Evaluating Survival After Hospitalization Due to Immune-Related Adverse Events From Checkpoint Inhibitors

Abstract Background As immune checkpoint inhibitors (CPI) are increasingly approved for cancer treatment, hospitalizations related to severe immune-related adverse events (irAE) will increase. Here, we identify patients hospitalized due to irAEs and describe survival outcomes across irAE, CPI, and cancer type. Methods We identified patients hospitalized at our institution from January 2012 to December 2020 due to irAEs. Survival was analyzed using Kaplan-Meier survival curves with log-rank tests. Results Of 3137 patients treated with CPIs, 114 (3.6%) were hospitalized for irAEs, resulting in 124 hospitalizations. Gastrointestinal (GI)/hepatic, endocrine, and pulmonary irAEs were the most common causes of irAE-related hospitalization. After CPI initiation, the average time to hospitalization was 141 days. Median survival from hospital admission was 980 days. Patients hospitalized due to GI/hepatic and endocrine irAEs had longer median survival than patients with pulmonary irAEs (795 and 949 days vs. 83 days [P < .001]). Patients with melanoma and renal cell carcinoma had longer median survival than patients with lung cancer (2792 days and not reached vs. 159 days [P < .001]). There was longer median survival in the combination group compared to the PD-(L)1 group (1471 vs. 529 days [P = .04]). Conclusions As CPI use increases, irAE-related hospitalizations will as well. These findings suggest that among patients hospitalized for irAEs, survival differs by irAE and cancer type, with worse survival for patients with irAE pneumonitis or lung cancer. This real-world data contributes to research pertaining to hospitalization due to severe irAEs, which may inform patient counseling and treatment decision-making.


Introduction
2][3] These medications block the interaction of immune checkpoint proteins on tumor cells, thus allowing for immune activation 4 ; they include inhibition of programmed cell death protein 1 (PD-1), programmed death-ligand 1 (PD-L1), cytotoxic T lymphocyte-associated 4 The Oncologist, 2023, Vol. 28, No. 10 e951 (CTLA-4), and most recently, lymphocyte activation gene-3 (LAG3). 5lthough these treatments are overall better tolerated than conventional chemotherapies, 6 the side-effect profile poses unique challenges.Side effects, called immune-related adverse events (irAEs), result from immune activation and can occur in almost any organ system. 7Rates of severe irAEs (Grade 3 or above) vary by treatment regimen, but up to 55% of patients treated with combination ipilimumab and nivolumab may develop a severe irAE. 8,9 subset of patients requires hospitalization for irAEs.A national study using insurance claims estimates 3.5% of patients initiating CPI therapy experience an irAE requiring hospitalization.10 Another study described 450 irAE-related hospitalizations; gastrointestinal (GI), pulmonary, hepatic, and endocrine irAEs were the most common.11 The irAE-specific mortality rate was 5.6% and highest for pulmonary and cardiac irAEs, higher than the overall reported incidence of fatal irAEs (0.3%-1.3%).12 Another study looked at 23 patients hospitalized for irAEs, of whom 3 (13%) died from the irAE.13 How the survival of patients hospitalized for irAEs might vary across multiple characteristics including CPI type, cancer type, and irAE remains poorly understood.
This study characterizes the spectrum of toxicities and survival of patients hospitalized for irAEs, as well as how survival varies across CPI type, cancer type, and irAE.Stratifying hospitalized patients in this way contributes to a more nuanced understanding of survival outcomes for patients hospitalized due to irAEs.

Materials and Methods
This study was approved by the UCSF Human Research Protection Program [#17-22987].

Patient Selection and Data Collection
Inclusion criteria are summarized in Fig. 1.The study used computational extraction to identify patients with solid tumor malignancies who received CPIs from 1/1/2012 to 12/31/2020 and hospitalized any time between treatment initiation and 6 months after the last CPI dose.Hospitalizations for surgical procedures or planned chemotherapy were excluded.The remaining charts were manually reviewed by either a trained medical student (F.W., D.K.), resident (J.S.), or faculty member (Z.Q.) to include only definite or likely irAE-related hospitalizations.Ambiguous cases were discussed with a board-certified oncologist (S.B.).A confirmed irAE hospitalization was defined as consensus around the irAE diagnosis between the inpatient oncology team, the outpatient primary oncologist, and exclusion of alternative diagnoses.In cases where a biopsy was taken, the biopsies were reviewed to confirm the diagnosis.Cases that remained unclear after manual chart review and discussion were excluded (n = 36).

Statistical Variables and Analysis
Relevant demographics, clinical history, and key admission characteristics were manually collected and stored using REDCap.Variables were summarized using means, medians, and interquartile ranges (IQRs) for continuous variables and proportions for categorical variables.Categorical and continuous variables were compared by CPI type using Fisher's exact tests and Wilcoxon rank-sum tests, respectively.For post-hospitalization survival, follow-up was measured from the date of hospital admission to the date of death or last follow-up, with survival censored at the last follow-up, by Kaplan-Meier survival curves and log-rank tests.Survival was assessed for the overall cohort and stratified by CPI type, cancer type, and irAE type.Only subgroups that were greater than or equal to 10 were included.Observed post-hospitalization patient survival and 95% confidence intervals (CIs) were estimated at 1, 3, and 5 years.A sensitivity analysis using the date of CPI initiation to death or last follow-up was also completed.Statistical analyses were performed using Stata.Statistical significance was defined as P-value of ≤.05.

irAE Characteristics
The average number of irAE-related hospitalizations per patient was 1.09 (ranging from 1 to 3).A summary of the Comparing Caucasian to all other races/ethnicities combined, some patients considered themselves both Hispanic and another race.*** P-value is categorical comparing 1 hospitalization versus more than 1 hospitalization.Abbreviations: BMI, body mass index; CPI, checkpoint inhibitor; Durva, durvalumab; HCC, hepatocellular carcinoma; Ipi, ipilimumab; irAE, immunerelated adverse event; Nivo, nivolumab; Pembro, pembrolizumab; RCC, renal cell carninoma; SCC, squamous cell carcinoma.In parenthesis, the number per irAE describes the total number of hospitalizations combining hospitalizations for single-system irAE with multisystem irAEs.For example, for cardiovascular there were 4 hospitalizations for cardiac irAE alone and 10 hospitalizations with cardiac irAEs combined with other system irAEs.A hospitalization within 1 system could have multiple specific irAEs within that system in 1 hospitalization.Number of hospitalizations in italics describe irAE hospitalizations with only that system.

Discussion
This study describes the experience at UCSF with irAE-related hospitalizations.The frequency and distribution of irAEs across organ systems are consistent with prior studies reporting irAE-related hospitalizations. 11,13The most common types of irAEs that cause hospitalization were GI/ hepatic, endocrine, and pulmonary, which are distinct from the most common irAEs of any grade which are pruritis, rash, and diarrhea not defined as colitis. 14This study adds nuance to the characterization of irAEs leading to hospitalization, reporting outcomes across cancer and CPI type in addition to irAE type.
This study characterized the type of patients hospitalized from irAEs at UCSF.In this study, 3.6% of patients treated with a CPI at UCSF were hospitalized from irAEs, which is consistent with a study that used national insurance claims. 10his number could underestimate the true percentage of patients hospitalized after irAEs since it only includes hospitalizations at UCSF.The catchment area of UCSF is broad, including 25 counties; therefore, it is possible patients could present to local hospitals for irAE treatment and not get transferred to UCSF, which would not be included in this study.Additionally, UCSF has created an urgent care for cancer center which promptly identifies and treats irAEs possibly preventing hospitalizations.The average age of patients A sensitivity analysis was performed with the same models but excluded patients with non-metastatic disease which had similar results to the overall cohort (Supplementary Figs.4A-4D).
The Oncologist, 2023, Vol. 28, No. 10 e957 in the cohort (62 years old) was similar to a population-level study of irAE-related hospitalizations. 10Most patients had metastatic or unresectable melanoma, lung cancer, or RCC at CPI initiation, which are most frequently treated with CPI therapy. 2 While the majority of patients did not have a prior autoimmune disorder, 13.7% did, which is larger than was reported in a previous study on irAE-related hospitalizations (4.0%). 11History of an autoimmune condition has been shown to increase the risk of irAEs and hospitalization from irAEs. 15,16Additionally, 20% of hospitalizations had multiple irAEs.Therefore, if a patient has one irAE, it is important to evaluate for others.This study assessed post-hospitalization survival after an irAE-related hospitalization stratified by CPI, cancer, and irAE type.The median survival following hospitalization was nearly 3 years, and 1-year post-hospitalization survival was 63%.Survival after hospitalization from irAEs differed by type of cancer type.In this study, patients with lung cancer who experienced an irAE hospitalization had worse OS.While this is likely due to the fact that lung cancer in general has worse OS, 17 the survival could be lower due to the irAE itself in lung cancer patients.The 1-year survival in this study for patients with lung cancer and any irAE hospitalization was 30%, compared to a prior large study of patients with all stages of lung cancer who initiated CPI therapy had a 1-year survival of 43%. 18This is in contrast to patients with melanoma and RCC, who despite having an irAE hospitalization had similar overall survival (1-year post-hospitalization survival: 80%) to studies of patients on CPIs without hospitalization, which ranged from 65% to 80%. 19,202][23][24][25][26][27][28] Therefore, even if the irAE leads to hospitalization, future studies with a proper control group and sample size that accounts for important clinical factors such as cancer type, are needed to determine if irAE hospitalization could be predictive of overall survival outcomes.
In this study, post-hospitalization survival not only differed by cancer type, but also by irAE type.Patients hospitalized for pulmonary irAEs (33%) were shown to have worse post-hospitalization survival than those with GI (64%), endocrine (87%), or multisystem irAEs (57%).3][34] Patients with lung cancer have the highest incidence of pneumonitis, so it is likely that the inferior survival for CPI-induced pneumonitis and lung cancer overlap (40% of hospitalizations for pneumonitis had lung cancer in this study). 357][38][39][40][41][42] The mechanism by which some irAEs are associated with improved survival is hypothesized to be increased immunological activation leading to improved antitumor immunity. 7n this cohort, irAE-related hospitalizations occurred earlier after CPI initiation with regimens including CTLA-4 inhibitors and with GI or pulmonary irAEs, and later with PD-(L)1 monotherapy regimens and with endocrine irAEs.The time to hospitalization after CPI initiation for the overall cohort averaged 138 days, similar to a population-level study of severe irAEs (148 days). 10Although not statistically significant, there was a trend toward earlier irAE hospitalization with CTLA-4 monotherapy (66 days) or combination therapy CTLA-4/PD-(L)1 (85 days) compared to PD-(L)1 monotherapy (170 days).Other studies have shown that CTLA-4 therapy is correlated with shorter time to severe irAEs. 43Pulmonary (mainly pneumonitis) and GI (colitis and hepatitis) irAEs had a shorter median time to hospitalization at 77 and 84 days respectively, while endocrine irAEs had a longer time to hospitalization at 111 days.Prior studies have shown that CPI-induced pneumonitis had a median onset of 79 days, 44 CPI colitis around 6-7 weeks, and CPI hepatitis around 6-14 weeks. 43,45The shorter onset of all-grade colitis in prior studies compared to this study of only high-grade colitis could reflect earlier lower-grade colitis preceding hospitalization and thus an opportunity to prevent hospitalization with early detection.Endocrine irAEs have been shown to occur later and take longer to resolve at up to 28 weeks after CPI initation. 43In this study, there was a large range of time to hospitalization for GI irAEs, with 1 patient experiencing a GI irAE resulting in hospitalization more than 3 years after CPI initiation.This suggests that severe irAEs can occur outside of the expected time window and thus clinicians should keep the possibility of an irAE in mind for all patients who have received CPI therapy.These patterns may help clinicians better recognize the expected time window for severe irAE onset, guiding earlier detection and prevention of hospitalization.
There are several limitations of the present study.Although this cohort was drawn from many years of patients treated at UCSF, the number of patients hospitalized for irAEs was small (n = 114), limiting the power to detect differences between subgroups and specifically by cancer type.Thus, only groups with greater than 10 patients were included in the subgroup analysis.Additionally, the findings of this singleinstitution study performed at an academic medical center may have limited generalizability to other practice settings.Since this study did not have a control group, further studies are needed to understand the difference in survival outcomes between patients who are treated with CPI therapy without a hospitalization compared to the patients in this study who are hospitalized for irAEs.Additionally, since there were only a small number of patients treated with CTLA-4 monotherapy, comparison by CPI type was limited.Lastly, this study cannot draw conclusions related to the cause of death because the study did not include the cause of death as a variable during chart review and data extraction.

Conclusions
As CPI therapy is used more frequently, hospitalizations from irAEs will continue to increase.This study's findings suggest that among patients hospitalized due to irAEs, survival may differ by irAE and cancer type, with particularly short survival for patients with lung cancer and/or irAE pneumonitis.This real-world data can contribute to clinical models that assess the outcomes of hospitalization and the risk of death due to severe irAEs, which may inform patient counseling and treatment decision-making.

Figure 1 .
Figure 1.Inclusion criteria for the study cohort. e952

Figure 2 .
Figure 2. (A) Distribution of cancer type for patients hospitalized for immune-related adverse events (irAEs).(B) Distribution of Hospitalized irAEs.In parenthesis, the number per irAE describes the total number of hospitalizations combining hospitalizations for single-system irAE with multisystem irAEs.For example, for cardiovascular there were 4 hospitalizations for cardiac irAE alone and 10 hospitalizations with cardiac irAEs combined with other system irAEs.A hospitalization within 1 system could have multiple specific irAEs within that system in 1 hospitalization.Number of hospitalizations in italics describe irAE hospitalizations with only that system.

Figure 3 .
Figure 3. (A) Overall survival from time of hospitalization of the entire cohort.(B) Overall survival of immune-related adverse event (irAE)-related hospitalizations by irAE type.(C) Overall survival for irAE-related hospitalizations by cancer type.(D) Overall survival of patients hospitalized for irAE by checkpoint inhibitor type.A sensitivity analysis was performed with the same models but excluded patients with non-metastatic disease which had similar results to the overall cohort (Supplementary Figs.4A-4D).

Table 1 .
Baseline demographic and clinical information by CPI type.