Five-Year Outcomes From the Randomized, Phase III Trials CheckMate 017 and 057: Nivolumab Versus Docetaxel in Previously Treated Non-Small-Cell Lung Cancer.

PURPOSE: Immunotherapy has revolutionized the treatment of advanced non-small-cell lung cancer (NSCLC). In two phase III trials (CheckMate 017 and CheckMate 057), nivolumab showed an improvement in overall survival (OS) and favorable safety versus docetaxel in patients with previously treated, advanced squamous and nonsquamous NSCLC, respectively. We report 5-year pooled efficacy and safety from these trials.
METHODS: Patients (N = 854; CheckMate 017/057 pooled) with advanced NSCLC, ECOG PS ≤ 1, and progression during or after first-line platinum-based chemotherapy were randomly assigned 1:1 to nivolumab (3 mg/kg once every 2 weeks) or docetaxel (75 mg/m2 once every 3 weeks) until progression or unacceptable toxicity. The primary end point for both trials was OS; secondary end points included progression-free survival (PFS) and safety. Exploratory landmark analyses were investigated.
RESULTS: After the minimum follow-up of 64.2 and 64.5 months for CheckMate 017 and 057, respectively, 50 nivolumab-treated patients and nine docetaxel-treated patients were alive. Five-year pooled OS rates were 13.4% versus 2.6%, respectively; 5-year PFS rates were 8.0% versus 0%, respectively. Nivolumab-treated patients without disease progression at 2 and 3 years had an 82.0% and 93.0% chance of survival, respectively, and a 59.6% and 78.3% chance of remaining progression-free at 5 years, respectively. Treatment-related adverse events (TRAEs) were reported in 8 of 31 (25.8%) nivolumab-treated patients between 3-5 years of follow-up, seven of whom experienced new events; one (3.2%) TRAE was grade 3, and there were no grade 4 TRAEs.
CONCLUSION: At 5 years, nivolumab continued to demonstrate a survival benefit versus docetaxel, exhibiting a five-fold increase in OS rate, with no new safety signals. These data represent the first report of 5-year outcomes from randomized phase III trials of a programmed death-1 inhibitor in previously treated, advanced NSCLC.

INTRODUCTION

Historically, 5-year survival rates of patients with advanced non–small-cell lung cancer (NSCLC) who received chemotherapy were < 5%.[1] Effective treatment options for patients without targetable molecular alterations, particularly for those who progressed after first-line chemotherapy, were limited until recently. With clinically meaningful survival benefits, durable responses, and favorable safety profiles versus chemotherapy, immune checkpoint inhibitors have become the standard of care for patients who progressed on or after platinum-based chemotherapy.[2-7] Immune checkpoint inhibitors are also effective as first-line treatment and are recommended, with or without chemotherapy, as the standard of care for treatment-naive patients with advanced NSCLC.[8-14]

CONTEXT

Key Objective

Immune checkpoint inhibitors have improved patient survival versus chemotherapy in previously treated, advanced non–small-cell lung cancer (NSCLC). However, data on long-term outcomes are limited. Using pooled data from two randomized phase III clinical trials (CheckMate 017 and CheckMate 057), we assessed 5-year efficacy and safety outcomes with nivolumab versus chemotherapy in this setting.

Knowledge Generated

Based on these first 5-year results of programmed death-1 inhibitors from phase III clinical trials in the previously treated, advanced NSCLC setting, patients derived long-term survival benefit and durable responses with nivolumab versus chemotherapy, regardless of histology and PD-L1 expression. Nivolumab maintained a favorable safety profile in this patient population; no new safety signals were identified. Furthermore, some patients experienced prolonged disease control even after stopping nivolumab.

Relevance

With the longest follow-up to date for randomized phase III trials of programmed death-1 inhibitors in previously treated, advanced NSCLC, these results represent an important advancement in the treatment of lung cancer and help inform clinical decisions.

Nivolumab, a fully human, monoclonal, antiprogrammed death-1 (PD-1) antibody, was the first PD-1 inhibitor to demonstrate clinically meaningful activity in NSCLC.[15] Nivolumab is approved in the United States, the European Union, and other countries for second-line treatment of advanced NSCLC, based on improved overall survival (OS) and a favorable safety profile versus docetaxel in two randomized, open-label, phase III trials in advanced squamous (CheckMate 017; NCT01642004) and nonsquamous (CheckMate 057; NCT01673867) NSCLC with disease progression following platinum-based chemotherapy.[4,5,16,17]

At 2-, 3-, and 4-year follow-ups, OS rate and progression-free survival (PFS) rate from these trials continued to favor nivolumab over docetaxel, with no new safety signals identified for nivolumab.[18-20] Here, we present the pooled 5-year survival and safety data from CheckMate 017 and 057, representing the longest follow-up to date for randomized phase III trials of an immune checkpoint inhibitor in previously treated, advanced NSCLC.

METHODS

Patients

Eligibility criteria for both trials have been previously described.[4,5]

Study Design

CheckMate 017 (previously treated squamous NSCLC) and CheckMate 057 (previously treated nonsquamous NSCLC) were international, randomized, open-label, phase III trials. Patients were randomly assigned 1:1 to receive nivolumab (3 mg/kg once every 2 weeks) or docetaxel (75 mg/m2 once every 3 weeks) in both trials (Appendix Fig A1, online only). Random assignment was stratified by prior paclitaxel use and geographical location in CheckMate 017 and by prior maintenance treatment and line of therapy (second v third) in CheckMate 057.

FIG A1.

Study design.a,b aNCT01642004; database lock: May 8, 2019; minimum follow-up for OS, 64.2 months; bNCT01673867; database lock: May 16, 2019; minimum follow-up for OS, 64.5 months; cOptional switch to nivolumab 480 mg every 4 weeks allowed as per the protocol amendment in September 2016; dAfter completion of the primary analyses, patients in the docetaxel arms who ended treatment at any time during the trials were allowed to cross over to nivolumab; eDefined by RECIST 1.1; patients receiving nivolumab may be treated beyond progression under protocol-defined circumstances; fAs assessed by investigator. ALK, anaplastic lymphoma kinase; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; IV, intravenous; NSCLC; non–small-cell lung cancer; NSQ, nonsquamous; ORR, objective response rate; OS, overall survival; PD-L1, programmed death ligand 1; PFS, progression-free survival; PROs, patient-reported outcomes; Q2W, every 2 weeks; Q3W, every 3 weeks; R, randomized; SQ, squamous; TKI, tyrosine kinase inhibitor. Reprinted from Annals of Oncology, 29(4), Waterhouse M, Domine M, Garassino LQM, et al, “Nivolumab Versus Docetaxel in Previously Treated Advanced Nonsmall-Cell Lung Cancer (CheckMate 017 and CheckMate 057): 3-Year Update and Outcomes in Patients With Liver Metastases,” 959-965, 2018, with permission from Elsevier.

Treatment continued until disease progression, unacceptable toxicity, or other protocol-specified reasons. Further details on treatment beyond progression in the nivolumab group and crossover in the docetaxel group are given in the Appendix (online only).

Both trials were conducted in accordance with the International Conference on Harmonisation Good Clinical Practice Guidelines and the Declaration of Helsinki. An institutional review board or independent ethics committee at each site approved the trial Protocols (online only). All patients provided written informed consent.

Assessments

Tumor assessments were performed by investigators according to RECIST v1.1 at baseline, at 9 weeks, every 6 weeks thereafter during the first year of treatment, and then every 12 weeks until disease progression or discontinuation of therapy in patients receiving nivolumab beyond progression. Patients were followed continuously for survival while receiving treatment and every 3 months after discontinuation.

Safety was assessed throughout the treatment period and at two follow-up visits, which occurred within 100 days of last dose or before the start of crossover treatment. Beyond 100 days from the last dose of treatment, patients with ongoing treatment-related adverse events (TRAEs) were followed until the TRAE resolved, returned to baseline, or was deemed irreversible. The severity of adverse events (AEs) was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.0). Select AEs were defined as having a potential immunologic cause that may require management through immune-modulating medication.

Archival or recent pretreatment tumor biopsy specimens were assessed for expression of PD-1 ligand 1 (PD-L1) protein at a central laboratory using a validated automated immunohistochemical assay (PD-L1 IHC 28-8 pharmDx; Dako, Carpinteria, CA) as previously described.[4,5]

Statistical Analyses

Efficacy and safety were assessed in all randomly assigned patients and in all patients who received at least one dose of the trial drug, respectively, using pooled data from CheckMate 017 and 057 studies. The primary end point was OS for both studies; secondary end points included objective response rate (ORR), PFS, and efficacy by tumor PD-L1 expression. Data for these end points have been previously reported.[4,5] To investigate the impact of progression-free status on long-term survival, an exploratory landmark analysis of OS at 5 years based on progression-free status at 2, 3, and 4 years was performed. The probability of patients remaining progression-free at later timepoints based on their progression-free status at 2, 3, and 4 years was also assessed.

Survival curves and rates, landmark analyses, and duration of response (DOR) were estimated using the Kaplan-Meier method. Hazard ratios (HRs) and CIs were estimated using a Cox proportional hazard model.

RESULTS

At the database locks (May 8, 2019, for CheckMate 017 and May 16, 2019, for CheckMate 057) for this analysis, the minimum follow-up was 64.2 months and 64.5 months, respectively; the corresponding median follow-up was 69.5 months and 69.4 months.

Patients and Treatment

Baseline characteristics of patients randomly assigned to nivolumab (CheckMate 017: n = 135; CheckMate 057: n = 292) and docetaxel (CheckMate 017: n = 137; CheckMate 057: n = 290) were generally well balanced (Appendix Table A1, online only).[4,5,18-20]

TABLE A1.

Baseline Characteristics in the Pooled CheckMate 017 and 057 Population and by Trial

Patient disposition is summarized in the Appendix Figure A2, online only. Following a protocol amendment, 23 nivolumab-treated patients transitioned to nivolumab 480 mg once every 4 weeks.[20] Of 427 docetaxel-treated patients, 23 crossed over to receive nivolumab 3 mg/kg once every 2 weeks. Two of these patients subsequently received nivolumab 480 mg once every 4 weeks per protocol amendments. At 5 years, 50 of 427 patients randomly assigned to nivolumab and 9 of 427 patients randomly assigned to docetaxel were still alive; 18 of 418 (4.3%) nivolumab-treated patients remained on treatment for ≥ 5 years; no patients remained on docetaxel. The median (range) number of nivolumab (3 mg/kg) and docetaxel doses in CheckMate 017 was 8.0 (1-151) and 3.0 (1-29), respectively, and in CheckMate 057, the median (range) was 6.0 (1-139) and 4.0 (1-23), respectively.

FIG A2.

CONSORT diagram of patient disposition for (A) CheckMate 017 and (B) CheckMate 057. AE, adverse event.

OS

In the pooled CheckMate 017/057 population, OS remained longer with nivolumab versus docetaxel (HR: 0.68; 95% CI, 0.59 to 0.78). Pooled 5-year OS rates were 13.4% (95% CI, 10.4 to 16.9) with nivolumab versus 2.6% (95% CI, 1.4 to 4.5) with docetaxel (Fig 1A). Consistent with previous reports, most deaths between 3 and 5 years with nivolumab (12 of 14 deaths) and docetaxel (20 of 23 deaths) were due to disease.[18,19]

FIG 1.

OS of all treated patients: (A) overall, (B) by SQ tumor histology, (C) by NSQ tumor histology, (D) by ≥ 1% PD-L1 expression, and (E) by < 1% PD-L1 expression. Minimum follow-up: CheckMate 017: 64.2 months; CheckMate 057: 64.5 months. HR, hazard ratio; mo, months; No., number; NSQ, nonsquamous; OS, overall survival; PD-L1, programmed death ligand 1; SQ, squamous.

Pooled OS rates at 5 years were similar with squamous and nonsquamous histology: 12.3% (95% CI, 7.4 to 18.5) and 14.0% (95% CI, 10.2 to 18.3) with nivolumab and 3.6% (95% CI, 1.4 to 7.8) and 2.1% (95% CI, 0.9 to 4.4) with docetaxel, respectively (Figs 1B and 1C). OS benefit continued to be observed with nivolumab versus docetaxel regardless of tumor PD-L1 expression (Figs 1D and 1E); 5-year OS rates were 18.3% (95% CI, 13.0 to 24.2) versus 3.4% (95% CI, 1.4 to 6.8) in patients with PD-L1 expression ≥ 1% and 8.0% (95% CI, 4.4 to 13.0) versus 2.0% (95% CI, 0.5 to 5.3) in those with PD-L1 expression < 1%.

OS benefit was observed with nivolumab across several subgroups, including patients with baseline liver metastases (HR, 0.67 [95% CI, 0.50 to 0.89]), adrenal metastases (HR, 0.41 [95% CI, 0.27 to 0.60]), neutrophil-to-lymphocyte ratio < median (HR, 0.63 [95% CI, 0.51 to 0.77]), lactate dehydrogenase ≥ upper limit of normal (HR, 0.74 [95% CI, 0.59 to 0.93]), and those with no baseline proton pump inhibitor use (HR, 0.61 [95% CI, 0.51 to 0.72]; Appendix Fig A3, online only).

FIG A3.

Forest plot of OS in predefined subgroups. Hazard ratios were not reported for subgroups with fewer than 10 patients per treatment group. aNot reported in two and one patients with nivolumab and docetaxel, respectively. bUnknown in seven and four patients with nivolumab and docetaxel, respectively. cMedian NLR was 4.80. NLR was not reported in two patients each in nivolumab and docetaxel arms. dNot reported in three and two patients with nivolumab and docetaxel, respectively. ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; LDH, lactate dehydrogenase; NLR, neutrophil-to-lymphocyte ratio; OS, overall survival; PD-L1, programmed death ligand 1; PPI, proton pump inhibitor; ULN, upper limit of normal.

PFS

PFS rates consistently favored nivolumab versus docetaxel over time (Fig 2A). Pooled 5-year PFS rates were 8.0% (95% CI, 5.4 to 11.2) with nivolumab and 0% with docetaxel. PFS rates by histology and for patients with PD-L1 expression ≥ 1% and < 1% are shown in the Appendix Figure A4, online only.

FIG 2.

(A) PFSa and (B) DORa among all treated patients. aPer local investigator; minimum follow-up: CheckMate 017: 64.2 months; CheckMate 057: 64.5 months. Since the primary analysis of the CheckMate 057 trial, one patient's response changed from SD to PR and one from PR to CR. DOR for these two patients was determined according to their latest response category. DOR, duration of response; HR, hazard ratio; mo, months; No., number; PFS, progression-free survival.

FIG A4.

PFS in patients with (A) SQ tumor histology, (B) NSQ tumor histology, (C) ≥ 1% PD-L1 expression, and (D) < 1% PD-L1 expression. aPer local investigator; minimum follow-up: CheckMate 017: 64.2 months and CheckMate 057: 64.5 months. HR, hazard ratio; NC, not calculable; No., number; NSQ, nonsquamous; PD-L1, programmed death ligand 1; PFS, progression-free survival; SQ, squamous.

Landmark Survival Analyses

Landmark analysis of PFS and OS by progression-free status at 2, 3, and 4 years showed that a high proportion of nivolumab-treated patients remained progression-free during subsequent years and had long-term OS benefits (Fig 3). Patients who were progression-free at 2 years (n = 45), 3 years (n = 29), and 4 years (n = 25) had a 59.6%, 78.3%, and 87.5% chance of being progression-free at 5 years, respectively, and an 82.0%, 93.0%, and 100.0% chance of survival at 5 years, respectively. In the docetaxel arm, patients who were progression-free at 2 years (n = 4) and 3 years (n = 1) had a 0% chance of being progression-free at 5 years and a 0% chance of survival at 5 years; no patients were progression-free at 4 years.

FIG 3.

PFS and OS landmark analyses by PFS at 2, 3, and 4 years. aBased on Kaplan-Meier estimates; bNumber of patients at risk. OS, overall survival; PFS, progression-free survival.

Tumor Response

Consistent with previous reports,[4,5,18,19] the pooled ORR was higher with nivolumab (19.7% [95% CI, 16.0 to 23.8]) than docetaxel (11.2% [95% CI, 8.4 to 14.6]; Appendix Table A2, online only). Since the primary analysis of CheckMate 057, one patient treated with nivolumab improved from stable disease to partial response (PR) and another, also treated with nivolumab, improved from PR to complete response (CR). No patients from CheckMate 017 experienced a change in the response since the primary analysis.

TABLE A2.

Tumor Response in All Randomly Assigned Patients and by Trial

Median DOR was longer with nivolumab (19.9 months [95% CI, 11.4 to 30.8]) versus docetaxel (5.6 months [95% CI, 4.4 to 7.0]) in the pooled population. Longer DOR with nivolumab was observed regardless of histology or tumor PD-L1 expression (Appendix Fig A5, online only). The pooled 5-year DOR rate with nivolumab was 32.2% (95% CI, 21.9 to 43.0); no patients in the docetaxel arm had ongoing responses at 5 years (Fig 2B).

FIG A5.

DORa in all treated patients with (A) SQ tumor histology, (B) NSQ tumor histology, (C) ≥ 1% PD-L1 expression, and (D) < 1% PD-L1 expression. aPer local investigator. DOR, duration of response; NC, not calculable. No., number; NSQ, nonsquamous; PD-L1, programmed death ligand 1; SQ, squamous.

5-Year Survivors

Baseline characteristics of patients who survived ≥ 5 years in the nivolumab arm (n = 50) and docetaxel arm (n = 9) were generally similar to the overall population and those who survived < 1 year (n = 222 and n = 282, respectively), despite numerical differences in ECOG PS 0 (in both arms), PD-L1 expression ≥ 1% (nivolumab arm), and stage IIIB NSCLC (docetaxel arm; Appendix Fig A3).

Among the 50 patients who survived ≥ 5 years in the nivolumab arm (including 18 who had switched to nivolumab 480 mg once every 4 weeks), 21 (42.0%) had not progressed by 5 years and 21 (42.0%) had progressed (Fig 5A; Appendix Fig A6, online only), and eight (16.0%) had been censored for PFS. The median (range) duration of treatment for nivolumab and docetaxel in 5-year survivors was 36.9 months (1.8-76.2+ months) and 3.5 months (0.7-20.0 months), respectively; 35 patients received nivolumab treatment for ≥ 2 years and 18 remained on nivolumab at 5 years. Of the 32 patients who had discontinued nivolumab, the median duration of treatment was 27.7 months. Aside from disease progression, reasons for discontinuation included TRAEs, AEs unrelated to the study drug, or maximum clinical benefit.

FIG 5.

Treatment status of ≥ 5-year survivors treated with (A) nivolumab and (B) docetaxel. Timing of only the first administered subsequent therapy is noted, patients may have received ≥ 1 subsequent treatment. aPatient received radiotherapy prior to discontinuation of nivolumab; bPatient received radiotherapy in addition to nivolumab; cPatient withdrew consent, and so subsequent treatment status is unknown. dPatient response note reported. CR, complete response; No., number; PD, progressive disease; PR, partial response.

FIG A6.

Treatment status of survivors at 5 years. aMedian (range) nivolumab treatment duration: 36.9 (1.8-76.2 +) months. bThe other 3 patients progressed (1.8, 4.5, and 44.2 months, respectively) after discontinuing nivolumab treatment. cAfter nivolumab treatment. dInformation on subsequent therapy was not available as of database lock. eNivolumab treatment durations for individual patients: 8.8, 21.7, 36.6, 43.3, and 43.5 months. fMedian (range) nivolumab treatment duration: 68.4 (62.9-76.2 +) months. gIncludes two patients treated with nivolumab as first subsequent therapy. hBecause of the receipt of subsequent therapy. AE, adverse event; TRAEs, treatment-related adverse events.

Of the patients who survived ≥ 5 years in the nivolumab arm (n = 50), 5 patients had CRs and 34 patients had PRs. A total of eight and three patients had stable and progressive disease, respectively. In the docetaxel arm, four of the 5-year survivors (n = 9) had a PR, two patients had stable disease, and three patients had progressive disease. No docetaxel-treated survivors had a CR.

A total of 24 nivolumab-treated patients were known to receive subsequent therapy, of whom 10 had subsequent immunotherapy (Appendix Tables A4 and A5, online only). At 5 years, 5 of 50 nivolumab-treated patients were progression-free and did not require subsequent therapy (Appendix Fig A6); reasons for discontinuing nivolumab (after 8.8-43.5 months of treatment) were TRAEs (n = 3), maximum clinical benefit (n = 1), and AE unrelated to study drug (n = 1). Among the 9 patients who survived ≥ 5 years in the docetaxel arm (including two patients who crossed over to receive nivolumab 3 mg/kg once every 2 weeks and one who received 3 mg/kg and 480 mg once every 4 weeks), eight had progressed and one was censored for PFS. All nine patients received subsequent therapy; four had subsequent immunotherapy (excluding patients who crossed over to nivolumab; Appendix Table A4; Fig 5B).

TABLE A4.

Subsequent Therapies Received by 5-Year Survivors

TABLE A5.

CheckMate 017 Investigators

Safety

No patients received treatment with docetaxel for more than 2 years; therefore, updated safety data as of the 5-year follow-up are presented only for patients who received nivolumab. At 5 years, 284 of 418 patients (67.9%) treated with nivolumab experienced TRAEs; 45 patients (10.8%) had grade 3-4 events. No new safety signals were observed. Between 3- and 5-year minimum follow-ups, eight of 31 patients (25.8%) still receiving nivolumab had TRAEs (Table 1), of whom one patient (3.2%) had a grade 3 event (increased lipase); there were no grade 4 events. A total of 13 different events were reported in these 8 patients between 3 and 5 years of treatment, of which three events were recurrent (nummular eczema, pruritus, and rash occurring in one patient each).

TABLE 1.

Treatment-Related Adverse Events With Nivolumab (Overall and at 3-5 Years' Follow-Up)

Overall, 27 (6.5%) nivolumab-treated patients experienced TRAEs of any grade leading to discontinuation; the most common (in ≥ 2 patients) were pneumonitis (n = 6; 1.4%) and interstitial lung disease (n = 3; 0.7%), and colitis, increased alanine aminotransferase, increased AST, and rash (n = 2; 0.5%). Since the 3-year follow-up, one patient in the nivolumab arm experienced a TRAE, leading to discontinuation (grade 2 nummular eczema).[19] At the time of database lock, no new treatment-related deaths had occurred since the primary analyses (n = 1 in the nivolumab arm and n = 4 in the docetaxel arm).[4,5]

Consistent with previous reports, few treatment-related select AEs occurred after the 3-year minimum follow-up (Fig 4).[18,19] Of the 31 patients who remained on treatment with nivolumab between 3 and 5 years of follow-up, five patients (16.1%) experienced treatment-related select AEs: four patients (12.9%) with skin or subcutaneous tissue disorders (one each of grade 1-2 erythema, pruritus, rash, and skin exfoliation) and two patients (6.5%) with a GI disorder (grade 1-2 diarrhea). A total of eight different events were reported in these five patients between 3 and 5 years of treatment, of which two events were recurrent (pruritus and rash occurring in one patient each).

FIG 4.

Patients on treatment with first treatment-related select AE by yeara,b. Median (range) nivolumab treatment duration: 2.8 (0-76.2 +) months. aIncludes events of any grade reported between the first dose and 30 days after the last dose of trial therapy; bSelect AEs were events with a potential immunological cause. AE, adverse event; No., number.

DISCUSSION

This is the longest follow-up to date for randomized phase III trials of a PD-1 inhibitor in previously treated, advanced NSCLC. After a 5-year minimum follow-up in the CheckMate 017 and 057 studies, nivolumab continued to demonstrate clinically meaningful OS, PFS, and DOR benefits versus docetaxel and maintained a favorable safety profile. The pooled 5-year OS rate was 13.4% with nivolumab, representing a five-fold increase over docetaxel (2.6%). These findings are consistent with previously reported 5-year and 6-year OS rates with nivolumab among patients with previously treated, advanced NSCLC in CheckMate 003 trial (15.6% and 14.7%, respectively).[21] The OS rates detailed here are also similar to the five-year OS rates observed in the single-arm, phase I trial of pembrolizumab in patients with previously treated NSCLC (15.5%).[22] OS benefit with nivolumab versus docetaxel was observed regardless of tumor histology. Notably, OS benefit (HR < 1) was observed with nivolumab versus docetaxel in patients with tumor PD-L1 expression ≥ 1% (5-year OS rates, 18.3% v 3.4%) or < 1% (8.0% v 2.0%) and across a variety of patient subgroups, demonstrating the potential for nivolumab to improve outcomes in a diverse patient population. In this analysis, no baseline clinical or tumor characteristics were identified to clearly distinguish long-term or short-term survivors in either treatment arm and, because of the disparity in sample sizes across treatment arms, multivariate analysis was not considered appropriate; only nine patients were alive in the docetaxel group at 5 years, making subgroup analysis unfeasible.

The pooled 5-year PFS rate with nivolumab versus docetaxel (8.0% and 0%, respectively) was consistent with previous analyses.[20] Notably, the majority of patients without disease progression at 2, 3, and 4 years after treatment with nivolumab remained progression-free at 5 years and survived ≥ 5 years. Although exploratory, these findings provide new information about the probability of remaining progression-free at subsequent timepoints and alive at 5 years, by progression-free status at 2, 3, and 4 years. This analysis provides insight into long-term efficacy outcomes and management of previously treated, advanced NSCLC following treatment with nivolumab. Consistent with the 2- and 3-year follow-ups, responses achieved with nivolumab were durable;[18,19] nearly one-third of patients who achieved an objective response had ongoing responses at 5 years versus none with docetaxel. The 5-year timepoint is considered a clinical landmark to evaluate long-term survival, and data beyond 5 years are scarce; a longer follow-up may be required to assess the outcomes of these patients.[23]

In both this analysis and a pooled analysis across four nivolumab trials in previously treated NSCLC, the proportion of nivolumab-treated patients who remained alive appeared to stabilize at approximately 3 years and plateau thereafter.[20] A similar observation was noted in a pooled analysis of ipilimumab in patients with unresectable or advanced melanoma, where the survival curve extended beyond 5 years.[24] This suggests that long-term survival beyond 5 years may also be possible in NSCLC; however, this remains to be addressed in future analyses. Indeed, patients with previously treated NSCLC who received nivolumab in CheckMate 003, which has the longest survival follow-up to date among trials of PD-1 inhibitors in previously treated, advanced NSCLC, exhibited similar OS rates at 4, 5, and 6 years (15.6%, 15.6%, and 14.7%, respectively).[20] Importantly, no new safety signals were observed with a 5-year follow-up; nivolumab maintained a favorable safety profile versus docetaxel, without long-term toxicity. No evidence of late-onset grade 3-4 treatment-related select AEs was observed.

Among ≥ 5-year survivors in the nivolumab arm (n = 50), the median duration of therapy was 36.9 months and 18 of 50 remained on nivolumab at 5 years, suggesting that some patients may achieve long-term survival with continuous nivolumab treatment. In contrast, median duration off-treatment among the 5-year survivors who had discontinued nivolumab was 41.9 months, and 10.0% (n = 5) of 5-year survivors in the nivolumab arm were off treatment, without subsequent therapy, and had not progressed, suggesting benefit even for patients who stopped nivolumab treatment. Meanwhile, exploratory data from CheckMate 153 suggested a survival benefit with continuous nivolumab treatment beyond 1 year versus stopping treatment at 1 year.[25] The optimal treatment duration of nivolumab and PD-1 inhibitors in general for patients with advanced NSCLC remains to be fully elucidated.

In conclusion, 5-year outcomes from the randomized phase III CheckMate 017 and 057 trials demonstrate that nivolumab can provide long-term survival benefit with durable responses and a tolerable safety profile in patients with previously treated, advanced NSCLC. Furthermore, some patients appear to maintain prolonged disease control even after stopping systemic therapy. These findings represent an important advancement in the treatment of lung cancer.

TABLE A3.

Baseline Characteristics of < 1-Year and ≥ 5-Year Survivors on Nivolumab and Docetaxel in the Pooled CheckMate 017 and 057 Population

TABLE A6.

CheckMate 057 Investigators