Impact of immune-related toxicity on survival and treatment response in non-small cell lung cancer patients.

Impact statements
Patients with NSCLC who develop irAEs during ICI therapy experience significantly improved survival outcomes and better treatment responses, highlighting the potential predictive value of irAEs for clinical benefit.
The findings suggest that monitoring and managing irAEs could become integral to optimizing therapeutic strategies, enabling clinicians to balance treatment efficacy with safety.
These results provide evidence supporting the use of irAEs as potential biomarkers to guide personalized treatment plans, improving patient stratification for immunotherapy.
Pharmacists and clinicians can leverage these insights to improve patient counseling and proactive management of irAEs, reducing unnecessary treatment discontinuations and enhancing overall outcomes.
The study underscores the importance of interdisciplinary collaboration in managing irAEs and tailoring care for NSCLC patients undergoing immunotherapy.

Introduction
Non-small cell lung cancer (NSCLC) accounts for approximately 85% of all lung cancer diagnoses and is a leading cause of cancer-related mortality on a global scale [1]. Despite significant advancements in the field of early diagnosis and the development of targeted therapies for specific mutations, the prognosis for advanced NSCLC patients remains limited, with high rates of disease progression and mortality [2]. The introduction of immunotherapy, particularly through immune checkpoint inhibitors, profoundly transformed the therapeutic landscape, offering an effective treatment option even for patients without actionable mutations.
Programmed cell death protein 1 (PD-1) and programmed death-ligand 1 (PD-L1) inhibitors block the interaction that allows tumor cells to evade immune system attack. These agents have significantly improved survival in various clinical trials, establishing a new standard of care. However, immunotherapy is not without challenges and prolonged immune activation can trigger immune-related events (irAEs). The frequency of irAEs is determined by both, the drug used and the specific characteristics of the patient, ranging from mild reactions to severe complications affecting multiple organs, mainly the endocrine, pulmonary, dermatologic, and gastrointestinal systems. Other less frequent immune-mediated adverse effects include renal failure, neurological effects, rheumatological effects, ocular effects and, rarely, cardiac effects [3]. Nevertheless, the incidence of fatal adverse events associated with ICIs is estimated to be between 0.3% and 1.3% [4]. This risk is lower than that associated with conventional chemotherapy. It has been described that the two anti PD-1 agents pembrolizumab and nivolumab have similar toxicity profile. Anti PD-1-related irAEs are less frequent and differ in the organs affected compared to anti CTLA-4, i.e. ipilimumab. Approximately 10% of patients on anti-PD-1 therapy suffer grade ≥ 3 irAEs and 5–20% have irAEs of any grade. Most irAEs occur in the first 6 months of treatment. Toxicities from anti-PD-1 or anti-PD-L1 drugs may take longer to resolve than those from ipilimumab, although there are no direct comparisons of the time to resolution of irAEs caused by the different drugs [5].
Recent studies suggest that the occurrence of these toxicities may be associated with greater immune activity, raising the hypothesis that certain toxicities could indicate improved survival and higher therapeutic response [6, 7]. It is hypothesized that this connection is the result of the underlying mechanism of immune checkpoint inhibitors (ICIs), which function by enhancing T-cell activation and preventing tumor immune evasion. When this heightened immune response effectively targets tumor cells, it may simultaneously induce off-target effects, leading to irAEs.
While the impact of immunotherapy has been extensively documented in terms of overall survival (OS) and progression-free survival (PFS), evaluating the objective response rate (ORR) and disease control rate (DCR) is increasingly relevant for understanding the real-world effectiveness of treatment [8, 9]. ORR, which includes complete and partial responses to treatment, and DCR, which also accounts for disease stabilization, have become key indicators for measuring the efficacy of immunotherapy. In several studies, ORR and DCR have shown correlations with better clinical outcomes, reinforcing their role as predictive factors for therapeutic success in NSCLC [8, 10].
Understanding these potential correlations is critical for optimizing clinical management, improving patient selection, and developing strategies that balance clinical benefit with treatment safety. Furthermore, these insights could facilitate the identification of patient subgroups with a higher likelihood of benefiting from immunotherapy, enabling more personalized medicine.

Aim
The present study analyzes the association between the presence of immune-related toxicities and outcomes in terms of survival (OS and PFS) and treatment response (DCR and ORR) in NSCLC patients treated with immunotherapy, assessing their impact on treatment effectiveness. This approach will provide real-world data to determine whether treatment toxicity can be considered not only as an indicator of clinical response but also as a potential predictive biomarker of effectiveness. This would open new perspectives for more personalized medicine.

Method
This retrospective observational study was conducted at the University Hospital of La Ribera (Alzira, Valencia, Spain) and included patients treated between September 2019 and December 2023. The study was performed in accordance with institutional standards for observational research. The study database was fully anonymized prior to analysis to ensure patient confidentiality.
Consecutive adult patients with advanced or metastatic NSCLC treated with ICIs, either as monotherapy or in combination with chemotherapy, were eligible for inclusion. All patients who received at least one dose of immunotherapy were included. Patients were excluded if they had incomplete baseline clinical data, lacked follow-up information necessary for survival analysis, or received immunotherapy outside the predefined study period.
Baseline demographic and clinical variables collected included age, sex, ECOG performance status, PD-L1 expression level, line of treatment, and treatment type (immunotherapy monotherapy versus immunotherapy combined with chemotherapy).
PD-L1 expression was assessed in tumor tissue using immunohistochemistry according to routine clinical practice at the institution. PD-L1 testing was performed using the 22C3 pharmDx antibody, and expression was quantified using the tumor proportion score (TPS). PD-L1 expression was categorized as < 1%, 1–49%, and ≥ 50%, in accordance with current clinical guidelines. Cases in which PD-L1 assessment was not feasible or results were unavailable were classified as not assessable.
IrAEs were identified through systematic review of medical records and classified according to the affected organ system. Toxicity severity was graded using the Common Terminology Criteria for Adverse Events v6.0 (CTCAE) [11]. Patients were categorized according to the presence or absence of irAEs, the number of irAEs (one versus ≥ 2), and toxicity severity (grades 1–2 versus grade ≥ 3).
The primary objective of the study was to evaluate the association between immune-related toxicity and survival outcomes, with PFS and OS as primary endpoints. Secondary objectives included assessing the relationship between immune-related toxicity and treatment response, measured by DCR and ORR.
Continuous variables were summarized as mean ± SD or median (IQR), as appropriate, while categorical variables were reported as frequencies and percentages and compared using chi-square tests. Survival curves were estimated using the Kaplan–Meier method and compared with the log-rank test. Multivariable Cox proportional hazards models were used to evaluate the independent association between irAEs and PFS and OS, adjusting for sex, age, ECOG, PD-L1 expression, line of treatment, and treatment type.
Given the retrospective nature of the study and the lack of precise information on the timing of irAE onset, immune-related toxicity was analyzed as a time-fixed covariate. However, follow-up was initiated at the time of first immune checkpoint inhibitor administration, aligning treatment assignment with time zero and thereby minimizing the risk of immortal time bias by design [12, 13].
Associations between immune-related toxicity and treatment response outcomes (DCR and ORR) were assessed using multivariable logistic regression models with identical adjustment variables. All statistical tests were two-sided, with p < 0.05 considered statistically significant. Statistical analyses were performed using SPSS software (version 25.0).

Results
The study included 174 NSCLC patients with a median age of 65.4 years (IQR 44.6), of whom 82.2% (n = 143) were men and 17.8% (n = 31) were women. Among the participants, 92.5% were smokers, and 70.1% were heavy smokers (> 40 pack-years).
The most common NSCLC histology was non-squamous (64.9%), followed by squamous (35.1%). Most patients present metastatic disease at treatment initiation (81.6%), with bone and lung metastases being the most frequent. Regarding PD-L1 expression, 35.1% of patients had PD-L1 expression < 1%, and 33.9% had PD-L1 ≥ 50% (Table 1).

Regarding treatment, most patients received pembrolizumab (63.8%), followed by nivolumab (19.5%), atezolizumab (13.8%), and cemiplimab (2.9%). Immunotherapy was predominantly administered as monotherapy (69.5%) and first-line setting (50.6%) (Table 1).
IrAEs occurred in 42.0% (n = 73) of patients treated with immunotherapy, with 83.6% (n = 61) being grade 1–2 and 16.4% (n = 12) grade ≥ 3. Endocrine toxicity was the most frequent (23.3%), particularly thyroid alterations (16.4%), followed by dermatologic (9.6%), pulmonary (9.6%), hepatic (6.8%), rheumatologic (5.5%), and gastrointestinal (colitis and dysgeusia) toxicities (5.5%). Renal (1.4%) and cardiac (1.4%) irAEs were less common. Treatment discontinuation due to irAEs occurred in 12.1% (n = 21) of patients, with no deaths attributed to immune-related toxicity.
A detailed analysis of the results was performed in order to assess whether the incidence of toxicity and its severity could be related to gender, age, smoking habit, presence of autoimmune diseases, tumor histology, PD-L1 expression level and brain metastases, IT treatment characteristics (line of treatment, type of treatment (IT monotherapy or in combination with QT) and type of drug (pembrolizumab, nivolumab or atezolizumab). However, no significant differences were observed in any case.
The median PFS in the overall cohort was 11.2 months (IQR 2.3; 95% CI 5.8–16.3), and the median OS was 16.1 months (IQR 2.0; 95% CI 11.9–20.2). Kaplan–Meier analyses showed significant differences in PFS and OS according to the occurrence of immune-related adverse events (log-rank p < 0.001; Figs. 1 and 2). Patients who developed irAEs showed substantially longer median PFS (32.3 vs. 6.2 months) and OS (37.4 vs. 9.8 months) than those without irAEs, with a median follow-up of 26.9 months.

In multivariable Cox proportional hazards models adjusted for sex, age, ECOG performance status, PD-L1 expression, line of treatment, and treatment type, the occurrence of immune-related toxicity remained independently associated with improved survival outcomes. Specifically, toxicity was associated with a 48% reduction in the risk of disease progression (HR 0.52) and a 43% reduction in the risk of death (HR 0.57) (Table 2).

Regarding treatment efficacy, patients who experienced irAEs demonstrated significantly improved response outcomes in multivariable analyses (Tables 2 and 3). The DCR was significantly higher among patients with toxicity (adjusted OR 3.21; 95% CI 1.34–9.46). A similar association was observed for ORR, which was also significantly increased in patients who developed toxicity compared with those who did not (adjusted OR 2.81; 95% CI 1.13–7.43).
Subgroup analyses were performed to explore the relationship between specific irAE categories and survival and treatment response. With the exception of gastrointestinal toxicity, most irAE subtypes were consistently associated with longer survival and improved treatment responses compared with the absence of toxicity (Table 2). Due to the limited number of events, formal statistical analyses were not feasible for renal and pericardial toxicities.
No statistically significant differences in survival or treatment response were observed when comparing patients with a single irAE to those experiencing two or more irAEs, irrespective of the organ system involved. Similarly, no statistically significant associations were identified between toxicity severity (grades 1–2 versus grade ≥ 3) and PFS, OS, DCR, or ORR (Table 2). However, the number of patients experiencing severe toxicity (grade ≥ 3) was limited (n = 12), which may have reduced the statistical power to detect clinically meaningful differences. Therefore, the absence of statistically significant associations should be interpreted with caution, as a potential Type II error cannot be excluded.

Discussion
The present study provides real-world evidence supporting a strong association between the development of irAEs and improved clinical outcomes in patients with advanced NSCLC treated with ICIs. In this retrospective cohort, the occurrence of irAEs was independently associated with prolonged PFS and OS, as well as with higher DCR and ORR, even after adjustment for clinically relevant confounders, including PD-L1 expression, ECOG performance status, line of therapy, and treatment type.
From a mechanistic perspective, this association is biologically plausible and increasingly supported by translational studies [14, 15]. ICIs enhance antitumor immunity primarily through the reinvigoration of exhausted CD8⁺ cytotoxic T lymphocytes and the modulation of CD4⁺ T-cell subsets, including Th1 and Th17 populations, which are essential for effective tumor cell recognition and elimination. However, this heightened immune activation may also disrupt peripheral immune tolerance, leading to off-target immune responses against normal tissues and the development of irAEs [16, 17]. Several studies have demonstrated that patients who develop irAEs exhibit enhanced immune activation, characterized by increased effector T-cell infiltration, greater T-cell receptor clonality, and shared antigen recognition between tumor tissue and irAE-affected organs [15, 18]. These findings support the concept that irAEs may reflect a state of sustained and effective antitumor immune engagement rather than merely treatment-related complications.
The survival outcomes observed in our cohort are consistent with previously published real-world studies and meta-analyses. The overall median PFS and OS were comparable to those reported in large observational datasets, such as those by Pérol et al. [19] and Cook et al. [20], reinforcing the external validity of our population. The incidence of irAEs in the present study (42.0%) also falls within the expected range reported in the literature, with most events being mild to moderate and a minority classified as grade ≥ 3. This distribution aligns with prior reports, including the meta-analyses by Berti et al. [21] and Huang et al. [22], which describe irAE rates of approximately 35–40% for any grade and 10–20% for severe toxicity.
Patients who developed irAEs exhibited substantially longer median PFS and OS compared with those who did not experience toxicity. These findings are in line with the majority of published meta-analyses demonstrating a consistent association between irAE occurrence and improved survival outcomes in NSCLC patients treated with ICIs [6, 7, 22–26]. Although some variability in the magnitude of benefit has been reported across studies, such differences are likely explained by heterogeneity in patient populations, treatment settings, and study design. Importantly, the association between irAEs and favorable outcomes remained statistically significant after multivariable adjustment in our analysis, indicating that this relationship is independent of established prognostic and predictive factors, including PD-L1 expression and line of therapy.
Consistent with the survival benefit observed, immune-related toxicity was also significantly associated with improved treatment response. Patients who developed irAEs exhibited higher rates of disease control and objective response, with adjusted odds ratios of 3.21 for DCR and 2.81 for ORR. These effect sizes are closely aligned with pooled estimates reported in previous meta-analyses and large observational studies, including those by Wang D et al. [23] and Lin L et al. [26], further supporting the robustness and reproducibility of the association between irAE occurrence and enhanced antitumor activity (Table 4).
PD-L1 expression is a well-established predictive biomarker for ICI efficacy in NSCLC and represents a potential confounder when evaluating treatment outcomes [27, 28]. In the present study, PD-L1 expression was systematically assessed using a standardized immunohistochemical assay and incorporated into all multivariable survival and response models. Importantly, the association between irAEs and improved PFS, OS, DCR, and ORR remained statistically significant after adjustment for PD-L1 expression levels. This finding suggests that the observed clinical benefit associated with irAEs is not merely a surrogate for PD-L1–high tumors, but rather reflects an independent immunological phenomenon.
Regarding irAE subtypes, endocrine, dermatologic, pulmonary, hepatic, and rheumatologic toxicities were the most frequently observed. Consistent with prior studies, several of these toxicity categories were associated with improved survival and treatment response [6, 7, 22–26]. However, gastrointestinal toxicity did not demonstrate a significant association with outcomes in our cohort, likely due to the low incidence of such events. Conversely, hepatic and pulmonary toxicities were associated with favorable outcomes in our analysis, a finding that contrasts with some previous meta-analyses [6, 7, 22–26] and should be interpreted cautiously given the limited sample size and low event frequency.
No statistically significant differences in survival or treatment response were observed according to the number of irAEs (one versus ≥ 2) or toxicity severity (grades 1–2 versus grade ≥ 3). However, the number of patients experiencing severe toxicity was small, limiting the statistical power to detect meaningful differences. Therefore, the absence of significant associations in these subgroup analyses should be interpreted with caution.
Several limitations of this study should be acknowledged. First, the retrospective observational design inherently entails risks of selection bias, residual confounding, and incomplete data capture, which may influence the interpretation of the results. Although multivariable adjustment was performed to control for relevant clinical confounders, the possibility of unmeasured variables affecting outcomes cannot be fully excluded.
Second, irAEs represent a time-dependent exposure, and treating their occurrence as a fixed baseline variable may introduce immortal time bias by inherently guaranteeing survival time for patients who develop toxicity. Due to the lack of precise individual-level data regarding the timing of irAE onset, it was not possible to implement a fully time-dependent Cox regression model or to perform a landmark analysis.
Finally, this was a single-center study, which may limit the generalizability of the findings to other populations or healthcare settings. Nevertheless, the real-world nature of the cohort and the consistency of the results with previously published studies support the external validity of the observed associations.

Conclusions
This study demonstrates a consistent association between the development of irAEs and improved survival and treatment response in patients with NSCLC receiving ICIs in a real-world setting. Although irAEs cannot be regarded as definitive surrogate markers of efficacy, their reproducible correlation with favorable outcomes suggests that they may reflect underlying immune activation processes relevant to therapeutic benefit. Further prospective and translational investigations are required to better characterize the immunological determinants of irAE development and to define their potential role in informing personalized immunotherapy strategies.

Supplementary Information