Relationship between immune-related adverse events and treatment outcome of immune checkpoint inhibitors for advanced non-small cell lung cancer: a retrospective cohort study.

Introduction
The treatment of advanced, driver-negative non-small cell lung cancer (NSCLC) has expanded with the development of immune checkpoint inhibitors (ICIs) (1-6). These cancer therapeutics were developed as a result of a deeper understanding of immune system checkpoints, such as programmed cell death 1/programmed death ligand 1 (PD-1/PD-L1) and cytotoxic T-lymphocyte antigen 4 (CTLA-4). Cancer cells use these checkpoints to avoid being detected by the immune system. ICIs work by blocking these pathways, boosting the immune system’s ability to find and destroy cancer cells (7-9).
ICIs like nivolumab, pembrolizumab, and atezolizumab have shown better results than traditional therapies in treating advanced, driver-negative NSCLC. These strategies have improved both the overall survival (OS) and progression-free survival (PFS) of patients (1-6). Additionally, some patients with advanced NSCLC experience a lasting response to ICIs, leading to long-term control of the disease and even complete remission (10).
Despite the benefits of ICIs, these treatments are associated with several challenges. ICIs can cause immune-related adverse events (irAEs), including inflammation of organs or tissues. These side effects require careful monitoring and management by healthcare providers. Furthermore, patients treated with ICIs may show a good response while also experiencing irAEs at the same time, because the treatment effects of ICIs and the occurrence of irAEs share a similar mechanism (11-13).
Here, we retrospectively analyzed the outcomes of NSCLC patients treated with ICIs, with a focus on the relationship between the incidence and severity of irAEs and patient outcomes. This study centers on the clinical implications of irAEs on NSCLC patient survival following ICI treatment.
Therefore, the objective of this study was to clarify the relationship between the occurrence and severity of (irAEs and clinical outcomes, including OS and treatment response, in patients with advanced NSCLC treated with ICIs.We hypothesized that patients who experienced irAEs would have better treatment outcomes than those who did not. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1434/rc).

Methods

Patients
This was a retrospective cohort study including consecutive patients who were treated with ICIs for advanced NSCLC at Oita University Faculty of Medicine (Yufu, Japan) between December 2015 and May 2022. Patients were eligible if they had (I) histologically confirmed NSCLC; (II) received at least one cycle of PD-1/PD-L1 inhibitor; and (III) had complete clinical and survival data available. The clinical and pathological stage was defined using the 8th edition of the American Joint Committee on Cancer lung cancer staging system (14). Histological classification was carried out following the 2021 World Health Organization Classification of Tumors of the Lung, Pleura, Thymus and Heart (15). Patients who received ICI for curative intention (e.g., adjuvant therapy after complete resection, consolidation therapy after chemoradiation) were excluded. Patients were followed up until death or last clinical contact, and clinical data were obtained from electronic medical records. Patients with incomplete key clinical data were excluded from analysis.
Data on clinicopathological factors, the presence or absence of driver mutation, tumor PD-L1 expression, regimen and lines for therapy, including ICI, irAEs, and prognosis, were collected and analyzed. Adverse events were determined using the Common Terminology Criteria for Adverse Events (CTC-AE) version 5.0. Treatment response was determined by physicians following the revised Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 (16).

Physician decision of treatment
PD-L1 expression was analyzed immunohistochemically using 22C3 antibody (Agilent Technologies, Santa Clara, CA, USA), and the tumor proportion score (TPS) was determined (SRL Inc., Tokyo, Japan). The presence or absence of driver mutation was determined by multiplexed analyses, including Oncomine DxTT (ThermoFisher Scientific, Waltham, MA, USA) or AmoyDx (Amoy Diagnostics, Shanghai, China); individual gene analysis, including EGFR gene mutation analysis, was performed using cobas v2 (Roche Diagnostics, Basel, Switzerland). Driver mutation test was not mandatory in cases with non-adenocarcinoma histology. Treatment decision was made by a multidisciplinary conference following Japanese clinical guidelines.
The irAEs observed in the patient group included endocrine disorders (hypothyroidism, hypopituitarism, adrenal insufficiency, and diabetes mellitus), rash, gastrointestinal events, liver dysfunction, dermatomyositis, arthritis, and interstitial lung disease (ILD). Diagnosis was determined by physicians on the basis of physical examination and laboratory tests.

Statistical analysis
Medical records were reviewed and collected, and Fisher’s exact test was used for comparing division tables. Quantitative variables such as age and PD-L1 expression were analyzed as continuous variables. For subgroup comparisons, PD-L1 TPS was categorized as <50% or ≥50% according to clinical relevance. Performance status (PS) was categorized as 0–1 versus ≥2. Clinical stage was classified according to the 8th edition of the TNM system and analyzed as stage II/III versus IV. The treatment line was categorized as first-line versus second or later lines. The Kaplan-Meier method was used to estimate survival curves, and log-rank test was used for comparison. A Cox proportional hazard model was used to determine independent factors for survival. A P value <0.05 indicated statistical significance. All statistical analyses were performed using EZR version 1.55 (17).
To minimize selection bias, all consecutive patients who received ICI therapy during the study period were included, therefore, no formal sample size calculation was performed.

Ethical information
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board at Oita University Faculty of Medicine (IRB No. 2437), and informed consent was obtained from each patient.

Results

Patient characteristics and irAE incidence
This study included 144 patients with advanced NSCLC who received treatment with ICIs and no patients were excluded due to missing data. The median patient age was 69 years (range, 35–91 years) and most patients were male (n=115) (Table 1). The histology was adenocarcinoma in 83 patients, squamous cell carcinoma in 44 patients, and others in 17 patients. Most cases (n=113) were clinical stage IV. EGFR mutation was present in 11 patients, and KRAS mutation was observed in 1 patient. The PS was 0 or 1 in 124 patients. PD-L1 expression was tested in 124 patients, and 63 patients showed high PD-L1 (TPS ≥50%).
The ICI used was pembrolizumab in 98 patients, nivolumab in 28 patients, and atezolizumab in 18 patients. ICIs were administered as monotherapy in 103 patients and in combination with other agents, mainly chemotherapy, in 41 patients. In the combination group, 4 patients received the combination of nivolumab and ipilimumab. Among the 144 patients, 87 patients were treated with ICIs as a first-line therapy, while 22 patients were treated with ICIs as third-line or later (Table 1).
The median follow-up time was 9.5 months (range, 0.4–73.3 months). Among the 144 patients, 35 patients (24%) experienced one or more irAE during the study period. The incidence of irAE was significantly higher in patients with PD-L1 ≥50% than those with PD-L1 <50% (38.1% and 16.4%, P<0.01) (Table 1). There were no significant relationships observed between irAE and age, sex, PS, stage, or histology.

Types and toxicity grades of irAEs
During the study period, 44 irAE events were observed in 35 patients (Figure 1). Among the 35 patients, 27 experienced only one type of irAE and 8 patients experienced two types of irAEs. One patient experienced rash (grade 3), hypothyroidism (grade 1), and colitis (grade 2) serially; the patient was on maintenance atezolizumab. The most common type of irAE was endocrine disorder (n=15, 10%), followed by ILD (n=13, 9%), rash (n=8, 6%), hepatic event (n=3, 2%), gastrointestinal event (n=2, 1%), dermatomyositis (n=2, 1%) and arthritis (n=1, 1%). Most irAE events were grade 3 or less (Table 2). A total of 12 grade ≥3 irAEs occurred in 10 patients (6.9%), of which ILD was the most frequent (n=7, including one grade 5 case). ILD accounted for the majority of treatment discontinuations due to toxicity (Table 2).

Treatment response to ICI in accordance with the occurrence of irAE
In the overall patient group, complete response was achieved in 10/144 patients (7%), partial response in 36/144 (25%), stable disease in 30/144 (21%), and disease progression in 57/144 (40%) (Table 3). The overall response rate (ORR) was 32% and disease control rate (DCR) was 55%.
We further analyzed the relationship between the occurrence of irAEs and treatment response. The treatment response was significantly better in patients with irAEs than in those without irAEs (P=0.008, Table 3); the ORR and DCR were 56% and 77% in patients with irAEs and 24% and 48% in patients without irAEs, respectively.
OS was also significantly longer in patients with irAEs than those without irAEs (median OS: not reached and 9.7 months, P<0.001) (Figure 2A). Multivariate analysis showed that the occurrence of irAE was an independent prognostic factor for better OS [hazard ratio (HR) 0.39, 95% confidence interval (CI): 0.22–0.70, P=0.002] (Table 4).
Finally, we performed survival analysis in patients stratified by the type of irAE. The patients were divided into the three most common irAE types: endocrine disorder (n=15), ILD (n=13), and rash (n=8). Regarding OS, endocrine disorder demonstrated a statistically significant association with better outcomes (P=0.02, Figure 2B), while rash did not reach statistical significance; however, it showed a favorable trend (P=0.08, Figure 2C). The presence of ILD was not predictive of better survival (P=0.17, Figure 2D). No sensitivity analyses were conducted.

Discussion
In this retrospective study of advanced, driver-negative NSCLC patients treated with ICIs, we observed that the development of irAEs was significantly associated with improved treatment responses and prolonged OS (Figure 2). Patients who experienced irAEs exhibited higher response rates (56% vs. 24%) and markedly superior survival outcomes compared with patients without irAEs (Table 3). Multivariate analysis confirmed that the occurrence of irAEs was an independent prognostic factor for survival, underscoring the potential role of irAEs as surrogate markers of an effective anti-tumor immune response (Table 4).
Our findings are in concordance with several recent reports that have suggested a positive correlation between irAE occurrence and favorable clinical outcomes in NSCLC patients receiving ICIs (18,19). The association of high PD-L1 expression with an increased incidence of irAEs observed in our cohort further supports the notion that heightened tumor immunogenicity may predispose ICI-treated patients to both robust anti-tumor activity and immune toxicity (20). This is consistent with emerging evidence indicating that an “activated” immune microenvironment, as reflected by high PD-L1 expression, may amplify both therapeutic efficacy and the risk of irAEs (21).
A particularly notable aspect of our study is the differential impact of irAE type on survival outcomes. Patients who developed endocrine disorders or rash tended to show improved survival, while the presence of ILD did not confer a similar benefit. This disparity may be attributed to the distinct immunopathological mechanisms underlying these toxicities. Endocrine and cutaneous irAEs may reflect a systemic immune activation that is beneficial for tumor control, whereas ILD may represent a more deleterious off-target effect that adversely impacts pulmonary function and overall patient health (22,23). Moreover, Japanese patients are known to have a higher susceptibility to ICI-related ILD compared with Western populations, making it particularly important to determine whether ILD-type irAEs truly reflect enhanced anti-tumor immunity or merely represent ethnicity-associated immune toxicity (24). Recent studies have similarly highlighted that the organ-specific manifestation of irAEs can differentially influence patient outcomes, thereby emphasizing the importance of detailed toxicity profiling in the era of immunotherapy (25).
Clinically, these results suggest that the occurrence of irAEs may serve as a useful biomarker for predicting treatment efficacy. However, the management of irAEs remains a significant challenge. While prompt recognition and intervention are crucial to mitigate toxicity, it is equally important to preserve the anti-tumor immune response. Recent evidence has indicated that early, judicious use of immunosuppressive agents such as corticosteroids does not compromise the therapeutic benefit of ICIs (26). Thus, developing standardized management protocols for irAEs will be critical to optimizing the therapeutic index of ICIs in NSCLC.
Despite the promising implications of our findings, several limitations must be acknowledged. The retrospective design and relatively small sample sizes of certain subgroups may have introduced selection bias and limited the statistical power to detect subtle differences among irAE types. Additionally, the heterogeneity of treatment regimens and the lack of standardized criteria for irAE assessment across centers may confound the interpretation of our results. And more, because patients with longer survival have a higher likelihood of developing irAEs over time, a potential reverse causation cannot be entirely excluded. Landmark analysis was not performed, and this should be considered when interpreting irAEs as prognostic indicators. Future prospective studies with larger cohorts and integrated translational analyses are warranted to validate our observations and to elucidate the underlying immunological mechanisms linking irAE development to improved clinical outcomes (27,28).
Although the present study was conducted at a single center in Japan, the results may be generalizable to similar real-world populations of patients with advanced NSCLC treated with ICIs in other settings.

Conclusions
Our study demonstrates that the occurrence of irAEs is associated with enhanced response and survival in patients with advanced NSCLC undergoing ICI therapy. The differential prognostic impact observed among various irAE types highlights the need for tailored management strategies. Continued investigation into the immunobiology of irAEs will not only refine patient selection and management but also potentially guide the development of combination strategies that maximize therapeutic efficacy while minimizing adverse effects.

Supplementary
The article’s supplementary files as