Immune checkpoint inhibitors as subsequent treatment in older adults with non-small cell lung cancer and synchronous brain metastases

Introduction

Immune checkpoint inhibitors (ICIs) have now become the mainstay treatment for non-small cell lung cancer (NSCLC). ICIs unleash the suppressed immune response by attaching to proteins (immune checkpoints) on T-cells of the immune system [cytotoxic T-lymphocyte antigen 4 (CTLA-4), programmed cell death-1 (PD-1)], or its ligand on cancer cell [programmed cell death ligand 1 (PD-L1)], resulting in immune-mediated destruction of cancer cells (1). For patients who progressed after platinum-doublet chemotherapy, pivotal phase 3 trials [CheckMate 017 (2), CheckMate 057 (3), KEYNOTE 010 (4), OAK (5)] comparing subsequent ICI use with docetaxel have reported prolonged OS of ICI in both squamous and non-squamous NSCLC. Based on these trials, the Food and Drug Administration (FDA) approved nivolumab and pembrolizumab in 2015 and atezolizumab in 2016 as a second- or later-line treatment for NSCLC (6). Current clinical guidelines recommend ICIs as subsequent treatment after the progression of NSCLC for patients without any mutations, if not treated with immunotherapy in first-line (7,8).

Brain metastases (BMs) are a common complication of cancer, accounting for 10–26% of cancer deaths (9). The most common cancer that metastasizes to the brain is lung cancer; 7–10% of patients with NSCLC are diagnosed with BM [“synchronous BM” (SBM)] and 20–30% develop BM subsequently (metachronous BM) (9). Patients with BM have poor morbidity and mortality with a median overall survival (OS) ranging from 3–15 months (10). Treatment for BM, however, remains an unmet need. Recent studies of central nervous system (CNS) lymphatics that demonstrated the ability of immune cells and related systemic therapies to cross the blood-brain barrier have provoked the discussion about intracranial effectiveness of ICIs (11,12). Despite the frequent occurrence of BM and associated deleterious morbidity and mortality, many patients with BM were excluded from ICI clinical trials, resulting in only a small number of patients with stable and asymptomatic BM remaining in each trial (2-5). Two recent phase 2 trials in patients with untreated or recurrent, or progressing BM have reported promising intracranial benefit, further highlighting the potential of pembrolizumab in the treatment of BMs (13,14). Meta-analyses pooling data on patients with BM across trials have reported superior effectiveness of ICIs over chemotherapy in NSCLC (15,16) when used in the first-line setting. However, no trials have examined the effectiveness of ICIs as subsequent treatment in patients with BM. Only one post hoc exploratory study of the OAK trial separately reported analysis of patients with a history of asymptomatic treated BM (n=115) and found no statistically significant difference for death for subsequent atezolizumab compared to docetaxel [hazard ratio (HR) =0.74; 95% confidence interval (CI): 0.49–1.13], likely due to small sample size (17).

However, patients in the aforementioned studies are younger (median age 61–63 years), with better performance status (PS) (ECOG 0–1) and fewer comorbidities than patients in the “real world” clinical practice (2-5). Although there are a few single-center (18,19) and multi-center (20-22) analyses of ICI in previously treated NSCLC patients, they are all conducted outside of the United States (US). One US study of previously treated patients with NSCLC found atezolizumab to have significantly longer OS compared to docetaxel (HR =0.79; 95% CI: 0.64–0.97) (23). Also, these real-world studies did not provide information related to BM or older adults separately. A recent meta-analysis of 550 patients with NSCLC and BM from 12 studies (including two trials) demonstrated the intracranial effectiveness of PD-1 inhibitors (nivolumab or pembrolizumab) used mostly as second-line therapy; however, OS was not reported (24).

Addressing these gaps, the objective of our study was to retrospectively analyze a population-based data to assess survival outcomes of subsequent treatment with ICI compared to chemotherapy in older patients with NSCLC and SBM who were previously treated with chemotherapy in the US. We hypothesize that older patients with NSCLC and SBM treated with subsequent ICI therapy exhibit improved OS compared to those treated with subsequent chemotherapy. Patients 65 years or older account for approximately 71% of patients with lung cancer (25), but are underrepresented in clinical studies due to advanced age, poor PS, lack of social support, or multiple comorbid conditions (26,27). Thus, there exists an urgent need to understand the treatment outcomes of ICIs in this patient population. We present this article in accordance with the STROBE reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-24-205/rc).

Methods

Data source

This retrospective cohort study used the Surveillance, Epidemiology, and End Results (SEER) cancer registry data linked to Medicare enrollment and fee-for-service (FFS) claims data (SEER-Medicare). The methodology, data collection strategy, and structure of SEER-Medicare have been described previously (28,29). Briefly, SEER, a population-based US cancer registry, collects data on patient demographics and tumor characteristics as well as the first course of treatment (29). Currently, SEER registries cover approximately 48% of the US population. SEER has been linked by the National Cancer Institute to Medicare enrollment and claims which provides information on inpatient/outpatient care and prescription utilization. Linking of these data sources occurs biannually and nearly 96% of individuals aged 65 or older documented in SEER are matched to their Medicare claims records (29). This study included SEER data from January 2010–December 2017 and Medicare enrollment and claims from January 2009–December 2019.

Study population

Patients diagnosed with NSCLC and SBM between 2010–2017 who received platinum-based doublets/taxane/pemetrexed/gemcitabine, or their combinations as a first-line treatment followed by ICI or a new chemotherapy regimen as subsequent treatment were included for analysis. Details regarding treatment exposure groups are defined below. Since the SEER only has information on diagnosis month and year, we imputed the date of cancer diagnosis using the date of the first Medicare claim with lung cancer diagnosis code (ICD-9/10-CM code: 162.x/C34.x) if the month and year matched to those reported in SEER. Additional information regarding the imputation process has been provided in the Appendix 1. We used International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) codes to identify patients with primary malignant neoplasm of bronchus and lung with NSCLC histology (30). Diagnosis of SBM was determined using an indicator in SEER for the presence of BM at diagnosis available from 2010 onwards. Prescription of FDA-approved ICIs (pembrolizumab, nivolumab, ipilimumab, atezolizumab, durvalumab, cemiplimab) and chemotherapy (platinum-based doublets with taxane/pemetrexed/gemcitabine) for lung cancer were identified from Medicare claims using Healthcare Common Procedural Coding System (HCPCS) codes (30).

Since patients who lived longer had a higher chance of receiving subsequent treatment, we matched each ICI patient with a unique chemotherapy patient to ensure patients in both groups initiated subsequent treatment around the same time and thus likely had similar disease progression risk at the time when the subsequent treatment started. Specifically, a second-line chemotherapy patient was matched to a second-line ICI patient if the time from diagnosis to initiation of the second-line chemotherapy treatment was within ±30 days of the time from diagnosis to second-line ICI treatment initiation of the matched ICI patient. If more than one chemotherapy patients were found within that time window, one patient was randomly chosen as the match. Third-line chemotherapy patients were matched to third-line ICI patients using the same criteria. No matches were found for fourth-line ICI or chemotherapy treatment in our data. The index treatment date was defined as the start of second-line treatment for matched second-line ICI or chemotherapy patients or the start of third-line treatment for matched third-line ICI or chemotherapy patients.

Key inclusion criteria were: (I) age ≥65 years at the time of diagnosis; (II) with continuous coverage of both Medicare Part A and B but no Part C (Medicare advantage) coverage for at least 6 months before diagnosis (to assess baseline comorbidities and PS) till index treatment date (for observation of subsequent treatment). Exclusion criteria were: (I) diagnosed with small cell/other lung cancer histology; (II) multiple primary cancers in SEER; (III) any Medicare Part C coverage or disenrollment from Medicare Part A or B from 6 months before diagnosis till index treatment date (to determine healthcare services use) because Medicare database does not have Part C claims; (IV) receipt of any oral targeted therapy medications (afatinib dimaleate, alectinib HCl, ceritinib, crizotinib, erlotinib HCl, cabozantinib s-m, ceritinib, crizotinib, dabrafenib mesylate, erlotinib HCl, gefitinib, osimertinib mesylate, sunitinib malate, trametinib dimethyl, vandetanib) to ensure uniformity in prognosis and tumor characteristics of the patient cohort; (V) died within 30 days of lung cancer diagnosis or diagnosed at autopsy; and (VI) previously treated with ICI as first-line treatment before second-line ICI. We further excluded patients (n<11) who received nivolumab and ipilimumab combinations since the combination is formally approved for first-line treatment only and the approval was on May 26, 2020 (31), which falls outside our study period.

Treatment exposure

Based on previous literature (32,33), and input from clinician co-authors, systemic agents received within 21 days from systemic treatment initiation after diagnosis were considered as part of the first-line regimen. Receipt of an ICI or a new chemotherapy agent not included in the first-line systemic regimen was labeled as a switch and the start of second-line treatment. Similarly, initiation of a new agent 21 days after the start of the second-line treatment was considered as a treatment switch and initiation of the third-line treatment. However, there were some exceptions, and further details about the determination of the line of treatment are provided in Appendix 1. Using this algorithm, patients who received ICIs as second- or third-line treatment were classified in the “ICI” group. Since the FDA first approved ICI as a subsequent treatment for NSCLC after platinum-based chemotherapy in 2015, the index dates for all patients in the ICI group were 2015 and beyond. For comparison, we identified a historical cohort of patients who received second- or third-line chemotherapy with index dates before 2015 and never received any ICIs during the study period (hereafter “chemotherapy” group). SEER does not have information on the proportion of PD-L1 expression on cancer cells, an important predictor of immunotherapy responses. Patients with higher tumor expression of PD-L1 may be more likely to receive subsequent ICI treatment. Thus using historical cohort of patients who initiated subsequent chemotherapy treatment before 2015 as the comparison group will help mitigate confounding by indication bias as these patients did not have the opportunity to receive ICI treatment at the time of subsequent treatment initiation regardless of their PD-L1 status.

Survival outcome

OS was measured as time (in days) from the index treatment date till death. Patients were censored at disenrollment from Medicare Part A or B, or enrollment in Medicare Part C, or the end of data availability (December 31, 2019), whichever occurred first.

Covariates

The following covariates were selected through a review of published literature (34-36) that used SEER-Medicare data, based on their relevance to the survival and/or receipt of treatment, and input from our clinical collaborators. Patients’ demographic characteristics at diagnosis included age, sex, race, ethnicity, marital status, percent poverty level in each patient’s census tract of residence, and rurality of each patient’s county of residence. Cancer and treatment-related covariates included the presence of any lung, bone, or liver metastasis at diagnosis, primary tumor grade and size at diagnosis, histology of NSCLC, receipt of cranial radiation before subsequent treatment (ICI/chemotherapy), and any neurosurgical resection within 1 year of diagnosis. Cranial radiation [stereotactic radiosurgery (SRS) or non-SRS] was identified using current procedural terminology (CPT)/HCPCS codes for radiation treatment delivery (30) combined with an ICD-9/10-CM code (198.3x/C79.31) for secondary malignant neoplasm of the brain to confirm radiation was received for BM. Baseline comorbidity and PS were assessed using the Charlson comorbidity index (CCI) (excluding cancer) (37,38) and a proxy of Eastern Cooperative Oncology Group (ECOG) PS scale (39,40) using Medicare claims within 6 months before the diagnosis date of NSCLC to account for different health status among the patient population. Additional information regarding the calculation of the proxy ECOG PS scale has been provided in the Appendix 1. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Statistical analysis

To compare patients’ characteristics between the matched groups (ICI and chemotherapy), we used the Wilcoxon signed-rank test for continuous variables (age) and Cochran’s Q test for categorical variables (all other covariates). We generated survival curves using the Kaplan-Meier (KM) method and compared them between treatment groups using a log-rank test. HR between groups were estimated using the Cox’s proportional hazards (CPH) model, adjusting for covariates. To account for the clustering effect of the matched patients (41), we used two approaches commonly used in the literature: (I) robust sandwich covariance matrix to account for correlations within matched pairs or (II) shared frailty model with a random effect for matched pairs (42). Choice between the approaches usually depends on the study perspective and relevance of the inferences as the former approach provides an average response of the entire population while the latter allows the hazard to vary by matched pairs (43). In our study, HRs obtained from the two approaches were similar (Table S1); hence only estimates from the robust sandwich method are included in the main text.

As a sensitivity analysis, we conducted propensity score matching using the 1:1 greedy matching method. The propensity score was estimated using the aforementioned covariates, line of treatment, and time from diagnosis to index treatment. Balance in these covariates before and after matching was assessed using absolute standardized differences (ASD), with ASD <0.1 indicating proper balance. After the appropriate balance was achieved, we used the KM method to estimate and compare survival curves. A CPH model was used to estimate the HR and 95% CI using the matched cohorts without covariates, adjusting for matched pairs using the robust sandwich method.

Several other secondary analyses were conducted. We observed the initial crossing of KM curves, similar to those presented in KM curves of clinical trials (2-5). Since crossing of survival curves indicates a violation of the proportional hazard assumption, we further compared the overall KM curves using the “MaxCombo” test, a combination weighted log-rank test (44,45), and additionally used a CPH model with a change point (46) of 114 days from the index date. The change point was selected by comparing the range of Akaike’s information criterion and Schwartz’s Bayesian Criterion values generated by models analyzed around the survival time when KM curves were seen to cross. Further, we separately compared ICI and chemotherapy patients in second-line and third-line groups using the KM method and log-rank test. Subgroup analyses by pre-selected demographic (age, sex, and race) and clinical characteristics (any bone/liver/lung metastasis at diagnosis and CCI score) were also conducted to assess treatment heterogeneity using unadjusted CPH models. Due to sample size concerns, non-White categories were combined for analysis. Sensitivity analysis excluding those who did not receive cranial radiation before subsequent treatment was conducted using adjusted CPH models from a robust sandwich method.

SAS statistical software (version 9.4; SAS Institute Inc., Cary, NC, USA) was used for all analyses.

Results

Figure 1 shows the sample selection flow chart. There were 14,661 patients aged 65 years or older with a diagnosis of NSCLC + SBM in SEER. After applying the inclusion and exclusion criteria, 716 patients remained with 350 patients in the ICI group and 366 patients in the chemotherapy group. After matching on time to treatment, the final sample included a total of 546 patients; 273 ICI patients were matched to 273 chemotherapy patients.

Figure 1 Study sample selection flowchart. According to the SEER-Medicare data use agreement, counts <11 are masked. NSCLC, non-small cell lung cancer; SBM, synchronous brain metastasis; SEER, Surveillance Epidemiology and End Results; ICI, immune checkpoint inhibitor; FDA, Food and Drug Administration.

The median age of the matched sample was 71 (range, 66–87) years. Table 1 displays the characteristics of the matched sample. Comparing to the chemotherapy group, patients receiving ICI treatment were older {mean 72.11 [standard deviation (SD): 5.03] vs. 71.25 (SD: 4.56) years, P=0.03}, more likely to be non-Hispanic (P=0.02), and have liver metastasis at diagnosis (20.15% vs. 11.36%, P=0.003). Majority of patients in both groups had adenocarcinoma (63.37% vs. 62.64%); however, patients receiving ICI treatment were more likely to have squamous cell carcinoma (16.85% vs. 9.16%, P=0.009) compared to the chemotherapy group, but similar concerning all other covariates. Median time to second-line (241 vs. 227 days, P=0.29) or third-line (259 vs. 267 days, P=0.80) treatment from diagnosis was similar between the ICI and chemotherapy groups.

The median follow-up time was 206 days (range, 2–1,607 days) for the ICI group and 151 days (range, 6–3,240 days) for the chemotherapy group after the index treatment date. KM survival curves show that patients receiving subsequent ICI treatment had longer survival than those receiving subsequent chemotherapy treatment [median OS: 209 days (95% CI: 160–275) for the ICI group and 155 days (95% CI: 135–187) for chemotherapy group, log-rank test P<0.001] (Figure 2A). After adjusting for covariates, subsequent ICI treatment had a lower risk of death than subsequent chemotherapy treatment (HR =0.63; 95% CI: 0.52–0.75) using the robust sandwich method (Table 2). Secondary analyses using the MaxCombo test also showed an overall significant difference in survival time between groups (P<0.001) and when applying higher weights to early (P=0.03) or later (P<0.001) survival times (Figure S1). Robust sandwich method CPH with a change point showed that subsequent ICI was associated with a significantly higher hazard of death (HR =1.48; 95% CI: 1.09–2.02) before 114 days from index date, and a significantly lower hazard of death (HR =0.53; 95% CI: 0.42–0.68) after 114 days from index date when compared to subsequent chemotherapy (Table S2).

Figure 2 Overall survival. (A) Kaplan-Meier survival curves of subsequent ICI vs. chemotherapy only; (B) hazard ratio (95% CI) for subsequent ICI vs. chemotherapy only in selected subgroups from unadjusted CPH model. Vertical dotted line in (B) denotes the hazard ratio in the overall population. *, non-White group has <30 patients with subsequent ICI treatment; thus, the estimate may not be reliable. All P values are significant at α <0.05 except non-White. According to the SEER-Medicare data use agreement, counts <11 are masked. Chemo, chemotherapy; ICI, immune checkpoint inhibitor; CI, confidence interval; CPH, Cox’s proportional hazards; SEER, Surveillance, Epidemiology, and End Results.

Using the propensity score matching method, we found 259 matched pairs. After matching, all variables were well balanced with ASD <0.1 (Table S3). KM survival curves comparing propensity matched groups show that patients receiving subsequent ICI treatment had longer survival than those receiving subsequent chemotherapy treatment [median OS: 209 days (95% CI: 161–284) for the ICI group and 152 days (95% CI: 133–172) for chemotherapy group, log-rank test P<0.001] (Figure S2). Cox’s regression model estimated a lower risk of death in subsequent ICI treatment vs. chemotherapy (HR =0.63; 95% CI: 0.52–0.76) using a robust sandwich method.

Stratified analysis by lines of subsequent treatment showed similar survival benefits in second-line ICI use over chemotherapy as in the main analysis (Figure S3, top figure). However, no statistically significant difference was found (HR =1.50; 95% CI: 0.58–3.90) for third-line ICI use vs. chemotherapy (Figure S3, bottom figure), a finding consistent with the subgroup analysis using clinical trial data (8). In subgroup analyses, HRs for OS favored subsequent ICI over chemotherapy across all pre-specified subgroups, although estimates for patients with any other (bone/live/lung) metastasis at diagnosis and CCI ≥2 were not statistically significant (Figure 2B). Sensitivity analyses excluding those who did not receive cranial radiation before subsequent treatment showed similar survival benefits as in the main analysis (Table S4).

Discussion

Using population-based real-world data of older patients, we found that subsequent ICI treatment was associated with improved survival compared to chemotherapy in NSCLC patients with SBM previously treated with chemotherapy. Although the survival benefit of subsequent ICI treatment over chemotherapy has been shown in previous clinical trials (2-5), patients with BMs, those with poor PS (ECOG PS ≥2), and older adults (especially those age ≥75 years) were under-represented.

In this cohort of older patients with NSCLC and SBM previously treated with chemotherapy, we found the median OS was 6.8 months in patients receiving ICI as second or third-line treatment. This median OS is shorter than those reported in previous trials involving second-line ICI use (9.2 months in CheckMate 017, 12.2 months in CheckMate 057, 10.4 months in KEYNOTE-010). This difference likely reflected the generally younger and healthier patients included in clinical trials. The median age in our study was 71 years, which was 8–10 years older than patients included in the clinical trials (2-5,27), but more comparable to the average age at diagnosis of real-world lung cancer patients (70 years) (47). Only 7–8% of the population in clinical trials were older than 70 years. Additionally, we focused on patients with SBM and included both patients who were previously treated with cranial radiation and those who were not, but trials only included a small number of patients (6–16%) with previously treated and stable BM. Thus, our patient population is likely more frail and has poorer prognoses. However, even among these older and more frail patients, our findings indicate that subsequent ICI treatment after failed chemotherapy can improve survival in patients with NSCLC and BM.

We found a 37% overall reduction and a 47% reduction after 114 days post-index treatment in OS of patients who received subsequent ICI treatment compared to chemotherapy after previous chemotherapy. These estimates are comparable to the 44% reduction in OS favoring subsequent atezolizumab treatment compared to chemotherapy in a subgroup of BM patients (n=85) from the OAK trial (5). It is also consistent with findings from a large meta-analysis of 102 trials that compared second-line ICI treatment with chemotherapy in lung cancer patients (n=36,058) (48). This study separately reported HRs for specific second-line ICI agents compared to docetaxel (nivolumab: HR =0.69, 95% CI: 0.56–0.83; pembrolizumab: HR =0.71, 95% CI: 0.56–0.90; atezolizumab: HR =0.73, 95% CI: 0.62–0.87). However, no separate analysis of patients with BM was reported and the analysis included both phase 2 and 3 trials, with some conducted only in SCLC patients, or with EGFR/KRAS gene mutations. Our finding of a significant OS benefit of subsequent ICI treatment over chemotherapy in older patients with NSCLC and BM further strengthened the evidence that ICI is possibly as effective as the subsequent treatment over chemotherapy in NSCLC, even in the presence of BM.

We found a lower hazard of death using subsequent ICI treatment compared to chemotherapy in both males and females, consistent with clinical trials (2-5). A survival benefit was also seen in patients 65–74 years and those ≥75 years. Only trials of subsequent nivolumab had conducted subgroup analysis in very old patients ≥75 years and reported a statistically insignificant HR in squamous (HR =1.53; 95% CI: 0.65–3.62) (7) and non-squamous (HR =0.90; 95% CI: 0.43–1.87) NSCLC compared to docetaxel (8). However, the number of patients ≥75 years was small in both trials (29 and 43 respectively) (7,8), compared to 115 patients in our study. We also found that subsequent ICI was associated with a survival benefit over chemotherapy among both White and non-White patients. None of the trials of subsequent ICI use had reported subgroup analysis by race (7-10). However, both clinical trials and our study have a small number of patients from racial minorities. This finding should be interpreted with caution and should be confirmed in larger studies.

Strength and limitation

The strength of our study is that it included a large sample of older NSCLC patients with BM aged 65 years or older, a population commonly underrepresented and more frail than those included in clinical trials. Nonetheless, some important limitations should be noted. First, we do not have information on some important prognostic factors such as the proportion of PD-L1 on cancer cells, patient PS, and smoking status. To reduce potential confounding bias due to missing PD-L1 status, we used a historical cohort of patients who received subsequent chemotherapy treatment before 2015, the year when the FDA first approved ICIs as a subsequent treatment for NSCLC, as the comparison group. However, using a historical cohort might introduce time bias as patients treated with ICIs more recently may benefit from improved supportive care, potentially influencing survival outcomes. We also do not have information on the driver mutation testing results, which were not in wide use during the time period of this study and therefore was not likely a consideration for physician’s treatment choice between ICI and chemotherapy. Although a direct measure of patients’ PS was unavailable, we used a proxy measure derived from using previously validated study (39). However, this measure can only categorize patients into ECOG 0–2 vs. 3–4 and could not distinguish ECOG =2 from ECOG 0–1; most trials of subsequent ICIs included only patients with ECOG 0–1 (7-10). Thus, patients with worse PS (ECOG ≥2), a population excluded from clinical trials, could not be studied separately. Second, although we used matched cohorts to ensure comparison of patients with similar prognoses to minimize survival bias, there may exist other biases and residual confounding that may lead to over or underestimation. Third, due to the inability to directly observe the line of therapy from SEER-Medicare data, we employed a modified algorithm to delineate the treatment line, the validity of which has not been evaluated. Fourth, in our matched sample, the number of patients who received ICI as third-line treatment or who did not receive cranial radiation before subsequent treatment was small. Further, subgroups by ethnicity or rurality could not be compared and some subgroups were combined to form larger groups for comparison [e.g., race, any (liver/lung/bone) metastasis]. Caution is warranted to interpret these findings and future larger studies with more patients in those subgroups should be conducted to confirm the findings. Finally, we used a variable in SEER to identify NSCLC patients who were presented with BM at diagnosis (SBM). Although additional NSCLC patients who developed BM later (metachronous BM) may be identified using diagnosis codes for secondary cancer in Medicare claims, these codes do not provide information on primary cancer. Literature suggests that SBM and metachronous BM have different prognoses and outcomes (49). Thus, our study may not generalize to NSCLC patients with metachronous BM.

Conclusions

In this population-based study of older patients ≥65 years with NSCLC and SBM who were previously treated with chemotherapy, we found subsequent ICI treatment was associated with improved survival compared to subsequent chemotherapy. These findings add to the evidence that ICIs likely confer a survival advantage in the second-line treatment of NSCLC over chemotherapy in older, more frail patients with BM.

Acknowledgments

The authors express gratitude to Dr. Cheng Peng for her contribution to the development of the SAS code for the matching algorithm methodology in this paper. Special thanks are extended to Adrian Jemal Williams for his technical writing assistance in finalizing and publishing the manuscript. It is important to note that the content is the sole responsibility of the authors and does not necessarily reflect the official views of the National Institutes of Health, National Cancer Institute, or Arkansas Biosciences Institute.

Funding: A.R. was supported by the Robert A. Winn Diversity in Clinical Trials Award Program and NIH (No. P50MD017319-03). C.L.’s effort on study was supported by the National Center for Advancing Translational Sciences under award number UL1 TR003107-02S2. Additionally, support for data acquisition was partially supplied by the Arkansas Biosciences Institute, a key research component of the Arkansas Tobacco Settlement Proceeds Act of 2000.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-24-205/rc

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-24-205/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-24-205/coif). B.C.M. receives royalties from Trestle Tree LLC for the commercialization of an opioid risk prediction tool that is unrelated to this investigation and receives honoraria for his service on the Midwest CEPAC which assesses the clinical and economic value of new health technologies. A.R. has obtained grant funding from Bristol Myers Squibb, Nico Corporation and the NIH which are not related to this work. She serves on multiple advisory boards (Nico Corporation, Medexus, and UCSC) which are not in conflict with this work. C.L. is supported by the American Cancer Society under award number RSGI‐23‐1039245‐01‐HOPS, and was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under award number UL1 TR003107-02S2. Support of data acquisition was provided by the Arkansas Biosciences Institute, the major research component of the Arkansas Tobacco Settlement Proceeds Act of 2000. C.L. received research support (payment to the university) for unrelated projects sponsored by University of Utah/AstraZeneca, and received honoraria from PCORI for grant review. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Johnson DB, Rioth MJ, Horn L. Immune checkpoint inhibitors in NSCLC. Curr Treat Options Oncol 2014;15:658-69. [Crossref] [PubMed]
2. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus Docetaxel in Advanced Squamous-Cell Non-Small-Cell Lung Cancer. N Engl J Med 2015;373:123-35. [Crossref] [PubMed]
3. Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus Docetaxel in Advanced Nonsquamous Non-Small-Cell Lung Cancer. N Engl J Med 2015;373:1627-39. [Crossref] [PubMed]
4. Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled trial. Lancet 2016;387:1540-50. [Crossref] [PubMed]
5. Rittmeyer A, Barlesi F, Waterkamp D, et al. Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial. Lancet 2017;389:255-65. [Crossref] [PubMed]
6. Vaddepally RK, Kharel P, Pandey R, et al. Review of Indications of FDA-Approved Immune Checkpoint Inhibitors per NCCN Guidelines with the Level of Evidence. Cancers (Basel) 2020;12:738. [Crossref] [PubMed]
7. Hanna N, Johnson D, Temin S, et al. Systemic Therapy for Stage IV Non-Small-Cell Lung Cancer: American Society of Clinical Oncology Clinical Practice Guideline Update. J Clin Oncol 2017;35:3484-515. [Crossref] [PubMed]
8. Ettinger DS, Wood DE, Aisner DL, et al. NCCN Guidelines® Insights: Non-Small Cell Lung Cancer, Version 2.2023. J Natl Compr Canc Netw 2023;21:340-50. [Crossref] [PubMed]
9. Amsbaugh MJ, Kim CS. Brain Metastasis. StatPearls. Published online August 7, 2021. Accessed November 24, 2021. Available online: https://www.ncbi.nlm.nih.gov/books/NBK470246/
10. Ali A, Goffin JR, Arnold A, et al. Survival of patients with non-small-cell lung cancer after a diagnosis of brain metastases. Curr Oncol 2013;20:e300-6. [Crossref] [PubMed]
11. Berghoff AS, Venur VA, Preusser M, et al. Immune Checkpoint Inhibitors in Brain Metastases: From Biology to Treatment. Am Soc Clin Oncol Educ Book 2016;35:e116-22. [Crossref] [PubMed]
12. Xiao G, Liu Z, Gao X, et al. Immune checkpoint inhibitors for brain metastases in non-small-cell lung cancer: from rationale to clinical application. Immunotherapy 2021;13:1031-51. [Crossref] [PubMed]
13. Goldberg SB, Schalper KA, Gettinger SN, et al. Pembrolizumab for management of patients with NSCLC and brain metastases: long-term results and biomarker analysis from a non-randomised, open-label, phase 2 trial. Lancet Oncol 2020;21:655-63. [Crossref] [PubMed]
14. Brastianos PK, Kim AE, Giobbie-Hurder A, et al. Pembrolizumab in brain metastases of diverse histologies: phase 2 trial results. Nat Med 2023;29:1728-37. [Crossref] [PubMed]
15. Yang K, Li J, Bai C, et al. Efficacy of Immune Checkpoint Inhibitors in Non-small-cell Lung Cancer Patients With Different Metastatic Sites: A Systematic Review and Meta-Analysis. Front Oncol 2020;10:1098. [Crossref] [PubMed]
16. Powell SF, Rodríguez-Abreu D, Langer CJ, et al. Outcomes With Pembrolizumab Plus Platinum-Based Chemotherapy for Patients With NSCLC and Stable Brain Metastases: Pooled Analysis of KEYNOTE-021, -189, and -407. J Thorac Oncol 2021;16:1883-92. [Crossref] [PubMed]
17. Gadgeel SM, Lukas RV, Goldschmidt J, et al. Atezolizumab in patients with advanced non-small cell lung cancer and history of asymptomatic, treated brain metastases: Exploratory analyses of the phase III OAK study. Lung Cancer 2019;128:105-12. [Crossref] [PubMed]
18. Chen M, Li Q, Xu Y, et al. Immunotherapy as second-line treatment and beyond for non-small cell lung cancer in a single center of China: Outcomes, toxicities, and clinical predictive factors from a real-world retrospective analysis. Thorac Cancer 2020;11:1955-62. [Crossref] [PubMed]
19. Ivanović M, Knez L, Herzog A, et al. Immunotherapy for Metastatic Non-Small Cell Lung Cancer: Real-World Data from an Academic Central and Eastern European Center. Oncologist 2021;26:e2143-50. [Crossref] [PubMed]
20. Martin C, Lupinacci L, Perazzo F, et al. Efficacy and Safety of Nivolumab in Previously Treated Patients With Non-Small-cell Lung Cancer: Real World Experience in Argentina. Clin Lung Cancer 2020;21:e380-7. [Crossref] [PubMed]
21. Figueiredo A, Almeida MA, Almodovar MT, et al. Real-world data from the Portuguese Nivolumab Expanded Access Program (EAP) in previously treated Non Small Cell Lung Cancer (NSCLC). Pulmonology 2020;26:10-7. [Crossref] [PubMed]
22. Furuya N, Nishino M, Wakuda K, et al. Real-world efficacy of atezolizumab in non-small cell lung cancer: A multicenter cohort study focused on performance status and retreatment after failure of anti-PD-1 antibody. Thorac Cancer 2021;12:613-8. [Crossref] [PubMed]
23. Ramagopalan S, Gupta A, Arora P, et al. Comparative Effectiveness of Atezolizumab, Nivolumab, and Docetaxel in Patients With Previously Treated Non-Small Cell Lung Cancer. JAMA Netw Open 2021;4:e2134299. [Crossref] [PubMed]
24. Teixeira Loiola de Alencar V, Guedes Camandaroba MP, Pirolli R, et al. Immunotherapy as Single Treatment for Patients With NSCLC With Brain Metastases: A Systematic Review and Meta-Analysis-the META-L-BRAIN Study. J Thorac Oncol 2021;16:1379-91. [Crossref] [PubMed]
25. Lung and Bronchus Cancer — Cancer Stat Facts. Published 2021. Accessed November 21, 2021. Available online: https://seer.cancer.gov/statfacts/html/lungb.html
26. Pang HH, Wang X, Stinchcombe TE, et al. Enrollment Trends and Disparity Among Patients With Lung Cancer in National Clinical Trials, 1990 to 2012. J Clin Oncol 2016;34:3992-9. [Crossref] [PubMed]
27. Sacher AG, Le LW, Leighl NB, et al. Elderly patients with advanced NSCLC in phase III clinical trials: are the elderly excluded from practice-changing trials in advanced NSCLC? J Thorac Oncol 2013;8:366-8. [Crossref] [PubMed]
28. Ambs A, Warren JL, Bellizzi KM, et al. Overview of the SEER--Medicare Health Outcomes Survey linked dataset. Health Care Financ Rev 2008;29:5-21.
29. Warren JL, Klabunde CN, Schrag D, et al. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care 2002;40:IV-3-18. [Crossref] [PubMed]
30. Mahashabde R, Bhatti SA, Martin BC, et al. Real-World Survival of First-Line Immune Checkpoint Inhibitor Treatment Versus Chemotherapy in Older Patients With Non-Small-Cell Lung Cancer and Synchronous Brain Metastases. JCO Oncol Pract 2023;19:1009-19. [Crossref] [PubMed]
31. FDA. FDA approves nivolumab plus ipilimumab and chemotherapy for first-line treatment of metastatic NSCLC. Accessed May 5, 2022. Available online: https://www.fda.gov/drugs/resources-information-approved-drugs/fda-approves-nivolumab-plus-ipilimumab-and-chemotherapy-first-line-treatment-metastatic-nsclc
32. Determining lines of therapy (LOT) in oncology in claims databases. Accessed December 4, 2021. Available online: https://www.optum.com/content/dam/optum3/optum/en/resources/white-papers/guidelines-for-determining-lines-of-therapy-whitepaper.pdf
33. Arunachalam A, Li H, Bittoni MA, et al. Real-World Treatment Patterns, Overall Survival, and Occurrence and Costs of Adverse Events Associated With Second-Line Therapies for Medicare Patients With Advanced Non-Small-Cell Lung Cancer. Clin Lung Cancer 2018;19:e783-99. [Crossref] [PubMed]
34. Youn B, Trikalinos NA, Mor V, et al. Real-world use and survival outcomes of immune checkpoint inhibitors in older adults with non-small cell lung cancer. Cancer 2020;126:978-85. [Crossref] [PubMed]
35. Lamba N, Kearney RB, Catalano PJ, et al. Population-based estimates of survival among elderly patients with brain metastases. Neuro Oncol 2021;23:661-76. [Crossref] [PubMed]
36. Sadetsky N, Hernandez A, Wallick CJ, et al. Survival outcomes in an older US population with advanced melanoma and central nervous system metastases: SEER-Medicare analysis. Cancer Med 2020;9:6216-24. [Crossref] [PubMed]
37. Stedman MR, Doria-Rose P, Warren JL, et al. Comorbidity Technical Report: The Impact of Different SEER-Medicare Claims-based Comorbidity Indexes on Predicting Non-cancer Mortality for Cancer Patients.
38. SEER-Medicare: Comorbidity SAS Macros. Accessed May 2, 2022. Available online: https://healthcaredelivery.cancer.gov/seermedicare/considerations/calculation.html
39. Davidoff AJ, Gardner LD, Zuckerman IH, et al. Validation of disability status, a claims-based measure of functional status for cancer treatment and outcomes studies. Med Care 2014;52:500-10. [Crossref] [PubMed]
40. Davidoff AJ, Zuckerman IH, Pandya N, et al. A novel approach to improve health status measurement in observational claims-based studies of cancer treatment and outcomes. J Geriatr Oncol 2013;4:157-65. [Crossref] [PubMed]
41. Gharibvand L, Liu L. Analysis of Survival Data with Clustered Events. Accessed June 13, 2022. Available online: https://support.sas.com/resources/papers/proceedings09/237-2009.pdf
42. Ying G, Liu C. Statistical Analysis of Clustered Data using SAS® System. Published online 2006. Accessed June 13, 2022. Available online: https://www.lexjansen.com/nesug/nesug06/an/da01.pdf
43. Difference between generalized linear models & generalized linear mixed models - Cross Validated. Accessed June 13, 2022. Available online: https://stats.stackexchange.com/questions/32419/difference-between-generalized-linear-models-generalized-linear-mixed-models/32421#32421
44. Mukhopadhyay P, Ye J, Anderson KM, et al. Log-Rank Test vs MaxCombo and Difference in Restricted Mean Survival Time Tests for Comparing Survival Under Nonproportional Hazards in Immuno-oncology Trials: A Systematic Review and Meta-analysis. JAMA Oncol 2022;8:1294-300. [Crossref] [PubMed]
45. Knezevic A, Patil S. Combination weighted log-rank tests for survival analysis with non-proportional hazards. SAS Global Forum 2020.
46. Liang KY, Self SG, Liu XH. The Cox proportional hazards model with change point: an epidemiologic application. Biometrics 1990;46:783-93.
47. American Cancer Society. Team TACS medical and editorial content. Key Statistics for Lung Cancer. Published 2021. Accessed October 23, 2021. Available online: https://www.cancer.org/cancer/lung-cancer/about/key-statistics.html
48. Créquit P, Chaimani A, Yavchitz A, et al. Comparative efficacy and safety of second-line treatments for advanced non-small cell lung cancer with wild-type or unknown status for epidermal growth factor receptor: a systematic review and network meta-analysis. BMC Med 2017;15:193. [Crossref] [PubMed]
49. Jiang K, Parker M, Materi J, et al. Epidemiology and survival outcomes of synchronous and metachronous brain metastases: a retrospective population-based study. Neurosurg Focus 2023;55:E3. [Crossref] [PubMed]