Cutaneous adverse events to immune checkpoint inhibitors in pediatric populations: A retrospective cohort study

Abstract

To the Editor: Cutaneous adverse events (cAEs) represent some of the most significant effects of immune checkpoint inhibitor (ICI) therapy, with evidence suggesting associations between cAE emergence and improved survival.1Coleman E. Ko C. Dai F. Tomayko M.M. Kluger H. Leventhal J.S. Inflammatory eruptions associated with immune checkpoint inhibitor therapy: a single-institution retrospective analysis with stratification of reactions by toxicity and implications for management.J Am Acad Dermatol. 2019; 80: 990-997Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar,2Phillips 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 (92) Google Scholar Although these toxicities are well characterized among adults, the patterns and prognostic significance among children remain unclear.1Coleman E. Ko C. Dai F. Tomayko M.M. Kluger H. Leventhal J.S. Inflammatory eruptions associated with immune checkpoint inhibitor therapy: a single-institution retrospective analysis with stratification of reactions by toxicity and implications for management.J Am Acad Dermatol. 2019; 80: 990-997Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar, 2Phillips 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 (92) Google Scholar, 3Moreno-Vicente J. Beers S.A. Gray J.C. PD-1/PD-L1 blockade in paediatric cancers: what does the future hold?.Cancer Lett. 2019; 457: 74-85Crossref PubMed Scopus (9) Google Scholar Given differences in cancer types and immune responses between adults and children, cAEs to ICI therapies may differ by age as well.3Moreno-Vicente J. Beers S.A. Gray J.C. PD-1/PD-L1 blockade in paediatric cancers: what does the future hold?.Cancer Lett. 2019; 457: 74-85Crossref PubMed Scopus (9) Google Scholar We addressed this gap by characterizing cAE patterns and assessing associations with survival outcomes in a 2-institution pediatric cohort. We retrospectively reviewed medical records of patients aged 0 to 21 years receiving ICIs at Massachusetts General and Boston Children's Hospitals between 2010 and 2019. We identified individuals with cAEs, defined as reactions emerging after ICI initiation, consistent with established descriptive categories and attributed to the ICI by the treating clinician.1Coleman E. Ko C. Dai F. Tomayko M.M. Kluger H. Leventhal J.S. Inflammatory eruptions associated with immune checkpoint inhibitor therapy: a single-institution retrospective analysis with stratification of reactions by toxicity and implications for management.J Am Acad Dermatol. 2019; 80: 990-997Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar,2Phillips 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 (92) Google Scholar We abstracted demographics, oncologic history, and cAE features, assessing differences by cAE status using Fisher exact and Mann-Whitney tests. We assessed the relationship between cAE emergence and progression-free survival by using a time-varying multivariate Cox regression, incorporating age, sex, cancer type, pre-ICI therapies, and cAE status. Overall survival was considered as a secondary outcome. Seventy-six patients (median age, 16 years; 42% female) received ICIs between 2010 and 2019, most commonly for central nervous system neoplasms (n = 26) (Tables I and II). A subanalysis of patients in this cohort treated for sarcomas was published elsewhere.4Thompson L.L. Chang M.S. McCormack L. et al.Patterns of cutaneous immune-related adverse events in adults and children with advanced sarcoma: a retrospective cohort study.Br J Dermatol. September 7, 2020; Google Scholar ICIs included programmed death-1 (PD-1)/programmed death-ligand 1 (PDL-1) (n = 53) and cytotoxic T-lymphocyte antigen-4 (CTLA-4) (n = 4) monotherapy as well as combined CTLA-4/PD-1 regimens (n = 19) (Supplementary Material, available on Mendeley at https://doi.org/10.17632/k4k4mxwrfp.1).Table ICharacteristics of pediatric patients receiving ICI therapy (N = 76)CharacteristicscAE (n = 24)No cAE (n = 52)P value∗P values for cancer features (type, stage, grade, pre-ICI treatments) reflect the statistical outputs of 2-sided Fisher exact tests (categorical variables) or Mann-Whitney tests (continuous variables). P values for survival outcomes reflect the statistical outputs of Cox proportional hazards models adjusted for age, sex, cancer type, and treatments before immune checkpoint therapy.Age, y, median (IQR)18 (14-19)16 (11-18)Female sex, n (%)11 (46)21 (40)Race, n (%) White18 (75)35 (67) Asian3 (13)6 (12) African American0 (0)3 (6) Hispanic or Latino0 (0)5 (9) Not reported3 (13)3 (6)Cancer type, n (%).207 Central nervous system neoplasm†For patients with a cAE, central nervous system neoplasms included glioblastoma multiforme (n = 4), glioma (n = 2), meningioma (n = 1), craniopharyngioma (n = 1), ependymoma (n = 1), dysembryoplastic neuroepithelial tumor (n = 1), nongerminomatous germ cell tumor (n = 1), and primary intracranial melanoma (n = 1). For patients without a cAE, central nervous system neoplasms included glioblastoma multiforme (n = 1), neuroblastoma (n = 2), glioma (n = 8), pineoblastoma (n = 2), and atypical teratoid rhabdoid tumor (1).12 (50)14 (27) Hematologic neoplasm‡For patients with a cAE, hematologic neoplasms included Hodgkin lymphoma (n = 1). For patients without a cAE, hematologic neoplasms included Hodgkin lymphoma (n = 4), non-Hodgkin lymphoma (n = 1), and acute myeloid leukemia (n = 1).1 (4)6 (12) Sarcoma§For patients with a cAE, sarcomas involving the bone (n = 2), soft tissues (n = 3), and notochord (n = 2) were present. For patients without a cAE, sarcomas involving the bone (n = 8), soft tissues (n = 6), and notochord (n = 1) were present.7 (29)15 (29) Gastrointestinal neoplasmǁFor patients with a cAE, gastrointestinal neoplasms included hepatocellular carcinoma (n = 2). For patients without a cAE, gastrointestinal neoplasms included hepatocellular carcinoma (n = 8), cholangiocarcinoma (n = 1), and poorly differentiated adenocarcinoma of unknown gastrointestinal primary (n = 1).2 (8)10 (19) Melanoma2 (8)3 (6) Other¶Other cancer types included cutaneous squamous cell carcinoma (n = 2), sinonasal squamous cell carcinoma (n = 1), and Wilms tumor (n = 1).0 (0)4 (6)Cancer stage, median (IQR)#Cancer grade at ICI start was collected for central nervous system neoplasms. Cancer stage at ICI start was collected for all other cancers.IV (III-IV)IV (II-IV).315Cancer grade, median (IQR)#Cancer grade at ICI start was collected for central nervous system neoplasms. Cancer stage at ICI start was collected for all other cancers.IV (III-IV)IV (IV).111Pre-ICI treatments, n (%) Radiation19 (80)30 (58).318 Traditional chemotherapy10 (42)37 (71).022 Targeted agent7 (29)17 (33).798 Non-ICI immunotherapy4 (17)4 (8).236ICI regimen, n (%) CTLA-41 (4)3 (6).831 PD-1 or PD-L116 (67)37 (71) CTLA-4 + PD-L1/PD-17 (29)12 (23)Number of noncutaneous AEs, median (IQR)For patients with a cAE, noncutaneous AEs included enterocolitis (n = 4) thyroiditis (n = 3), hepatotoxicity (n = 3), pneumonitis (n = 2), neutropenia (n = 1), hemolytic anemia (n = 1), and recurrent pyrexia (n = 1). For those without a cAE, noncutaneous AEs included enterocolitis (n = 2), hepatotoxicity (n = 2), pneumonitis (n = 1), nephritis (n = 1), and pancreatitis (n = 1)..002 None13 (54)43 (90) 1 or more11 (46)5 (10)AE, Adverse event; cAE, cutaneous adverse event; ICI, immune checkpoint inhibitor; IQR, interquartile range.∗ P values for cancer features (type, stage, grade, pre-ICI treatments) reflect the statistical outputs of 2-sided Fisher exact tests (categorical variables) or Mann-Whitney tests (continuous variables). P values for survival outcomes reflect the statistical outputs of Cox proportional hazards models adjusted for age, sex, cancer type, and treatments before immune checkpoint therapy.† For patients with a cAE, central nervous system neoplasms included glioblastoma multiforme (n = 4), glioma (n = 2), meningioma (n = 1), craniopharyngioma (n = 1), ependymoma (n = 1), dysembryoplastic neuroepithelial tumor (n = 1), nongerminomatous germ cell tumor (n = 1), and primary intracranial melanoma (n = 1). For patients without a cAE, central nervous system neoplasms included glioblastoma multiforme (n = 1), neuroblastoma (n = 2), glioma (n = 8), pineoblastoma (n = 2), and atypical teratoid rhabdoid tumor (1).‡ For patients with a cAE, hematologic neoplasms included Hodgkin lymphoma (n = 1). For patients without a cAE, hematologic neoplasms included Hodgkin lymphoma (n = 4), non-Hodgkin lymphoma (n = 1), and acute myeloid leukemia (n = 1).§ For patients with a cAE, sarcomas involving the bone (n = 2), soft tissues (n = 3), and notochord (n = 2) were present. For patients without a cAE, sarcomas involving the bone (n = 8), soft tissues (n = 6), and notochord (n = 1) were present.ǁ For patients with a cAE, gastrointestinal neoplasms included hepatocellular carcinoma (n = 2). For patients without a cAE, gastrointestinal neoplasms included hepatocellular carcinoma (n = 8), cholangiocarcinoma (n = 1), and poorly differentiated adenocarcinoma of unknown gastrointestinal primary (n = 1).¶ Other cancer types included cutaneous squamous cell carcinoma (n = 2), sinonasal squamous cell carcinoma (n = 1), and Wilms tumor (n = 1).# Cancer grade at ICI start was collected for central nervous system neoplasms. Cancer stage at ICI start was collected for all other cancers.∗∗ For patients with a cAE, noncutaneous AEs included enterocolitis (n = 4) thyroiditis (n = 3), hepatotoxicity (n = 3), pneumonitis (n = 2), neutropenia (n = 1), hemolytic anemia (n = 1), and recurrent pyrexia (n = 1). For those without a cAE, noncutaneous AEs included enterocolitis (n = 2), hepatotoxicity (n = 2), pneumonitis (n = 1), nephritis (n = 1), and pancreatitis (n = 1). Open table in a new tab Table IIFeatures of cutaneous adverse events (n = 24)CharacteristicsValueFeatures of cAEs Time to cAE, d, median (IQR)53 (16-176) Duration of cAE, d, median (IQR)27 (12-133) CTCAE v5.0 severity grade, median (IQR)1 (1-2)Clinical diagnosis of cAE, n (%) Maculopapular eruption∗This included 5 cases of “rash” with limited descriptive details most consistent with a maculopapular eruption.12 (50) Isolated xerosis3 (13) Pruritus without visible rash2 (8) Acneiform eruption2 (8) Vitiligo2 (8) Other†“Other” cAE types included psoriasiform eruption (n = 1), urticarial eruption (n = 1), and mucositis (n = 1).3 (13)Referral to dermatology department for cAE, n (%)‡Two of 13 grade 1 cAEs, 2 of 9 grade 2 cAEs, and 1 of 2 of grade 3 cAEs prompted referral to the dermatology department.5 (21)Biopsy of cAE, n (%)2 (8)Most potent treatment for cAE, n (%) None or emollient only§Eleven of 14 cases of no treatment and 3 of 14 cases with emollient treatment only.14 (58) Oral antipruritic only3 (13) Topical corticosteroid§Eleven of 14 cases of no treatment and 3 of 14 cases with emollient treatment only.,ǁIn several cases, patients received other therapies alongside a topical corticosteroid: 1 case of concomitant emollient, 1 case of concomitant oral antibiotic, 1 case of concomitant oral antipruritic, and 1 case of concomitant oral acitretin.7 (29)Treatment response, n (%)¶Consistent with prior work, the following definitions were used to assess cAE response to treatment: significant improvement, improvement by 2 CTCAE grades or resolution; moderate improvement, improvement by 1 grade; no improvement, no change or increase in grade. Nonapplicable, no treatment provided11 (46) Treated with significant improvement11 (46) Treated with moderate improvement1 (4) Treated with no improvement1 (4)ICI changes due to cAE, n (%)#The cAE prompting ICI discontinuation was a grade 2 psoriasiform reaction that was treated with a high-potency topical steroid. The cAE prompting temporary ICI suspension was a grade 2 maculopapular reaction that was treated with an oral antihistamine. ICI not changed due to cAE22 (92) ICI suspended, then restarted due to cAE1 (4) ICI discontinued due to cAE1 (4)cAE, Cutaneous adverse event; CTCAE v5.0, Common Terminology Criteria of Adverse Events version 5.0; CTLA-4, cytotoxic T-lymphocyte antigen-4; ICI, immune checkpoint inhibitor; IQR, interquartile range; PD-1, programmed death-1; PDL-1, programmed death-ligand 1.∗ This included 5 cases of “rash” with limited descriptive details most consistent with a maculopapular eruption.† “Other” cAE types included psoriasiform eruption (n = 1), urticarial eruption (n = 1), and mucositis (n = 1).‡ Two of 13 grade 1 cAEs, 2 of 9 grade 2 cAEs, and 1 of 2 of grade 3 cAEs prompted referral to the dermatology department.§ Eleven of 14 cases of no treatment and 3 of 14 cases with emollient treatment only.ǁ In several cases, patients received other therapies alongside a topical corticosteroid: 1 case of concomitant emollient, 1 case of concomitant oral antibiotic, 1 case of concomitant oral antipruritic, and 1 case of concomitant oral acitretin.¶ Consistent with prior work, the following definitions were used to assess cAE response to treatment: significant improvement, improvement by 2 CTCAE grades or resolution; moderate improvement, improvement by 1 grade; no improvement, no change or increase in grade.# The cAE prompting ICI discontinuation was a grade 2 psoriasiform reaction that was treated with a high-potency topical steroid. The cAE prompting temporary ICI suspension was a grade 2 maculopapular reaction that was treated with an oral antihistamine. Open table in a new tab AE, Adverse event; cAE, cutaneous adverse event; ICI, immune checkpoint inhibitor; IQR, interquartile range. cAE, Cutaneous adverse event; CTCAE v5.0, Common Terminology Criteria of Adverse Events version 5.0; CTLA-4, cytotoxic T-lymphocyte antigen-4; ICI, immune checkpoint inhibitor; IQR, interquartile range; PD-1, programmed death-1; PDL-1, programmed death-ligand 1. A total of 24 (32%) individuals developed a cAE, and 52 (68%) did not. cAE clinical descriptions varied but commonly included maculopapular reaction (n = 12; 50%), isolated pruritus (n = 2; 8%), and vitiligo (n = 2; 8%). Few cAEs were referred to the dermatology department (n = 5; 21%) or biopsied (n = 2; 8%). cAEs were untreated in 46% (n = 11) of cases. Individuals with and without cAEs had similar progression free survival (hazard ratio, 0.887; 95% confidence interval CI, 0.452-1.741; P = .728) and overall survival (hazard ratio, 0.817; 95% CI, 0.361-1.849). In this study, we characterized cAE features and associations with survival in a 2-institution pediatric cohort. Although previous trials reported cAEs in 3% of children, we observed cAEs in 30% of our cohort, suggesting possible underdiagnosis.2Phillips 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 (92) Google Scholar,3Moreno-Vicente J. Beers S.A. Gray J.C. PD-1/PD-L1 blockade in paediatric cancers: what does the future hold?.Cancer Lett. 2019; 457: 74-85Crossref PubMed Scopus (9) Google Scholar Clinical features were similar to those among adults, with a high proportion of “maculopapular” reactions.1Coleman E. Ko C. Dai F. Tomayko M.M. Kluger H. Leventhal J.S. Inflammatory eruptions associated with immune checkpoint inhibitor therapy: a single-institution retrospective analysis with stratification of reactions by toxicity and implications for management.J Am Acad Dermatol. 2019; 80: 990-997Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar,2Phillips 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 (92) Google Scholar,5Sibaud V. Dermatologic reactions to immune checkpoint inhibitors.Am J Clin Dermatol. 2018; 19: 345-361Crossref PubMed Scopus (262) Google Scholar The “maculopapular” descriptor may conflate different cAE subtypes, highlighting the need for enhanced clinical categorization.5Sibaud V. Dermatologic reactions to immune checkpoint inhibitors.Am J Clin Dermatol. 2018; 19: 345-361Crossref PubMed Scopus (262) Google Scholar We observed infrequent treatment for cAEs, warranting further investigation. Unlike cAEs among adults, cAEs were not associated with survival in our pediatric cohort.2Phillips 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 (92) Google Scholar Differences in predominant cancer type across age groups may partially account for this. Sample size may have limited our power to detect differences. Limitations include small sample size and cAE characterizations provided by nondermatologists. Although our retrospective design may have contributed to reporting and ascertainment biases, we found that cAEs to ICIs are common, although infrequently evaluated by dermatologists, in children. Future studies better characterizing cutaneous reactions and responses to therapy could help shape guidelines in the pediatric population. None disclosed.