Recent treatment patterns and real-world survival following first-line anti-PD-L1 treatment for extensive-stage small cell lung cancer

Abstract

Background

The landscape of small cell lung cancer (SCLC) has changed since the 2019 and 2020 approvals of anti-PD-L1 atezolizumab and durvalumab for first-line (1L) treatment in combination with chemotherapy. We studied treatment patterns and real-world overall survival (rwOS) following 1L-3L therapy.

Patients and Methods

A nationwide electronic health record (EHR)-derived de-identified database was used to describe treatment patterns, characteristics, and survival of patients with extensive-stage (ES)-SCLC by 1L anti-PD-L1 treatment. Patients with ES-SCLC who initiated ≥1 line of systemic therapy from 2013 to 2021, with potential follow-up through 2022, were included.

Results

Among 9952 patients with SCLC, there were 4308 patients with ES-SCLC treated during the study period who met eligibility criteria. Etoposide + platinum (EP) chemotherapy was most common in the 1L, with addition of anti-PD-L1 therapy to most regimens by 2019. Second-line regimens varied by platinum sensitivity status and shifted from topotecan to lurbinectedin over time. Median rwOS following 1L therapy was 8.3 months (95% CI, 7.9-8.8) in those treated with 1L anti-PD-L1 and 8.0 months (95% CI, 7.8-8.2) in those who were not. Following 2L and 3L, median rwOS was 5.6 (95% CI, 4.9-6.3) and 4.9 months (95% CI, 3.4-6.0), respectively, among 1L anti-PD-L1-treated, and 4.5 (95% CI, 4.2-4.9) and 4.0 months (95% CI, 3.7-4.5), respectively, among those who were not.

Conclusion

Despite the introduction of frontline anti-PD-L1 therapy, survival remains dismal among patients with ES-SCLC treated in the real-world setting.

Introduction

Lung cancer is the second most common cancer and leading cause of cancer death worldwide, representing 11% of cancers diagnosed and 18% of cancer deaths.¹ Small cell lung cancer (SCLC) is a high-grade neuroendocrine tumor that accounts for 10%-15% of all lung cancers.² SCLC is traditionally staged based on the size and spread of the tumor as limited-stage (LS) disease or extensive-stage (ES) disease.³ Prognosis is poor, with a 5-year overall survival (OS) rate of <10% including LS-SCLC and ES-SCLC.⁴

For many years, first-line (1L) standard-of-care for patients with ES-SCLC consisted of etoposide plus platinum (EP)-based chemotherapy.^5,6 Within the last 5 years, the addition of anti-programmed death ligand 1 (anti-PD-L1) immunotherapy to chemotherapy in the 1L setting was approved for patients diagnosed with ES-SCLC.⁴ Based on results from the CASPIAN⁷ and IMpower133 trials,⁸ 1L treatment of SCLC with anti-PD-L1 atezolizumab or durvalumab combined with chemotherapy followed by maintenance therapy with checkpoint inhibitor alone, led to durable but modest survival gains for a subset of patients with ES-SCLC. In the CASPIAN trial, median OS was 12.9 months (95% CI, 11.3-14.7) for durvalumab plus EP with either cisplatin or carboplatin versus 10.5 months (95% CI, 9.3-11.2) for EP alone, with 36-month OS rates of 17.6% (95% CI, 13.3-22.4) versus 5.8% (95% CI, 3.4-9.1), respectively.⁷ In IMpower 133, median OS was 12.3 months (95% CI, 10.8-15.8) for atezolizumab plus EP versus 10.3 months (95% CI, 9.3-11.3) for placebo plus EP, with 18-month OS rates of 34.0% versus 21.0%, respectively.⁸

While anti-PD-L1 therapy has shown effectiveness in the first line, the majority of patients acquire resistance to frontline therapy.⁹ Second-line treatment decisions depend on various factors such as performance status, prior therapy, and durability of response. For patients with relapsed ES-SCLC, current guidelines recommend lurbinectedin, while platinum-based rechallenge is often given in those who are platinum sensitive.¹⁰ A systematic literature review reported a median OS in the real world following 2L therapy of 4.9 months.¹¹ There are currently no approved therapies in the third-line (3L) setting and beyond, illustrating an unmet need for therapeutic options.^4,5,12

In 2019, a real-world effectiveness and tolerability systematic literature review of SCLC treatments in the United States (immunotherapy, single-agent or combination chemotherapy, or radiotherapy) between the years 2006 and 2018 found poor rates of OS following any line of therapy (LOT), with no treatment option found to be superior.¹¹ Since the approval of atezolizumab in March 2019 and durvalumab in March 2020 for 1L treatment in combination with chemotherapy, the landscape of SCLC has changed.⁴ Current published real-world data on treatment patterns in the United States are limited beyond 2019 and thus do not reflect the shift in standard-of-care post-approval of atezolizumab and durvalumab in the 1L setting.

Evaluation of the treatment landscape over the last decade will demonstrate how standard of care has evolved and will elucidate recent treatment patterns in the real world by line of therapy, following the approval of frontline anti-PD-L1 therapy. Additionally, it is important to understand whether there has been an impact of 1L anti-PD-L1 therapy on real-world outcomes. Herein we report treatment patterns over time across therapy lines and real-world overall survival (rwOS) after 1L-3L therapy of patients with ES-SCLC from 2013 to 2022.

Patients and methods

Study design and data source

This was a retrospective, descriptive analysis of patients with SCLC using de-identified electronic health record-derived data from the US-based, nationwide, Flatiron Health database. The Flatiron Health database contains demographic, clinical, treatment, and outcome information on patients who received care within the Flatiron Health network. This longitudinal database comprises de-identified patient-level structured and unstructured data, curated via technology-enabled abstraction. During the study period of 2013-2022, de-identified data originated from ~280 cancer clinics (~800 sites of care), that were primarily community-based oncology settings. The de-identified data were subject to obligations to prevent re-identification and protect patient confidentiality.

Study population

For the assessment of treatment patterns, eligible patients were diagnosed with SCLC, initiated ≥1 systemic therapy from January 01, 2013 to December 31, 2021, and had ES disease at diagnosis. Patients with a gap of >90 days from the initial SCLC diagnosis date to the start of structured data activity after diagnosis were excluded to ensure continuous enrollment and patients with a diagnosis of non–small cell lung cancer (NSCLC) on or before the time a patient was first included in the SCLC cohort were excluded. For outcome assessment, the cohort was further restricted to exclude patients with any exposure to a clinical study drug, and those who received 1L anti-PD-1 therapy (pembrolizumab, nivolumab, or cemiplimab), to ensure a homogeneous population when stratifying by 1L anti-PD-L1 treatment. The data cutoff date was December 31, 2022 to allow for potential follow-up of at least 12 months after 1L therapy initiation.

Data were included for patients from the date of SCLC diagnosis until the end of follow-up. Patients were followed until the first of death, loss to follow-up, or data cutoff date (December 31, 2022).

Variables

Structured treatment data on individual drug administrations, as well as unstructured documents (eg, oncology visit notes), were used in Flatiron Health’s derivation of LOTs, which were algorithmically derived by oncologist-defined, rule-based LOT definitions. Palliative or supportive care were not included as a LOT. The first 3 LOTs were further categorized into the most common regimens for ES-SCLC as follows for LOT1: etoposide + platinum (EP) without anti-PD-L1, EP plus anti-PD-L1, and all other regimens. LOT2 and LOT3 regimen categories were defined in the following hierarchical order: EP without anti-PD-L1, EP plus anti-PD-L1, any other immunotherapy-containing regimens, any other platinum regimens, topotecan, lurbinectedin, taxane-containing regimens, and all other regimens. Platinum agents included carboplatin or cisplatin, and anti-PD-L1 therapy included atezolizumab or durvalumab. First-line anti-PD-L1 therapy was defined as receipt of atezolizumab or durvalumab at any time during LOT1 (induction with or without maintenance therapy). Platinum sensitivity status was defined by the length of patients’ chemotherapy-free interval (CTFI) from the end of LOT1 to the start of LOT2 among those platinum-treated in the first line who had at least 2 LOTs. It was assessed categorically by a CTFI < 90 days (resistant), CTFI of 90-180 days, and CTFI ≥180 days to indicate the degree of sensitivity to platinum therapy. SCLC stage (limited vs extensive) was also derived by rule-based definitions through chart abstraction at the time of initial diagnosis.

Outcomes

Most common treatment regimens were described for each LOT over time. Second-line (2L) treatments were also described separately by platinum sensitivity status (defined according to a chemotherapy-free interval <90 days, 90-180 days, or ≥180 days) among those who were 1L platinum-treated.

Real-world overall survival (rwOS) was defined as the time from the index date (ie, start date of a given LOT) to the date of death or censorship for an individual patient. Patients without evidence of death at the end of the data cutoff were censored at the time of their last confirmed activity date (ie, last encounter of any type) in the database or the data cutoff date (December 31, 2022), whichever came first. Death was captured using Flatiron Health’s composite mortality variable which has been shown to have high sensitivity and specificity in multiple cancer types including SCLC.¹³

Data analysis

Study objectives were descriptive in nature and only included estimations; no hypotheses were tested. Treatment patterns, patient demographics, and disease characteristics were described for all study patients and separately by use of anti-PD-L1 treatment in 1L. Frequencies and proportions were presented for categorical variables, while means, medians, standard deviations, and interquartile ranges were presented for continuous variables (eg, age at index date).

Real-world OS was estimated following the start date of each LOT (1L-3L) in groups defined by use of anti-PD-L1 treatment in 1L. Median rwOS and corresponding 95% CIs were calculated using Kaplan-Meier estimates. The estimated survival probabilities for rwOS and corresponding 95% CIs were presented for patients at various timepoints (6, 12, 24, and 36 months). Subgroup analyses were also performed by select subgroups including age, race, sex, Eastern Cooperative Oncology Group (ECOG) performance status at index date, and 1L maintenance therapy.

Results

Treatment demographics and patient characteristics

Overall, among 9952 patients diagnosed with SCLC from January 2013 to December 2022, 8191 (82%) received at least 1 LOT during that time and 7268 received their first LOT from January 1, 2013 to December 31, 2021 (Figure 1). Of 6978 with non-missing stage information, a final cohort of 4593 (66%) with ES-SCLC at diagnosis were used to describe treatment patterns (Figure 2 only). An additional 285 patients were excluded for exposure to a clinical study drug or 1L treatment with a PD-1 inhibitor, totaling 4308 patients with ES-SCLC for all other analyses.

Of 4308 patients with ES-SCLC, the mean (SD) age at diagnosis was 67.5 (9.0) years (Table 1). A total of 2087 patients (48%) were female, 3158 (73%) were White, 3704 (86%) were treated in community healthcare settings, and 4221 (98%) had a history of smoking. Additionally, 1009 patients (23%) had an ECOG performance status ≥2 at index, 808 (19%) had received radiation therapy prior to 1L initiation, and 475 (11%) had received prophylactic cranial irradiation prior to 1L initiation. Over the study period from 2013 to 2021, 1149 (27%) patients with ES-SCLC received 1L anti-PD-L1 treatment, and 3159 (73%) did not. Demographics and disease characteristics were similar regardless of 1L anti-PD-L1 treatment group. Demographics were also assessed in the untreated cohort and there were no discernible differences with the treated cohort other than patients being slightly older (Supplementary Table S1).

Treatment patterns

Among patients with ES-SCLC (N = 4593), 1L treatment with EP chemotherapy was used in most patients (91%) before 2018. First-line treatment with EP shifted to include anti-PD-L1s (atezolizumab or durvalumab) beginning in 2018 (7%), increased sharply in 2019 to 68%, and were used in the majority (~80%) of 1L regimens by 2021 (Figure 2A).

Among ES-SCLC patients who received EP plus anti-PD-L1 in the 1L, 61% went on to receive single-agent atezolizumab or durvalumab maintenance therapy. Switching of anti-PD-L1 (ie, atezolizumab to durvalumab or vice versa) during induction, from induction to maintenance, or during the maintenance phase, occurred in <1% of patients (data not shown). Nearly three-quarters of 2L patients who received 1L platinum plus anti-PD-L1 treatment were platinum sensitive (42% with a CTFI of 90-180 days and 30% with CTFI ≥180 days), and nearly half of those who received 1L platinum with no anti-PD-L1 were platinum sensitive (27% with CTFI 90-180 days and 17% with CTFI ≥180 days; Table 2).

Among platinum-resistant (CTFI < 90 days) patients with ES-SCLC, topotecan was the most common 2L therapy prior to 2018 (31%) but declined over time, while lurbinectedin use increased from 10% in 2020 to be the most common regimen in 2022 (40%; Figure 2B). Similarly, in the platinum-sensitive group in which CTFI was 90-180 days, topotecan was most common in 2L prior to 2018 (36%) but declined over time, while lurbinectedin increased from 24% of 2L regimens in 2020 to be the most common regimen (69%) in 2022 (Figure 2C). Those more sensitive to platinum (with a CTFI ≥180 days) were more commonly treated with EP with or without anti-PD-L1 in the 2L over time until most recently in 2022 when lurbinectedin rose to 54% of regimens (Figure 2D). The proportions treated with “other platinum” regimens declined over time and most commonly included carboplatin plus irinotecan, carboplatin plus paclitaxel, or cisplatin plus irinotecan. Other immunotherapy regimens were used frequently in 2018-2019 and most commonly included nivolumab plus ipilimumab, or nivolumab or pembrolizumab monotherapy.

No notable trends were observed for 3L treatments of SCLC as treatments varied (Figure 2E). In 2022, lurbinectedin (33%) or topotecan (22%) were the most common 3L treatments for ES-SCLC.

Real-world overall survival

Overall, the median rwOS after 1L therapy of patients with ES-SCLC was 8.1 months (95% CI, 7.9-8.2), 4.8 months (95% CI, 4.5-5.1) following 2L therapy, and 4.1 months (95% CI, 3.7-4.6) following 3L therapy (Figure 3A–C). When stratified by 1L anti-PD-L1 treatment, median rwOS was 8.3 months (95% CI, 7.9-8.8) in those treated with an anti-PD-L1 in the 1L, and 8.0 months (95% CI, 7.8-8.2) among those not treated with anti-PD-L1 (Figure 4A). When limited to patients treated in 2018 or later, median rwOS in the 1L anti-PD-L1 treated group remained the same (all were treated in 2018+), while median rwOS was 8.0 (95% CI, 7.6-8.4) months among the 763/3159 (24%) patients who were not treated with anti-PD-L1 in 2018 and beyond (data not shown). Additionally, among the 1L anti-PD-L1 treated, those who received 1L maintenance with an anti-PD-L1 had a median rwOS of 11.4 months (95% CI, 10.6-12.2), while those who did not receive maintenance therapy had a median rwOS of 3.8 (95% CI, 3.3-4.3) months (Supplementary Figure S1).

The 12-month rwOS probabilities after 1L therapy of patients with ES-SCLC were 33.0% (95% CI, 30.2-35.9) with anti-PD-L1 therapy, and 28.1% (95% CI, 26.5-29.8) among those without anti-PD-L1 therapy (Figure 4A). At 24 months after 1L therapy, rwOS was 14.9% (95% CI, 12.7-17.5) of those who received anti-PD-L1 therapy, and 9.3% (95% CI, 8.2-10.4) of those without anti-PD-L1 therapy. In both treatment groups (with or without 1L anti-PD-L1), younger age (<65 years) and better ECOG performance status were associated with longer survival (Supplementary Figure S1). Within almost all subgroups of age and ECOG performance status, overall survival was numerically longer among patients who received versus did not receive 1L anti-PD-L1 therapy, and these differences were more pronounced with increased duration of follow-up (eg, 12 and 24 months following 1L initiation).

Following 2L, median rwOS was 5.6 months (95% CI, 4.9-6.3) among those with 1L anti-PD-L1 and 4.5 months (95% CI, 4.2-4.9) among those without (Figure 4B). Finally, following 3L, patients treated with anti-PD-L1 in 1L had a median rwOS of 4.9 months (95% CI, 3.4-6.0) and those not treated with anti-PD-L1 in 1L had a median rwOS of 4.0 months (95% CI, 3.7-4.5; Figure 4C).

Discussion

In this retrospective cohort study using a real-world database in the United States through 2022, we evaluated recent treatment patterns over time and real-world outcomes among patients with ES-SCLC. We explored the impact of exposure to anti-PD-L1 treatment in the 1L setting on these outcomes.

Treatment patterns were in line with what was expected. Etoposide plus platinum remained the mainstay of 1L treatment, with the use of anti-PD-L1s increasing over time to now being added to the majority of 1L regimens. Second-line regimens varied by platinum sensitivity status and shifted from topotecan to lurbinectedin over time. Patients most sensitive to platinum therapy in the 1L (CTFI ≥180 days) had proportionally more platinum use in the 2L setting. The accelerated approvals of nivolumab in 2018 and pembrolizumab in 2019, and later withdrawals in 2021 after subsequent confirmatory studies did not meet their primary endpoints of OS,^14–16 are reflected by a 2018 spike and subsequent reduction in “other immunotherapies” in the 2L and 3L settings. Third-line treatments vary because there are currently no approved therapies in 3L and beyond.

Between the years 2006 and 2018, platinum-based regimens were found to be the most common 1L therapy across a number of captured publications.¹¹ From 2014 to 2016, the standard of 1L treatment for ES-SCLC in the United States was EP (87%), followed by other platinum therapy (7%), non-platinum therapy (4%), and platinum plus irinotecan (2%).¹⁷ A systematic literature review of real-world treatment patterns and outcomes outside of the United States reported a similar rate of EP use in >80% of 1L patients.¹⁸ The current study aligns with these previous studies before 2018, where platinum-based regimens were common before the approval of anti-PD-L1s in 2019-2020.⁴ Second-line treatment options reported in real-world studies were limited.^11,18 A systematic literature review on treatment patterns from 2000 to 2020 found limited data in the 2L setting and beyond but consistency with guidelines where most patients were receiving a topoisomerase inhibitor (irinotecan or topotecan) or retreatment with their previous 1L regimen.¹⁸ This was similarly observed in our study with a shift in preference to lurbinectedin after 2020.

Real-world OS rates following 1L-3L therapy among patients receiving or not receiving 1L anti-PD-L1-treatment were marginally different but still remained poor in both groups, with larger numeric differences observed at longer durations of follow-up in this descriptive study. Median rwOS rates from our study were numerically lower than those observed in the clinical trial setting,^7,8,18 and in other real-world studies, which ranged from 8.9 to 11.6 months,^19–21 although no formal comparisons can be made. Lower OS rates in real-world cohorts in general are not unexpected given real-world cohorts are generally sicker than the patients enrolled in clinical studies (eg, no requirements for minimum life expectancy, suitability for treatment, or ECOG performance status requirements of 0-1 in clinical trials). Patients treated with 1L anti-PD-L1 therapy during induction and maintenance had a numerically longer median rwOS of 11.4 months and those who did not receive maintenance therapy had a median rwOS of 3.8 months. In our analysis, 61% of patients who received frontline anti-PD-L1 therapy received single-agent anti-PD-L1 maintenance. Those who received maintenance therapy in the current study were slightly younger (median age: 67 vs 69 years) and had numerically lower ECOG performance status (74% vs 61% with ECOG performance status 0-1; data not shown).

In an exploratory analysis of the IMpower133 trial, there were 154/201 (77%) patients who went on to receive atezolizumab maintenance therapy.²² The median OS of the group who received maintenance was 15.7 months,²² in contrast to the median OS of 12.3 months⁸ among all 201 atezolizumab patients (results for the nonmaintenance group not reported). An ECOG performance status of 0, younger age, and the absence of liver metastases at baseline were associated with reaching maintenance.²² The higher median rwOS observed in patients who received maintenance in our study may be a reflection of the baseline characteristics of that subgroup.

Real-world overall survival was assessed by other key subgroups of the study population. In the current study, approximately 31% of patients with available ECOG scores had ECOG performance status of ≥2 at index, and performance status was balanced between treatment groups at 1L initiation (68.9% and 68.8% had performance status 0-1 in the 1L anti-PD-L1-treated vs no 1L anti-PD-L1 group, respectively). Restricting the sample to the subgroup of patients with ECOG performance status 0-1 resulted in a median rwOS of 9.2 (95% CI, 8.6-10.2) months in the 1L anti-PD-L1-treated group and 8.6 (95% CI, 8.2-9.1) months in those not treated with anti-PD-L1 therapy. Shorter rwOS rates were observed in the performance status ≥2 groups, highlighting performance status as a known prognostic factor, with median rwOS following 1L anti-PD-L1 therapy of 6.6 months and 5.9 months in those not treated with an anti-PD-L1. Longer survival was observed in the younger age groups, with numerically longer survival observed in the younger group treated with 1L anti-PD-L1 (Supplementary Figure S1). Information on some other prognostic factors, such as metastases, was not available.

Few real-world studies have investigated the impact of the introduction of anti-PD-L1 therapy to the 1L setting in real-world clinical practice. One retrospective study of 102 patients with ES-SCLC treated at a single institution in Israel from October 2017 to July 2021 compared patients who received chemoimmunotherapy versus chemotherapy alone and found that those receiving chemoimmunotherapy had significantly longer median OS from diagnosis (11.6 vs 6.4 months; HR, 0.40; P < .0001).¹⁹ Patients in the chemoimmunotherapy group were significantly younger, had lower Ki-67 protein proliferation rate, and better performance status, and most (87%) began chemotherapy-only in the first cycle with immunotherapy being added in future cycles (median timing, second cycle). Models were adjusted for ECOG, congestive heart failure, lactate dehydrogenase, and albumin levels, but not for age, metastatic burden, or other known confounders. Other studies have estimated OS following chemoimmunotherapy in the real-world setting without comparison to a chemotherapy-only group. One study of 200 patients with ES-SCLC from 10 cancer centers in Switzerland treated with 1L anti-PD-L1 plus chemotherapy from October 2018 to October 2021 estimated a median OS of 11.0 months²⁰; while a study in the United Kingdom of 192 patients with ES-SCLC treated from January 2020 to September 2021 estimated a median OS of 8.9 months.²¹

Although SCLC has a high tumor mutational burden related to the carcinogenic activity of tobacco exposure, this does not translate to a similar favorable response to PD-L1 and PD-1 inhibitors as in NSCLC,²³ and other alternative immunotherapeutic approaches may be of interest. Clinical evaluation of novel agents alone and in combination with chemotherapy for the treatment of SCLC is ongoing.^24–28

Although our study is strictly descriptive in nature, the potential for missing data in healthcare records and selection bias may pose limitations for the analysis. Data were not available for all known or unknown confounders of survival in this patient population, such as metastases. Although treatment groups were similar demographically, there may have been other differences between the 2 groups that we were not able to characterize in this study. Time periods of treatment differed between the 2 groups such that the majority of non-anti-PD-L1 treated patients (76%) were treated prior to 2018 and all anti-PD-L1 patients were treated in 2018 or later; however, similar median rwOS results were observed when we limited the sample to patients treated in 2018 or later. Additionally, inferences made from this population may not be generalizable to treatment practices and outcomes of patients with SCLC outside of the United States, or to patients from academic centers or clinical trial populations, which implement strict enrollment criteria that many patients in the real world would not meet.

Despite these limitations, Flatiron Health data quality is high and undergoes rigorous quality control processes. Patients with SCLC in Flatiron have been found to be comparable to those in SEER data with respect to demographics and outcomes²⁹ and include real-world patients who may be poorly represented in clinical trials. Additionally, this study had the largest sample size to date of real-world patients with SCLC in this current treatment scenario. Finally, the mortality endpoint in the Flatiron Health EDM has been previously validated against a gold standard, and sensitivity and specificity are high.¹³

Conclusion

Despite the introduction of anti-PD-L1 therapies for 1L treatment of ES-SCLC, patient outcomes remain dismal. There is a critical need to develop novel approaches to improve survival outcomes for patients with ES-SCLC.

Supplementary Material

Supplementary material is available at The Oncologist online.

Acknowledgments

The authors thank Lisa R. Denny, PhD, and Maryann T. Travaglini, PharmD (ICON, Blue Bell, PA), whose work was funded by Amgen Inc. for medical writing assistance in the preparation of this manuscript.

Author Contributions

Jaime Shaw: conception/design, provision of study material or patients, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript. Xerxes Pundole: conception/design, provision of study material or patients, collection and/or assembly of data, data analysis and interpretation, final approval of manuscript. Akhila Balasubramanian: conception/design, data analysis and interpretation, final approval of manuscript. Erik S. Anderson: data analysis and interpretation, final approval of manuscript. Malaika Pastel: data analysis and interpretation, final approval of manuscript. Gwyn Bebb: data analysis and interpretation, final approval of manuscript. Tony Jiang: data analysis and interpretation, final approval of manuscript. Pablo Martinez Rodriguez: data analysis and interpretation, final approval of manuscript. Suresh S. Ramalingam: data analysis and interpretation, final approval of manuscript. Hossein Borghaei: data analysis and interpretation, final approval of manuscript.

Funding

The study was sponsored and funded by Amgen Inc.

Conflict of Interest

H.B. reports research support (clinical trials) from BMS, Lilly, and Amgen; advisory board/consultant from BMS, Lilly, Genentech, Pfizer, Merck, EMD-Serono, Boehringer Ingelheim, Astra Zeneca, Novartis, Genmab, Regeneron, BioNTech, Amgen, Axiom, PharmaMar, Takeda, Mirati, Daiichi, Guardant, Natera, Oncocyte, Beigene, iTEO, Jazz, Janssen, Da Volterra, Puma, BerGenBio, Bayer, and Iobiotech; Data and Safety Monitoring Board from University of Pennsylvania: CAR T Program, Takeda, Incyte, Novartis, and Springworks; employment from Fox Chase Cancer Center; Scientific Advisory Board from Sonnetbio (Stock Options), Inspirna (formerly Rgenix, Stock Options), and Nucleai (Stock Options); honoraria from Amgen, Pfizer, Daiichi, and Regeneron; travel from Amgen, BMS, Merck, Lilly, EMD-Serono, Genentech, and Regeneron. J.S., X.P., A.B., E.S.A., T.J., M.P., and P.M. are Amgen employees with stock ownership in Amgen. G.B. is a former employee of Amgen with stock ownership in Amgen. S.S.R.: research support to institution: Amgen, Astra Zeneca, BMS, Merck, Pfizer, and Takeda

Data Availability

Qualified researchers may request data from Amgen clinical studies. Complete details are available at the following: http://www.amgen.com/datasharing. The data that support the findings of this study originated by Flatiron Health, Inc. and Foundation Medicine, Inc. Requests for data sharing by license or by permission for the specific purpose of replicating results in this manuscript can be submitted to [email protected] and [email protected].

References

Author notes