Melanoma Associated with TNFα Inhibitors: a Research on Adverse Drug events And Reports (RADAR) Project

Abstract

Background

Tumor necrosis factor-alpha inhibitors (TNFαIs) are used for treatment of inflammatory disorders. There is evidence linking these agents with occurrence of malignancies. For four out of five TNFαIs the Food and Drug Administration (FDA) label states, “melanoma has been reported in patients treated with these agents.”

Objectives

Determine whether a statistically-significant association exists between administration of TNFαIs and development of malignant melanoma.

Methods

We searched the FDA Adverse Events Reporting System (FAERS) database for terms related to melanoma and TNFαIs for detection of safety signals. We also searched a large urban academic electronic medical record (EMR) database for which we calculated the relative risk (RR) of melanoma in subjects exposed to TNFαIs vs. non-exposed subjects.

Results

There were 972 reports of melanoma associated with a TNFαIs identified in the FAERS database, with 69 reports among individuals using more than one TNFαI. A safety signal was detected for infliximab (I) golimumab (G), etanercept (E), and adalimumab (A). Cetrolizumab pegol (CP) had no detectible safety signal. For TNFαIs as a class of drugs, a safety signal was detectable in the FAERS database, and RR was significant in the EMR database. For the EMR cohort, 6,045 patients were exposed to TNFαIs and 35 cases of melanoma were detected. Significance for RR was detected for A (RR = 1.8, p = 0.02) and E (RR 2.35, p = 0.0004).

Conclusions

We identified a significant association between exposure to TNFαIs and malignant melanoma in two different analyses. Our findings add to existing evidence linking these agents with the occurrence of malignant melanoma. Additional investigations are required to further explore this association and the risk of melanoma with TNFαI therapy.

Introduction

Tumor Necrosis Factor-α inhibitors (TNFαIs) are widely-used monoclonal antibodies for the treatment of several dermatologic, rheumatologic, and gastrointestinal inflammatory disorders. TNFα is an important cytokine that regulates systemic inflammation, and its inhibition results in immunosuppression and therapeutic benefit in the above systemic inflammatory disorders. Despite the potential therapeutic benefit of TNFαIs, there is some evidence that links TNFαIs with the occurrence of malignancies. Recently-reported retrospective cohort and case-control data have indicated that patients with inflammatory bowel disease who were treated with biologics have an increased risk of melanoma.¹ Moreover, a prospective cohort study showed an increased risk of melanoma in rheumatoid arthritis patients treated with TNFαIs, ² and there are at least 10 case reports of patients developing melanoma after TNFαI treatment. Currently, for four out of five TNFαIs, the FDA package insert states that “melanoma has been reported in patients treated with these agents.” We elected to determine whether a statistically-significant association or safety signal exists between administration of TNFαIs and development of melanoma, regardless of concurrent diagnosis. Our local institution’s electronic medical record (EMR) database (2.2 million individuals) was used for assessing statistical association and the FDA Adverse Event Reporting System (FAERS) database was used for assessing safety signals.

Materials and Methods

Using RADAR methodology, ³ we searched a large urban academic center EMR database and calculated the relative risk (RR) of melanoma in subjects exposed to TNFαIs compared with non-exposed subjects. Again using RADAR methodology, we also searched the FAERS database for terms related to melanoma, including malignant melanoma, malignant melanocytic tumor, melanoma in situ, amelanotic melanoma, metastatic melanoma, lentigo maligna, Hutchinson’s melanocytic freckle, uveal melanoma, and others, combined with all five currently FDA-approved TNFαIs. The search was carried out from the date of each drug’s FDA approval through August 2012. We performed a disproportionality analysis by calculating empirical Bayes geometric means (EBGM) with accompanying 95% confidence intervals (CIs) for detection of safety signals, specifically to determine whether melanoma was reported at a significantly higher rate when compared to reporting for all other drugs in the entire FAERS database.⁴

Results

For the EMR cohort, 6,045 patients were exposed to TNFαIs and 35 cases of melanoma were detected (I, n = 3; G, n = 1; E, n = 17; A, n = 14; CP, n = 0). Significance for RR was detected for A (RR = 1.8, 95% CI 1.06–3.00, p = 0.02) and E (RR 2.35, 95% CI 1.46 to 3.77, p = 0.0004) for prevalence rate of 58 patients per 10,000 patients receiving TNFαIs.

There were 972 reports of melanoma associated with a TNFαI identified in the FAERS database, with 69 reports among individuals using more than one TNFαI. A safety signal was detected for infliximab (I), n = 434 (EBGM 7.90, 95% CI 7.13–8.60), golimumab (G), n = 10 (EBGM 5.34, 95% CI 2.41–9.88), etanercept (E), n = 347 (EBGM 2.49, 95% CI 2.24–2.76), and adalimumab (A), n = 237 (EBGM 2.49, 95% CI 2.19–2.83). Cetrolizumab pegol (CP), n = 13 had no detectible safety signal. For TNFαIs as a class of drugs, a safety signal was detectable in the FAERS database (EBGM 3.30, 95% CI 3.10–3.52) and RR was significant in the EMR database (RR = 1.75, 95% CI 1.25–2.43, p < 0.0009).

Discussion

We identified a significant association between exposure to TNFαIs and development of melanoma for all drugs except CP in the FAERS database. Concordantly, unlike the other four approved TNFαIs, the current FDA labeling for CP does not include melanoma as an adverse effect. We also identified a similar association for drugs A and E in the academic EMR dataset and for the TNFαI class of drugs taken as a whole. Our findings add to existing evidence linking these agents with the occurrence of melanoma, and are the first to analyze the linkage between TNFαIs and melanoma across all concurrent diagnoses. We estimated a prevalence rate associated with TNFαI drugs. However, malignant melanoma and other malignancies are associated with chronic inflammatory conditions in general. Additionally, immunosuppression itself is known to be associated with particular cutaneous malignancies, notably, malignant melanoma. Moreover, limitations of this study include the lack of known melanoma risk factors data, especially phototherapy commonly used in psoriasis patients who represent one of the target population for these drugs. Thus, we did not ascertain TNF alpha inhibitors as an independent risk factor for development of melanoma. Our findings imply that additional investigations are essential to further explore this association and the risk of melanoma with TNFαI therapy. Additionally, from a regulatory standpoint, it may be suggested that the FDA strengthen its labeling for TNFαIs to indicate that they as a drug class have been statistically associated with the development of melanoma.

What is already known about this topic?

• Tumor necrosis factor-alpha inhibitors (TNFαIs) have been linked with the occurrence of malignancies, including melanoma. Statistical association has not been clarified yet

What does this study add?

• A safety signal has been detected in the Food and Drug Administration Adverse Event Reporting System (FAERS) database.

• Enhanced monitoring for patients receiving TNFαIs is therefore warranted.