Impact of Immune Checkpoint Inhibitor Therapy on The Overall Survival in Elderly Patients With Non-Small Cell Lung Cancer

Purpose: We analyzed the relationship between a history of immune checkpoint inhibitor and overall survival in patients with non-small cell lung cancer (NSCLC) aged ≥ 70 years. Methods: We conducted a retrospective analysis of the data of patients with NSCLC aged ≥ 70 years old who had received systemic anticancer therapy at our institution between 2015 and 2019. Results: The analysis included the data of a total of 63 patients, including 36 patients who had received immune checkpoint inhibitor therapy and 27 patients who had not received treatment with an immune checkpoint inhibitor. Univariate analysis revealed a longer overall survival in patients who had received treatment with an immune checkpoint inhibitor as compared to those who had not received treatment with an immune checkpoint inhibitor (median: 17.2 vs. 9.8 months; p = 0.026, log-rank test). Multivariate analysis revealed a signicant association between a history of treatment with immune checkpoint inhibitors and the overall survival (hazard ratio, 95% condence interval: 0.42, 0.20-0.86; p = 0.019, Cox proportional hazards model). A signicant interaction was also observed between a history of treatment with an immune checkpoint inhibitor and the tumor histology (p = 0.006), the association between the overall survival and a history of immune checkpoint inhibitor therapy being stronger in the non-small cell lung cancer patients with squamous cell carcinoma than in those with adenocarcinoma. Conclusion: A signicant association between history of immune checkpoint inhibitor therapy and the overall survival was detected in elderly NSCLC patients aged ≥ 70 years old in a clinical practice setting. Our results also suggested that the impact of immune checkpoint inhibitor therapy on the survival differed depending on the tumor histology.


Introduction
In patients with advanced non-small cell lung cancer (NSCLC), systemic drug therapy is the standard of care. Recent studies have shown that treatment with immune checkpoint inhibitors (ICIs) yields a longer survival than treatment with cytotoxic agents in patients with advanced NSCLC [1][2][3][4][5]. In addition, although immune-related adverse events are problematic, ICI therapy has been shown to be associated with a lower frequency of severe adverse reactions as compared to cytotoxic chemotherapy [1][2][3][4][5], and ICIs are expected to be useful agents for the treatment of elderly patients with NSCLC. However, the e cacy and safety of ICI therapy in elderly patients with NSCLC has not yet been clearly evaluated, although subset analysis in patients classi ed by age has been conducted in some previous trials [2][3][4][5].
Pooled analysis in the KEYNOTE study revealed the superiority of pembrolizumab over a cytotoxic agent in terms of the survival in elderly NSCLC patients [6], and a single-arm phase II study showed the e cacy and safety of pembrolizumab treatment in elderly NSCLC patients aged ≥ 75 years old [7]. Furthermore, an analysis of real-world data showed that the e cacy of ICIs did not differ among patient groups divided by age [8]. These previous studies suggest that ICI therapy is effective even in elderly patients with NSCLC in clinical settings.
We conducted this retrospective study and analyzed the effect of immune checkpoint inhibitor therapy on the overall survival in patients with non-small cell lung cancer aged ≥ 70 years old.

Clinical information
We conducted a retrospective analysis of the data of patients with NSCLC who had received anticancer treatment at Toyama University Hospital. The patient inclusion criteria were as follows: 1) cytologically or histologically diagnosed NSCLC; 2) received systemic anticancer treatment, excluding adjuvant chemotherapy and chemoradiotherapy, between 2015 and 2019; 3) aged ≥70 years at the start of the systemic anticancer treatment. The exclusion criteria were as follows: 1) presence of tumor driver mutations; 2) synchronous multiple cancer; 3) history of treatment with an ICI as part of a clinical trial or of combined ICI therapy with cytotoxic agents. The disease stage was evaluated according to the 8th edition of the UICC TNM classi cation and patients who had undergone surgery for NSCLC were reclassi ed based on the ndings at the start of the systemic anticancer treatment. Information about the tumor PD-L1 status was retrieved from the medical charts, testing of which was commissioned to BML (Tokyo, Japan) and performed using 22C3 antibody. Any preexisting interstitial pneumonia was evaluated based on the ndings of chest computed tomography and from the medical charts.
This study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects (Ministry of Health, Labour and Welfare of Japan), after obtaining approval from the Ethics committee, University of Toyama (Reference number: R2019146). We disclosed information about the study to the subjects but did not obtain written informed consent as it was a retrospective study.

Statistical analysis
Statistical analysis was performed using JMP ver. 14.0.2 (SAS, Cary, NC), and P value <0.05 was considered as being indicative of signi cance.
Patients were divided into subgroups according to each of the clinical parameters and the overall survival (OS) among the groups were compared by the log-rank test. A Cox proportional hazards model was used to evaluate the association between clinical parameters, including history of treatment with ICIs, and the OS. We planned to include performance status (PS), disease stage, history of treatment with an ICI, history of treatment with a platinum doublet, and variables identi ed by univariate analysis as being signi cant by the log-rank test with P values of <0.05 as independent variables.
A Cox proportional hazards model was also used to analyze the interaction between history of ICI therapy and the patient background characteristics for the OS. The model included history of ICI therapy (yes or no), each of the patient background characteristics, and interaction variables (product term of two variables) as independent variables.

Results
A total of 209 patients with NSCLC received systemic anticancer treatment between 2015 and 2019 at Toyama University Hospital. Of these, 79 patients with tumor driver mutations, 58 patients aged < 70 years old, and 9 other patients (reasons for exclusion indicated in the gure) were excluded. Finally, data of a total of 63 patients were included in the analyses( Figure 1) Table 1 shows the patient characteristics. Twenty (31.7%) and 36 (57.1%) of the NSCLC patients were diagnosed as having squamous cell carcinoma and adenocarcinoma, respectively. Other histologic types included adenosquamous carcinoma, large cell neuroendocrine cell carcinoma, and NSCLC, not otherwise speci ed. In all, 36 patients (57.1%) had received treatment with ICIs. Thirteen (36.1%), 19 (52.8%), and 4 (11.1%) patients received ICI therapy as 1st, 2nd, and 3rd line treatment, respectively, and 11 (30.6%), 21 (58.3%), and 4 (11.1%) patients received nivolumab, pembrolizumab, and atezolizumab, respectively. ICI, Immune checkpoint inhibitor; ILD, interstitial lung disease; PD-L1, programmed death ligand-1; PS, performance status; TPS, tumor proportion score Table 2 shows the association between the patient characteristics and the OS analyzed by the log-rank test. Univariate analysis identi ed the tumor PD-L1 status,disease stage and history of treatment with an ICI as being signi cantly associated with the OS. Figure 2 shows the Kaplan-Meier curve for the OS in the patients with and without a history of treatment with ICIs. ICI, Immune checkpoint inhibitor; ILD, interstitial lung disease; OS, overall survival; PD-L1, programmed death ligand-1; PS, performance status; TPS, tumor proportion score Table 3 shows the association between a history of treatment with an ICI and the OS evaluated using the Cox proportional hazards model. The PS, disease stage, history of treatment with an ICI, and history of treatment with a platinum doublet were selected as the independent variables. The tumor PD-L1 status was excluded, because there was a signi cant overlap between the tumor PD-L1 status and history of treatment with an ICI, even though the tumor PD-L1 status was found to be signi cantly associated with the OS by the log-rank test. Analysis using the Cox proportional hazards model revealed a signi cant association between history of treatment with an ICI and the OS.  Table 4 shows the results of a subset analysis in patients divided into subgroups by the patient background characteristics to assess the association between history of treatment with an ICI and the OS. There was a signi cant interaction between the tumor histology and history of treatment with an ICI on the effect, with the association between ICI therapy and OS not being seen in patients with adenocarcinoma, but evident in those with squamous cell carcinoma and carcinoma of other histologic types. No signi cant differences in the OS were observed among patients who received ICI therapy as 1st (median: 12.6, 95% con dence interval: 5.5-NE months), 2nd (median: 29.3, 95% con dence interval: 12.8-53.9 months) and 3rd (median: 9.7, 95% con dence interval: 8.4-17.2 months) line treatment (p = 0.073, log-rank test). Furthermore, the OS was also not signi cantly different among patients who received ICI therapy (p = 0.807, log-rank test), including treatment with nivolumab (median: 19.5, 95% con dence interval: 3.2-53.9 months), pembrolizumab (median: 27.4, 95% con dence interval: 10.7-NE months), and atezolizumab (median: 12.9, 95% con dence interval: 10.0-NE months).

Discussion
In the present study, multivariate analysis revealed a signi cant independent association between a history of treatment with an ICI and a longer OS in patients with NSCLC aged ≥ 70 years old. However, because of the retrospective nature of the study, it was di cult to rule out the possible confounding effects of patient background factors on the results of the analyses. Especially, patients who showed long survival durations might have received ICI therapy in the late phase because they had survived for longer. Thus, we cannot draw any de nitive conclusion about the causative relationship between history of treatment with an ICI and a longer OS in patients with NSCLC aged ≥ 70 years old. However, the proportion of patients that received ICI therapy as 3rd line treatment was relatively low. Given that pooled analysis in the KEYNOTE study [6] and a single-arm phase 2 study [7] demonstrated the e cacy of treatment with an ICI in elderly patients with NSCLC, we believe that it is possible that treatment with an ICI contributed signi cantly to the longer survival in elderly patients with NSCLC.
There was a signi cant interaction between ICI therapy and the histology for the OS, the association between ICI therapy and OS being stronger in patients with squamous cell carcinoma and carcinoma of other histologic types than in patients with adenocarcinoma. It has been reported that nivolumab is effective in patients with squamous cell carcinoma, regardless of the tumor PD-L1 expression (1), while was less effective in patients with non-squamous cell carcinoma with no tumor PD-L1 expression [2]. A possible explanation for the interaction between ICI therapy and tumor histology in terms of the OS in the present study could be the lower lesser e cacy of ICI therapy against PD-L1-negative tumors in NSCLC patients with adenocarcinoma.
The limitations of the present study were the retrospective nature of the study and the small sample size, which could have reduced the statistical power of the study; random errors and selection bias could also have in uenced the results of the analysis. In addition, confounding factors might not have been fully adjusted for.
In conclusion, the present study demonstrated that NSCLC patients aged ≥ 70 years old who had received ICI therapy showed a longer overall survival than those who had not received ICI therapy. The association was stronger in NSCLC patients with squamous cell carcinoma and other histologic types of carcinoma than in those with adenocarcinoma. Patient selection. Data of NSCLC patients aged ≥70 years old with no tumor driver mutations were analyzed.