Immune-Related Adverse Events and Survival Among Patients With Metastatic NSCLC Treated With Immune Checkpoint Inhibitors

Key Points Question Are immune-related adverse events associated with improved overall survival in patients with locally advanced or metastatic non–small cell lung cancer receiving immune checkpoint inhibitor (ICI) therapy? Findings In this cohort study of 803 patients, developing an immune-related adverse event mandating delay or discontinuation of ICI therapy and/or systemic corticosteroids for management of toxic effects was associated with significantly improved median overall survival. This association was not compromised by hospitalization for management of toxic effects of severe immune-related adverse events. Meaning These findings suggest that immune-related adverse events are associated with improved overall survival in patients receiving ICI therapy, even in the context of hospitalization for management of toxic effects.


Introduction
Globally, lung cancer remains the leading cause of cancer-related death. 1 Improvements are being made, however, and overall mortality has declined over the past decade.This decline has largely been driven by treatment advances in non-small cell lung cancer (NSCLC), with immune checkpoint inhibitors (ICIs) improving clinical outcomes. 2Several clinical trials examining ICIs as monotherapy [3][4][5][6][7] or in combination with chemotherapy, [8][9][10][11] angiogenesis inhibitors, 12 and a second ICI agent [13][14][15] have demonstrated improved time to event end points and response rates.The treatment landscape in metastatic NSCLC has thus been transformed.
It is well established that although ICIs improve clinical outcomes in NSCLC, they are complicated by immune-related adverse events (irAEs), the incidence of which varies by study setting, patient selection, treatment parameters, and irAE grading.  Theris growing recognition across solid malignant neoplasms that irAEs affect overall survival (OS).In advanced or metastatic NSCLC, the data suggest that developing an irAE is associated with improved OS.]52 We examined a contemporary multicenter cohort of patients with locally advanced or metastatic NSCLC treated with ICI alone or combined with chemotherapy, agnostic to treatment line.
We investigated the association of survival outcomes with (1) irAEs mandating delay or discontinuation of ICI therapy and/or systemic corticosteroids for management of toxic effects (hereinafter termed clinically meaningful irAEs) and (2) hospitalization for irAEs.

Study Design
In this cohort study, we examined data from the Alberta Immunotherapy Database, a contemporary provincial multicenter observational cohort nested in routine clinical practice.Using a standardized template, the Alberta Immunotherapy Database captures information from patients treated with ICI, including baseline demographic characteristics (age and gender), tumor histology, and clinical, laboratory, and imaging data. 53Study approval was obtained from the Health Research Ethics Board of the Alberta Cancer Committee, which waived the need for patient consent given the low-risk, deidentified, retrospective nature of the study.The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Patients were included if they had metastatic NSCLC or unresectable locally advanced disease ineligible for curative intent chemoradiotherapy and/or durvalumab.Additional study inclusion criteria were age 18 years or older and completion of at least 1 cycle of ICI therapy (atezolizumab, nivolumab, or pembrolizumab alone or combined with chemotherapy) of noncurative intent, agnostic to treatment line.Age and gender are included in the baseline demographic data captured through the Alberta Immunotherapy Database, which relies on information collected through our electronic health records system and currently does not include race or ethnicity.Response Evaluation Criteria in Solid Tumours, version 1.1, prior to data analysis. 54Where treatment duration data were missing, patients were excluded from survival analysis.

Outcomes of Interest
We investigated patients with locally advanced or metastatic NSCLC who met the study inclusion criteria.Clinically meaningful irAEs were identified.As detailed in Figure 1, irAEs were categorized as thyroiditis, pneumonitis, myositis, adrenal insufficiency, hepatitis, dermatitis, colitis, carditis, and hypophysitis.Less commonly occurring irAEs were collectively categorized as "other."Clinically meaningful irAEs were defined as mandating delay or discontinuation of ICI therapy and/or  Electronic medical records were reviewed to identify irAEs during ICI therapy.Only clinically meaningful irAEs as described above were recorded.Where multiple irAEs occurred in the same patient, the most clinically meaningful irAE was included.

Statistical Analysis
Statistical analysis was completed April 26, 2023.Both the χ 2 test and the Fisher exact test were used to identify baseline characteristics that were prognostic factors associated with developing an irAE.
To mitigate immortal time bias from patients with a poor prognosis who may have died before developing an irAE, a 12-week cutoff was used for OS and TTNT analyses.This cutoff was selected because it corresponds to the median time (3.4 [IQR, 1.4-8.4]months) to irAE development in our study.Thus, patients dying within 12 weeks of ICI therapy initiation were excluded from Kaplan-Meier analysis (compared using a log-rank test).We defined OS as time from ICI therapy initiation to date of death or last censored follow-up.We defined TTNT as time from ICI initiation to starting the next line of systemic therapy, death, or last censored follow-up.A multivariable Cox proportional hazards regression model was applied to OS and TTNT analysis to adjust for significant differences in baseline characteristics, including treatment line and use of ICI alone or combined with chemotherapy.
Where data were missing, the case deletion method was used.All statistical tests were 2 sided.
Results were considered statistically significant at P Յ .05.SAS Cloud of Academics, version 9.4 (SAS Institute Inc) was used for analysis.

Characterization of irAEs
A total of 297 patients were diagnosed with a clinically meaningful irAE (37.0%).The median time to developing an irAE was 3.

Survival Outcomes and Time to Next Treatment After irAE Development and Hospitalization
The median OS for the total cohort was 15.  3A).For patients with a clinically meaningful irAE, no association in median TTNT was seen between those hospitalized and those treated as outpatients (15.3 CI, 9.6-24.8]vs 15.9 [95% CI, 13.6-18.0]months; P = .67)(Figure 3B).

Variables Associated With OS by Cox Proportional Hazards Regression Analysis
To further evaluate the association between potential prognostic factors and OS among patients surviving to 12 weeks, Cox proportional hazards regression analysis was performed (

Discussion
To our knowledge, this study represents the largest contemporary clinical dataset evaluating the association between irAEs and survival in locally advanced or metastatic NSCLC, agnostic to ICI agent, delivery as monotherapy or combined with chemotherapy, and treatment line.In our multicenter, retrospective population-based cohort, developing a clinically meaningful irAE was Analysis includes patients who survived to at least 12 weeks.A, Full cohort of patients who did or did not develop an immune-related adverse event (irAE) (median OS, 23.   antibodies with CTLA-4 found on healthy tissue, intensifying complement-mediated inflammation.
Ultimately irAE development suggests the host immune system has been activated. 55Whether and how this translates to activation of antitumoral host immunity, with clinical outcomes subsequently affected, remains uncertain across solid malignant neoplasms.7][58] Certainly, in our large cohort, irAE development was associated with improved OS.This supports previously published studies in NSCLC, [31][32][33][49][50][51] recognizing they have largely been small, mainly examined nivolumab monotherapy, and inconsistently defined clinical impact of irAEs.
The role of irAE severity in the survival of patients with NSCLC remains contentious.This may reflect differences in toxicity grading, immunosuppressant use, and hospitalization thresholds.Like other published experiences, 32,33 we found that irAE severity (reflected by management of toxic effects as an outpatient or hospitalized inpatient) did not compromise the survival gains seen with irAE development.However, this finding is not universally supported.A single study 52 demonstrated only high-grade irAEs improve OS in NSCLC.In contrast, pooled data from phase 3 randomized clinical trials evaluating atezolizumab 31 reported only low-grade irAEs are associated with improved survival in advanced NSCLC.Patients with high-grade irAEs had inferior survival except at a 12-month landmark analysis, where median OS was longer than among patients without irAEs. 31Improved OS being limited to patients with low-grade irAEs has been corroborated by several other groups 34,35,59 and may be explained by the life-threatening potential of high-grade events.High-grade irAEs can result in death, albeit rarely.They may also require (permanent) treatment cessation and/or highdose corticosteroids that can suppress antitumor efficacy of ICIs. 31,557][38] However, we only recorded clinically meaningful irAEs per our definition, and studies have shown low-grade events develop earlier than high-grade events. 34,39The incidence of reported irAEs in NSCLC varies widely, ranging from 24.0% to 70.5%.
This reflects heterogeneity including in study setting (observational vs randomized clinical trial), patient characteristics (cancer stage, PD-L1 status, geographical location, race and ethnicity), treatment parameters (ICI agent used either alone or combined with chemotherapy, treatment line), and irAE grading.  In o cohort, the incidence of clinically meaningful irAEs was 37.0%.Nearly one-third of all clinically meaningful irAEs resulted in hospitalization (30.3%) for the management of toxic effects.Given our definition, grade 1 toxic effects, topically treated dermatitis, and mild hypothyroidism would not have been captured.Our incidence rates are thus best compared with irAEs higher than grade 2 and hospitalization rates with irAEs higher than grade 3.
Why irAEs occur in some patients but not others remains poorly understood, but a variety of predictive factors in NSCLC have been described.][20][21][22][23][24][25][26][27][28][29][30]48,[60][61][62] In our cohort, we identified several baseline patient characteristics associated with irAE development: 60 years or older, ECOG performance status 0, high expression of PD-L1, absence of bone metastases, DNLR of 3 or less, and levels of hemoglobin, albumin, and LDH within reference range.Other than response to ICI therapy, irAE development was not associated with treatment-related characteristics (ICI agent, ICI alone or in combination with chemotherapy, and treatment line) in our study population.

JAMA Network Open | Oncology
irAEs and Survival Among Patients With Metastatic NSCLC Treated With ICIs irAEs and Survival Among Patients With Metastatic NSCLC Treated With ICIs Patients received treatment between March 1, 2014, and November 30, 2021.The date of last follow-up was March 31, 2023.Study investigators assessed the objective imaging response using

Figure 1 .
Figure 1.Relative Frequency and Type of Clinically Meaningful Immune-Related Adverse Event by Agent Received

Figure 2 .
Figure 2. Median Overall Survival for Patients With Metastatic Non-Small Cell Lung Cancer

Figure 3 .
Figure 3. Median Time to Next Treatment for Patients With Metastatic Non-Small Cell Lung Cancer

Table 1 .
Baseline Characteristics of Patients With NSCLC With or Without irAEs A total of 803 patients with locally advanced or metastatic NSCLC treated with ICIs with noncurative intent during the period of interest were included in the analysis.A clinically meaningful irAE was confirmed in 297 patients.The median follow-up was 51.9 (95% CI, 49.4-56.2) months.Patient baseline characteristics are summarized in Table 1.The median age of patients who developed an irAE was 69.7 (IQR, 63.1-75.2) years compared with 67.5 (IQR, 60.4-73.3)years among those who did not.Sex distribution was comparable (139 of 295 men [47.1%] and 156 of 295 women [52.(152 of 212 [71.7%] vs 217 of 358 [60.6%];P = .007).Developing an irAE was not associated with patient sex, tumor histology, or the presence of an underlying driver mutation.Regarding the ICI agent, 509 patients received pembrolizumab (of whom 62 also received chemotherapy), 280 received nivolumab, and 9 received atezolizumab.Development of an irAE was not associated with the ICI agent or the treatment line.Whether the ICI was used as monotherapy or combined with chemotherapy was also not associated with developing an irAE.A complete or partial response to ICI treatment was associated with developing an irAE (104 of 255 [40.8%] vs 52 of 315 [16.5%];P < .001).

Table 1 .
Baseline Characteristics of Patients With NSCLC With or Without irAEs (continued) 4 (IQR, 1.4-8.4)months.The most common irAEs were pneumonitis (82 [27.6%]), dermatitis (63 [21.2%]), and colitis (42 [14.1%]).Figure 1 further details the types of irAEs and their relative frequency per ICI agent used in our cohort.b Two-tailed P Յ .05indicates statistical significance.c Lower scores indicate fewer restrictions and capable of more activity.d Patients may have multiple metastatic sites.e Best observed response was categorized per Response Evaluation Criteria in Solid Tumours, version 1.1.

Table 2 .
Multivariable Adjusted Cox Proportional Hazards Regression Analysis of Factors Associated With Overall Survival a Two-tailed P Յ .05indicated statistical significance.