Rare immune-related adverse events in patients with melanoma: incidence, spectrum, and clinical presentations

ABSTRACT

Immune-related adverse events (irAEs) are side effects of immune checkpoint inhibitor therapy (ICI). While common irAEs have been well characterized, there are more limited data on rare immune related adverse events (RirAEs) due to low incidence. Lack of characterization of these entities has led to difficulties in accurate diagnosis and management. Here, we conducted a multi-institution analysis of all patients with stage III/IV melanoma who developed RirAEs after being treated with ICIs (anti-PD-1/L1, anti-CTLA-4, and combination PD-1/CTLA-4 blockade) at three institutions (Vanderbilt University Medical Center, Massachusetts General Hospital, and Melanoma Institute of Australia). RirAEs were defined as those occurring in approximately <1% of patients treated with anti-PD-1 or <2% with combination. Of 2834 patients who received ICIs, 82 developed RirAEs and were more common with combination PD-1/CTLA-4 blockade (4.6%) vs. anti-PD-1/L1 agents (2.8%). Overall median time from ICI start to RirAE was 86 days (interquartile range 42–235 days) with significantly earlier onset in combination therapy (p < 0.001). The spectrum of RirAEs spanned across several organ systems. Most RirAEs were grade 2 (57 [41.3%]) and grade 3 (40 [29.0%]) with relatively few grade 4 (11 [8.0%]) or 5 (5 [3.6%]) events. Steroid re-escalation (21.4%) or additional immunosuppression (13.8%) were commonly required. RirAE recurrence occurred in 22.6% with ICI rechallenge; 37.1% had new irAEs with rechallenge. In conclusion, RirAEs associated with ICIs in melanoma patients occurred, in aggregate, in 2–5% of patients treated with anti-PD-1-based therapy. Steroid re-escalation and alternative immunosuppression use were frequently required but fatal irAEs were fairly uncommon.

KEYWORDS:

• Immune checkpoint inhibitor toxicity
• immune checkpoint inhibitors
• immune related adverse events
• melanoma
• rare immune related adverse events
• uncommon immune related adverse events

Introduction

Immune checkpoint inhibitors (ICIs) target tumor cells by upregulating antitumor immunity. However, this immune stimulation can provoke off target, autoimmune-like immune related adverse events (irAEs) as “collateral damage.” With increasing utilization of ICIs, a wide spectrum of irAEs have been reported and characterized, ultimately affecting most patients on therapy. The clinical features of common irAEs have been generally well-characterized along with the development of clinical consensus management guidelines. ^{1, 2} These common irAEs affect the colon, liver, thyroid, skin, joints, lungs, and pituitary; more uncommonly affected areas include the eyes, heart, nervous system, and kidneys.^3,4 However, data regarding rare irAEs (RirAEs), or those impacting <1% of patients, remains less well explored.

Characterizing RirAEs is critical as these events are often associated with high rates of morbidity and mortality, including neurotoxicities, cardiomyotoxicities, and hematologic events.^5–7 This association may be due to the difficulty in appropriate diagnosis and management of RirAEs, leading to negative outcomes.⁵ Current consensus clinical guidelines include more limited (albeit increasing) information regarding the diagnosis and management of RirAEs.^1,2,8,9 As the clinical indications for ICIs expand in different tumor types and settings, clinicians are increasingly encountering these less common events. Further, given the wide spectrum of irAEs caused by ICI, and the idiosyncrasies of adverse event reporting in clinical trials (e.g. many trials only report events that occur in ≥5% of patients), RirAEs may actually be more common than realized, at least when considered in aggregate. Thus, this study aims to characterize the incidence, spectrum, and clinical presentations of RirAEs in patients with advanced melanoma.

Materials and methods

Patients & Data Collection

This study was a retrospective, multi-institutional study of patients with stage III/IV melanoma that developed RirAEs after being treated with ICIs (anti-PD-1/L1, anti-CTLA-4, and combination PD-1/CTLA-4 blockade) at three institutions (Vanderbilt University Medical Center, Massachusetts General Hospital, and Melanoma Institute of Australia) who began therapy between January 2014 and April 2021. We screened all patients treated with these agents at all centers to identify patients who developed RirAEs (inflammatory clinical events thought to be due to ICI therapy). IRB approval was obtained from VUMC Human Research Protections Program.

We defined RirAEs as irAEs with an approximate incidence of ≤1% of patients treated with anti-PD-1/PD-L1 monotherapy or ≤2% with combination ipilimumab/nivolumab from published literature (Table 1). Notably, terminology and definitions vary between studies; some pharmacovigilance studies technically define “rare” events occurring even less frequently (e.g. 0.01–0.1%) with “uncommon” events occurring in the 0.1–1% range)²¹, however for this study we used the former definitions. Events without probable and direct immune-relation (e.g. fatigue) were not included. Given the inconsistent reporting across clinical trials and case series that potentially over-report RirAEs, our estimates of these incidence rates were necessarily approximate. Data collection for each patient included patient demographics, treatment characteristics, RirAE characteristics (including type, grade, outcome, and treatment), outcomes with ICI rechallenge, steroid use, treatment with non-steroidal immunomodulators, and survival outcomes. Patient demographics and baseline characteristics collected included: gender, race/ethnicity, age, BMI, cancer stage, presence of brain metastases, preexisting autoimmune diseases, and past anti-tumor treatment history. RirAEs were assessed and graded using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.

Table 1. Events considered as Rare Immune Related Adverse Events by System^1,8.

System	Rare irAEs
Ocular^Citation10	Diplopia (1), ocular sicca (3), vitreitis (1), optic neuritis (1), uveitis (10), keratitis, extraocular muscle myopathy, blepharitis (3), conjunctivitis (2), scleritis, retinitis, retinal vasculitis (1)
Gastrointestinal^Citation11	Oral mucositis (8), pancreatitis (10), esophagitis (2), gastritis (1), xerostomia (3), celiac disease,
Rheumatologic^Citation12	Sjogren (1), sarcoidosis (2), scleroderma, myositis (4), anhidrosis (1), polymyalgia rheumatica (2)
Cardiovascular^Citation13,Citation14	Myocarditis (11), vasculitis, dilated cardiomyopathy, pericardial effusion
Cutaneous^Citation15	Bullous pemphigoid (3), neutrophilic dermatoses, SJS/TEN, erythema nodosum (1), psoriasis (2), vaginitis (1), alopecia (5)
Hematologic^Citation16	Immune thrombocytopenia purpura (1), aplastic anemia (1), hemolytic anemia, pure red cell aplasia (1), neutropenia, lymphopenia, hemophilia, hemophagocytic lymphohistiocytosis
Endocrine^Citation17	Type 1 diabetes (12), primary adrenal insufficiency* (6)
Neurologic^Citation18	Seizures (1), dysphagia (1), paresthesia (1), myasthenia gravis (1), neuropathy (6), myelopathy (1), limbic encephalopathy (1), Guillain-Barre, posterior reversible encephalopathy, transverse myelitis, Bell’s palsy (1), encephalitis (4), aseptic meningitis (3)
Pulmonary	None
Renal/Genitourinary^Citation19,Citation20	Nephritis (17), acute tubular necrosis, nephrotic syndrome (2), orchitis (1)
Hepatologic	Cholecystitis, cholestatic liver injury (2)

Note. Number in parentheses are the number of patients observed for each side effect (events not observed in our cohort have no number listed). *Primary adrenal insufficiency without evidence of hypophysitis or hypopituitarism.

Statistical analysis

Continuous variables were reported as median and interquartile range (IQR) and categorical variables were reported as frequencies and percentages. Survival time was calculated from date of ICI initiation to time of progression (for progression-free survival) or last follow-up (for overall survival) using the Kaplan–Meyer method and compared with the logrank test; patients were censored at last follow-up. Correlation matrix was visualized using R package “ggcorrplot”. As an exploratory analysis, we performed latent cluster analysis (LCA) to assess whether different patterns of toxicity were associated with clinical features or survival and grouped patients into three classes. LCA is a probabilistic modeling method of unsupervised clustering that uses EM and Newton-Raphson algorithms to maximize the latent class model log-likelihood function (R package: “poLCA”). To balance fit and parsimony, we limited our comparison to a three-class analysis. All statistical tests were two-sided, and p-values <0.05 were regarded as statistically significant. All statistical analyses were conducted using SPSS version 28.0.1.0 and R 4.2.1.

Results

Patient characteristics and treatment

To assess the incidence of RirAEs, we assessed the data from two centers where we had an accurate assessment of the total number of patients treated (VUMC and MIA). We screened the following cohorts: from VUMC, 782 total patients received ICIs, of which, 471 received anti-PD-1/L1, 172 received anti-CTLA-4/anti-PD-1/L1 combination, and 139 received anti-CTLA-4. From MIA (n = 2052) patients, 1086 received anti-PD-1/L1, 630 received anti-CTLA-4/anti-PD-1/L1 combination, and 336 received anti-CTLA-4. Of the two-center cohort consisting of 2,834 patients with advanced melanoma treated with immune checkpoint inhibitors, 82 developed RirAEs for an overall incidence of 2.9%. For anti-PD-1/L1 agents, combination agents, and anti-CTLA-4 agents, the incidences of rare toxicities were 2.8%, 4.6%, and 0.2%, respectively (Table 2).

Table 2. RirAE Incidences by Agent.

ICI Class	Rare Tox Cohort	Denominator	Incidence
MIA
Anti-PD-1/L1	19	1086	1.75%
Combination	22	630	3.49%
Anti-CTLA-4	1	336	0.30%
VUMC
Anti-PD-1/L1	25	471	5.31%
Combination	15	172	8.72%
Anti-CTLA-4	0	139	0.00%
Total
Anti-PD-1/L1	44	1557	2.83%
Combination	37	802	4.61%
Anti-CTLA-4	1	475	0.21%

Note. Values represent unique patients rather than unique treatments. ICI; Immune Checkpoint Inhibitor. MIA; Melanoma Institute of Australia. VUMC; Vanderbilt University Medical Center. This assessment does not include the additional 30 patients from MGH.

An additional 30 RirAE patients were added from MGH for further characterization of rare toxicities; however since the total number screened (“denominator”) was unclear, they were not included in the incidence calculation. Of the full RirAE cohort of 112 patients, the median age was 64 years (IQR, 52–74 years) (Table 3). Most patients were male (80 [71.4%]) and Caucasian (110 [98.2%]). Among patients with advanced disease, 21 (19.3%) had brain metastases at the start of treatment. Fourteen patients (12.5%) had preexisting autoimmune disease including type 1 diabetes, myasthenia gravis, dermatomyositis, psoriasis, gout, hypothyroidism, asthma, hypogonadotropic hypogonadism, multiple sclerosis, and idiopathic thrombocytopenia.

Table 3. Patient Demographics and Treatment Characteristics.

Characteristic	Number of patients (%)
Age (median)	64 (52–74)
Sex (male)	80 (74.1%)
BMI (median)	27.0 (24.9–30.7)
Initial stage at start of treatment*
III	20 (18.3%)
M1a	13 (11.9%)
M1b	14 (12.8%)
M1c	24 (22.0%)
M1d	21 (19.3%)
IV (not specified)	17 (15.6%)
Pre-existing autoimmune disease	14 (12.5%)
ICI treatment
Anti-PD-1/PD-L1	56 (50.0%)
Anti-CTLA-4	3 (2.7%)
Combination	53 (47.3%)
Response to treatment*
Complete response	32 (29.4%)
Partial response	40 (36.7%)
Stable disease	13 (11.9%)
Progressive disease	24 (22.0%)
ICI discontinuation	69 (61.6%)
Reason for discontinuation
Rare toxicity	48 (69.6%)
Other toxicity	10 (14.5%)
Progressive disease	3 (4.3%)
Completed treatment	8 (11.6%)

Note. Continuous variables are reported as medians.

*Missing values that were unreported.

ICI treatments in patients who had RirAEs were anti-PD-1/PD-L1 (56 [50.0%]), anti-CTLA-4 (3 [2.7%]), or combination (53 [47.7%]) therapy (Table 3). Responses to treatments included: complete response (29.4%), partial response (36.7%), stable disease (11.9%), and progressive disease (22.0%). ICI was discontinued in 61.6% of patients for the following reasons: rare toxicity (48 [69.6%]), other toxicity (10 [14.5%]), progressive disease (3 [4.3%]), and completed treatment (8 [11.6%]).

RirAE incidence, spectrum, and timing by organ system

One hundred and twelve patients developed a total of 145 RirAEs (1.3 RirAEs per patient). Most patients experienced temporally or clinically distinct events (n = 106) and six experienced concurrent RirAEs. Overall median days from ICI start to RirAE was 86 (42–235); with medians of 137, 58, and 122 for anti-PD-1/L1, combination therapy, and anti-CTLA-4 treatments, respectively (Figure 1, Table 4). Combination therapy was associated with significantly earlier onset of RirAE (p < 0.001). The spectrum of RirAEs spanned across several organ systems with the most commonly affected systems being gastrointestinal (26 [17.9%]) followed by neurologic and renal (20 [13.8%] and 19 [13.1%] respectively) (Tables 1, 4). Most RirAEs were grade 2 (57 [41.3%]) and grade 3 (40 [29.0%]) with few grade 4 (11 [8.0%]) or 5 (5 [3.6%]) events. Among all high-grade (grade 3–5) events, endocrine RirAEs were most frequent (20.7%; three adrenal insufficiency, eight type 1 DM). Endocrine (25.0%) and neurological (20.0%) RirAEs were most common among grade 3 events (Supplementary Table 1). Gastrointestinal RirAEs were most common among grade 4 events (27.3%). Of the five fatal events, four were myocarditis (one associated with single-agent anti-PD-1 and three with combination). RirAEs; the fifth fatal event was limbic encephalitis.

Figure 1. Time to RirAE for patients treated with single agent anti-PD-1 vs. those treated with combination PD-1/CTLA-4 blockade.

Table 4. Rare irAE Incidence and Grades.

Characteristic	Days (IQR)
Days to rare irAE (median)	86 (42–235)
Anti-PD-1/L1	137 (63–341)
Combination	58 (21–134)
Anti-CTLA-4	122 (-)
Rare irAE Type	Number of patients (%)
Ocular	23 (15.9%)
Hematologic	3 (2.1%)
Gastrointestinal	26 (17.9%)
Hepatologic	2 (1.4%)
Rheumatologic	11 (7.6%)
Cardiac	11 (7.6%)
Cutaneous	12 (8.3%)
Endocrine	18 (12.4%)
Neurological	20 (13.8%)
Respiratory	0 (0.0%)
Renal	19 (13.1%)
Additional irAE (non-RirAE)	72 (63.4%)
Grade of RirAE*
1	25 (18.1%)
2	57 (41.3%)
3	40 (29.0%)
4	11 (8.0%)
5	5 (3.6%)

Note. Continuous variables are reported as medians. IQR; interquartile range.

*Missing values that were unreported

Over half of patients developed additional non RirAE toxicities (63.4%), most commonly rash, pruritus, fatigue, and arthritis. Notably, 25 patients developed 2 or more RirAEs (including 7 with 3 RirAEs and 1 with 4). Of these, 18 also had non-RirAEs (72% with a median of 3 non-RirAEs, range 1–5). Of 87 patients with a single RirAE, 51 had a non-RirAE as well (59%). We assessed whether particular organ system irAEs co-occurred in this cohort (including both rare and non-rare events); several systems had modest positive correlations (Pearson’s ρ 0.16–0.19, including gastrointestinal with hepatic [p = 0.05–0.1] or cutaneous irAEs [p < 0.05], and cutaneous and ocular irAEs [p = 0.05–0.1]), whereas others had modest anti-correlations (e.g. neurologic and gastrointestinal [p = 0.05–0.1]; cardiac and almost all other systems [p = 0.05–0.1]) (Figure 2).

Figure 2. Correlation between organ system irAes (both rare and non-rare irAes).

Management and outcomes associated with RirAEs

Of the 145 RirAEs that patients experienced, 100 (69.0%) required corticosteroids for management (Table 5). Median steroid dose was 60 mg daily of prednisone equivalents and often required re-escalation during steroid taper (21.4%) or additional immunosuppression (13.8%). The following 20 RirAEs required additional immune suppression: hepatologic (n = 1, 50%), neurological irAEs (n = 7, 35%), rheumatologic (n = 3, 27%), cardiac (n = 3, 27%), cutaneous (n = 3, 25%), renal (n = 1, 5%), and gastrointestinal (n = 2, 8%). Additional immunosuppressive agents included: TNFα inhibitors (Adalimumab, Infliximab), IL-4 receptor α inhibitors (Dupilumab, Omalizumab), VEGF inhibitors (Ranibizumab), mycophenolate mofetil, methotrexate, and rituximab (see Supplementary Table 1 for indications). Although over half of RirAEs resolved at last follow-up (57.2%), a substantial number of patients had persistent symptomatic RirAEs (31.0%), were asymptomatic with residual complications (10.3%), or experienced other related complications (0.7% - specifically a myocardial infarction following development of type 1 diabetes). Residual complications included need for maintenance steroids or insulin, and chronic kidney disease following nephritis. Rates of complete resolution by system were as follows: hepatologic (2/2, 100%), ocular (21/23, 91.3%), gastrointestinal (21/26, 80.8%), hematologic (2/3, 66.7%), rheumatologic (7/11, 63.6%), cardiac (7/11, 63.6%), neurological (10/20, 50%), cutaneous (5/12, 41.7%), renal (7/19, 36.8%), and endocrine (2/18, 11.1%).

Table 5. Rare immune related adverse events management and outcomes.

Characteristic	N (%)
Steroid treatment^†	100 (69.0%)
Steroid dose^†	60 (25–80)
Steroid escalation during taper^†	31 (21.4%)
Required additional immunosuppression^†	20 (13.8%)
Resolved at last follow-up^†	83 (57.2%)
Unresolved at last follow-up^†	62 (42.8%)
Complete resolution	83 (57.2%)
Chronic RirAE	45 (31.0%)
Asymptomatic w/residual issues	15 (10.3%)
Additional complications*	1 (0.7%)
Rechallenge with ICI^†	62 (42.8%)
Rare toxicity flare	14 (22.6%)
New toxicity	23 (37.1%)
No flare or toxicity	28 (45.2%)
Progressed at last follow-up^‡	35 (31.3%)
Follow up time (median years)^‡	2.9 (1.7–4.7)
Alive at last follow-up^‡	82 (73.2%)
Reason for death^‡
Disease	16 (53.3%)
RirAE	5 (16.7%)
Other	9 (30.0%)

Note. Continuous variables are reported as medians.

*Myocardial infarction.

^†Values represent unique RirAEs rather than unique patients.

^‡Values represent unique patients rather than unique RirAEs or treatments.

ICI rechallenge occurred in almost half of patients (42.8%); most patients developed either rare toxicity flares (22.6%) or de novo toxicities (37.1%). Median follow-up time was 2.9 years and 73.2% (n = 82) of patients were alive at last follow-up with relatively low rates of progressive disease (31.3%). There was no significant difference between monotherapy and combination therapy in terms of overall survival (p = 0.452). Among patients that died, most died from disease (n = 16, 53.3%) or other/unknown causes (n = 9, 30%) rather than RirAE (n = 5, 16.7%).

To assess whether various patterns of irAEs correlated with patient characteristics or clinical outcomes, we performed latent cluster analysis as an exploratory analysis, and classified patients based on patterns of co-occurring irAEs (Supplementary Figure 1). This system classified patients into three groups; these classes differed by overall survival (p = 0.02; Supplementary Figure 2) improved response rate (81% vs. 66% vs. 38% for class 1, 2, and 3; p = 0.002) despite similar treatments (combination in 47%, 54%, and 49%, p = 0.88), age, and stage (Supplementary Table 2), though class 1 had marginally lower rates of preexisting autoimmune disorders and baseline BMI. Organ systems were markedly different between classes; Class 1 was associated gastrointestinal, cutaneous, and/or renal involvement, class 2 had more frequent neurologic and respiratory involvement, and class 3 had more rheumatologic, cardiac, and multi-gland endocrine involvement (Supplementary Table 2). Notably, while the rates of steroid administration were fairly similar across classes (52% vs. 69% vs. 66% for classes 1, 2, and 3, p = 0.23), the median steroid doses appeared lower in class 1 (median 40 mg vs. 65 mg vs. 60 mg, p = 0.037).

Discussion

To our knowledge, this study is the largest retrospective analysis of RirAEs published to date. Overall, we found an RirAE incidence of 2.83% with single-agent anti-PD-1 and 4.61% with combination PD-1/CTLA-4 blockade. These toxicities required re-escalation of high-dose steroids or alternative immunosuppressants in approximately half of cases. However, they were rarely fatal (within the full cohort), and ICI rechallenge did not usually result in RirAE flare, albeit in only a modest number of cases attempted. Further, patients had anti-cancer outcomes that appeared favorable compared with general populations.

There are no standardized definitions of specific RirAEs; we thus used a somewhat arbitrary and highly approximate cutoff of <1% for single-agent anti-PD-1 and <2% for combination treatment.²² Within this RirAE cohort, the more commonly affected systems included gastrointestinal, ocular, neurological, endocrine, cutaneous, and renal. Even among this highly selected group of toxicities, hematologic and cardiac events were fairly uncommon.

Most RirAEs tended to be low to moderate grade, with few high-grade events. Only five patients died due to RirAE, with most deaths having occurred due to disease progression. This potentially suggests that high volume centers experienced in irAE management are usually able to salvage patients even with severe, fulminant RirAEs, and effectively manage even patients with potential delays in diagnosis due to unusual presentations. The vast majority of patients were alive at last follow-up with relatively long progression-free and overall survival times without sequelae; previous studies have also suggested a link between irAEs and superior survival, including in patients with multi-system involvement.^23,24 It is certainly possible that such a similar link is present in these RirAEs, though the heterogeneity of events and potential confounding bias (e.g. those on therapy longer have more toxicity) could also cloud such associations. Notably, latent cluster association was able to identify a population of patients with very poor survival outcomes (class 3), characterized by cardiac, rheumatologic, and multi-gland endocrine involvement. While cardiac events may have partially explained this finding, these patients also had low response rates to therapy. Further study in larger cohorts is needed to validate this classification approach.

Consistent with general irAE severity estimates, most moderate (3) and high grade (4 or 5) events were cardiac, neurological, gastrointestinal, renal, and endocrine.²⁵ Most grade 5 events were myocarditis, which has been characterized as the irAE with the highest mortality rate.⁶ Overall, the rate of fatal RirAEs within the full cohort was low (5 of 2,834) although 17% of the deaths within the RirAE cohort were from toxicities. Thus, the risk of death from RirAEs largely appears driven by myocarditis.

Despite the overall low incidence of 2.8–4.6% within the full cohort, many patients experienced multiple RirAEs (1.3 per patient), indicating that there may be underlying risk factors associated with irAE development. Adding credence to this argument, most patients also experienced concurrent non-rare irAEs, some with up to 5 additional toxicities, and 12% had preexisting autoimmune disease, a number that is likely higher than in an unselected population (e.g. 4.7% from VUMC patients without RirAEs). Although several systems had modest co-occurrence, they tended to be more common events (e.g. gastrointestinal and cutaneous irAEs). However, the association of irAE patterns with survival warrants further study. A number of efforts are underway to assess biomarkers for irAEs, with possible markers including preexisting autoantibodies, effector memory CD4 T cells, microbiome profiles, and genetic factors.²⁶ This study suggests that this population may provide particularly relevant clues, given the frequency of multi-system involvement. As with other studies, combination therapy was a risk factor: RirAEs occurred earlier (median 58 days) with combination therapy compared to anti-PD-1/L1 monotherapy, and with higher incidence (4.6% vs. 2.8%).²⁷ Interestingly, RirAEs appeared more common at VUMC than in Australia. This could be related to differences in screening, or possibly dietary, lifestyle, or genetic factors.

A better understanding of RirAEs is critical as these toxicities are encountered with more frequency with the increasing usage of ICIs. However, the current understanding of RirAEs is limited due to their low overall incidence, further complicated by challenges in diagnosis. We observed several trends in the clinical presentations and management of these RirAEs including: 1) RirAEs generally presented similarly to their non-ICI induced counterparts (e.g. uveitis, type 1 DM; although presumably at a higher rate compared with untreated patients²⁸, 2) RirAEs were often effectively managed with corticosteroids although rates of steroid re-escalation and alternative immunosuppression were high compared with other studies,²⁹ and 3) fatal events were uncommon. Although broadly similar to common toxicities, the rate of steroid-refractory irAEs suggests that more personalized management strategies are needed for various RirAEs. As expected, irAE systems traditionally associated with higher mortality were associated with additional immunosuppression requirements such as neurological and cardiac irAEs, while gastrointestinal and renal irAEs were less refractory to steroids.

High grade irAEs and RirAE are relative contraindications for rechallenge. However, this study found that while rechallenge was common and was often safe. Rates of irAEs with rechallenge were 22.6% and 37.1% for rirAE flares and de novo irAEs, respectively. Thus, cautious rechallenges may be considered on a case-by-case basis, particularly for treatable RirAEs with low risk of sequelae and mortality in patients who need therapeutic rechallenge.

Limitations

This study has several limitations. First, it only included patients with advanced melanoma and three centers, limiting generalizations to other cancer types and centers. Second, the lack of standardized clinical and literature-based criteria for the definition of some RirAEs may have led to variable interpretations of the results based on expert opinion. This may have led to the differences in incidences between the two institutions with differences in diagnoses of RirAEs. Previous reports of RirAEs in the literature are likely underestimates of their true incidences. Third, the survival outcomes of our cohort may be skewed by the inclusion of a small subset of stage III disease patients who received ICIs. Several patients experienced more than one RirAE during treatment, making it difficult to distinguish between specific clinical courses and management for this subset of patients.

Conclusion

RirAEs associated with ICIs in melanoma patients occur in 3–6% of patients in aggregate and present as diagnostic and management challenges to clinicians. Our study found that most RirAEs were of moderate grade and had favorable long-term outcomes, although rates of steroid re-escalation and need for alternative immunosuppression were high, and myocarditis continues to have a high fatality rate. Clinicians should be aware of these general patterns of incidence and clinical presentations to support diagnosis and effective management.

Availability of data and material

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors’ contributions

BCP, SN, LZ and DBJ collected the data. BCP, DBJ, FY, and RF analyzed and interpreted the data. AG, RJS, GB, KLR, JMB, MSC, GVL, AMM, BCP, and DBJ reviewed and revised the manuscript. All authors read and approved the manuscript.

Ethics approval and consent to participate

This study was conducted in accordance with IRB guidelines and approved by the Vanderbilt University IRB committee #161485. Informed consent was received from all participants.

Disclosure statement

DBJ has served on advisory boards or as a consultant for BMS, Catalyst Biopharma, Iovance, Jansen, Mallinckrodt, Merck, Mosaic ImmunoEngineering, Novartis, Oncosec, Pfizer, Targovax, and Teiko, and has received research funding from BMS and Incyte. AMM has served on advisory boards for BMS, MSD, Novartis, Roche, Pierre-Fabre and QBiotics. MSC: consultant advisor: Amgen, Bristol-Myers Squibb, Eisai, Ideaya, Merck Sharp and Dohme, Nektar, Novartis, Oncosec, Pierre-Fabre, Qbiotics, Regeneron, Roche; honoraria: Bristol-Myers Squibb, Merck Sharp and Dohme, Novartis. GVL is consultant/advisor for Agenus, Amgen, Array Biopharma, Boehringer Ingelheim, Bristol Myers Squibb, Evaxion, Hexal AG (Sandoz Company), Highlight Therapeutics S.L., Innovent Biologics USA, Merck Sharpe & Dohme, Novartis, OncoSec, PHMR Ltd, Pierre Fabre, Provectus, Qbiotics, Regeneron. JMB receives research support from Genentech/Roche, Bristol Myers Squibb, and Incyte Corporation, has received consulting/expert witness fees from Novartis, and is an inventor on patents regarding immunotherapy targets and biomarkers in cancer..

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/2162402X.2023.2188719.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

DBJ receives funding from R01 HL155990, R01CA227481, Susan and Luke Simons Directorship, James C. Bradford Melanoma Fund, Van Stephenson Melanoma Fund, and the Melanoma Research Foundation. GVL and AMM are supported by National Health and Medical Research Council (NHMRC) Investigator Grants. GVL is supported by the University of Sydney Medical Foundation.