Drug-induced skin ulcers: A disproportionality analysis from the WHO pharmacovigilance database

Abstract

To the Editor: Skin ulcerations have occasionally been reported as adverse drug reactions.1D'Epiro S. Salvi M. Luzi A. et al.Drug cutaneous side effect: focus on skin ulceration.Clin Ter. 2014; 165: e323-329PubMed Google Scholar However, to our knowledge, no pharmacovigilance study has been performed, and little is known about drug-related skin ulcers.We therefore performed combined disproportionality analyses using data from VigiBase, the World Health Organization's pharmacovigilance database, and from the biomedical literature through MEDLINE, to identify pharmacologic classes suspected of inducing skin ulcers.2Bate A. Lindquist M. Edwards I.R. et al.A Bayesian neural network method for adverse drug reaction signal generation.Eur J Clin Pharmacol. 1998; 54: 315-321Crossref PubMed Scopus (665) Google Scholar,3Xu R. Wang Q. Large-scale combining signals from both biomedical literature and the FDA Adverse Event Reporting System (FAERS) to improve post-marketing drug safety signal detection.BMC Bioinformatics. 2014; 15: 17Crossref PubMed Scopus (60) Google Scholar Then, all disproportionality signals were independently reviewed by a committee of experts in pharmacology and dermatology, and selected according to their clinical relevance, the strength of evidence, and previous knowledge of possible associations (Supplemental methods available via Mendeley at 10.17632/sgnzg8k9tt.1).4European Medicines Agency (EMA). Guideline on good pharmacovigilance practices (GVP) - Module IX – Signal management (Rev 1). Accessed April 26, 2021. Available at: https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-good-pharmacovigilance-practices-gvp-module-ix-signal-management-rev-1_en.pdfGoogle ScholarFrom the 21,421,352 cases available at extraction (March 3, 2020), we identified 19,887 reports of “skin ulcer,” 4440 of “decubitus ulcer,” 596 of “diabetic ulcer,” 103 of “varicose ulceration,” and 79 of “ischemic skin ulcer.” Sixty-one drugs displayed a disproportionality signal for both “skin ulcer” and at least 1 other specific skin ulcer term (Supplemental Fig 1). Among 814 records of chemically induced skin ulcers identified in the literature after screening of titles and abstracts, 241 reports regarding 59 drugs were selected, among which 10 had a disproportionality signal.Among the 61 drugs selected for signal validation, 22 were excluded for obvious indication or protopathic bias, or were combined substances (Supplemental Table I). Then, after retrieval of references databases, Food and Drug Administration and European Medicines Agency labels, and safety alerts, 21 drugs were considered as “already known” and the plausibility of the remaining 19 drugs was assessed through discussion during expert meetings based on cases analyses, potential bias, and pharmacologic mechanisms (Supplemental Tables II and III). Finally, 13 drugs were considered as having potential signals for skin ulcer: 3 protein kinase inhibitors (afatinib, ibrutinib, and tofacitinib), 2 immunomodulators (thalidomide and lenalidomide), 2 interferons (alfa and beta), 1 bisphosphonate (alendronic acid), and leflunomide, rofecoxib, mitomycin, ocrelizumab, and digoxin (Table I and Supplemental Table IV).Table IResults of the disproportionality analysesDrug classesDrugsSDR in WHO pharmacovigilance databaseSDR in MEDLINEKnown associationsProtein kinase inhibitorsSorafenib2.19Ulceration∗Food and Drug Administration summary of product characteristics.Sunitinib1.99Cracking of skin∗Food and Drug Administration summary of product characteristics.Afatinib†New safety signals.1.02Erlotinib2.12Skin fissure∗Food and Drug Administration summary of product characteristics.Ibrutinib†New safety signals.2.10Cabozantinib4.63Skin ulcer‡EMA summary of product characteristics.Lenvatinib2.56Impaired wound healing§Micromedex database.Pazopanib1.10Skin ulcer∗Food and Drug Administration summary of product characteristics.Tofacitinib†New safety signals.0.51Regorafenib0.97Impaired wound healing§Micromedex database.Ponatinib0.77Skin painmTOR inhibitorsEverolimus1.15Impaired wound healing§Micromedex database.Sirolimus1.66Impairment of wound healing∗Food and Drug Administration summary of product characteristics.AntimetabolitesMethotrexate1.521.84Skin ulcer∗Food and Drug Administration summary of product characteristics.Hydroxycarbamide5.855.95Leg ulcer∗Food and Drug Administration summary of product characteristics.Teriflunomide†New safety signals.0.02Calcineurin inhibitorTacrolimus0.16Skin ulcer∗Food and Drug Administration summary of product characteristics.ImmunomodulatorsLenalidomide†New safety signals.0.19Thalidomide†New safety signals.0.15Antitumor antibioticMitomycin†New safety signals.2.520.39CD-20 antibodyOcrelizumab†New safety signals.0.12BisphosphonatesAlendronic acid†New safety signals.1.49VEGF inhibitorsBevacizumab1.33Wound healing complications∗Food and Drug Administration summary of product characteristics.TNF-α inhibitorsAdalimumab0.65Signal of skin ulcerPharmacovigilance Risk Assessment Committee (PRAC) safety signal assessment.Etanercept1.11Leg ulcer‡EMA summary of product characteristics.InterferonsInterferon beta†New safety signals.0.141.59Interferon alfa†New safety signals.0.990.81Cardiac glycosidesDigoxin†New safety signals.0.77Nonopioid analgesicZiconotide0.06Skin ulcer∗Food and Drug Administration summary of product characteristics.COX-2 inhibitorsRofecoxib†New safety signals.0.96GlucocorticoidsPrednisone1.38Impairs wound healing∗Food and Drug Administration summary of product characteristics.SGLT-2 inhibitorsCanagliflozin4.41Ulcers of the lower limbs∗Food and Drug Administration summary of product characteristics.Potassium-channel activatorsNicorandil5.024.85Skin ulceration∗Food and Drug Administration summary of product characteristics.Opiate analgesicsPentazocine1.133.75Ulceration of the skin∗Food and Drug Administration summary of product characteristics.The lower boundary of the Information Component (IC025) in the World Health Organization pharmacovigilance database and in MEDLINE (if significant) is provided for each drug. Warnings about skin ulcers, retrieved from references databases, EMA, and Food and Drug Administration summary product characteristics, are indicated.COX, Cyclooxygenase; SDR, signal of disproportionate reporting; SGLT, sodium-glucose cotransporter; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor; WHO, World Health Organization.∗ Food and Drug Administration summary of product characteristics.† New safety signals.‡ EMA summary of product characteristics.§ Micromedex database.|| Pharmacovigilance Risk Assessment Committee (PRAC) safety signal assessment. Open table in a new tab Four main mechanisms have been proposed as being implicated in the pathogenesis of iatrogenic skin ulcers: angiogenesis inhibition, direct skin toxicity, decreased skin perfusion through vasoconstriction, or decreased skin perfusion through increased blood viscosity.1D'Epiro S. Salvi M. Luzi A. et al.Drug cutaneous side effect: focus on skin ulceration.Clin Ter. 2014; 165: e323-329PubMed Google Scholar Except for ocrelizumab, for which we did not find any pharmacologic hypothesis, all identified drugs could be related to these main mechanisms (Fig 1 and Supplemental Table V).This study has limitations. First, selective reporting of adverse drug reactions in pharmacovigilance databases, inherent to the spontaneous nature of notifications, may distort the associations found in this study.5Bate A. Evans S.J.W. Quantitative signal detection using spontaneous ADR reporting.Pharmacoepidemiol Drug Saf. 2009; 18: 427-436Crossref PubMed Scopus (430) Google Scholar In addition, the development of a skin ulcer is often multifactorial, and clear causality with drug exposure might be difficult to assess, given the high rate of missing clinical data.To the best of our knowledge, this is the first pharmacovigilance study identifying iatrogenic etiologies of skin ulcers. The identified signals represent plausible adverse drug reactions by considering the understanding of iatrogenic skin ulcers, biological and pharmacologic drug activity, and reported case characteristics. However, they have yet to be evaluated by purpose-designed epidemiologic studies. To the Editor: Skin ulcerations have occasionally been reported as adverse drug reactions.1D'Epiro S. Salvi M. Luzi A. et al.Drug cutaneous side effect: focus on skin ulceration.Clin Ter. 2014; 165: e323-329PubMed Google Scholar However, to our knowledge, no pharmacovigilance study has been performed, and little is known about drug-related skin ulcers. We therefore performed combined disproportionality analyses using data from VigiBase, the World Health Organization's pharmacovigilance database, and from the biomedical literature through MEDLINE, to identify pharmacologic classes suspected of inducing skin ulcers.2Bate A. Lindquist M. Edwards I.R. et al.A Bayesian neural network method for adverse drug reaction signal generation.Eur J Clin Pharmacol. 1998; 54: 315-321Crossref PubMed Scopus (665) Google Scholar,3Xu R. Wang Q. Large-scale combining signals from both biomedical literature and the FDA Adverse Event Reporting System (FAERS) to improve post-marketing drug safety signal detection.BMC Bioinformatics. 2014; 15: 17Crossref PubMed Scopus (60) Google Scholar Then, all disproportionality signals were independently reviewed by a committee of experts in pharmacology and dermatology, and selected according to their clinical relevance, the strength of evidence, and previous knowledge of possible associations (Supplemental methods available via Mendeley at 10.17632/sgnzg8k9tt.1).4European Medicines Agency (EMA). Guideline on good pharmacovigilance practices (GVP) - Module IX – Signal management (Rev 1). Accessed April 26, 2021. Available at: https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-good-pharmacovigilance-practices-gvp-module-ix-signal-management-rev-1_en.pdfGoogle Scholar From the 21,421,352 cases available at extraction (March 3, 2020), we identified 19,887 reports of “skin ulcer,” 4440 of “decubitus ulcer,” 596 of “diabetic ulcer,” 103 of “varicose ulceration,” and 79 of “ischemic skin ulcer.” Sixty-one drugs displayed a disproportionality signal for both “skin ulcer” and at least 1 other specific skin ulcer term (Supplemental Fig 1). Among 814 records of chemically induced skin ulcers identified in the literature after screening of titles and abstracts, 241 reports regarding 59 drugs were selected, among which 10 had a disproportionality signal. Among the 61 drugs selected for signal validation, 22 were excluded for obvious indication or protopathic bias, or were combined substances (Supplemental Table I). Then, after retrieval of references databases, Food and Drug Administration and European Medicines Agency labels, and safety alerts, 21 drugs were considered as “already known” and the plausibility of the remaining 19 drugs was assessed through discussion during expert meetings based on cases analyses, potential bias, and pharmacologic mechanisms (Supplemental Tables II and III). Finally, 13 drugs were considered as having potential signals for skin ulcer: 3 protein kinase inhibitors (afatinib, ibrutinib, and tofacitinib), 2 immunomodulators (thalidomide and lenalidomide), 2 interferons (alfa and beta), 1 bisphosphonate (alendronic acid), and leflunomide, rofecoxib, mitomycin, ocrelizumab, and digoxin (Table I and Supplemental Table IV). The lower boundary of the Information Component (IC025) in the World Health Organization pharmacovigilance database and in MEDLINE (if significant) is provided for each drug. Warnings about skin ulcers, retrieved from references databases, EMA, and Food and Drug Administration summary product characteristics, are indicated. COX, Cyclooxygenase; SDR, signal of disproportionate reporting; SGLT, sodium-glucose cotransporter; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor; WHO, World Health Organization. Four main mechanisms have been proposed as being implicated in the pathogenesis of iatrogenic skin ulcers: angiogenesis inhibition, direct skin toxicity, decreased skin perfusion through vasoconstriction, or decreased skin perfusion through increased blood viscosity.1D'Epiro S. Salvi M. Luzi A. et al.Drug cutaneous side effect: focus on skin ulceration.Clin Ter. 2014; 165: e323-329PubMed Google Scholar Except for ocrelizumab, for which we did not find any pharmacologic hypothesis, all identified drugs could be related to these main mechanisms (Fig 1 and Supplemental Table V). This study has limitations. First, selective reporting of adverse drug reactions in pharmacovigilance databases, inherent to the spontaneous nature of notifications, may distort the associations found in this study.5Bate A. Evans S.J.W. Quantitative signal detection using spontaneous ADR reporting.Pharmacoepidemiol Drug Saf. 2009; 18: 427-436Crossref PubMed Scopus (430) Google Scholar In addition, the development of a skin ulcer is often multifactorial, and clear causality with drug exposure might be difficult to assess, given the high rate of missing clinical data. To the best of our knowledge, this is the first pharmacovigilance study identifying iatrogenic etiologies of skin ulcers. The identified signals represent plausible adverse drug reactions by considering the understanding of iatrogenic skin ulcers, biological and pharmacologic drug activity, and reported case characteristics. However, they have yet to be evaluated by purpose-designed epidemiologic studies. We thank VigiBase for giving us access to the data. We also thank Alison Foote (Grenoble Alpes University Hospital) for critically editing our manuscript.