Prognostic significance of cutaneous immune-related adverse events in patients with melanoma and other cancers on immune checkpoint inhibitors

Abstract

To the Editor: Cutaneous immune-related adverse events (cirAEs) are heterogenous toxicities that impact nearly one-third of patients on immune checkpoint inhibitor (ICI) monotherapy.1Geisler A.N. Phillips G.S. Barrios D.M. et al.Immune checkpoint inhibitor-related dermatologic adverse events.J Am Acad Dermatol. 2020; 83: 1255-1268Abstract Full Text Full Text PDF PubMed Scopus (85) Google Scholar Although previous works suggest that cirAEs may correlate with improved survival among patients with melanoma, the prognostic significance of specific cirAE morphologies, particularly in populations without melanoma, remains unclear.2Nakano E. Takahashi A. Namikawa K. et al.Correlation between cutaneous adverse events and prognosis in patients with melanoma treated with nivolumab: a single institutional retrospective study.J Dermatol. 2020; 47: 622-628Crossref PubMed Scopus (4) Google Scholar, 3Min Lee C.K. Li S. Tran D.C. et al.Characterization of dermatitis after PD-1/PD-L1 inhibitor therapy and association with multiple oncologic outcomes: a retrospective case-control study.J Am Acad Dermatol. 2018; 79: 1047-1052Abstract Full Text Full Text PDF PubMed Scopus (66) Google Scholar, 4Haratani K. Hayashi H. Chiba Y. et al.Association of immune-related adverse events with nivolumab efficacy in non-small-cell lung cancer.JAMA Oncol. 2018; 4: 374-378Crossref PubMed Scopus (534) Google Scholar We therefore sought to investigate the relationships between initial cirAE morphologies and survival outcomes among patients with both melanoma and other nonmelanoma cancer diagnoses. In this retrospective cohort study, we screened patients who were started on ICIs at our institution between January 1, 2016 and March 8, 2019 (n = 2459) for possible cirAEs, manually confirming cirAE status as previously defined (Supplemental Material 1 available via Mendeley at https://doi.org/10.17632/bjntzjsdxb.1).5Phillips G.S. Wu J. Hellmann M.D. et al.Treatment outcomes of immune-related cutaneous adverse events.J Clin Oncol. 2019; 37: 2746-2758Crossref PubMed Scopus (100) Google Scholar We collated demographics, oncologic history, and cirAE features (Table I, Supplemental Material 1).5Phillips G.S. Wu J. Hellmann M.D. et al.Treatment outcomes of immune-related cutaneous adverse events.J Clin Oncol. 2019; 37: 2746-2758Crossref PubMed Scopus (100) Google Scholar Time-varying Cox proportional hazard (CPH) models were used to assess relationships between the initial cirAE subtype and progression-free survival (PFS), with consideration of overall survival (OS) as a secondary outcome. CPH models were adjusted for age, sex, and other covariates with P < .05.Table IDemographic and clinical featuresCharacteristicMelanoma cancer (N = 150)Other nonmelanoma cancer (N = 208)P value#P values reflect Fischer exact or Mann-Whitney tests.Age, years—median (IQR)∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.64 (54-72)66 (56-74).010∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.Female sex, no. (%)92 (44.2)53 (35.3).102Cancer type, no. (%)∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation. Melanoma150 (100.0)0 (0.0)- Thoracic0 (0.0)59 (16.5)- GI0 (0.0)34 (9.5)- Head or neck0 (0.0)32 (8.9)- Other‡Other cancer types included genitourinary (29), hematologic (11), breast (11), ovarian (10), nonmelanoma skin cancer (10), Merkel cell carcinoma (8), central nervous system (5), thyroid (4), neuroendocrine (3), adrenal (4), sarcoma (1), and cervical (1) cancer.0 (0.0)83 (23.2)-Cancer stage – median (IQR)∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.4 (4-4)4 (4-4)<.001∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.Cancer metastases, no. (%)∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.33 (22.0)34 (16.8).289ECOG-PS—median (IQR)∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.0 (0-1)1 (0-1)<.001∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.No. of pre-ICI treatments—median (IQR)§Pre-ICI treatments included (melanoma no. (%), other cancer no. (%): radiation (n = 24 (16.0%), n = 117 (56.3%), systemic chemotherapy (n = 4 (2.7%), n = 146 (70.2%), targeted therapy (n = 12 (8.0%), n = 55 (26.4%), and non-ICI immunotherapy (n = 14 (9.3%), n = 5 (2.4%).0 (0-1)2 (1-2)<.001∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.ICI regimen precipitating cirAE<.001∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation. PD-1/PDL-1101 (67.3)174 (83.7)- CTLA-414 (9.3)0 (0.0)- PD-1/PDL-1 + CTLA-435 (23.3)34 (16.4)-Concomitant antineoplastic therapyTypes of concomitant systemic antineoplastic therapy included (melanoma no., other cancer no.): targeted therapy (n = 15, n = 30), traditional chemotherapy (n = 1, n = 9), and non-ICI immunotherapy (i.e. interleukin-2) (n = 1, n = 7).17 (11.3)46 (22.1).011Noncutaneous irAE—median (IQR)1 (1-2)0 (0-1)<.001∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.Time to initial cirAE, days—median (IQR)32 (12-82)44 (19-140).816Types of initial cirAE, no. (%).023∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation. Macular and papular80 (53.3)78 (37.5)- Eczematous7 (4.7)18 (8.7)- Lichenoid6 (4.0)8 (3.9)- Isolated pruritus‖The term “isolated pruritus” denotes cases of pruritus without visible cutaneous manifestations, emerging after ICI initiation, or attributed to ICI by the treating clinician.29 (19.3)37 (17.8)- Psoriasis3 (2.0)9 (4.3)- Vitiligo5 (3.3)4 (1.9)- Morbilliform10 (6.7)22 (10.6)- Other¶Other cirAE types included (melanoma no., other cancer no.): mucositis (n = 1, n = 8), bullous pemphigoid (n = 2, n = 5), erythema multiforme (n = 1, n = 4), acneiform reaction (n = 3, n = 2), urticarial reaction (n = 0, n = 4), macular erythema (n = 1, n = 3), follicular reaction (n = 0, n = 3), panniculitis (n = 1, n = 1), drug rash with eosinophilia and systemic symptoms (n = 1, n = 0), Sweet's syndrome (n = 0, n = 1), and Steven-Johnson Syndrome (n = 0, n = 1).10 (6.7)32 (15.4)-CTCAE v5.0 severity of initial cirAE, median (IQR)1 (1-2)1 (1-2).178ICI disruption due to initial cirAEICI disruptions due to cirAE included (melanoma no., other cancer no.): discontinuation (n = 3, n = 16), temporary suspension (n = 6, n = 17), and change in ICI class (n = 0, n = 1).9 (6.0)34 (16.4).003∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.SCS treatment for initial cirAE13 (8.7)37 (17.8).014∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.Multiple cirAE, no. (%)56 (37.3)26 (7.3)<.001∗P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.CTCAE v5.0, Common terminology criteria for adverse events version 5.0; cirAE, cutaneous immune-related adverse event; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; GI, gastrointestinal; ICI, immune checkpoint inhibitor; irAE, immune-related adverse event; IQR, interquartile range; NOS, not otherwise specified; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; SCS, systemic corticosteroid.∗ P values with an asterisk were significant with two-sided P value < .05. †Data collected for date of ICI initiation.‡ Other cancer types included genitourinary (29), hematologic (11), breast (11), ovarian (10), nonmelanoma skin cancer (10), Merkel cell carcinoma (8), central nervous system (5), thyroid (4), neuroendocrine (3), adrenal (4), sarcoma (1), and cervical (1) cancer.§ Pre-ICI treatments included (melanoma no. (%), other cancer no. (%): radiation (n = 24 (16.0%), n = 117 (56.3%), systemic chemotherapy (n = 4 (2.7%), n = 146 (70.2%), targeted therapy (n = 12 (8.0%), n = 55 (26.4%), and non-ICI immunotherapy (n = 14 (9.3%), n = 5 (2.4%).‖ The term “isolated pruritus” denotes cases of pruritus without visible cutaneous manifestations, emerging after ICI initiation, or attributed to ICI by the treating clinician.¶ Other cirAE types included (melanoma no., other cancer no.): mucositis (n = 1, n = 8), bullous pemphigoid (n = 2, n = 5), erythema multiforme (n = 1, n = 4), acneiform reaction (n = 3, n = 2), urticarial reaction (n = 0, n = 4), macular erythema (n = 1, n = 3), follicular reaction (n = 0, n = 3), panniculitis (n = 1, n = 1), drug rash with eosinophilia and systemic symptoms (n = 1, n = 0), Sweet's syndrome (n = 0, n = 1), and Steven-Johnson Syndrome (n = 0, n = 1).# P values reflect Fischer exact or Mann-Whitney tests.∗∗ ICI disruptions due to cirAE included (melanoma no., other cancer no.): discontinuation (n = 3, n = 16), temporary suspension (n = 6, n = 17), and change in ICI class (n = 0, n = 1).†† Types of concomitant systemic antineoplastic therapy included (melanoma no., other cancer no.): targeted therapy (n = 15, n = 30), traditional chemotherapy (n = 1, n = 9), and non-ICI immunotherapy (i.e. interleukin-2) (n = 1, n = 7). Open table in a new tab CTCAE v5.0, Common terminology criteria for adverse events version 5.0; cirAE, cutaneous immune-related adverse event; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; GI, gastrointestinal; ICI, immune checkpoint inhibitor; irAE, immune-related adverse event; IQR, interquartile range; NOS, not otherwise specified; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; SCS, systemic corticosteroid. CirAEs developed in 358 patients (median age, 64 years; 40.5% female), including 150 (41.9%) with melanoma and 208 (58.1%) with other cancer types. The most common types of other cancers were thoracic (n = 59, 16.5%), gastrointestinal (n = 34, 9.5%), and genitourinary (n = 39, 8.1%) cancers. Across all patients, a “macular and papular” reaction was the most common initial cirAE (melanoma: n = 87, 58.0%, other cancer: n = 93, 44.7%), followed by isolated pruritus (melanoma: n = 29, 19.3%, other cancer: n = 37, 17.8%). The initial development of a lichenoid cirAE was associated with improved PFS across cancer types (HR, 0.323; 95% CI, 0.154-0.674; P = .003) and among the subset of patients with other cancers (HR, 0.270; 95% CI, 0.120-0.607; P = .002, Table II). The initial development of a lichenoid or psoriasiform cirAE was also found to be significantly associated with OS, both overall and among the subgroup of other cancers. An initial eczematous cirAE finding trended toward an association with PFS, both overall (HR, 0.490; 95% CI, 0.218-1.099; P = .084) and among the subgroup of other cancers (HR, 0.397; 95% CI, 0.157-1.005; P = .051), but these relationships were insignificant.Table IIRelationship between cancer type, cirAE subtype and progression-free and overall survivalHR95% CIP value‡The time-varying Cox models examining associations between cirAE type and PFS across cancer types and in the other cancer subgroup each incorporated age, sex, cirAE severity, cancer stage, cancer type, and ECOG-PS at ICI start as covariates. cirAE severity and cancer type were significant with P < .05 in both models. Age was significant at P < .05 in other cancer analysis only.,§The time-varying Cox models examining associations between cirAE type and OS across cancer types and in the other cancer subgroup each incorporated age, sex, cirAE severity, cancer stage, cancer type, and ECOG-PS at ICI start as covariates. cirAE severity and cancer type were significant at P < .05 for both models.All cancer types cirAE subtype – progression-free survivalMacular and papularReferenceReferenceReferenceEczematous0.4900.218-1.099.084Lichenoid0.3230.154-0.674.003∗Indicates that the P values are significantIsolated pruritus1.1580.754-1.778.504Psoriasis0.4320.116-1.608.211Vitiligo0.6520.184-2.315.509Morbilliform0.6530.360-1.183.160Other0.8600.494-1.494.592 cirAE subtype—overall survivalMacular and papularReferenceReferenceReferenceEczematous0.5310.237-1.191.125Lichenoid0.1730.038-0.793.024∗Indicates that the P values are significantIsolated pruritus0.9880.633-1.541.958Psoriasis0.2300.073-0.728.012∗Indicates that the P values are significantVitiligo0.6450.137-3.055.583Morbilliform0.5660.311-1.023.061Other1.3450.794-2.274.272Other nonmelanoma cancer cirAE subtype—progression-free survivalMacular and papularReferenceReferenceReferenceEczematous0.3970.157-1.005.051Lichenoid0.2700.120-0.607.002∗Indicates that the P values are significantIsolated pruritus1.25900.721-2.307.391Psoriasis0.2940.065-1.323.111Morbilliform0.6110.296-1.261.183Other†Because the number of progression events and cases of vitiligo were relatively small, cases of vitiligo were condensed into the “other” cirAE category for the time-varying Cox model examining the association between cirAE subtypes and progression-free survival in other cancers.0.5390.259-1.122.099 cirAE subtype—overall survivalMacular and papularReferenceReferenceReferenceEczematous0.4600.191-1.108.083Lichenoid0.1830.038-0.883.034∗Indicates that the P values are significantIsolated pruritus1.0370.625-1.720.889Psoriasis0.1570.045-0.546.004∗Indicates that the P values are significantVitiligo0.4040.041-4.000.438Morbilliform0.5560.295-1.056.075Other1.1710.640-2.111.599CI, Confidence interval; cirAE, cutaneous immune-related adverse event; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; HR, hazards ratio; IQR, interquartile range; NOS, not otherwise specified.∗ Indicates that the P values are significant† Because the number of progression events and cases of vitiligo were relatively small, cases of vitiligo were condensed into the “other” cirAE category for the time-varying Cox model examining the association between cirAE subtypes and progression-free survival in other cancers.‡ The time-varying Cox models examining associations between cirAE type and PFS across cancer types and in the other cancer subgroup each incorporated age, sex, cirAE severity, cancer stage, cancer type, and ECOG-PS at ICI start as covariates. cirAE severity and cancer type were significant with P < .05 in both models. Age was significant at P < .05 in other cancer analysis only.§ The time-varying Cox models examining associations between cirAE type and OS across cancer types and in the other cancer subgroup each incorporated age, sex, cirAE severity, cancer stage, cancer type, and ECOG-PS at ICI start as covariates. cirAE severity and cancer type were significant at P < .05 for both models. Open table in a new tab CI, Confidence interval; cirAE, cutaneous immune-related adverse event; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; HR, hazards ratio; IQR, interquartile range; NOS, not otherwise specified. In this study, we characterized the associations between initial cirAE subtypes and survival outcomes, with stratification by cancer type. Across different cancer types and among patients with nonmelanoma cancers, we observed relationships between lichenoid cirAEs and PFS and between lichenoid and psoriasiform cirAEs and OS. Interestingly, broader and more common cirAE morphologies, such as macular and papular reactions and isolated pruritus, were not meaningfully associated with survival in any patient group.5Phillips G.S. Wu J. Hellmann M.D. et al.Treatment outcomes of immune-related cutaneous adverse events.J Clin Oncol. 2019; 37: 2746-2758Crossref PubMed Scopus (100) Google Scholar Vitiligo was also not associated with survival across cancer types, though this may have been due to the limited number of patients presenting with this late-appearing toxicity as their first cirAE. Collectively, our findings suggest that specific cirAE morphologies may hold prognostic significance, highlighting the importance of precise cirAE evaluation in the management of patients with melanoma and nonmelanoma cancer types receiving ICIs. The limitations of this study include its retrospective single-center design, reliance on information documented in electronic medical records, and small sample size, limiting the generalizability of our findings. Despite these limitations, our findings highlight the vital role that dermatologists can play in precisely diagnosing cirAEs to improve the collaborative management of ICI patients. None disclosed.