SD age of 73.8±6.4 years in the methotrexate group and 73.7±6.5 years in the hydroxychloroquine group. Nearly 80% were female and 84% white. All measured baseline characteristics were adequately balanced with standardized differences <10% (Table 1).
Risk of Malignant Melanoma and NMSC
During a mean follow-up of 388.96 days in the PS-matched cohort (412.57 days for methotrexate, 365.35 days for hydroxychloroquine), 728 methotrexate initiators (IR 26.21 per 1,000 person-years) and 620 hydroxychloroquine initiators (IR 25.20 per 1,000 person-years) developed the primary outcome, i.e., any skin cancer including malignant melanoma, BCC, SCC, or carcinoma in situ of skin (Table 2). The RD between methotrexate and hydroxychloroquine was 1.00 (95% CI −1.75, 3.75) per 1,000 person-years. The HR for any skin cancer associated with methotrexate was 1.03 (95% CI 0.92, 1.14) compared with hydroxychloroquine. The Kaplan-Meier curve of any skin cancer also showed no difference in the risk between the two groups (Figure 2).
Table 2.
Incidence rates and hazard ratios (95% CI) for primary and secondary outcomes in the 1:1 PS matched cohort
Methotrexate (N =24,577) | Hydroxychloroquine (N =24,577) | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
N of events | PY | IR/1,000 PY | RD/1,000 PY (95% CI) | HR (95% CI) | N of events | PY | IR/1,000 PY | RD/1,000 PY (95% CI) | HR (95% CI) | |
Primary outcome | ||||||||||
Any skin cancer * | 728 | 27,780 | 26.21 | 1.00 (-1.75, 3.75) | 1.03 (0.92, 1.14) | 620 | 24,600 | 25.20 | Ref | Ref |
Secondary outcomes | ||||||||||
Malignant melanoma | 46 | 28,938 | 1.59 | 0.45 (-0.16, 1.07) | 1.39 (0.87, 2.21) | 29 | 25,540 | 1.14 | Ref | Ref |
NMSC | 700 | 27,819 | 25.16 | 0.54 (-2.16, 3.25) | 1.01 (0.90, 1.12) | 606 | 24,616 | 24.62 | Ref | Ref |
BCC | 264 | 28,543 | 9.25 | 2.61 (1.11, 4.11) | 1.37 (1.13, 1.66) | 168 | 25,301 | 6.64 | Ref | Ref |
SCC | 140 | 28,741 | 4.87 | −1.14 (-2.39, 0.11) | 0.79 (0.63, 0.99) | 152 | 25,307 | 6.01 | Ref | Ref |
Control outcome | ||||||||||
Ophthalmologic exam | 4,342 | 23,263 | 186.64 | −362.38 (-375.51, −349.26) | 0.39 (0.38, 0.41) | 8,181 | 14,901 | 549.03 | Ref | Ref |
Includes malignant melanoma, non-melanoma skin cancer and carcinoma in situ of skin
Abbreviations: BCC, basal cell carcinoma; CI, confidence interval; HR, hazard ratio; IR, incidence rate; N, number; NMSC, non-melanoma skin cancer; PS, propensity score; PY, person-years; RD, rate difference; SCC, squamous cell carcinoma;
Figure 2.
Cumulative incidence of any skin cancer in the 1:1 propensity score-matched cohort
In the secondary outcome analysis (Table 2), methotrexate was not associated with a higher RD or HR for malignant melanoma or NMSC. However, methotrexate had a higher IR of BCC compared with hydroxychloroquine [RD 2.61 (95% CI 1.11, 4.11) per 1,000 person-years; HR 1.37 (95% CI 1.13, 1.66)]. In addition, methotrexate initiators had a lower IR of SCC versus hydroxychloroquine [RD −1.14 (95% CI −2.39, 0.11) per 1,000 person-years; HR 0.79 (95% CI 0.63, 0.99)]. In the analysis using the receipt of ophthalmic examination as a positive control for hydroxychloroquine use, the occurrence of eye exams in the methotrexate users was lower than hydroxychloroquine users as expected [HR 0.39 (95% CI 0.38,0.41)].
Subgroup and Sensitivity analyses
There was no difference in the risk of any skin cancer stratified by baseline use of steroids or NSAIDs, gender, or race (Supplemental Table S4). For both MTX and HCQ groups, the incidence rate of any skin cancer was higher in males (43.38 per 1,000 PYs for methotrexate, 41.36 per 1,000 PYs for hydroxychloroquine) than in females (22.18 per 1,000 PYs for methotrexate, 20.36 per 1,000 PYs for hydroxychloroquine) and greater in whites (29.38 per 1,000 PYs for methotrexate, 28.16 per 1,000 PYs for hydroxychloroquine) than in non-whites (1.91 per 1,000 PYs for methotrexate, 1.94 per 1,000 PYs for hydroxychloroquine). In the sensitivity analysis with a 90-day latency period, there was no difference in the risk of any skin cancer [HR 1.03 (95% CI 0.92, 1.16)] (Supplemental Table S5). When patients were stratified according to their time of follow-up, the distribution of patients for each follow-up time tier was as follows: 72.64% for methotrexate and 68.15% for hydroxychloroquine between 0–365 days; 13.29% for methotrexate and 15.27% for hydroxychloquine between 366–730 days; 14.07% for methotrexate and 16.58% for hydroxychloroquine beyond 731 days (Supplemental Table S6). In the follow-up time stratified analysis, the incidence rate was highest during the first year of follow-up (46.20 per 1,000 PYs for methotrexate, 41.48 per 1,000 PYs for hydroxychloroquine) and decreased as follow-up time lengthened (Supplemental Table S6). HR for all three time intervals were similar to the primary outcome [HR 1.07 (95% CI 0.91,1.25) for follow-up between 0–365 days; HR 1.03 (95% CI 0.83, 1.28) for follow-up between 366–730 days; HR 1.03(95% CI 0.84, 1.25) for follow-up beyond 731 days. (Supplemental Table S6).
DISCUSSION
In this observational cohort of 49,154 older RA patients, 2.74% of patients developed any skin cancer during a mean follow-up of 388 days after initiating methotrexate or hydroxychloroquine. We observed no difference in the risk of any skin cancer consisting of malignant melanoma and NMSC (BCC, SCC, and carcinoma in situ) between methotrexate and hydroxychloroquine initiators. The results for individual subsets of skin cancer were different, however, with a 37% increased risk for BCC and a 21% lower risk for SCC in the methotrexate group. The risk for malignant melanoma did not differ.
In a previous study based on self-reported questionnaires in RA patients, the risk of NMSC was not elevated in patients on methotrexate without TNFi (HR 1.15; 95% CI 0.81–1.64).(11) Yet in a recent randomized placebo-controlled trial, CIRT, patients without RA or any systemic inflammatory disease who received low dose methotrexate had a 2-fold increase in the risk of skin cancer compared to placebo (HR 2.04; 95% CI 1.28–3.26) (17, 18). The CIRT result was largely driven by the increased risk for SCC (HR 3.31; 95% CI 1.63–6.71), but the risk for BCC (HR 1.35; 95% CI 0.68–2.68) and melanoma (HR 2.33; 95% CI 0.60–9.04) were also numerically increased. Our study did not observe elevated risk of composite skin cancer with methotrexate (HR 1.03; 95% CI 0.92–1.14), and the risk of SCC was lower in the methotrexate users than in hydroxychloroquine users (HR 0.79; 95% CI 0.63,0.99).
Studies which compared methotrexate against other DMARDs as a group also showed different findings. Analyzing RA patients in Medicare, the risk of recurrent NMSC was higher when methotrexate was used in combination with other DMARDs (HR 1.49; 95% CI 1.08–2.37).(19) Also, in a Consortium of Rheumatology Researchers of North America (CORRONA) registry based study, the risk of NMSC was lower in conventional DMARDs than in methotrexate (HR 0.11; 95% CI 0.01–0.91).(20)
To better understand our results, key differences in the study design and population should be reiterated. First, unlike CIRT which compared methotrexate against placebo in patients with no systemic inflammatory disease, our study compared methotrexate with an active comparator, hydroxychloroquine, among older patients with RA. To our knowledge, no studies have directly compared methotrexate against a single conventional DMARD agent, and studies on the effect of methotrexate and hydroxychloroquine on keratinocytes are limited. Both drugs exert immunosuppressive actions which can lead to increased susceptibility to malignancy. However, the relative risk of skin cancer when two drugs are directly compared is unknown. There is a growing body of literatures on hydroxychloroquine’s antitumor activity in non-small cell lung cancer(27) and bladder cancer(28) through its effects on inhibiting autophagy,(29) but the limited evidence provides so far only modest support for a relative benefit of hydroxychloroquine versus methotrexate on malignancy risk.
Second, the differences in the study population and outcome definition can impact the results. The outcome definitions in the previous Medicare RA cohort study by Scott et al.,(19) and our work were similarly based on ICD diagnosis and procedures codes. However, Scott et al. included only patients who already had an incident NMSC event, and the follow-up for the cohort began 1 year after the first NMSC study to assess the outcome of the second NMSC. Consequently, the population captured in Scott et al. study are patients with a history of prior skin cancer who are inevitably at a higher risk of NMSC than our cohort who were malignancy-naïve during the covariate assessment period.(30)
Our subgroup analyses found no significant modification of the overall effect across several prespecified clinical factors. Glucocorticoids possess immunosuppressive effects which could increase the risk of skin cancer, (30–32) and NSAIDs are linked with a lower risk of malignancy by inhibiting the production of COX-2 (33, 34). However, some NSAIDs are not captured completely in insurance claims datab due to the availability of over-the-counter NSAIDs, and a single claims-based history of steroid use may not accurately represent patients who are at increased risk of skin cancer due to prolonged exposure. On the other hand, increased risk of skin cancer in males and whites has been observed consistently throughout past studies. (35),(36, 37) Our study reaffirmed these findings which can be attributed to differential surveillance in addition to other genetic or pathophysiological factors.
In a case-control study using Taiwan National Health Insurance Research Database, RA patients had a higher risk of NMSC (Odds ratio =2.23) compared to those without RA, and the risk was even higher in those with higher cumulative doses of corticosteroids and methotrexate (38). We also assessed the potential association between the duration of the treatment and risk of skin cancer by conducting a follow-up time (i.e., time on treatment) stratified analysis. Most outcomes (NMSC and malignant melanoma) occurred during the first year of the follow-up, and the incidence rate decreased as follow-up time lengthened although the hazard ratios remained similar in all three time intervals.
We also conducted a sensitivity analysis by applying a 90-days latency period based on the rationale that cutaneous malignancy will take time to develop after the initiation of drugs. The results did not differ between our main analysis and the 90-days latency period analysis.
Our study is subject to several limitations. First, we are unable to distinguish whether an apparent difference in risk is due to greater hazard of one treatment or a protective effect of the other. However, we purposefully designed our study as an active comparator, new user design without a nonuser comparator group because non-user comparator designs in observational studies are prone to critical methodological flaws such as confounding by indication, immortal time bias, and healthy user bias.(39) Second, we could not assess RA disease activity and duration which might impact risk for malignancy. Inability to fully control for these factors could lead to confounding by indication (i.e. less severe RA patients may be assigned to hydroxychloroquine which may lead to a outcome rate).(40) But, as noted, we utilized a new user, active comparator design(41) and PS matching (42) to minimize potential bias and balance covariates between treatment groups. In addition to RA disease activity and duration, risk factors for skin cancer (e.g., family history, smoking, amount of UV exposure) were not measurable in claims data. As a proxy for extreme UV exposure, we looked at the proportion of people with ICD diagnosis codes for sunburn which was well-balanced. However, we could not determine the characteristic of sun exposure that can confound an individual’s risk for skin cancer. For example, intermittent overexposure is associated with development of melanoma and BCC, and cumulative sun exposure is associated with SCC.(43) Third, our outcome definition of skin cancer is subject to misclassification with its reliance on claims-based diagnosis and procedure codes. However, our evaluation of these definitions yielded PPVs of 82% for malignant melanoma and 84% for NMSC.
In conclusion, in our large population-based cohort of older RA patients, 2.74% of patients developed any skin cancer during the follow-up. The risk for overall skin cancer did not differ between methotrexate and hydroxychloroquine. However, the secondary outcome analysis showed 37% higher risk of BCC and 21% lower risk of SCC in the methotrexate group than the hydroxychloroquine group. We could not distinguish whether observed differences indicated enhanced risk of one treatment or protective effects of the alternative. Nevertheless, differences in the risk of skin cancer subtypes signify a need to monitor development of cutaneous malignancy in patients who are initiating nonbiologic DMARDs for treatment of RA.
Supplementary Material
Acknowledgements:
This study was funded by internal resources in the Division of Pharmacoepidemiology and Pharmacoeconomics at Brigham and Women’s Hospital. Also Dr. Kim is in part supported by the NIH K24AR078959.
Potential Conflicts of Interest:
Chen reports current employment at Gilead
Desai reports serving as the Principal Investigator on research grants from Vertex, Bayer and Novartis to the Brigham and Women’s Hospital for unrelated projects.
Glynn reports research grants to the Brigham and Women’s Hospital from AstraZeneca, Kowa, Novartis, and Pfizer for unrelated studies.
Weinblatt reports research grants from Amgen, Crescendo Bioscience, Bristol Myers Squibb, and Sanofi for unrelated studies and stock options from Lycera, Canfite, Scipher, Vorso, and Inmedix. Weinblatt has served as a consultant for AbbVie, Amgen, Bristol Myers Squibb, Canfite, Corrona, Crescendo Bioscience, GlaxoSmithKline, Gilead, Eli Lilly, Lycera, Merck, Novartis, Pfizer, Roche, Samsung, Set Point, and Scipher.
Kim has received research grants to the Brigham and Women’s Hospital from Roche, Pfizer, AbbVie, and Bristol-Myers Squibb for unrelated studies.
Gautam, He, Vine, and Lee have nothing to disclose.
References
- 1.In: Ward WH, Farma JM, editors. Cutaneous Melanoma: Etiology and Therapy. Brisbane (AU): Codon Publications; Copyright © 2017 Codon Publications.; 2017. [PubMed] [Google Scholar]
- 2.U.S. Cancer Statistics Working Group. U.S. Cancer Statistics Data Visualizations Tool bosd-USDoHaHS. [updated released in June 2020.]. Available from: www.cdc.gov/cancer/dataviz.
- 3.Christenson LJ, Borrowman TA, Vachon CM, Tollefson MM, Otley CC, Weaver AL, et al. Incidence of basal cell and squamous cell carcinomas in a population younger than 40 years. Jama. 2005;294(6):681–90. Epub 2005/08/11. doi: 10.1001/jama.294.6.681. [DOI] [PubMed] [Google Scholar]
- 4.Rogers HW, Weinstock MA, Feldman SR, Coldiron BM. Incidence Estimate of Nonmelanoma Skin Cancer (Keratinocyte Carcinomas) in the U.S. Population, 2012. JAMA Dermatol. 2015;151(10):1081–6. Epub 2015/05/01. doi: 10.1001/jamadermatol.2015.1187. [DOI] [PubMed] [Google Scholar]
- 5.MacKie RM, Freudenberger T, Aitchison TC. Personal risk-factor chart for cutaneous melanoma. Lancet. 1989;2(8661):487–90. Epub 1989/08/26. doi: 10.1016/s0140-6736(89)92097-7. [DOI] [PubMed] [Google Scholar]
- 6.Kubica AW, Brewer JD. Melanoma in immunosuppressed patients. Mayo Clin Proc. 2012;87(10):991–1003. doi: 10.1016/j.mayocp.2012.04.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kimmel JN, Taft TH, Keefer L. Inflammatory Bowel Disease and Skin Cancer: An Assessment of Patient Risk Factors, Knowledge, and Skin Practices. J Skin Cancer. 2016;2016:4632037-. Epub 2016/02/29. doi: 10.1155/2016/4632037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chakravarty EF, Farmer ER. Risk of skin cancer in the drug treatment of rheumatoid arthritis. Expert Opin Drug Saf. 2008;7(5):539–46. Epub 2008/09/02. doi: 10.1517/14740338.7.5.539. [DOI] [PubMed] [Google Scholar]
- 9.Rheumatoid Arthritis. Annals of Internal Medicine. 2019;170(1):ITC1–ITC16. doi: 10.7326/aitc201901010 %m 30596879. [DOI] [PubMed] [Google Scholar]
- 10.Wolfe F, Michaud K. Biologic treatment of rheumatoid arthritis and the risk of malignancy: analyses from a large US observational study. Arthritis Rheum. 2007;56(9):2886–95. Epub 2007/08/31. doi: 10.1002/art.22864. [DOI] [PubMed] [Google Scholar]
- 11.Chakravarty EF, Michaud K, Wolfe F. Skin cancer, rheumatoid arthritis, and tumor necrosis factor inhibitors. J Rheumatol. 2005;32(11):2130–5. Epub 2005/11/03. [PubMed] [Google Scholar]
- 12.Mercer LK, Green AC, Galloway JB, Davies R, Lunt M, Dixon WG, et al. The influence of anti-TNF therapy upon incidence of keratinocyte skin cancer in patients with rheumatoid arthritis: longitudinal results from the British Society for Rheumatology Biologics Register. Ann Rheum Dis. 2012;71(6):869–74. Epub 2012/01/14. doi: 10.1136/annrheumdis-2011-200622. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Raaschou P, Simard JF, Asker Hagelberg C, Askling J. Rheumatoid arthritis, anti-tumour necrosis factor treatment, and risk of squamous cell and basal cell skin cancer: cohort study based on nationwide prospectively recorded data from Sweden. Bmj. 2016;352:i262. Epub 2016/01/30. doi: 10.1136/bmj.i262. www.icmje.org/coi_disclosure.pdf [DOI] [PMC free article] [PubMed] [Google Scholar]; (available on request from the corresponding author) and declare: support for the submitted work as described above; JA has received research grants from Pfizer and AstraZeneca as part of a public-private research consortium (COMBINE, www.combinesweden.se) and a research grant from Merck; no other relationships of activities that could appear to have influenced the submitted work.
- 14.Buchbinder R, Barber M, Heuzenroeder L, Wluka AE, Giles G, Hall S, et al. Incidence of melanoma and other malignancies among rheumatoid arthritis patients treated with methotrexate. Arthritis Rheum. 2008;59(6):794–9. Epub 2008/06/03. doi: 10.1002/art.23716. [DOI] [PubMed] [Google Scholar]
- 15.Stern RS, Zierler S, Parrish JA. Methotrexate used for psoriasis and the risk of noncutaneous or cutaneous malignancy. Cancer. 1982;50(5):869–72. Epub 1982/09/01. doi: . [DOI] [PubMed] [Google Scholar]
- 16.Tseng HW, Huang WC, Lu LY. The influence of immunosuppressants on the non-melanoma skin cancer among patients with systemic lupus erythematosus and primary Sjögren’s syndrome: a nationwide retrospective case-control study in Taiwan. Clin Exp Rheumatol. 2019;37(6):946–52. Epub 2019/05/11. [PubMed] [Google Scholar]
- 17.Solomon DH, Sparks JA. Adverse Effects of Low-Dose Methotrexate. Ann Intern Med. 2020;173(2):167. Epub 2020/07/21. doi: 10.7326/l20-0525. [DOI] [PubMed] [Google Scholar]
- 18.Vanni KMM, Berliner N, Paynter NP, Glynn RJ, MacFadyen J, Colls J, et al. Adverse Effects of Low-Dose Methotrexate in a Randomized Double-Blind Placebo-Controlled Trial: Adjudicated Hematologic and Skin Cancer Outcomes in the Cardiovascular Inflammation Reduction Trial. ACR Open Rheumatol. 2020;2(12):697–704. Epub 2020/11/18. doi: 10.1002/acr2.11187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Scott FI, Mamtani R, Brensinger CM, Haynes K, Chiesa-Fuxench ZC, Zhang J, et al. Risk of Nonmelanoma Skin Cancer Associated With the Use of Immunosuppressant and Biologic Agents in Patients With a History of Autoimmune Disease and Nonmelanoma Skin Cancer. JAMA Dermatol. 2016;152(2):164–72. Epub 2015/10/29. doi: 10.1001/jamadermatol.2015.3029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Solomon DH, Kremer JM, Fisher M, Curtis JR, Furer V, Harrold LR, et al. Comparative cancer risk associated with methotrexate, other non-biologic and biologic disease-modifying anti-rheumatic drugs. Semin Arthritis Rheum. 2014;43(4):489–97. Epub 2013/09/10. doi: 10.1016/j.semarthrit.2013.08.003. [DOI] [PubMed] [Google Scholar]
- 21.Kim SY, Servi A, Polinski JM, Mogun H, Weinblatt ME, Katz JN, et al. Validation of rheumatoid arthritis diagnoses in health care utilization data. Arthritis Res Ther. 2011;13(1):R32–R. doi: 10.1186/ar3260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lin KJ, Singer DE, Glynn RJ, Murphy SN, Lii J, Schneeweiss S. Identifying Patients With High Data Completeness to Improve Validity of Comparative Effectiveness Research in Electronic Health Records Data. Clin Pharmacol Ther. 2018;103(5):899–905. Epub 2017/09/03. doi: 10.1002/cpt.861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64(7):749–59. Epub 2011/01/07. doi: 10.1016/j.jclinepi.2010.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kim DH, Glynn RJ, Avorn J, Lipsitz LA, Rockwood K, Pawar A, et al. Validation of a Claims-Based Frailty Index Against Physical Performance and Adverse Health Outcomes in the Health and Retirement Study. J Gerontol A Biol Sci Med Sci. 2019;74(8):1271–6. Epub 2018/08/31. doi: 10.1093/gerona/gly197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.ROSENBAUM PR RUBIN DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70(1):41–55. doi: 10.1093/biomet/70.1.41. [DOI] [Google Scholar]
- 26.Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083–107. doi: 10.1002/sim.3697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Malhotra J, Jabbour S, Orlick M, Riedlinger G, Guo Y, White E, et al. Phase Ib/II study of hydroxychloroquine in combination with chemotherapy in patients with metastatic non-small cell lung cancer (NSCLC). Cancer Treat Res Commun. 2019;21:100158. Epub 2019/09/15. doi: 10.1016/j.ctarc.2019.100158. [DOI] [PubMed] [Google Scholar]
- 28.Lin YC, Lin JF, Wen SI, Yang SC, Tsai TF, Chen HE, et al. Chloroquine and hydroxychloroquine inhibit bladder cancer cell growth by targeting basal autophagy and enhancing apoptosis. Kaohsiung J Med Sci. 2017;33(5):215–23. Epub 2017/04/24. doi: 10.1016/j.kjms.2017.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Olsen NJ, Schleich MA, Karp DR. Multifaceted effects of hydroxychloroquine in human disease. Semin Arthritis Rheum. 2013;43(2):264–72. Epub 2013/03/14. doi: 10.1016/j.semarthrit.2013.01.001. [DOI] [PubMed] [Google Scholar]
- 30.Baibergenova AT, Weinstock MA. Oral prednisone use and risk of keratinocyte carcinoma in non-transplant population. The VATTC trial. J Eur Acad Dermatol Venereol. 2012;26(9):1109–15. Epub 2011/09/20. doi: 10.1111/j.1468-3083.2011.04226.x. [DOI] [PubMed] [Google Scholar]
- 31.Jensen A, Thomsen HF, Engebjerg MC, Olesen AB, Friis S, Karagas MR, et al. Use of oral glucocorticoids and risk of skin cancer and non-Hodgkin’s lymphoma: a population-based case-control study. Br J Cancer. 2009;100(1):200–5. Epub 2008/11/27. doi: 10.1038/sj.bjc.6604796. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Sørensen HT, Mellemkjaer L, Nielsen GL, Baron JA, Olsen JH, Karagas MR. Skin cancers and non-hodgkin lymphoma among users of systemic glucocorticoids: a population-based cohort study. J Natl Cancer Inst. 2004;96(9):709–11. Epub 2004/05/06. doi: 10.1093/jnci/djh118. [DOI] [PubMed] [Google Scholar]
- 33.Khan YS, Gutiérrez-de-Terán H, Åqvist J. Molecular Mechanisms in the Selectivity of Nonsteroidal Anti-Inflammatory Drugs. Biochemistry. 2018;57(7):1236–48. Epub 2018/01/19. doi: 10.1021/acs.biochem.7b01019. [DOI] [PubMed] [Google Scholar]
- 34.Ma Y, Yu P, Lin S, Li Q, Fang Z, Huang Z. The association between nonsteroidal anti-inflammatory drugs and skin cancer: Different responses in American and European populations. Pharmacological Research. 2020;152:104499. doi: 10.1016/j.phrs.2019.104499. [DOI] [PubMed] [Google Scholar]
- 35.Liu-Smith F, Farhat AM, Arce A, Ziogas A, Taylor T, Wang Z, et al. Sex differences in the association of cutaneous melanoma incidence rates and geographic ultraviolet light exposure. J Am Acad Dermatol. 2017;76(3):499–505 e3. Epub 2017/04/18. doi: 10.1016/j.jaad.2016.08.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Bradford PT. Skin cancer in skin of color. Dermatol Nurs. 2009;21(4):170–8. [PMC free article] [PubMed] [Google Scholar]
- 37.CDC Melanoma Incidence and Mortality, United States—2012–2016. Available from: https://www.cdc.gov/cancer/uscs/about/data-briefs/no9-melanoma-incidence-mortality-UnitedStates-2012-2016.htm.
- 38.Tseng HW, Lu LY, Lam HC, Tsai KW, Huang WC, Shiue YL. The influence of disease-modifying anti-rheumatic drugs and corticosteroids on the association between rheumatoid arthritis and skin cancer: a nationwide retrospective case-control study in Taiwan. Clin Exp Rheumatol. 2018;36(3):471–8. Epub 20171215. [PubMed] [Google Scholar]
- 39.Lund JL, Richardson DB, Stürmer T. The active comparator, new user study design in pharmacoepidemiology: historical foundations and contemporary application. Curr Epidemiol Rep. 2015;2(4):221–8. Epub 2016/03/10. doi: 10.1007/s40471-015-0053-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wilton KM, Matteson EL. Malignancy Incidence, Management, and Prevention in Patients with Rheumatoid Arthritis. Rheumatol Ther. 2017;4(2):333–47. Epub 2017/05/17. doi: 10.1007/s40744-017-0064-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Yoshida K, Solomon DH, Kim SC. Active-comparator design and new-user design in observational studies. Nat Rev Rheumatol. 2015;11(7):437–41. Epub 2015/03/25. doi: 10.1038/nrrheum.2015.30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Jackson JW, Schmid I, Stuart EA. Propensity Scores in Pharmacoepidemiology: Beyond the Horizon. Curr Epidemiol Rep. 2017;4(4):271–80. Epub 2018/02/20. doi: 10.1007/s40471-017-0131-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Kaae J, Boyd HA, Hansen AV, Wulf HC, Wohlfahrt J, Melbye M. Photosensitizing medication use and risk of skin cancer. Cancer Epidemiol Biomarkers Prev. 2010;19(11):2942–9. Epub 2010/09/24. doi: 10.1158/1055-9965.Epi-10-0652. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.