irAEs indicates immune-related adverse events.
eTable 1. Patient and disease characteristics
eTable 2. Acute irAEs arising on treatment
eTable 3. Chronic irAEs
eTable 4. Chronic irAEs in patients that did not require additional systemic therapy and had evaluable follow-up >6 months
Customize your JAMA Network experience by selecting one or more topics from the list below.
Patrinely JR, Johnson R, Lawless AR, et al. Chronic Immune-Related Adverse Events Following Adjuvant Anti–PD-1 Therapy for High-risk Resected Melanoma. JAMA Oncol. 2021;7(5):744–748. doi:10.1001/jamaoncol.2021.0051
What is the incidence, time course, and spectrum of chronic immune-related adverse events (irAEs) arising from adjuvant treatment with anti–programmed cell death 1 (anti–PD-1) for advanced melanoma?
In this multicenter cohort study of 387 patients with stage III to IV melanomas, 43% of patients treated with anti–PD-1 developed a chronic irAE, with only 14% resolving at last follow-up. Chronic irAEs most commonly affected nonvisceral organs (endocrine, joints, salivary glands, eye, and peripheral nerves).
Chronic irAEs associated with anti–PD-1 therapy are more common than previously recognized and frequently persist even with prolonged follow-up.
Agents targeting programmed cell death 1 (PD-1)/PD ligand 1 (PD-L1) improve long-term survival across many advanced cancers and are now used as adjuvant therapy for resected stage III and IV melanomas. The incidence and spectrum of chronic immune-related adverse events (irAEs) have not been well defined.
To determine the incidence, time course, spectrum, and associations of chronic irAEs arising from adjuvant anti–PD-1 therapy.
Design, Setting, and Participants
This retrospective multicenter cohort study was conducted between 2015 and 2020 across 8 academic medical centers in the United States and Australia. Patients with stage III to IV melanomas treated with anti–PD-1 in the adjuvant setting were included.
Main Outcomes and Measures
Incidence, types, and time course of chronic irAEs (defined as irAEs persisting at least 12 weeks after therapy cessation).
Among 387 patients, the median (range) age was 63 (17-88) years, and 235 (60.7%) were male. Of these patients, 267 (69.0%) had any acute irAE, defined as those arising during treatment with anti–PD-1, including 52 (19.5%) with grades 3 through 5 events; 1 patient each had fatal myocarditis and neurotoxicity. Chronic irAEs, defined as those that persisted beyond 12 weeks of anti–PD-1 discontinuation, developed in 167 (43.2%) patients, of which most (n = 161; 96.4%) were mild (grade 1 or 2) and most persisted until last available follow-up (n = 143; 85.6%). Endocrinopathies (73 of 88; 83.0%), arthritis (22 of 45; 48.9%), xerostomia (9 of 17; 52.9%), neurotoxicities (11 of 15; 73.3%), and ocular events (5 of 8; 62.5%) were particularly likely to become chronic. In contrast, irAEs affecting visceral organs (liver, colon, lungs, kidneys) had much lower rates of becoming chronic irAEs; for example, colitis became chronic in 6 of 44 (13.6%) cases, of which 4 of 6 (66.7%) resolved with prolonged follow-up. Age, gender, time of onset, and need for steroids were not associated with the likelihood of chronicity of irAEs.
Conclusion and Relevance
In this multicenter cohort study, chronic irAEs associated with anti–PD-1 therapy appear to be more common than previously recognized and frequently persisted even with prolonged follow-up, although most were low grade. The risks of chronic irAEs should be integrated into treatment decision-making.
Adjuvant treatment with anti–programmed cell death 1 (anti–PD-1) improves relapse-free survival (RFS) in patients with resected stage III through IV melanomas.1,2 Beyond melanoma, indications for neoadjuvant, adjuvant, and maintenance therapy for anti–PD ligand 1 (PD-L1) are proliferating, also increasing the number of long-term survivors treated with these agents.3 Thus, characterizing long-term toxic effects is paramount and remains a key knowledge gap.
Anti–PD-L1 agents cause acute and chronic immune-related adverse events (irAEs). Most present within the first 12 weeks of therapy and resolve with use of glucocorticoids.4 However, delayed, chronic, or even fatal events may also occur with long-term or permanent ramifications.5-7
Characterizing long-term events is challenging in the metastatic setting given the frequency of subsequent therapies and disease-related complications and deaths. Therefore, studying patients treated with adjuvant anti–PD-1 for melanoma allows more accurate characterization of persistent irAEs. In this large, multicenter cohort study, we identify the spectrum, incidence, and kinetics of chronic irAEs.
After institutional review board approval with waiver of informed consent, retrospective deidentified data were collected from 8 participating institutions in the United States and Australia. Patients receiving 1 or more doses of adjuvant anti–PD-1 (pembrolizumab or nivolumab) were included. Patient demographics (melanoma subtype, comorbidities, genetic variation status, and stage) and treatment details (dose, frequency, duration, and reason for discontinuation) were noted. Acute irAEs were defined as those developing during treatment, as delayed if arising after completion of treatment, and as chronic if extending 12 or more weeks past treatment discontinuation. Patients without 12 weeks of follow-up after treatment discontinuation were excluded. All irAEs were described with grade (Common Terminology Criteria for Adverse Events, version 5.0), time of onset, symptoms, treatment required, and resolution.8
Categorical and continuous variables were analyzed with descriptive statistics. Kaplan-Meier curves assessed survival; groups were compared using log-rank testing. The χ2 and t tests assessed associations of chronic irAEs with anti–PD-1 therapy. A 2-sided P < .05 was considered statistically significant. Analyses were conducted using GraphPad, version 8.2.1 (Prism).
Of 387 patients included in the cohort, the median (range) age was 63 (17-88) years, 235 (60.7%) were male, 290 (74.9%) had preexisting comorbidities (30 [7.8%] with autoimmune conditions), 332 (85.8%) had cutaneous primary sites, 198 (51.2%) had BRAF/NRAS wild-type variants, and 128 (33.1%) had resected stage IIIb or 153 (39.5%) with stage IIIc melanomas (AJCC Cancer Staging Manual, version 8) (eTable 1 in the Supplement). A total of 326 (84.2%) patients received nivolumab and 61 (15.8%) received pembrolizumab monotherapy, with a median (range) treatment duration of 306 (1-1049) days.
Treatment was discontinued for therapy completion (n = 193; 50.0%), irAEs (n = 98; 25.3%), and disease progression (n = 81; 20.9%). Most patients had no disease recurrence (n = 268; 69.3%); 69 (17.8%) and 50 (12.9%) patients had metastatic and regional-only recurrences, respectively. Median overall survival and RFS were not reached, and 359 (93.0%) patients were alive at median follow-up of 529 days (530 days among survivors) (Figure, A and B). Patients with acute or chronic irAEs had superior RFS compared with patients lacking irAEs (Figure, C and D).
Of the 387 included patients, 276 (69.0%) developed acute irAEs, including 171 (44.2%) with grade 2 or higher and 52 (13.4%) with grades 3 through 5. Two (0.5%) patients died of irAEs: 1 with myocarditis and 1 with Guillain-Barré–like syndrome. Among the cohort, acute irAEs included dermatitis/pruritus (n = 100; 25.8%), thyroiditis/hypothyroid (n = 63; 16.3%), arthralgias (n = 41; 10.6%), colitis/diarrhea (n = 38; 9.8%), hepatitis (n = 24; 6.2%), pneumonitis (n = 17; 4.4%), and xerostomia (n = 16; 4.1%), and required glucocorticoids in 109 (28.2%) patients. Of the patients who required glucocorticoids, 60 (55.0%) were for grade 2 irAEs and 49 (45.0%) were for grades 3 through 5 irAEs (Table 1 and eTable 2 in the Supplement).
We assessed how often acute irAEs developed into chronic events (Table 2 and eTable 3 in the Supplement). Patients with endocrinopathies (73 of 88; 83.0%) arthritis (22 of 45; 48.9%), xerostomia (9 of 17; 52.9%), neurotoxicities (8 of 8; 100%), and ocular events (5 of 8; 63.0%) were more likely to experience chronic events compared with patients with colitis (6 of 44; 13.6%), hepatitis (4 of 25; 16.0%), and pneumonitis (6 of 18; 33.3%). Among the full cohort, 167 (43.2%) patients developed chronic irAEs. Only 24 (14.4%) of these resolved during the median 529-day follow-up; the remainder persisted (eFigure 1 in the Supplement). Approximately half of irAEs were symptomatic (n = 82; 49.1%). Systemic glucocorticoids were required in 55 (32.9%) patients, including 12 patients taking replacement-dose steroids for pituitary/adrenal insufficiency.
Most chronic irAEs were grade 1 or 2 (161 of 167; 96.4%). In the full cohort, they were most commonly hypothyroidism (n = 54; 14.0%), arthralgias (n = 22; 5.7%), dermatitis/pruritus (n = 19; 6.6%), adrenal insufficiency (n = 12; 3.1%), and xerostomia (n = 9; 2.3%). However, diverse persistent irAEs affected nearly every organ system (Table 2 and eTable 3 and eFigure 2 in the Supplement).
Age (median, 63.0 vs 63.0 years in patients who did and did not develop chronic irAEs, respectively; P = .67) and gender (97 [41.2%] male vs 70 [46.1%] female; P = .31) were not associated with chronic irAE development. Median time of onset of acute irAEs was not associated with chronicity (84 vs 73 days; P = .95). Excluding endocrinopathies, patients who received glucocorticoids for acute irAEs did not develop chronic irAEs more often than those who were not treated with glucocorticoids (62 of 104 [59.6%] vs 63 of 126 [50.0%]; P = .15).
We then assessed each type of chronic irAE for resolution (median duration of follow-up after treatment discontinuation, 287 days) (Table 2). Endocrinopathies (73 of 73; 100%), arthralgias (22 of 22; 100%), ocular events (5 of 5; 100%), xerostomia (8 of 9; 88.9%), and cutaneous events (17 of 19; 89.5%) remained persistent at last follow-up. In contrast, colitis (2 of 6; 33.3%), neuropathies (1 of 3; 33.3%), and nephritis (1 of 3; 33.3%) had lower rates of prolonged persistence. To rule out confounding contributions, we assessed 295 patients who did not receive additional systemic agents with 6 or more months of follow-up after anti–PD-1 discontinuation. Chronic irAE rates, glucocorticoid requirement, irAE resolution, and rates of grade 2 and higher and 3 to 4 events were similar to the full cohort (eTable 4 in the Supplement).
To our knowledge, this is the first study to systematically examine anti–PD-1–related chronic irAEs in patients with high-risk, resected melanomas. Results demonstrated that nearly half of patients developed a chronic irAE, with most unresolved at last follow-up. Adjuvant anti–PD-1 therapy for previously resected advanced melanoma significantly extends RFS.1,9 The adjuvant patient population presents unique considerations; they may have been cured by surgery alone and have longer or normal life expectancies. Thus persistent, life-altering, or life-threatening irAEs should be characterized in detail and integrated into patient counseling and treatment decision-making.
We observed that grade 3 and higher acute irAEs occurred in 13.4% of patients, which is similar to prior studies.1,9 However, treatment discontinuation secondary to acute irAEs occurred more often (25.8%).1,9 This could be due to a real-world population that included more patients with comorbidities, including autoimmune disease. Patients may have also electively discontinued early for low-grade, bothersome irAEs rather than only for predefined high-grade irAEs. Two (0.5%) patients died of irAEs, similar to our previous estimate of a 0.36% death rate with anti–PD-1 monotherapy.5 These results reinforce the rare but serious nature of these uncommon events.
Importantly, chronic irAEs (lasting >12 weeks beyond treatment discontinuation) occurred in 40% of patients, were largely grade 1 and 2 (96%), and usually persisted to last follow-up (85%). These data are juxtaposed with other studies suggesting that most acute events resolve.10 On closer inspection, however, we observed that most irAEs involving visceral organs did resolve (eg, only 6 of 44 colitis events became chronic, of which 4 ultimately resolved with further follow-up).
Most chronic irAEs occurred in nonvisceral systems.11,12 Endocrinopathies, arthritis, xerostomia, ocular events, and neurotoxicities were likely to evolve into a chronic phenotype. In aggregate, 20% of patients experienced symptomatic chronic irAEs (nonendocrinopathies), including some that required persistent immunosuppression. One perhaps overly simplistic explanation for this differential chronicity may be the small volume of affected nonvisceral organs (eg, hormone-producing endocrine cells, salivary ducts, synovium, eyes, distal peripheral nerves) compared with the larger volume of the colon, lungs, liver, etc. Differential regenerative capacity could also play a role.
There are several limitations to this study, including its retrospective nature. While few chronic irAEs resolved, more follow-up time could show more resolution. For example, several patients with xerostomia experienced slow improvement over months to years. Additionally, we observed that both acute and chronic irAEs were associated with improved RFS, which may be subject to unavoidable biases (eg, no chronic irAEs in patients with early progression and death).13-15
Patients treated with anti–PD-1 therapy increasingly experience long-term survival. To our knowledge, this is the largest characterization of chronic irAEs from anti–PD-1 to date. We conclude that chronic irAEs, while usually low grade, occur more frequently than previously reported and particularly affect nonvisceral organs.1,9
Accepted for Publication: December 3, 2020.
Published Online: March 25, 2021. doi:10.1001/jamaoncol.2021.0051
Correction: This article was corrected on May 20, 2021, to add a missing funder to the Funding/Support section.
Corresponding Author: Douglas B. Johnson, MD, 2220 Pierce Ave, 777 PRB, Nashville, TN 37232 (email@example.com).
Author Contributions: Dr Johnson and Mr Patrinely had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Patrinely, D. Johnson.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Patrinely, Mehnert, D. Johnson.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Patrinely, Palmeri, Fan, Ye, D. Johnson.
Obtained funding: D. Johnson.
Administrative, technical, or material support: Bhave, Palmeri, Davis, Long, Osman, Carlino, Menzies.
Supervision: Rapisuwon, Haydon, D. Johnson.
Conflict of Interest Disclosures: Dr Johnson serves on advisory boards for Array Biopharma, Bristol Myers Squibb, Iovance, Janssen, Merck, OncoSec, and Novartis, and receives research funding from Bristol Myers Squibb and Incyte. Dr Haydon serves on advisory boards for Novartis, Bristol Myers Squibb, MSD, and Pierre Fabre. Dr Carlino has served on advisory boards for Bristol Myers Squibb, MSD, Amgen, Novartis, Pierre Fabre, Roche, Sanofi, Merck, Ideaya Biosciences, Regeneron, Nektar, Eisai, and QBiotics, and receives honoraria from Bristol Myers Squibb, MSD, and Novartis. Dr Sullivan serves on advisory boards for Asana BioSciences, Bristol Myers Squibb, Novartis, Pfizer, Iovance, Eisai, and Merck, and has received research funding from Amgen and Merck. Dr Bhave has received travel support from MSD. Dr Menzies has served on advisory boards for Bristol Myers Squibb, MSD, Novartis, Roche, Pierre Fabre, and QBiotics. Ms Johnson has received honoraria from MSD. Dr Davis receives research funding from BMS, Karyopharm, Five Prime Therapeutics, Genentech, Actuate, Incyte, and Top Alliance Biosciences. Dr Long is a consultant advisor for Array BioPharma, Aduro, Amgen, Boehringer Ingelheim International, Bristol Myers Squibb, Highlight Therapeutics, MassArray, Merck, MSD, Novartis, OncoSec Medical, Pierre Fabre, Regeneron Pharmaceuticals, Roche, QBiotics, SkylineDx, and Sandoz. Dr Mehnert receives personal fees from Regeneron and Bristol Myers Squibb, as well as grants from Incyte and Merck. Dr Rapisuwon receives grants from Bristol Myers Squibb. No other disclosures were reported.
Funding/Support: This study was supported by a Cancer Institute of New South Wales fellowship and the Melanoma Institute of Australia (Dr Menzies); a National Health and Medical Research Council practitioner fellowship and the University of Sydney Medical Foundation (Dr Long); and a National Comprehensive Cancer Network Young Investigators Award, an American Cancer Society Institutional Research Grant, and grants from the National Institutes of Health (K23CA204726 and R01CA227481; Dr Johnson). This study was also supported by a grant from the New York University’s Specialized Program of Research Excellence in Melanoma (P50 CA225450).
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.