Weights and between-subgroup heterogeneity test are from random-effects model. PM indicates Paule-Mandel estimate of tau squared.
Weights are from random-effects model. PM indicates Paule-Mandel estimate of tau squared.
A, For patients with rheumatoid arthritis. B, For patients with psoriasis/psoriatic arthritis. Weights are from random-effects model. PM indicates Paule-Mandel estimate of tau squared.
eFigure. PRISMA 2020 flow diagram for new systematic reviews which included searches of databases and registers only
eTable 1. Search Strategy for MEDLINE/EMBASE/CENTRAL via OVID MEDLINE 12th May 2022
eTable 2. Cochrane risk of bias 2 tool – randomized trials
eTable 3. Joanna Briggs Institute checklist – cohort studies
eTable 4. Joanna Briggs Institute checklist – case-control studies
Customize your JAMA Network experience by selecting one or more topics from the list below.
Yan MK, Wang C, Wolfe R, Mar VJ, Wluka AE. Association Between Low-Dose Methotrexate Exposure and Melanoma: A Systematic Review and Meta-analysis. JAMA Dermatol. 2022;158(10):1157–1166. doi:10.1001/jamadermatol.2022.3337
Is exposure to low-dose methotrexate associated with a higher risk of melanoma development?
In this systematic review and meta-analysis pooling 16 642 cases of melanoma, individuals exposed to low-dose methotrexate had a 15% increased risk of melanoma compared with the risk in unexposed individuals. However, the melanoma risk associated with methotrexate exposure may not translate to a significant association at a population level, as the number needed to harm was estimated to be large, even for populations with high melanoma incidence rates.
Findings suggest that although low-dose methotrexate may be associated with higher risk of malignant melanoma, the true effect could be considered negligible.
Methotrexate is widely used for the treatment of inflammatory disorders, including rheumatoid arthritis. Studies suggest that methotrexate may be associated with an increased risk of melanoma.
To determine whether methotrexate exposure is associated with an increased risk of cutaneous melanoma.
MEDLINE, Embase, Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov were searched from inception to May 12, 2022, for eligible studies.
Case-control studies, cohort studies, or randomized clinical trials (RCTs) were included if they examined the odds or risk of cutaneous melanoma in individuals exposed to low-dose methotrexate in comparison with individuals unexposed. No language limitations were applied.
Data Extraction and Synthesis
Two independent reviewers extracted data on study characteristics and outcome data. The Meta-analysis of Observational Studies in Epidemiology guidelines were followed. To assess study quality, the Cochrane risk of bias tool was used for RCTs, and the Joanna Briggs Institute Checklist was used for cohort and case-control studies. Odds ratio from case-control studies and relative risk or hazard ratio from cohort studies or RCTs were pooled, and a random-effects model meta-analysis was conducted.
Main Outcomes and Measures
Prespecified outcome was the odds ratio, hazard ratio, or risk ratio of cutaneous melanoma comparing low-dose methotrexate exposure with nonexposure.
Seventeen studies (8 RCTs, 5 cohort studies, 4 case-control studies) were eligible for inclusion, and of these, 12 studies with 16 642 cases of melanoma were pooled in the primary analysis. Indications for methotrexate included rheumatoid arthritis, psoriasis, psoriatic arthritis, and inflammatory bowel disease and were unknown in 5 studies. Compared with unexposed individuals, study participants with methotrexate exposure had a small increased risk of melanoma (pooled relative risk, 1.15; 95% CI, 1.08-1.22), but this did not persist in a sensitivity analysis excluding the largest study (pooled relative risk, 1.11; 95% CI, 1.00-1.24). Subgroup analyses according to comparator group (comparing methotrexate exposure with either immunomodulator alone vs immunomodulator and methotrexate) or the indication for methotrexate being rheumatoid arthritis provided similar risk estimates. Using geographical population melanoma incidence rates, a number needed to harm of 18 630 was calculated in Australia, and 41 425 in North America.
Conclusions and Relevance
In this systematic review and meta-analysis, low-dose methotrexate exposure was associated with an increased melanoma risk, but the absolute risk increase could be considered negligible.
Globally, the incidence of melanoma is increasing, and it is projected to become the second most common cancer in the US by 2040.1,2 Melanoma is an aggressive form of skin cancer that tends to metastasize, and although it accounts for less than 5% of all skin cancer cases, it is responsible for the majority of skin cancer deaths.3,4 Additionally, treatment costs for melanoma are also rapidly rising, particularly with the advent of new pharmacotherapies for metastatic melanoma.5 The substantial public health and financial burdens of melanoma highlight the need to identify high-risk individuals to target surveillance efforts.
Low-dose methotrexate is a widely used medication to treat immune-mediated inflammatory conditions recommended as the primary therapy for inflammatory arthritis6 and to treat a wide array of inflammatory dermatoses.7 It suppresses T-cell activation and affects other immune responses including antimetabolite and cytotoxicity by inhibiting dihydrofolate reductase and thymidylate synthetase, increasing extracellular adenosine (by inhibiting aminoimadzole-4-carboxamide ribonucleotide transformylase [ATIC]), and inhibiting intracellular signaling pathways, including nuclear factor-kB and JAK-STAT pathways.8
Despite decades of clinical use, some studies suggest that low-dose methotrexate may be associated with an increased risk of skin cancer.9 Prospective studies, including clinical trials, have shown a possible association with increased risk of melanoma, although these studies are limited by a very low number of melanoma events and short duration of follow-up.10,11 Some population-based observational studies have reported an association between melanoma and low-dose methotrexate, but these studies could not adjust for established melanoma risk factors such as UV light exposure and personal and family melanoma history.12,13 Furthermore, these studies defined the general population as the comparator, raising the possibility of confounding by indication.
Comprehensively defining the risk of melanoma associated with low-dose methotrexate is important for clinicians across several medical disciplines. To our knowledge, no systematic review and meta-analyses has examined this association. Thus, we performed a systematic review and meta-analysis to determine whether exposure to low-dose methotrexate is associated with higher risk of malignant melanoma.
This systematic review was prospectively registered with PROSPERO (CRD42021251001) and reported in accordance with the Meta-analysis of Observational Studies in Epidemiology (MOOSE) reporting guideline14 and the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline.15 This study was exempt from institutional review board approval because it is a systematic review and meta-analysis of existing literature.
Our search strategy was designed with specialist institutional librarian support. Two study investigators (M.K.Y., C.W.) undertook electronic searches in 3 biomedical databases—MEDLINE, Embase, and Cochrane Central Register of Controlled Trials—from their dates of inception until May 12, 2022. We combined the terms melanoma and methotrexate (and its synonyms) as either key words or subject headings (eFigure and eTable 1 in the Supplement). We reviewed the reference lists of all retrieved articles and searched melanoma events reported as adverse events in all completed methotrexate trials registered with ClinicalTrials.gov, as rare safety events may not be explicitly stated.
We included all observational (case-control, cohort studies) and interventional studies (randomized clinical trials [RCTs]) in which melanoma occurrence was identified in a cohort of patients exposed to low-dose methotrexate and compared with an unexposed individually defined comparator group. Low-dose methotrexate was defined as a dose of 5 to 30 mg weekly. We included only the most completely reported results relating to our review question from studies with multiple published analyses. Exclusion criteria included cohorts of fewer than 50 participants in each comparator group, nonhuman studies, pediatric cohorts, conference abstracts, case reports, and noncomparative studies. Language was not an exclusion criterion. Eligibility was independently assessed by 2 investigators (M.K.Y. and C.W.); initially, titles and abstracts were screened for potentially relevant studies. Their full texts were accessed and read to determine final eligibility. Disagreements about eligibility were resolved by discussion and consensus.
All data were extracted independently by 2 investigators (M.K.Y., C.W.). The data recorded included study characteristics (author, year of publication, study design, methotrexate indication, exposure duration, study size, comparator group, and duration of follow-up) and outcome data (number/proportion of melanoma cases in the methotrexate and comparator groups, melanoma definition/diagnosis) and statistical measures of risk (risk estimates, 95% CIs, adjustment for confounding). Attempts to contact study authors were made if the required data were not reported.
To assess the risk of bias, we used the Cochrane risk-of-bias tool for RCTs16 and the Joanna Briggs Institute Checklist for cohort and case-control studies (eTables 2, 3, and 4 in the Supplement).17 Discrepancies were resolved by discussion and consensus.
A random-effects meta-analysis model and the Paule-Mandel estimator were used to calculate the pooled relative risk (RR) and account for heterogeneity between studies to determine the effect of methotrexate on melanoma risk (RR with corresponding 95% CI). If studies did not report a measure of association, an unadjusted odds ratio (OR) was calculated for case-control studies, and RR for cohort studies and RCTs. The RR was used as the measure of association because melanoma is a rare outcome, and in this situation, different measures of effect size (OR, RR, and hazard ratio [HR]) yield mathematically similar estimates.18 Studies that compared methotrexate vs an active or unspecified comparator were included in the primary analysis. In studies including more than 2 comparator arms, data were extracted to compare methotrexate alone vs immunomodulator alone. If no melanoma events occurred in both arms, the study was excluded from the analysis. If no event occurred in 1 comparator arm, ½ was added to each cell to allow computation of a RR. The results were reported as forest plots of the RR with the 95% CI, grouped by study design.
The Cochran Q test was applied to test for statistical heterogeneity among the pooled estimates, and the I2 statistic was used to quantify the proportion of heterogeneity in individual studies, with more than 50% considered high.19 All analyses were conducted using Stata statistical software, version 17.0 (StataCorp).
Prespecified subgroup analyses were performed according to the comparator group, comparing patients exposed to immunomodulators vs those taking immunomodulator in addition to methotrexate, and comparing patients exposed to an immunomodulator vs methotrexate alone. In studies including more than 2 comparator arms (eg, methotrexate alone, methotrexate plus other immunomodulators, or other immunomodulator alone), data were extracted for both of these analyses where possible. Multiplicity adjustments and power analyses were not performed because these were exploratory in nature. Subgroup analyses were also performed according to indication for methotrexate using studies that published relevant data.
Sensitivity analysis was performed in which RRs were estimated with inclusion of studies with high risk of bias. A post hoc sensitivity analysis, excluding the largest study, was performed to examine whether this study may have dominated the overall results.12
A number needed to harm (NNH) was computed from the pooled RR estimated in the primary analysis and an assumed comparator risk based on the Global Cancer Observatory melanoma risk for Australia, North America, and globally.20
We identified a total of 1412 references from the electronic database search and 515 search results from the ClinicalTrials.gov registry (eFigure in the Supplement). A total of 1806 articles were excluded after screening of title and abstracts and removal of duplicates. The full text was read for 100 articles, which yielded 16 eligible studies. An additional study was identified after review of reference lists. Thus, 17 studies were included.
Table 111-13,21-34 describes the 17 included studies: 8 were RCTs, 5 were cohort studies, 2 were nested case-control studies, and 2 were case-control studies. The studies included a total of 889 799 participants with 17 513 cases of melanoma. Six studies included patients with rheumatoid arthritis (RA),21,22,24,26,27,29 1 with psoriatic arthritis,25 1 with granulomatosis with polyangiitis,23 2 with inflammatory bowel disease,28,34 and 1 with psoriasis,32 and the others were unspecified regarding the indication for methotrexate or the presence of underlying disease in the control group.
All RCTs, except 1,11 included participants with active immune-mediated inflammatory diseases and had multiple intervention arms (ie, methotrexate vs methotrexate with other immunomodulator vs other immunomodulator alone). The follow-up time in most RCTs was relatively short, generally around 1 to 2 years. Melanoma events were reported as adverse events and occurred in low numbers. Most observational studies had longer follow-up and used a general population as the comparator.
Including studies with multiple comparator arms that had sufficient melanoma event numbers to allow analysis (ie, not 0 events in both arms), 6 studies examined methotrexate vs other immunomodulator,21,23,24,26-28 6 studies examined methotrexate vs unspecified comparator, and 3 studies examined methotrexate plus other immunomodulator vs other immunomodulator.22,25,29 The latter 3 studies were not included in the primary analysis given the presence of another immunomodulator in the methotrexate-exposed arm.
The risk of bias assessment for included studies is detailed in eTables 2, 3, and 4 in the Supplement and was judged to be at low risk, unclear risk, or some concerns of bias in 15 studies. Common critiques in RCTs included lack of allocation concealment, although this was considered less significant given that melanoma is a hard end point. Inadequate adjustment of confounders was commonly identified in cohort studies. One case-control study deemed at high risk of bias was a data mining exercise, and concerns about the controls being dissimilar to cases were raised because they represented individuals who reported adverse drug reactions to other medications. Additionally, it was unclear whether cases and controls were matched appropriately, and details about the duration of methotrexate exposure were not mentioned. Confounding factors were also not identified or adjusted for.33 Concerns were raised regarding another case-control study also assessed to be at high risk of bias because cases and controls were not matched, cases were more likely than controls to have more extensive inflammatory bowel disease, controls were selected from a separate population cohort compared with cases, details about methotrexate exposure definition and duration were unclear, and confounding factors were not adjusted for.34
Of the 17 included studies, 2 studies were excluded due to high risk of bias,33,34 and 3 because the exposed arm included exposure to another immunomodulator.22,25,29 Thus, 12 studies with 16 642 cases of melanoma were pooled to examine the association between methotrexate and melanoma risk (Table 2). Compared with no methotrexate use, exposure was associated with a small increased melanoma risk (RR, 1.15; 95% CI, 1.08-1.22). Heterogeneity was minimal (I2 = 0%) (Figure 1). Further analyses including 10 studies11-13,21,23,28-32 that had at least 2 years of follow-up time yielded similar results: RR, 1.14; 95% CI, 1.04-1.25. Inclusion of 2 studies deemed to be at high risk of bias demonstrated a greater increased risk; however, introducing the 2 studies created moderate heterogeneity: RR, 1.24; 95% CI, 1.03-1.49; I2 = 48.4%.
Considering the potential for a large study to dominate the primary analysis,12 an additional sensitivity analysis excluding this study was conducted in which the association was statistically nonsignificant (RR, 1.11; 95% CI, 1.00-1.24).
Three studies22,25,29 that compared the effect of methotrexate with a concomitant immunomodulator (etanercept, abatacept, baricitinib) vs immunomodulator alone were pooled. Although the combination of methotrexate with an immunomodulator showed an increased melanoma risk compared with an immunomodulator alone, this estimate had low precision (RR, 1.23; 95% CI, 0.56-2.68) (Figure 2A).
Six studies21,23,24,26-28 comparing the effect of methotrexate monotherapy against another immunomodulator (thiopurine, JAK inhibitor, tumor necrosis factor-α inhibitors) were pooled. An increased melanoma risk was observed with methotrexate therapy but was estimated with low precision (RR, 1.38; 95% CI, 0.48-3.97) (Figure 2B).
Pooling studies including only participants with RA, the RR for the effect of methotrexate exposure on melanoma risk was similar in magnitude to the primary analysis, but the estimate had a wide CI (RR, 1.13; 95% CI, 0.55-2.33) (Figure 3A).
Two studies included participants with either psoriasis or psoriatic arthritis. No association between melanoma risk and exposure to methotrexate was observed (RR, 1.01; 95% CI, 0.79-1.28) (Figure 3B).
Based on results of the primary analysis with an estimated 15% increased melanoma risk, an NNH was calculated. An NNH of 196 078, 41 425, and 18 630 was calculated using global, North American, and Australian melanoma incidence rates, respectively.20 This translated to corresponding absolute risk increases in those populations of 0.0005%, 0.002%, and 0.005%, respectively.
This systematic review and meta-analysis is, to our knowledge, the first to specifically examine the association between methotrexate and melanoma risk. Low-dose methotrexate exposure was associated with a potential 15% increased risk of melanoma, though this estimate should be viewed with caution. Nevertheless, this risk may be insignificant on a population level, as the calculated NNH was greater than 18 000 in high-incidence regions such as Australia,3 and greater than 195 000 based on global melanoma incidence rates.
It has been unclear whether methotrexate increases the risk of melanoma. Some immune-mediated inflammatory conditions have been associated with an elevated risk of numerous cancers, including melanoma.35,36 Thus, the use of the community baseline risk of melanoma as the comparator may overestimate the risk. We noted that larger studies did not exclude those without immune-mediated inflammatory conditions. However, such conditions are not common; thus, most of the comparison group were likely to be free of this potential cause of increased risk. In subgroup analyses of studies including patients with RA only, a similarly elevated risk was observed but was not statistically significant; the magnitude of risk in patients with psoriasis/psoriatic arthritis was lower and also not significant. Subgroup analyses did not show a higher risk of melanoma when methotrexate was used in conjunction with, or compared to, other immunomodulators. A priori power calculations were not performed for subgroup analyses nor were multiplicity adjustments made, and hence these results should be interpreted with caution.
Methotrexate may contribute to melanoma development via its immunosuppressive effects and photosensitizing properties.37 Melanoma is among the most immunogenic tumors, with immune cells playing a pivotal role in tumor development, progression, and development.38 Methotrexate is thought to exert systemic anti-inflammatory effects through the reduction of purine pools required for T-lymphocyte proliferation as well as increased adenosine signaling due to the disruption of ATIC.8 In the skin, methotrexate may have additional immunosuppressive effects, such as the inhibition of tissue-specific lymphocytes.39 Furthermore, methotrexate has been associated with photosensitivity reactions; in particular, photo recall of a recent sunburn has been well described in case series.40
Our study has some limitations. First, most included articles did not adjust for melanoma risk factors such as family history of melanoma, cumulative UV radiation exposure, history of blistering sunburns, skin phototype, and total nevus count. Additionally, most studies did not specify whether reported melanoma events included cases of in situ as well as invasive melanoma. Furthermore, we were also only able to assess confounding by indication using the few studies that included only patients with RA or psoriasis, with reduced power. In observational studies, individuals receiving methotrexate may be more engaged with the health care system and therefore more likely to be diagnosed with a melanoma, raising the possibility of detection bias. This is unlikely to affect RCTs, given that all participants have the disease. A higher point estimate of melanoma risk of lower precision was observed in the RCTs. Overall, as most participants were drawn from observational studies, it is possible that differential health care engagement contributed to the observed increased risk. Lastly, considering the latency of melanoma, the short follow-up time for RCTs may be insufficient to detect a true effect.
Nevertheless, our study has several strengths. It used a comprehensive search strategy including multiple databases and registers without language restrictions and applied strict eligibility criteria to identify relevant articles addressing our clinically relevant question. We examined disease-specific cohorts and population comparator cohorts separately and found similar results.
This study provides valuable, and we believe reassuring, information to clinicians to guide prescribing and surveillance practices. Previous studies have suggested an increased risk. We have quantified this risk and found that although methotrexate was associated with a statistically significant 15% increased melanoma risk, the absolute risk increase, which takes into account population melanoma incidence rates, was very low to minimal.41 The number needed to treat to cause 1 additional melanoma was high and estimated to be 18 000 even in Australia, where the estimated lifetime risk of melanoma is particularly high at 1 in 15 people.42 In other words, in a country such as Australia, which has one of the world’s highest melanoma incidence rates, 18 000 individuals would need to be treated with methotrexate for 1 additional melanoma to occur. In North America, where the incidence of melanoma is relatively lower, 41 425 individuals would need to be treated for 1 additional melanoma to occur. A higher NNH is usually required for more serious outcomes such as melanoma, with triple-digit values considered acceptable.43 An NNH of 18 000 is considerable, and comparable, for example, to the lifetime cancer risk of plain hip radiographs performed in young adults with hip pain (NNH = 16 667).44 In a sensitivity analysis excluding a large study that used the general population as the comparator, the melanoma risk estimate was not statistically significant. Thus, while the results of our primary analysis are statistically significant, these results must still be viewed with caution.
Defining the carcinogenic potential of low-dose methotrexate is crucial in the long-term care of patients with immune-mediated disease. There remains a need for large, prospective studies in disease populations with longer periods of follow-up. Studies assessing the effect of low-dose methotrexate therapy following a melanoma diagnosis, examining the risk of disease recurrence or subsequent primary melanoma, would also be prudent. The decision to continue or cease methotrexate therapy after a melanoma diagnosis is a common clinical question that has not been adequately addressed in the literature.
In this systematic review and meta-analysis, results show an association of methotrexate with a higher risk of incident malignant melanoma. However, the absolute increase in risk is small, even in populations at relatively high risk of melanoma. This finding provides information to better inform patients when initiating methotrexate therapy. To distinguish the degree of risk associated with methotrexate in populations with inflammatory disease, large prospective studies with longer follow-up are required. Future studies examining the effect of methotrexate continuation after the diagnosis of melanoma are required to inform practice.
Accepted for Publication: June 22, 2022.
Published Online: August 31, 2022. doi:10.1001/jamadermatol.2022.3337
Corresponding Author: Anita E. Wluka, MBBS, PhD, School of Public Health and Preventive Medicine, Monash University, 553 St Kilda Rd, Melbourne, VIC 3004, Australia (email@example.com).
Author Contributions: Dr Yan 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.
Concept and design: Yan, Wang, Mar, Wluka.
Acquisition, analysis, or interpretation of data: Yan, Wang, Wolfe, Wluka.
Drafting of the manuscript: Yan, Wang.
Critical revision of the manuscript for important intellectual content: Wang, Wolfe, Mar, Wluka.
Statistical analysis: Yan, Wang.
Administrative, technical, or material support: Yan.
Supervision: Wolfe, Mar, Wluka.
Conflict of Interest Disclosures: Dr Yan reported receiving personal fees (stipend scholarship during PhD enrollment) from Australian Government Research Training Program during the conduct of the study. Dr Mar reported receiving grants from National Health and Medical Research Council Early Career Fellowship during the conduct of the study; grants from MoleMap paid to institution for clinical trial and personal fees from Bristol Myers Squibb, Merck, Novartis, and Janssen outside the submitted work. Dr Wluka reported receiving grants from a Royal Australasian College of Physicians Career Development Fellowship outside the submitted work. No other disclosures were reported.
Funding/Support: Drs Yan and Wang are supported by Australian Government Research Training Program Scholarships. Dr Mar is supported by a National Health and Medical Research Council Early Career Fellowship.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: The authors thank Ms Lorena Romano, MBIT, from the Alfred Health Ian Potter Library, for her assistance in developing the database search strategy. She was not compensated for this work.
Additional Information: The data extracted from included studies and used for analyses are presented in the Supplement. The templates used for data collection are publicly available from the Cochrane Methods Bias and Joanna Briggs Institute Critical Appraisal Tools websites.