CADTH Reimbursement Review

Nivolumab (Opdivo)

Sponsor: Bristol Myers Squibb Canada

Therapeutic area: Gastric, gastroesophageal junction, or esophageal adenocarcinoma

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Stakeholder Input

Clinical Review

Executive Summary

An overview of the submission details for nivolumab is provided in Table 1.

Introduction

Esophagogastric cancers, including gastric cancer (GC), gastroesophageal junction cancer (GEJC), and esophageal cancer (EC), are a heterogeneous group of highly lethal malignancies of the upper gastrointestinal (GI) tract.¹ Most esophagogastric cancers are adenocarcinomas (gastric adenocarcinoma [GAC], gastroesophageal junction adenocarcinoma [GEJAC], and esophageal adenocarcinoma [EAC]). These diseases occur more often in older men² and are often diagnosed at an advanced stage in North America.³ Approximately 40% of patients have advanced or metastatic disease at diagnosis.⁴ In the small proportion (approximately 25%)⁵ of patients who are candidates for potentially curative resection, there is a high probability of recurrence within 1 year to 2 years.⁶ Symptoms — including dysphagia, pain, dyspepsia, reflux, weight loss, bleeding, and anemia⁷ — negatively impact health-related quality of life (HRQoL).⁸

Following diagnosis of de novo, recurrent advanced, or metastatic GAC, GEJAC, or EAC, a subset of patients will be eligible for systemic therapy with palliative intent (typically combination chemotherapy regimens, including a fluoropyrimidine and/or a platinum drug). According to the clinical experts consulted by CADTH for this review, the most important goals of treatment in these patients are prolonging survival, controlling symptoms, and maintaining HRQoL. The clinical experts indicated that leucovorin and 5-fluorouracil plus oxaliplatin (FOLFOX) is the most common regimen used in Canada for first-line therapy for advanced or metastatic GAC, GEJAC, or EAC, although capecitabine plus oxaliplatin (XELOX) and leucovorin and 5-fluorouracil plus irinotecan (FOLFIRI) may also be used in some patients. Existing therapies have limited efficacy, and responses are both infrequent and short-lived. In patients with advanced or metastatic disease, 5-year survival is approximately 4%,^2,9 and median overall survival (OS) when first-line therapy is administered is approximately 7.7 months.¹⁰ In recent years, some patients have received immunotherapies (nivolumab and pembrolizumab) in combination with chemotherapy through registration in trials or special access programs. Programmed death ligand 1 (PD-L1) expression levels and combined positive scores (CPSs) vary in GACs, GEJACs, and EACs, and this may have prognostic significance in patients receiving immunotherapies.¹¹

Nivolumab is a human immunoglobulin (Ig)G4 monoclonal antibody that binds programmed cell death protein 1 (PD-1) on T cells, blocking its interaction with its ligands PD-L1 and programmed death ligand 2 (PD-L2), which are upregulated on tumour cells and potentiate antitumour immune responses. Nivolumab (10 mg/mL) is supplied as 40 mg and 100 mg single-use vials and is administered through IV infusion at a dosage of 360 mg every 3 weeks or 240 mg every 2 weeks (depending on the backbone chemotherapy regimen). The Health Canada–approved indication is for nivolumab, in combination with fluoropyrimidine- and platinum-containing chemotherapy, for the treatment of adult patients with human epidermal growth factor receptor 2 (HER2)-negative advanced or metastatic GAC, GEJAC, or EAC. The indication also includes the following statement: “A positive association was observed between PD-L1 CPS score and the magnitude of treatment benefit.” The reimbursement request initially submitted by the sponsor for review by CADTH, |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||, was for nivolumab, in combination with fluoropyrimidine- and platinum-containing chemotherapy, for the treatment of adult patients with advanced or metastatic GAC, GEJAC, or EAC. During the CADTH review, a Notice of Compliance (NOC) was issued by Health Canada. The approved indication differs from the initially submitted reimbursement request, in that HER2-negative status is specified and the approved indication is therefore narrower. After the NOC was issued, the sponsor confirmed with CADTH that the reimbursement request should be updated to match the approved indication. The sponsor estimated that the total population of patients with HER2-negative GAC, GEJAC, or EAC eligible for treatment with nivolumab plus chemotherapy under the Health Canada–approved indication was approximately 1,551 patients in Canada (excluding Quebec) in 2021.¹² Nivolumab has been previously reviewed by CADTH for a variety of indications (melanoma, classical Hodgkin lymphoma, hepatocellular carcinoma, non–small cell lung cancer, malignant pleural mesothelioma, renal cell carcinoma, and squamous cell cancer of the head and neck) and is currently under CADTH review for adjuvant treatment of completely resected GEJC or EC.

The objective of this report was to perform a systematic review of the beneficial and harmful effects of nivolumab (IV injection over 30 minutes of 360 mg every 3 weeks or 240 mg every 2 weeks), in combination with fluoropyrimidine- and platinum-containing chemotherapy, for the treatment of adult patients with HER2-negative advanced or metastatic GAC, GEJAC, or EAC.

Stakeholder Perspectives

The information in this section is a summary of input provided by the patient groups that responded to CADTH’s call for patient input and from clinical experts consulted by CADTH for the purpose of this review.

Patient Input

Input was provided by 1 patient group for this review (My Gut Feeling – Stomach Cancer Foundation of Canada). My Gut Feeling distributed a survey via email, social media, and online forums to patients with GC, GEJC, or EC, as well as their caregivers, between August 20 and September 9, 2021. Among the 62 respondents (one-half patients and one-half caregivers), most (79%) were female, resided in Canada or the US (63% and 29%, respectively), and had received or were caring for someone who had received a diagnosis of GC (74.2%) and adenocarcinoma (82.3%). The number of respondents with advanced or metastatic disease was unclear. Most respondents (90.3%) reported a significant impact of their cancer on HRQoL, with adverse effects on physical health, mental health, ability to eat, ability to work, finances, social life, identity, and self-image. Some of these impacts extended to caregivers and families as well. Symptoms frequently included weight loss, change in appetite, pain, fatigue, reflux, nausea/vomiting, difficulty swallowing, shortness of breath, bleeding, anemia, ascites, and dumping syndrome. Patients highlighted the limited treatment options for GC, GEJC, and EC and their experiences with prior therapies (surgery, radiation, chemotherapy, and immunotherapy), including variable effectiveness in delaying progression and in controlling symptoms, as well as significant side effects affecting HRQoL (e.g., fatigue, nausea and/or vomiting, appetite changes exacerbating weight loss).

According to patients, an ideal therapy for GC, GEJC, and GC would prolong survival while maintaining or improving HRQoL compared with standard of care. Delaying recurrence or progression and having manageable side effects were also important factors for patients. Patients identified an unmet need for equitable access to therapies that may prolong life, improve symptoms, reduce risk of recurrence, and have improved tolerability. Such treatment options should be available barrier-free for all Canadian patients with GC, GEJC, or EC who could benefit.

Clinician Input

Input From Clinical Experts Consulted by CADTH

Two clinical specialists with expertise in the diagnosis and management of advanced or metastatic GAC, GEJAC, and EAC provided input for this review. According to the experts, current systemic therapies, including combination chemotherapy, are palliative in nature. Only a minority of patients have responses, and responses are often short-lived. Survival is typically less than 1 year; among responding patients, few live beyond 15 months to 18 months. There is an unmet need for more effective therapies with similar or lower toxicity than current chemotherapy options.

According to the experts, nivolumab would be administered in combination with fluoropyrimidine- and platinum-based chemotherapy for first-line treatment in most patients with HER2-negative advanced or metastatic GAC, GEJAC, or EAC who can tolerate chemotherapy. According to the experts, it would not be appropriate for patients to receive other therapies before nivolumab plus chemotherapy because the most effective treatment should be used in the first-line setting. The clinical experts stated that patients with good performance status (PS; Eastern Cooperative Oncology Group [ECOG] PS 0 or 1), patients with higher PD-L1 CPSs, and patients with high microsatellite instability (MSI-H) are most likely to respond to nivolumab plus chemotherapy. Patients with poor PS and PD-L1 CPS less than 1 are least likely to derive benefit from therapy, but, in the opinion of the clinical experts consulted by CADTH for this review, nivolumab plus chemotherapy could be considered for all patients with advanced or metastatic GAC, GEJAC, or EAC for whom local or curative treatment is not possible, regardless of tumour PD-L1 expression or PD-L1 CPS. Patients with active autoimmune diseases are least suitable for treatment with nivolumab due to safety concerns. Treatment would be initiated as soon as possible following diagnosis of advanced or metastatic GAC, GEJAC, or EAC, and response would be assessed by imaging approximately every 3 months. Objective tumour response on imaging, improvement in symptoms, HRQoL, weight, and PS are important parameters in response assessment. Treatment should be discontinued in patients with clear, objective tumour progression assessed by imaging. Treatment intolerance or significant toxicity may also require discontinuation of therapy.

Clinician Group Input

Two clinician groups provided input for this review: the Ontario Health (Cancer Care Ontario) Gastrointestinal Cancer Drug Advisory Committee (5 clinicians) and the Canadian Gastrointestinal Oncology Evidence Network (CGOEN) in collaboration with other physicians treating gastroesophageal cancers (7 clinicians). No major contrary views were presented. Both clinician groups echoed the limited efficacy of available systemic therapies for GAC, GEJAC, or EAC, and the short duration of response in many patients. By contrast with the clinical experts consulted by CADTH for this review, clinicians from the CGOEN felt that patients with PD-L1 CPS less than 5 (rather than < 1) would be least suitable for treatment with nivolumab plus chemotherapy, while those from the Ontario Health (Cancer Care Ontario) Gastrointestinal Cancer Drug Advisory Committee felt that, although patients with PD-L1 CPS of 5 or higher and 1 or higher are more likely to respond, all patients with GAC, GEJAC, or EAC can benefit from addition of nivolumab to chemotherapy.

Drug Program Input

The Provincial Advisory Group (PAG) identified several jurisdictional implementation issues. The PAG asked how FOLFOX and XELOX compared with other chemotherapy regimens for first-line therapy of GAC, GEJAC, or EAC and whether the available evidence could be generalized to combinations of nivolumab with non–platinum-based regimens. The clinical experts consulted by CADTH for this review responded that, with some jurisdictional variation, FOLFOX and XELOX, and to a lesser extent FOLFIRI, are generally preferred in Canada due to lower toxicity and more convenient administration, and that generalizability to other chemotherapy regimens was uncertain. The PAG inquired whether re-treatment with nivolumab (with or without chemotherapy) would be an option for patients who have received first-line nivolumab plus chemotherapy for 2 years and whose disease progresses while off therapy; the clinical experts confirmed that these patients would be offered re-treatment but that, in the absence of evidence, there would be variation in clinical practice in administration of nivolumab alone versus nivolumab plus chemotherapy for re-treatment. The PAG also inquired whether nivolumab plus chemotherapy would be offered to patients who had previously received nivolumab for adjuvant treatment of resected GAC, GEJAC, or EAC; the clinical experts confirmed that these patients would be eligible to receive nivolumab plus chemotherapy. The PAG asked whether patients treated with nivolumab plus chemotherapy could continue monotherapy if either component was discontinued; the clinical experts replied that yes, treatment with either nivolumab or chemotherapy alone would be continued, although it would be more common for patients to discontinue chemotherapy and continue nivolumab. The PAG raised the issue of how prescribers would choose which immunotherapy (e.g., pembrolizumab versus nivolumab) to administer in patients with GAC, GEJAC, or EAC; the clinical experts responded that this would depend on multiple factors, including funding of both immunotherapies, availability of PD-L1 testing, tumour site (GAC versus GEJAC versus EAC), PD-L1 CPS, and familiarity of the oncologist with each drug. The PAG asked whether patients with ECOG PS of 2 or higher would be eligible for nivolumab plus chemotherapy; the clinical experts confirmed that yes, some patients with ECOG PS of 2 or higher would be appropriate candidates for nivolumab plus chemotherapy. For patients currently receiving fluoropyrimidine- and platinum-based chemotherapy, PAG asked what time frame would be appropriate for addition of nivolumab to therapy; the clinical experts replied that 2 cycles to 3 cycles, or at least before the first scan at 3 months, would be appropriate cut-offs.

Clinical Evidence

Pivotal Studies and Protocol-Selected Studies

Description of Studies

CheckMate-649 was a phase III, open label (OL), multi-centre randomized controlled trial (RCT) (N = 1,581)¹³ with a primary objective to compare the efficacy of first-line therapy with nivolumab plus FOLFOX or XELOX versus FOLFOX or XELOX in prolonging OS and progression-free survival (PFS) per blinded independent central review (BICR) in patients with advanced or metastatic GAC, GEJAC, or EAC (all with PD-L1 CPS ≥ 5). Secondary objectives included comparing OS and PFS by BICR in patients with PD-L1 CPS of 1 or higher (OS: hierarchically tested), all randomized patients (OS: hierarchically tested), and patients with PD-L1 CPS of 10 or higher and comparing objective response rates (ORRs) in patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, all randomized patients, and patients with PD-L1 CPS of 10 or higher. Changes in HRQoL (measured using the patient-reported EQ-5D-3L and functional assessment of cancer therapy – gastric [FACT-Ga] instruments, including the FACT-Ga gastric cancer subscale [GaCS]), were assessed in exploratory fashion. Patients had to be 18 years of age or older with inoperable advanced or metastatic, HER2-negative or HER2-unreported GAC, GEJAC, or EAC previously untreated in the advanced/metastatic setting and have ECOG PS 0 or 1. Patients were enrolled at 175 sites in 29 countries. Patients were randomized 1:1:1 to receive nivolumab 360 mg plus XELOX every 3 weeks or nivolumab 240 mg plus FOLFOX every 2 weeks; XELOX (every 3 weeks) or FOLFOX (every 2 weeks); or 4 cycles of nivolumab (1 mg/kg) plus ipilimumab (3 mg/kg) every 3 weeks followed by nivolumab monotherapy 240 mg every 2 weeks. The nivolumab plus ipilimumab and nivolumab monotherapy group was closed to recruitment on June 5, 2018, |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||; data for these patients were not relevant to the objective of this report and are not included. Patients were treated until disease progression, unacceptable toxicity, study withdrawal, or death, whichever came first. Treatment with nivolumab plus chemotherapy beyond initial, investigator-assessed progressive disease was allowed if the patient had investigator-assessed clinical benefit and was tolerating treatment. Further progression (increase in tumour burden ≥ 10%) resulted in discontinuation of nivolumab plus chemotherapy. For patients receiving nivolumab plus chemotherapy, the maximum treatment period was 24 months. Following treatment discontinuation, patients entered survival follow-up (every 3 months until study withdrawal, death, or data cut-off, whichever came first).

The mean ages of study participants were 60.3 years and 59.9 years in the nivolumab plus chemotherapy and chemotherapy arms, respectively. Approximately 70% of patients were male, approximately 70% were White, and approximately 60% were enrolled at sites outside of North America and Asia. Approximately 70% of patients had GAC and ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||. Almost all patients (approximately 96%) had metastatic disease, while a minority (approximately 4%) had locally advanced or recurrent disease. Only a minority of patients (10% to 20%) had received prior surgery, ||||||||||||||||||||||||||||||||||||||||||||||||||||||||. Approximately 60% had HER2-negative tumours while HER2 status was not reported in approximately 40% of patients. Approximately 3% of patients were MSI-H. Approximately 83%, 61%, and 49% of patients had PD-L1 CPS of 1 or higher, 5 or higher, and 10 or higher, respectively. Baseline demographic and disease characteristics were generally well balanced between study arms.

Efficacy Results

Key efficacy results of the CheckMate-649 trial are summarized in Table 2. At the database lock of 10 July 2020 (minimum follow-up 12.1 months; mean follow-up ||||| months [standard deviation (SD) |||| months] in the nivolumab plus chemotherapy arm and ||||| months [SD |||| months] in chemotherapy arm), the co-primary efficacy analyses of OS and PFS in patients with PD-L1 CPS of 5 or higher showed that patients in the nivolumab plus chemotherapy arm had longer OS and PFS than those in the chemotherapy arm. Median OS was 14.39 months (95% confidence interval [CI], 13.11 to 16.23 months) in the nivolumab plus chemotherapy arm versus 11.10 months (95% CI, 10.02 to 12.09 months) in the chemotherapy arm (P < 0.0001). The hazard ratio (HR) for OS comparing nivolumab plus chemotherapy with chemotherapy was 0.71 (98.4% CI, 0.59 to 0.86). Median PFS was 7.69 months (95% CI, 7.03 to 9.17 months) in the nivolumab plus chemotherapy arm versus 6.05 months (95% CI, 5.55 to 6.90 months) in the chemotherapy arm (P < 0.0001). The HR for PFS comparing nivolumab plus chemotherapy with chemotherapy was 0.68 (98% CI, 0.56 to 0.81). The hierarchically tested secondary analyses of OS in patients with PD-L1 CPS of 1 or higher and all randomized patients also showed that patients in the nivolumab plus chemotherapy arm had longer OS than those in the chemotherapy arm. Among patients with PD-L1 CPS of 1 or higher, median OS was 13.96 months (95% CI, 12.55 to 14.98 months) in the nivolumab plus chemotherapy arm versus 11.33 months (95% CI, 10.64 to 12.25 months) in the chemotherapy arm (P < 0.0001). The HR comparing nivolumab plus chemotherapy with chemotherapy was 0.77 (99.3% CI, 0.64 to 0.92). Among all randomized patients, median OS was 13.83 months (95% CI, 12.55 to 14.55 months) in the nivolumab plus chemotherapy arm versus 11.56 months (95% CI, 10.87 to 12.48 months) in the chemotherapy arm (P = 0.0002). The HR comparing nivolumab plus chemotherapy with chemotherapy was 0.80 (99.3% CI, 0.68 to 0.94). The results of the co-primary and hierarchically tested secondary OS analyses were clinically relevant, according to the clinical experts consulted by CADTH for this review, based on their judgment that a 6-week improvement in survival represents a clinically meaningful improvement in this patient population.

Table 2: Summary of Key Efficacy Results From the CheckMate-649 Study

chemo = chemotherapy; CI = confidence interval; CPS = combined positive score; DOR = duration of response; HR = hazard ratio; Nivo = nivolumab; ORR = objective response rate; OS = overall survival; PD-L1 = programmed death ligand 1; PFS = progression-free survival.

Note: Efficacy analyses were conducted in all randomized patients, which consisted of the set of patients randomized concurrently to receive nivolumab plus chemotherapy or chemotherapy.

^aAnalysis based on database lock of July 10, 2020.

^bAnalysis based on database lock of February 16, 2021.

^cBased on Kaplan-Meier estimates.

^dStratified Cox proportional hazards model; HR for nivolumab + chemotherapy compared with chemotherapy. Stratification factors were region, ECOG PS, tumour cell PD-L1, and chemotherapy regimen. For analysis of OS in patients with PD-L1 CPS ≥ 1 at the February 16, 2021, database lock, an unstratified HR is reported.

^eTwo-sided P value from stratified log-rank test. Value is from analysis based on database lock of July 10, 2020.

^fStrata adjusted difference in response rate (nivolumab + chemotherapy versus chemotherapy) based on DerSimonian and Laird method of weighting.

^gStratified by region, ECOG PS, tumour cell PD-L1, and chemotherapy regimen.

Source: CheckMate-649 Clinical Study Report¹⁴ and CADTH review submission for nivolumab.¹²

OS and PFS analyses were conducted as secondary end points across other PD-L1 CPS cut-offs. Median OS in patients with PD-L1 CPS of 10 or higher was ||||||| months (95% CI, ||||||| to ||||||| months) in the nivolumab plus chemotherapy arm and ||||||||| months (95% CI, ||||||||| to ||||||||| months) in the chemotherapy arm; the HR for OS comparing nivolumab plus chemotherapy with chemotherapy was ||||||| (95% CI, ||||||| to |||||||). Among patients with PD-L1 CPS of 10 or higher, median PFS was ||||||| months (95% CI, ||||||| to ||||||| months) in the nivolumab plus chemotherapy arm and ||||||| months (95% CI, ||||||| to ||||||| months) in the chemotherapy arm; the HR for PFS comparing nivolumab plus chemotherapy with chemotherapy was ||||||| (95% CI, ||||||| to |||||||). Among patients with PD-L1 CPS of 1 or higher, median PFS was 7.49 months (95% CI, 7.03 to 8.41 months) in the nivolumab plus chemotherapy arm and 6.90 months (95% CI, 6.08 to 7.03 months) in the chemotherapy arm; the HR for PFS comparing nivolumab plus chemotherapy with chemotherapy was 0.74 (95% CI, 0.65 to 0.85). Among all randomized patients, median PFS was 7.66 months (95% CI, 7.10 to 8.54 months) in the nivolumab plus chemotherapy arm and 6.93 months (95% CI, 6.60 to 7.13 months) in the chemotherapy arm; the HR for PFS comparing nivolumab plus chemotherapy with chemotherapy was 0.77 (95% CI, 0.68 to 0.87). Subgroup analyses of OS by PD-L1 status, showed decreasing treatment effects of nivolumab plus chemotherapy with lower PD-L1 CPS cut-offs as follows: PD-L1 CPS less than 10, HR = 0.94 (95% CI, 0.80 to 1.10); PD-L1 CPS of 10 or higher, HR = 0.65 (95% CI, 0.55 to 0.78); PD-L1 CPS less than 5, HR = 0.94 (95% CI, 0.78 to 1.13); PD-L1 CPS of 5 or higher, HR = 0.70 (95% CI, 0.60 to 0.81); PD-L1 CPS less than 1, HR = 0.92 (95% CI, 0.70 to 1.23); and PD-L1 CPS of 1 or higher, HR = 0.76 (95% CI, 0.67 to 0.87). Subgroup analyses of PFS followed a similar pattern.

EQ-5D-3L utility index scores, EQ visual analogue scale (VAS) scores, FACT-Ga total scores, and GaCS scores at baseline were similar among all randomized patients in the 2 treatment groups. Mean values for EQ-5D-3L utility index scores, EQ VAS scores, FACT-Ga total scores, and GaCS scores were numerically higher (improved) at post-baseline assessments during the treatment period compared with the baseline assessment among all randomized patients in both treatment groups. Interpretation of changes in patient-reported HRQoL outcomes was limited by high rates of missing data at later times post-baseline.

Comparisons of ORR and duration of responses (DOR) also favoured nivolumab plus chemotherapy over chemotherapy alone. According to the clinical experts consulted by CADTH for this review, differences in these outcomes, which were outside the statistical hierarchy, had uncertain clinical significance on their own but supported the clinically meaningful difference in OS in favour of nivolumab plus chemotherapy.

Harms Results

Key efficacy results of the CheckMate-649 trial are summarized in Table 3. Adverse events (AEs) occurred in almost all patients treated with nivolumab plus chemotherapy and chemotherapy alone (99.2% versus 98.0%). Serious AEs and withdrawals due to AEs occurred in larger proportions of patients receiving nivolumab plus chemotherapy compared with chemotherapy alone (54.1% versus 43.7% and 47.4% versus 32.7%, respectively). For death rates, 68.8% of patients treated with nivolumab plus chemotherapy and 74.6% of patients treated with chemotherapy alone died during the study period.

Select AEs, immune-mediated AEs (IMAEs), and other events of special interest (protocol-defined to capture the expected toxicity profile of nivolumab) occurred more frequently in the nivolumab plus chemotherapy arm than in the chemotherapy arm. Select AEs affecting the gastrointestinal system (40.3% in the nivolumab plus chemotherapy arm and 33.9% in the chemotherapy arm), the hepatic system (34.1% and 24.3%), the skin (33.5% and 17.9%), and the endocrine system (15.0% and 1.8%), as well as hypersensitivity and infusion reactions (15.1% and 5.9%), were the most common selected AEs in the nivolumab plus chemotherapy arm. Hypothyroidism and thyroiditis (9.5% in the nivolumab plus chemotherapy arm and 0.8% in the chemotherapy arm), rash (6.5% and 0.5%), pneumonitis (4.2% and 0%), diarrhea and colitis (3.3% and 0%), hyperthyroidism (2.9% and 0.3%), and hepatitis (2.4% and 0%) were the most common IMAEs in the nivolumab plus chemotherapy arm.

Table 3: Summary of Key Harms Results From the CheckMate-649 Study

AE = adverse event; Chemo = chemotherapy; GI = gastrointestinal; IMAE = immune-mediated adverse event; Nivo = nivolumab; OESI = other event of special interest; SAE = serious adverse event; WDAE = withdrawal due to adverse event.

Source: CheckMate-649 Clinical Study Report.¹⁴

Critical Appraisal

A notable limitation of the CheckMate-649 study was its OL RCT design and potential bias associated with this design. Randomization was stratified by tumour PD-L1 expression (≥ 1% versus < 1%), while the co-primary OS and PFS analyses were conducted in patients with PD-L1 CPS of 5 or higher, eliminating the protection of stratified randomization. Although outcome assessment of tumour response and progression was performed by BICR using objective response evaluation criteria in solid tumours (RECIST) 1.1 criteria, patient-reported HRQoL data and assessment of harms outcomes may have been affected to some degree by knowledge of treatment allocation. The OL design may also have resulted in more frequent discontinuation before receiving any study therapy (nivolumab plus chemotherapy arm 0.9% versus chemotherapy arm 3.2%), discontinuation of therapy during the treatment phase (patient request: nivolumab plus chemotherapy arm 1.7% versus chemotherapy arm 4.6%; withdrawal of consent: nivolumab plus chemotherapy arm 2.6% versus chemotherapy arm 5.3%), and discontinuation from the study (withdrawal of consent: nivolumab plus chemotherapy arm 2.6% versus chemotherapy arm 4.7%) by patients randomized to the chemotherapy arm. The OL design could have altered treatment exposure in either or both study arms due to investigator biases, especially since treatment beyond progression was allowed for nivolumab plus chemotherapy but not chemotherapy alone. According to the clinical experts consulted by CADTH for this review, treatment with nivolumab plus chemotherapy beyond progression is clinically appropriate in some patients, although, in most patients, therapy would be discontinued at the first objective determination of progressive disease. According to the clinical experts consulted for this review, the frequency of treatment beyond progression in the CheckMate-649 trial (||||% of all randomized patients) was higher and the duration of treatment beyond progression was longer than expected based on current clinical practice in Canada. The impact of extended administration of nivolumab on OS was uncertain, although the clinical experts were of the opinion that post-progression treatment was unlikely to significantly influence OS or the interpretation of OS data. The absence of formal statistical comparison and high rates of missing HRQoL data (due to deaths and low questionnaire completion rates following treatment discontinuation) limited interpretation of these end points. In addition, the GaCS has not been validated as a stand-alone scale, and the degree to which it specifically measures changes in symptoms versus general HRQoL changes was unclear. The study had very high power for the co-primary efficacy analyses and would likely have been capable of detecting smaller treatment effects than originally anticipated with uncertain clinical relevance. The magnitude of OS differences between the nivolumab plus chemotherapy and chemotherapy arms in the primary analysis population (PD-L1 CPS ≥ 5) was statistically and clinically significant, according to the clinical experts consulted by CADTH for this review, but smaller differences in PFS were of uncertain clinical relevance.

The demographic and disease characteristics of the CheckMate-649 study population broadly reflected the Canadian population with GAC, GEJAC, or EAC. However, there were major unresolved questions of generalizability to some patient groups that would be covered by the Health Canada indication and the reimbursement request submitted for CADTH review. While the CheckMate-649 study enrolled patients with ECOG PS 0 or 1 and no prior systemic therapy in the advanced or metastatic setting, this review identified no evidence regarding administration of nivolumab plus chemotherapy in other types of patients (e.g., ECOG PS ≥ 2, non–first-line therapy). Critically, the study was not designed to conclusively identify the PD-L1 expression thresholds required for therapeutic benefit. Analyses of OS and PFS using different PD-L1 CPS cut-offs, as well as subgroup analyses by tumour cell PD-L1 expression, pointed toward potentially important differences in efficacy according to PD-L1 status.

Indirect Comparisons

Description of Studies

One sponsor-submitted indirect treatment comparison (ITC)¹⁵ contributed evidence to this review. The purpose of the ITC was to compare the efficacy of nivolumab plus chemotherapy to relevant comparators (chemotherapy regimens: fluoropyrimidine, fluoropyrimidine plus platinum, taxane plus platinum drug, fluoropyrimidine plus topoisomerase inhibitor, fluoropyrimidine plus taxane, platinum drug plus topoisomerase inhibitor, taxane plus topoisomerase inhibitor, fluoropyrimidine plus platinum drug and taxane, or fluoropyrimidine plus platinum drug and anthracycline) for first-line treatment of advanced or metastatic GC, GEJC, or EAC. Pembrolizumab plus chemotherapy and trastuzumab plus chemotherapy were not considered relevant comparators by the ITC authors.

Following literature searching, 31 studies presenting data on OS and PFS with relevant treatment comparisons were considered for inclusion in the network meta-analysis, of which 23 were used in the PFS network and 28 were used in the OS network. The ATTRACTION-4 study¹⁶ was excluded from the main ITC. Studies were connected in drug class–based networks for OS and PFS outcomes to indirectly compare nivolumab plus chemotherapy to other relevant therapies among the all-comers population (defined by the ITC authors as all patients studied in the selected trials, regardless of their PD-L1 status or other characteristics). A Bayesian framework was conducted with non-informative priors. As both fixed- and random-effects models were used, models were compared using the deviance information criterion. Scenario analyses were conducted based on the heterogeneity observed across trials included in the networks.

Efficacy Results

Pairwise comparisons for OS and PFS did not show differences between nivolumab plus fluoropyrimidine and platinum and the following treatments of interest: fluoropyrimidine plus platinum, fluoropyrimidine plus topoisomerase inhibitor, and fluoropyrimidine plus platinum and anthracycline. Scenario analyses were generally consistent with the primary analyses for PFS and OS for all relevant comparisons.

Harms Results

The sponsor-submitted ITC did not assess harms outcomes.

Critical Appraisal

Studies of pembrolizumab plus chemotherapy were not included in the ITC. While pembrolizumab is currently not funded outside special access programs across Canadian jurisdictions, it was still considered a clinically relevant comparator by the clinical experts consulted for this review. Substantial heterogeneity was observed across patient and trial characteristics. While multiple scenario analyses were conducted to explore the impact of certain effect modifiers, others could not be investigated. A risk of bias assessment conducted by the sponsor revealed that most studies included in the ITC were of low to medium quality, and scenario analyses that excluded low-quality studies produced more precise estimates. The sponsor’s ITC did not include outcomes other than OS and PFS, such as toxicities or HRQoL, both of which were important outcomes to patients. Overall, the ITC had limitations associated with clinical and statistical heterogeneity, which increased the uncertainty of estimates and may have prevented detection of differences among treatments.

Other Relevant Evidence

No other relevant evidence was identified for this review.

Conclusions

Evidence from the CheckMate-649 study suggested that, compared with FOLFOX or XELOX alone, first-line administration of nivolumab plus FOLFOX or XELOX contributed to statistically significant and clinically meaningful prolongation of OS among patients with HER2-negative GAC, GEJAC, or EAC. This finding was consistent across patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, and all randomized patients. Administration of nivolumab plus FOLFOX or XELOX also resulted in statistically significant prolongation of PFS among patients with PD-L1 CPS of 5 or higher, although the clinical relevance of the difference in PFS was unclear. Other analyses of PFS, ORR, and DOR across different PD-L1 CPS cut-offs also numerically favoured nivolumab plus chemotherapy and supported the OS results. Results for patient-reported HRQoL and symptom scores (EQ-5D-3L, FACT-Ga) could not be interpreted due to absence of formal statistical testing, potential for bias in an OL trial, and high rates of missing data after baseline. There were signals from the trial that the comparative efficacy of nivolumab plus chemotherapy versus chemotherapy in patients with GAC, GEJAC, or EAC depended on PD-L1 status. Despite this, prolongation of OS by nivolumab plus chemotherapy, which was acknowledged as the most important outcome of therapy by both patients and clinicians, was statistically and clinically significant among all randomized patients. A sponsor-submitted ITC did not provide evidence of differences in efficacy between nivolumab plus fluoropyrimidine- and platinum-based chemotherapy and other chemotherapy regimens and did not include pembrolizumab plus chemotherapy as a comparator. Notable harms associated with nivolumab (including IMAEs) were appreciable but were expected and generally manageable with supportive care in most patients.

Introduction

Disease Background

Esophagogastric cancers, including GC, GEJC, and EC, are a heterogeneous group of highly lethal malignancies of the upper gastrointestinal tract.¹ Tumours form in the squamous cells of the upper esophagus (squamous cell carcinoma) and, more frequently, in the glandular epithelial cells of the distal esophagus, gastroesophageal junction (GEJ), and stomach (adenocarcinoma; GAC, GEJAC, and EAC, respectively). Approximately 2-thirds to 3-quarters of cases occur in men, typically between the ages of 50 years to 70 years but sometimes in younger individuals.² GAC, GEJAC, and EAC are often asymptomatic in the early stages of disease and thus go undetected.³ As the tumour grows, patients may experience dysphagia, pain, dyspepsia, reflux, weight loss, bleeding, and anemia.⁷ The disease severely negatively affects HRQoL because of physical symptoms, difficulty eating, financial difficulties arising from limited ability to work, and social impacts.⁸

In some regions with a high incidence of esophagogastric cancers, screening is routine.¹⁷ However, in North America, diagnosis is often made at a more advanced stage of disease. In approximately 40% of patients, disease extends beyond locoregional confines at diagnosis,⁴ and only about one-quarter of patients are eligible for potentially curative resection with adjuvant or neoadjuvant therapy.⁵ Curable esophagogastric cancers are infrequently detected outside of screening programs, although these programs are also rare in North America. Diagnosis is typically made by a gastroenterologist or surgeon based on endoscopic biopsy and imaging findings. Approximately 15% to 20% of patients with GAC, GEJAC, or EAC have HER2-positive tumours.⁵ PD-L1 expression levels vary in GAC, GEJAC, or EAC, which may have prognostic significance in patients receiving immunotherapies.¹¹

Five-year survival among patients with stage IV GAC, GEJAC, or EAC is approximately 4%,^2,9 and the median OS of patients receiving first-line treatment for metastatic or advanced GAC, GEJAC, or EAC is approximately 7.7 months.¹⁰ In Canada, the combined incidence of GC and GEJC was approximately 4,200 cases in 2020 and that of EC was approximately 2,400 cases.^2,18 Approximately 90% of GCs and GEJCs¹⁹ and approximately 75% of ECs are adenocarcinomas.²⁰ Approximately 40% of patients present with metastatic disease at diagnosis,⁴ and approximately 35% of incident cases of advanced or metastatic GC, GEJC, or EC represent recurrent disease following a prior diagnosis at an earlier stage.²¹ Approximately 75% of patients are eligible for systemic therapy,²¹ and approximately 77% have HER2-negative tumours.²² Based on these figures, the sponsor estimated a total population eligible for treatment with nivolumab plus chemotherapy of 1,551 patients in Canada (outside of Quebec) in 2021.¹²

Standards of Therapy

According to clinical experts consulted by CADTH for this review, a variety of non-drug approaches are used to manage symptoms associated with GAC, GEJAC, or EAC, such as bleeding and dysphagia, including surgery, radiation, and gastric or esophageal stents. Patients with resectable disease typically undergo radical esophagogastrostomy with adjuvant or neoadjuvant therapy in the hopes of cure. Unfortunately, unless resection is performed at an early stage, recurrence or relapse at locoregional or distant sites occurs in more than half of patients within 1 year to 2 years.⁶ Patients with de novo advanced, metastatic, or recurrent disease are transferred to the care of a medical oncologist for palliative systemic therapy.

According to the clinical experts, systemic therapy (typically combination chemotherapy) is used in patients with adequate PS and organ function. Because of the similar tumour locations of GAC, GEJAC, or EAC, similar combination chemotherapy regimens are used. These regimens typically include a fluoropyrimidine and/or a platinum drug. Folinic acid and 5-fluorouracil plus oxaliplatin (FOLFOX) is the most common first-line regimen used in Canada, although the regimen is not completely standardized. Capecitabine plus oxaliplatin (XELOX) may also be used, although this is not the regimen of choice for most clinicians because patients may have difficulty swallowing capecitabine pills. According to the clinical experts, folinic acid and 5-fluorouracil plus irinotecan (FOLFIRI) is used in patients who are unlikely to tolerate or have contraindications to platinum drugs. The clinical experts indicated that, in the approximately 15% of patients whose GAC or GEJAC is HER2-positive, trastuzumab plus chemotherapy can be administered. However, there are limited data on use of trastuzumab plus chemotherapy to treat EAC. Finally, some patients receive immunotherapies (nivolumab and pembrolizumab) in combination with chemotherapy through registration in trials or special access programs; both drugs are currently under review by CADTH. Responses to current combination chemotherapy regimens are both infrequent and short-lived. The clinical experts emphasized the importance of using the most active drugs for first-line treatment, because patients deteriorate rapidly, and few will receive second or subsequent lines of therapy (generally other chemotherapy regimens). Second-line regimens include paclitaxel, with or without ramucirumab (ramucirumab is not used for EAC), and trifluridine/tipiracil, while third-line regimens include irinotecan and trifluridine/tipiracil.

The clinical experts stated that the most important goals of treatment with palliative intent in patients with advanced or metastatic GAC, GEJAC, or EAC are controlling symptoms, maintaining HRQoL, and prolonging survival. These goals can conflict with one another, as the side effects of systemic therapy can worsen HRQoL. Thus, selecting appropriate interventions for each individual patient is critical.

Drug

Nivolumab is a human IgG4 monoclonal antibody directed against PD-1. Binding of nivolumab to PD-1 blocks its interaction with its ligands PD-L1 and PD-L2. Upregulation of PD-L1 and PD-L2 occurs in some tumours and inhibits antitumour T-cell responses. Thus, nivolumab can release PD-1 pathway-mediated inhibition of antitumour immunity, decreasing tumour growth.

The reimbursement request initially submitted by the sponsor for review by CADTH, |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||, was for nivolumab, in combination with fluoropyrimidine- and platinum-containing chemotherapy, for the treatment of adult patients with advanced or metastatic GAC, GEJAC, or EAC. Nivolumab underwent an expedited Health Canada review for this indication through Project Orbis. During the CADTH review, an NOC was issued by Health Canada. The updated, Health Canada–approved indication is for nivolumab, in combination with fluoropyrimidine- and platinum-containing chemotherapy, for the treatment of adult patients with HER2-negative advanced or metastatic GAC, GEJAC, or EAC. The approved indication differs from the initially submitted indication in that HER2 status is specified and the approved indication is therefore narrower. The indication also includes the following statement: “A positive association was observed between PD-L1 CPS score and the magnitude of treatment benefit.” After the NOC was issued, the sponsor confirmed with CADTH that the reimbursement request should be updated to match the approved indication.

Key characteristics of nivolumab are shown in Table 4. Nivolumab is administered at a dose of 360 mg every 3 weeks or 240 mg every 2 weeks (depending on the chemotherapy backbone) through IV infusion. Nivolumab is also indicated for the treatment of melanoma, classical Hodgkin lymphoma, hepatocellular carcinoma, non–small cell lung cancer, malignant pleural mesothelioma, renal cell carcinoma, and squamous cell cancer of the head and neck. The drug has been previously reviewed by CADTH for all of these indications. Nivolumab is currently under CADTH review for adjuvant treatment of completely resected GEJC or EC.

Table 4: Key Characteristics of Nivolumab and Fluoropyrimidine- and Platinum-Based Chemotherapy for GAC, GEJAC, or EAC

EAC = esophageal adenocarcinoma; 5-FdUMP = 5-fluoro-2′-deoxyuridylate; FOLFOX = leucovorin and 5-fluorouracil plus oxaliplatin; 5-FU = 5-fluorouracil; GAC = gastric adenocarcinoma; GEJAC = gastroesophageal junction adenocarcinoma; HER2 = human epidermal growth factor receptor 2; IM = intramuscular; PD-1 = programmed cell death protein 1; PD-L1 = programmed death ligand 1; PD-L2 = programmed death ligand 2; XELOX = capecitabine plus oxaliplatin.

^aNivolumab: Health Canada–proposed indication and indication submitted for CADTH reimbursement review; FOLFOX components: Health Canada–approved indications.

Source: CADTH review submission¹² and product monograph²³ for nivolumab and product monographs for leucovorin,²⁴ 5-fluorouracil,²⁵ and oxaliplatin.²⁶

Stakeholder Perspectives

Patient Group Input

This section was prepared by CADTH staff based on the input provided by a patient group. The original patient group submission can be found in the Stakeholder Input section.

Input was provided by 1 patient group for this review (My Gut Feeling – Stomach Cancer Foundation of Canada). My Gut Feeling is the first non-profit organization in Canada dedicated to providing support, awareness, education, information, and advocacy to patients with GC, GEJC, and EC as well as survivors and caregivers. My Gut Feeling distributed a survey via email, social media, and online forums to patients with GC, GEJC, and EC, as well as their caregivers, between 20 August and 9 September 2021. Among the 62 respondents (one-half patients and one-half caregivers), most (79%) were female, resided in Canada or the US (63% and 29%, respectively), and had received or were caring for someone who had received a diagnosis of GC (74.2%) and adenocarcinoma (82.3%). The number of respondents with advanced or metastatic disease was unclear.

Most respondents (90.3%) reported that cancer had a significant impact on HRQoL, with adverse effects on physical health, mental health, ability to eat, ability to work, finances, social life, identity, and self-image. Some of these impacts extended to caregivers and families as well. Common symptoms included weight loss, change in appetite, pain, fatigue, reflux, nausea/vomiting, difficulty swallowing, shortness of breath, bleeding, anemia, ascites, and dumping syndrome. Patients highlighted the limited treatment options for GC, GEJC, or EC and their experiences with prior therapies (surgery, radiation, chemotherapy, and immunotherapy), including variable effectiveness in delaying progression and controlling symptoms as well as significant side effects affecting HRQoL (e.g., fatigue, nausea/vomiting, appetite changes exacerbating weight loss). Twelve respondents had experience with nivolumab and felt that the drug controlled disease, improved HRQoL, and was more convenient and tolerable than surgery or chemotherapy.

According to patients, an ideal therapy for GC, GEJC, and GC would prolong survival while maintaining or improving HRQoL compared with standard of care. Delaying recurrence or progression and having manageable side effects were also important factors for patients. Patients identified an unmet need for equitable access to therapies that may prolong life, improve symptoms, reduce risk of recurrence, and have improved tolerability. Such treatment options should be available barrier-free for all Canadian patients with GC, GEJC, and EC who could benefit.

Clinician Input

Input From Clinical Experts Consulted by CADTH

All CADTH review teams include at least 1 clinical specialist with expertise regarding the diagnosis and management of the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., providing guidance on the development of the review protocol; assisting in the critical appraisal of clinical evidence; interpreting the clinical relevance of the results; and providing guidance on the potential place in therapy). The following input was provided by 2 clinical specialists with expertise in the diagnosis and management of advanced or metastatic GAC, GEJAC, or EAC.

Unmet Needs

According to the clinical experts consulted by CADTH for this review, all current treatment approaches for advanced or metastatic GAC, GEJAC, or EAC are palliative in nature, and survival is typically less than 1 year. Only a minority of patients respond to current combination chemotherapy regimens. These responses are usually short-lived, and very few patients live beyond 15 months to 18 months. There is clearly an unmet need for more effective therapies for advanced or metastatic GAC, GEJAC, or EAC that can be administered with similar or lower toxicity than current chemotherapy options.

Place in Therapy

According to the clinical experts, prior experience in other cancers has shown that, when immunotherapy alone is ineffective, combining it with chemotherapy may lead to better treatment outcomes with no or limited additional toxicity. According to the experts, as well as the sponsor, nivolumab would be administered in combination with fluoropyrimidine- and platinum-based chemotherapy for first-line treatment in most patients with HER2-negative advanced or metastatic GAC, GEJAC, or EAC who are able to tolerate chemotherapy. It would not be appropriate for patients to receive other treatments before nivolumab plus chemotherapy. According to the clinical experts, the most effective interventions should be used first-line. There is significant attrition, and many patients do not receive second or subsequent lines of therapy. Nivolumab would not be used in patients who are intolerant to or have failed chemotherapy. In patients with HER2-positive tumours, the clinical experts noted that trastuzumab would often be administered first-line and that, if the combination of pembrolizumab and trastuzumab with chemotherapy were available, this would be the preferred option for patients with HER2-positive tumours. The clinical experts stated that all patients with EAC would be eligible for nivolumab if funded, irrespective of the availability of HER2 testing. Nivolumab addresses the underlying disease process by potentiating antitumour immune responses, but there are many shortcomings to this approach. Reimbursement of nivolumab plus chemotherapy would not shift the treatment paradigm, as most patients would receive chemoimmunotherapy rather than fluoropyrimidine- and platinum-based chemotherapy alone in the first line, and second and subsequent lines of therapy would remain the same.

Patient Population

Unfortunately, the patients who are most in need of intervention have the most advanced disease, poor PS, and do not generally respond well to immunotherapy. As a result, they are excluded from clinical trials. According to the clinical experts consulted by CADTH for this review, the available data from trials provide evidence for use of nivolumab plus chemotherapy in patients with advanced or metastatic GAC, GEJAC, or EAC and ECOG PS 0 and 1, but administration in additional patients (ECOG PS 2 or potentially even 3) may be possible if judged appropriate by the treating clinician. In the opinion of the clinical experts, nivolumab plus chemotherapy should be available for all patients with advanced or metastatic GAC, GEJAC, or EAC for whom local or curative treatment is not possible and in whom chemotherapy is a treatment option. The site of metastasis or presence of symptoms would not affect patient selection. Diagnosis and staging (based on biopsy and CT or PET imaging) is standard, and misdiagnosis is unlikely.

According to the clinical experts, patients with good PS (ECOG PS 0 or 1) are most likely to respond to nivolumab plus chemotherapy. In addition, the clinical experts emphasized that PD-L1 expression is an established biomarker of response and patients with higher PD-L1 CPS are more likely to respond to nivolumab. PD-L1 CPS testing of biopsy specimens is now routinely performed for other cancer types and could easily be adapted for patients with GAC, GEJAC, or EAC, although this is not routinely done at present. According to the clinical experts, a small proportion of patients (3% to 5%) with MSI-H are also much more likely to respond to immunotherapy, including nivolumab. The clinical experts viewed patients with poor PS and PD-L1 CPS less than 1 as least likely to benefit from nivolumab, while patients with active autoimmune diseases are least suitable for treatment due to safety concerns.

Assessing Response to Treatment

Imaging (e.g., CT or PET scans) is used to evaluate response to therapy. In clinical practice, imaging assessments are performed approximately every 3 months (by contrast with the trial setting in which they are performed more frequently, i.e., every 6 weeks). Survival is the most important indicator of response, with improvement in symptoms and HRQoL also being important parameters in assessing response. Weight and PS may also give an indication of treatment response and are evaluated at each clinic visit.

Discontinuing Treatment

Treatment should be discontinued in patients with clear objective progressive disease assessed by imaging. Treatment intolerance or significant toxicity may also require discontinuation of therapy.

Prescribing Conditions

Diagnosis of GAC, GEJAC, or EAC and initial workup is typically performed by surgeons and gastroenterologists. Patients are then transferred to the care of a medical oncologist for systemic therapy. Palliative and supportive care specialists, as well as dietitians, would also follow these patients. Treatment would be administered in outpatient centres with experience in delivering systemic therapy (including chemotherapy and immunotherapy). Most centres already have significant experience with nivolumab and the accompanying chemotherapy regimens.

Additional Considerations

The clinical experts emphasized that most oncologists already have significant experience with administration of nivolumab, including combinations with chemotherapy, for other indications.

Clinician Group Input

This section was prepared by CADTH staff based on the input provided by clinician groups. The original clinician group input can be found in the Stakeholder Input section.

Two clinician groups provided input for this review: the Ontario Health (Cancer Care Ontario) Gastrointestinal Cancer Drug Advisory Committee (5 clinicians) and the CGOEN along with other physicians treating gastroesophageal cancers (7 clinicians). No major contrary views were presented. Both clinician groups echoed the limited efficacy of available systemic therapies for advanced or metastatic GAC, GEJAC, or EAC and short duration of response in many patients. By contrast with the clinical experts consulted by CADTH for this review, clinicians from the CGOEN felt that patients with PD-L1 CPS less than 5 (rather than < 1) would be least suitable for treatment with nivolumab plus chemotherapy, while those from the Ontario Health (Cancer Care Ontario) Gastrointestinal Cancer Drug Advisory Committee felt that, although patients with PD-L1 CPS of 5 or higher and 1 or higher are more likely to respond, all patients with GAC, GEJAC, or EAC can benefit from addition of nivolumab to chemotherapy.

Drug Program Input

The drug programs provide input on each drug being reviewed through CADTH’s reimbursement review processes by identifying issues that may affect their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical experts consulted by CADTH are summarized in Table 5. Since CADTH solicited drug plan input before the sponsor notifying CADTH of the change in wording of the Health Canada indication, drug plans raised implementation issues concerning HER2-positive disease. Since the Health Canada–approved indication specifies that patients must have HER2-negative disease, those issues no longer required consideration.

Clinical Evidence

The clinical evidence included in the review of nivolumab is presented in 2 sections. The first section, the systematic review, includes pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those studies that were selected according to an a priori protocol. The second section includes indirect evidence from the sponsor; no indirect evidence selected from the literature met the selection criteria specified in the review. No long-term extension studies or additional relevant studies were identified that addressed important gaps in the evidence included in the systematic review.

Systematic Review of Pivotal and Protocol-Selected Studies

Objectives

To perform a systematic review of the beneficial and harmful effects of nivolumab (IV injection over 30 minutes of 360 mg every 3 weeks or 240 mg very 2 weeks), in combination with fluoropyrimidine- and platinum-containing chemotherapy, for the treatment of adult patients with HER2-negative advanced or metastatic GAC, GEJAC, or EAC.

Methods

Studies selected for inclusion in the systematic review included pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those meeting the selection criteria presented in Table 6. Outcomes included in the CADTH review protocol reflect outcomes considered to be important to patients, clinicians, and drug plans. The systematic review protocol was established before Health Canada granted an NOC .

The literature search for clinical studies was performed by an information specialist using a peer-reviewed search strategy according to the PRESS Peer Review of Electronic Search Strategies resource.²⁷

Published literature was identified by searching the following bibliographic databases: MEDLINE All (1946–) via Ovid and Embase (1974–) via Ovid. All Ovid searches were run simultaneously as a multi-file search. Duplicates were removed using Ovid deduplication for multi-file searches, followed by manual deduplication in Endnote. The search strategy comprised both controlled vocabulary, such as the National Library of Medicine’s MeSH (Medical Subject Headings), and keywords. The main search concepts were Opdivo (nivolumab) and gastric, gastroesophageal junction, or esophageal adenocarcinoma. Clinical trials registries were searched: the US National Institutes of Health’s clinicaltrials.gov, the WHO International Clinical Trials Registry Platform (ICTRP) search portal, Health Canada’s Clinical Trials Database, and the European Union Clinical Trials Register.

No filters were applied to limit the retrieval by study type. Retrieval was not limited by publication date or by language. Conference abstracts were excluded from the search results. See Appendix 3 for the detailed search strategies.

The initial search was completed on September 16, 2021. Regular alerts updated the search until the meeting of the CADTH pan-Canadian Oncology Drug Review Expert Committee (pERC) on January 12, 2022.

Grey literature (literature that is not commercially published) was identified by searching relevant websites from the Grey Matters: A Practical Tool For Searching Health-Related Grey Literature checklist.²⁸ Included in this search were the websites of regulatory agencies (US FDA and European Medicines Agency). Google was used to search for additional internet-based materials. See Appendix 3 for more information on the grey literature search strategy. These searches were supplemented by reviewing bibliographies of key papers and through contacts with appropriate experts. In addition, the manufacturer of the drug was contacted for information regarding unpublished studies.

Two CADTH clinical reviewers independently selected studies for inclusion in the review based on titles and abstracts, according to the predetermined protocol. Full-text articles of all citations considered potentially relevant by at least 1 reviewer were acquired. Reviewers independently made the final selection of studies to be included in the review, and differences were resolved through discussion.

Findings From the Literature

One study was identified from the literature for inclusion in the systematic review (Figure 1). The included study is summarized in Table 7. A list of excluded studies is presented in Appendix 4.

Figure 1: Flow Diagram for Inclusion and Exclusion of Studies

Description of Studies

CheckMate-649 was a phase III, randomized, OL, multi-centre RCT (N = 1,581) funded by the sponsor (Bristol Myers Squibb). The primary objective was to compare the efficacy of first-line therapy with nivolumab plus FOLFOX-XELOX versus FOLFOX-XELOX in prolonging OS and PFS per BICR in patients with advanced or metastatic GAC, GEJAC, or EAC (all with PD-L1 CPS ≥ 5). Secondary objectives included comparing OS and PFS by BICR in patients with PD-L1 CPS of 1 or higher (OS: hierarchically tested), all randomized patients (OS: hierarchically tested), and patients with PD-L1 CPS of 10 or higher; and to compare ORR in patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, all randomized patients, and patients with PD-L1 CPS of 10 or higher. A summary of the design of the CheckMate-649 study is shown in Figure 2. Patients 18 years of age or older with inoperable advanced or metastatic, HER2-negative or HER2-unreported GAC, GEJAC, or EAC, and ECOG PS of 1 or lower were enrolled from April 17, 2017, to May 27, 2019, at 175 sites in 29 countries (8 sites in Canada; ||||||||||||||||||||||||||||||||||||||||||||||||||). Patients were screened for eligibility within 4 weeks of initiating protocol therapy.

Patients were randomized 1:1:1, using an interactive web response system, to receive nivolumab (360 mg) plus XELOX every 3 weeks or nivolumab (240 mg) plus FOLFOX every 2 weeks; XELOX every 3 weeks or FOLFOX every 2 weeks; or 4 cycles of nivolumab (1 mg/kg) plus ipilimumab (3 mg/kg) every 3 weeks followed by nivolumab monotherapy (240 mg) every 2 weeks. Chemotherapy regimens (FOLFOX versus XELOX) were selected at the discretion of investigators before randomization. The randomization procedure was not explicitly stated. The nivolumab plus ipilimumab-nivolumab monotherapy arm was closed to recruitment on June 5, 2018, due to concerns of the data monitoring committee (DMC) regarding increased early death rates and high toxicity rates; data for these patients are not yet available and were not relevant to the objective of this report. Randomization was stratified by region (Asia versus North America versus rest of world), ECOG PS (0 versus 1), chemotherapy regimen (FOLFOX versus XELOX), and PD-L1 expression (≥ 1% versus < 1% of tumour cells). Patients were treated until progressive disease (with some exceptions in the nivolumab plus chemotherapy arm), unacceptable toxicity, study withdrawal by patients or physicians, or death, whichever came first. Treatment delays of up to 6 weeks also led to discontinuation. Following treatment discontinuation, patients entered survival follow-up (every 3 months until study withdrawal, death, or data cut-off, whichever came first). The database was closed on May 27, 2020, and locked on July 10, 2020 (minimum follow-up 12.1 months). Data for a second database lock of February 16, 2021 (minimum follow-up 19.4 months) were made available during EMA review.

Figure 2: Design of the CheckMate-649 Trial

chemo = chemotherapy; ECOG PS = Eastern Cooperative Oncology Group performance status; FOLFOX = leucovorin and 5-fluorouracil plus oxaliplatin; F/U = follow-up; GC = gastric cancer; GEJC = gastroesophageal junction cancer; IC = informed consent; Ipi = ipilimumab; Mono = monotherapy; Nivo = nivolumab; PD = progressive disease; PD-L1 = programmed death ligand 1; Q2W = every 2 weeks; Q3W = every 3 weeks; ROW = rest of world; XELOX = capecitabine plus oxaliplatin.

Source: CheckMate-649 Clinical Study Report.¹⁴

Populations

Inclusion and Exclusion Criteria

Key inclusion and exclusion criteria for the CheckMate-649 study are shown in Table 7. Adult patients (18 years of age or older) with inoperable advanced, locally advanced, or metastatic GC, GEJC, or distal EC, and histologically confirmed predominant adenocarcinoma were eligible if they had ECOG PS 0 or 1, had at least 1 measurable lesion or evaluable disease by CT/PET, had adequate hematological and organ function, and had not previously received systemic therapy (including HER2 inhibitors) for advanced or metastatic disease. Prior adjuvant or neoadjuvant chemotherapy, radiotherapy, and/or chemoradiotherapy completed 6 months or more before randomization and palliative radiotherapy completed 2 weeks or more before randomization were permitted. Patients had to have recovered from all toxicities associated with prior therapies, and tumour tissue had to be available for PD-L1 analysis before randomization. Patients with HER2-positive tumours were excluded, although known HER2 status was not required for study eligibility. Patients with untreated central nervous system metastases, uncontrolled ascites, peripheral neuropathy of Grade 2 or higher, and active autoimmune diseases were excluded, as were patients receiving botanical preparations, systemic corticosteroids (> 10 mg daily prednisone or equivalent), or other immunosuppressive medications. Patients who had previously received immunotherapies, had allergy or hypersensitivity to study drug components, or had contraindications to XELOX or FOLFOX were excluded.

Baseline Characteristics

The baseline demographic and disease characteristics of randomized patients with PD-L1 CPS ≥ 5 and all randomized patients in the CheckMate-649 study are shown in Table 8. The mean ages of patients in the nivolumab plus chemotherapy and chemotherapy arms were 60.3 years and 59.9 years, respectively. Approximately 70% of patients were male, approximately 70% were White, and approximately 60% were enrolled at sites outside of North America and Asia. Approximately 70% of patients had GC and approximately ||||% had stage IV disease at diagnosis. Almost all patients (approximately 96%) had metastatic disease, while a minority (approximately 4%) had locally advanced or recurrent disease. Only a minority of patients (10% to 20%) had received prior surgery, ||||||||||||||||||||||||||||||||||||||||||||||||||||||||. Approximately 60% of patients had HER2-negative tumours, while HER2 status was not reported in approximately 40% of patients. Approximately 3% of patients were MSI-H. Approximately 83%, 61%, and 49% of patients had PD-L1 CPS ≥ 1, ≥ 5, and ≥ 10, respectively. Baseline demographic and disease characteristics were generally well balanced between study arms.

Table 8: Summary of Baseline Characteristics in the CheckMate-649 Trial

CPS = combined positive score; EAC = esophageal adenocarcinoma; ECOG = Eastern Cooperative Oncology Group; GC = gastric cancer; GEJC = gastroesophageal junction cancer; HER2 = human epidermal growth factor receptor 2; MSI = microsatellite instability; MSI-H = microsatellite instability-high; MSS = microsatellite stable; NR = not reported; PD-L1 = programmed death ligand 1.

^aMeasured among patients with quantifiable PD-L1.

^bAnticancer therapy (platinum-based or other chemotherapy) in the adjuvant, neo-adjuvant, or definitive chemoradiation setting.

Source: CheckMate-649 Clinical Study Report.¹⁴

Interventions

Patients were randomized 1:1:1 to receive nivolumab 360 mg plus XELOX every 3 weeks or nivolumab 240 mg plus FOLFOX every 2 weeks; XELOX every 3 weeks or FOLFOX every 2 weeks; or 4 cycles of nivolumab 1 mg/kg plus ipilimumab 3 mg/kg every 3 weeks, followed by nivolumab monotherapy 240 mg every 2 weeks. Only data for the first 2 arms are presented in this report. Investigators were free to choose either FOLFOX or XELOX, depending on local standards. Chemotherapy regimens (FOLFOX versus XELOX) were selected at the discretion of investigators before randomization. The study was OL, and protocol therapy included a mixture of IV and oral drugs. All therapy was administered by study personnel at site visits, except for capecitabine, which was self-administered.

Nivolumab was administered at a dose of 360 mg every 3 weeks or 240 mg every 2 weeks on day 1 of 3- or 2-week treatment cycles. The 240 mg every 2 weeks dosage is equivalent to 3 mg/kg every 2 weeks used across multiple indications, and the 360 mg every 3 weeks dosage was expected to result in similar exposure, based on pharmacokinetic modelling. In patients who discontinued chemotherapy, nivolumab 240 mg every 2 weeks, 360 mg every 3 weeks, or 480 mg every 4 weeks could be administered as monotherapy, depending on local standard of care. Nivolumab infusions were administered over 30 minutes.

FOLFOX and XELOX were administered as per local standard of care. For FOLFOX, oxaliplatin 85 mg/m², leucovorin 400 mg/m², and 5-fluorouracil 400 mg/m² were administered IV on day 1 of each 2-week treatment cycle, and 5-fluorouracil 1,200 mg/m² was administered by IV continuous infusion over 24 hours (or per local standard) daily on days 1 and 2 of each 2-week treatment cycle. For XELOX, oxaliplatin 130 mg/m² was administered IV on day 1 of each treatment 3-week treatment cycle, and capecitabine 1,000 mg/m² was administered orally twice daily on days 1 to 14 of each 3-week treatment cycle.

For patients receiving nivolumab plus chemotherapy, the maximum treatment period was 24 months. Treatment with nivolumab plus chemotherapy beyond initial, investigator-assessed progressive disease was allowed if the patient had investigator-assessed clinical benefit and was tolerating treatment. Further progression (increase in tumour burden ≥ 10%) resulted in discontinuation of nivolumab plus chemotherapy. Treatment beyond progressive disease was not allowed for patients receiving chemotherapy alone. In the nivolumab plus chemotherapy arm, either nivolumab or chemotherapy could be administered alone if the other component was discontinued due to toxicity or per local standard of care. The rules for nivolumab and chemotherapy dose reduction, dose interruption/delay, and discontinuation were as follows. Dose reductions of nivolumab were not permitted (although dose reductions of chemotherapy for toxicity were permitted according to local guidelines or package insert). Dose delays of nivolumab of up to 6 weeks were permitted for any AEs Grade 2 or higher or laboratory abnormalities until resolution of toxicities to Grade 1 or lower and discontinuation or tapering of corticosteroids (if administered) to 10 mg or less of prednisone daily or equivalent. Prior to Revised Protocol 07 (Amendment 23, September 14, 2018), all components of therapy were delayed together, while, after this amendment, dose delays of nivolumab and chemotherapy could be made independently. Infusion reactions (Grade 2) were treated with diphenhydramine, acetaminophen, corticosteroids, and decreased (50%) infusion rate for 30 minutes. Grade 3 or 4 infusion reactions that persisted following dose interruption resulted in permanent discontinuation of nivolumab. Nivolumab was also discontinued in patients who experienced a variety of AEs of Grade 3 or higher.

Immunosuppressive drugs, immunosuppressive doses of systemic corticosteroids (> 10 mg prednisone daily or equivalent), and concurrent antineoplastic therapies were prohibited. Topical and inhaled corticosteroids, adrenal replacement steroid dosages of more than 10 mg daily prednisone (or equivalent), and brief (< 3 weeks) courses of corticosteroids for prophylaxis or treatment of non-autoimmune conditions were permitted. In participants who experienced infusion reactions to nivolumab in any cycle, secondary prophylaxis with diphenhydramine, acetaminophen, and if necessary, corticosteroids (up to 25 mg hydrocortisone) were recommended as premedication in future cycles.

Outcomes

A list of efficacy end points identified in the CADTH review protocol that were assessed in the CheckMate-649 study is provided in Table 9. These end points are further summarized in this section. A detailed discussion and critical appraisal of the outcome measures is provided in Appendix 5. OS, PFS, and ORR are standard outcome measures in oncology trials; according to the clinical experts consulted by CADTH for this review, differences in survival outcomes of at least 6 weeks would be clinically meaningful in this patient population.

The EQ-5D-3L²⁹ is a generic, preference-based HRQoL instrument that has been applied to a wide range of health conditions and treatments. It is a self-administered questionnaire consisting of a descriptive system with 5 dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), each with 3 response options, as well as a VAS. A scoring function can be used to assign a population-preference weighted utility index value based on responses to descriptive system items. Utility index scores vary depending on the utility function (e.g., –0.59 for the UK algorithm and –0.109 for the US algorithm); utility index scores less than 0 represent health states worse than death, while score of 0 and 1 represent “dead” and “perfect health,” respectively. For the EQ VAS, the end points of 0 and 100 represent “worst imaginable health state” and “best imaginable health state,” respectively. The instrument has been extensively validated and shown to be reliable and responsive to change in many cancers; the minimal important difference (MID) for the utility index score has been estimated at approximately 0.09 to 0.12 (UK) and 0.06 to 0.09 (US), and the MID for the EQ VAS has been estimated at approximately 7 to 12 points. This instrument has not been validated specifically in patients with GAC, GEJAC, or EAC, and a disease-specific MID has not been determined.

The FACT-Ga³⁰ combines the functional assessment of cancer therapy – general (FACT-G; a self-administered, cancer-specific, preference-based HRQoL instrument consisting of 27 items evaluated on a 5-point Likert scale in 4 domains: physical, social/family, emotional, and functional well-being) and a 19-item GaCS. The FACT-G7 is an abbreviated form of the FACT-G, consisting of only 7 items in 4 domains (physical, emotional, and functional well-being) that was designed to quickly capture the most relevant issues to patients with cancer. In the CheckMate-649 study, the FACT-Ga instrument was used during the treatment period, and the FACT-G7 was used during survival follow-up. The GaCS of the FACT-Ga is a disease-specific subscale that assesses symptoms and impacts relating to pain, reflux, dysphagia, eating difficulties, tiredness, weakness, interference, and difficulty planning. FACT-Ga total scores range from 0 to 184, with higher values representing better HRQoL. FACT-Ga GaCS scores range from 0 to 76, with higher values representing better HRQoL. The FACT-Ga has been extensively validated and found to be reliable and responsive to change in patients with GC; however, the GaCS has not been validated as a stand-alone scale, and the degree to which it specifically measures changes in symptoms versus general HRQoL is unclear. The MID for the total FACT-Ga score was estimated at 15.1 to 22.6 points, and the MID for the GaCS subscale was estimated at 8.2 to 12.3 points.

Refer to Appendix 6 for a detailed appraisal of HRQoL outcomes.

Table 9: Summary of Outcomes of Interest Identified in the CADTH Review Protocol

BICR = blinded independent central review; CPS = combined positive score; DOR = duration of response; FACT-Ga = functional assessment of cancer therapy – gastric; FACT-G7 = 7-item version of the FACT – general; GaCS = gastric cancer subscale; HRQoL = health-related quality of life; ORR = objective response rate; OS = overall survival; PD-L1 = programmed death ligand 1; PFS = progression-free survival; TTSD = time to symptom deterioration; VAS = visual analogue scale.

^aDefined as time from randomization until death from any cause. For patients without documented death, OS was censored at the date last known alive. OS was not censored for patient who initiated subsequent anticancer therapies.

^bDefined as time from randomization to date of first progressive disease or death from any cause, whichever came first. Patients without progressive disease or death were censored at the last evaluable tumour assessment before initiation of subsequent anticancer therapy.

^cDefined as number of patients with a best overall response of complete or partial response by RECIST 1.1 based on BICR assessments divided by the number of randomized patients. Best overall response was measured between randomization and date of progressive disease or initiation of subsequent anticancer therapy, whichever came first.

^dDefined as time between first confirmed response and first documented progression by RECIST 1.1 or death from any cause, whichever occurred first.

^eDefined as time from randomization to a clinically meaningful decline (8.2 points) from baseline in GaCS score. Only events before treatment discontinuation were taken into account. Patients who did not deteriorate before treatment discontinuation were censored at the date of last GaCS assessment.

Source: CheckMate-649 Clinical Study Report.¹⁴

During the treatment period (until progressive disease, unacceptable toxicity, or 24 months for the nivolumab plus chemotherapy arm), tumour response was assessed by CT/PET at baseline (within 4 weeks before randomization), every 6 (± 1) weeks for the first 48 weeks, and then every 12 (± 1) weeks. For each patient, the same imaging modality was used to assess tumour response throughout the study. Patients who discontinued treatment due to toxicity continued with tumour response assessments at the protocol-required schedule until progressive disease or withdrawal of consent. Designation of response was based on the response of target and non-target lesions and the appearance of any new lesions using RECIST, version 1.1. Progressive disease was defined as a predefined increase (20%), taking as reference the smallest sum on study, in the sum of target lesions or the appearance of new non-target lesions; the sum must also demonstrate an absolute increase of at least 5 mm. Partial response (PR) was defined as at least a 30% decrease in the sum of diameters of target lesions, taking as reference the baseline sum diameters. Complete response (CR) was defined as disappearance of all target lesions, with reduction of the short axis of any pathological lymph nodes to less than 10 mm. Stable disease was defined as neither sufficient shrinkage (compared to baseline) to qualify for PR nor sufficient increase (taking as reference the smallest sum diameters while on study) to qualify for progressive disease.

||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.

The co-primary outcomes were OS (defined as time from randomization to death from any cause) and PFS per BICR (defined as time from randomization to date of first objective progressive disease or death from any cause, whichever came first) among patients with PD-L1 CPS of 5 or higher. PD-L1 CPS was calculated as the number of PD-L1 positive cells (tumour cells, lymphocytes, and macrophages) divided by the total number of viable tumour cells within the evaluated tumour area, multiplied by 100. For patients without documentation of death, OS was censored at the last date known alive but was not censored when patients initiated other anticancer therapy. For patients without objective progressive disease or death, PFS was censored at the last tumour assessment (or randomization date if no assessments) before or on the date of initiation of other anticancer therapy. OS among patients with PD-L1 CPS of 1 or higher and all randomized patients were hierarchically tested secondary outcomes. OS among patients with PD-L1 CPS of 10 or higher, PFS among patients with PD-L1 CPS of 10 or higher, PFS among patients with PD-L1 CPS of 1 or higher, and PFS among all randomized patients were secondary outcomes. ORR (proportion of patients with PR or CR by RECIST 1.1) per BICR was evaluated in patients with PD-L1 CPS of 10 or higher, PD-L1 CPS of 5 or higher, PD-L1 CPS of 1 or higher, and all randomized patients; all were secondary outcomes.

DOR, defined as time from first objective response per BICR until first objective progressive disease per BICR or death from any cause, whichever occurred first, was evaluated as an exploratory outcome. During the treatment period, patients completed the EQ-5D-3L and FACT-Ga every 6 weeks. During survival follow-up, patients completed the EQ-5D-3L, the GaCS, and FACT-G7 every 3 months. Time to symptom deterioration (TTSD) was defined as time from randomization until a decline from baseline in GaCS score of 8.2 points. TTSD was only evaluated before treatment discontinuation. EQ-5D-3L index scores were determined using a scoring algorithm based on the UK Time-Trade-Off value set.

Following treatment discontinuation, survival follow-up occurred 30 days and 84 days after the last dose of protocol therapy and then every 3 months (in person or by phone).

Harms outcomes included treatment-emergent AEs; serious AEs (SAEs); AEs requiring dose interruption, dose delay, or dose reduction; withdrawals due to AE (WDAEs); select AEs; IMAEs; and other events of significant interest (OESIs). Definitions of select AEs, IMAEs, and OESIs are listed in Table 10. AEs that began or worsened on or after the start of protocol therapy until 30 days after the last dose of study drug were captured, except for deaths (no time limit) and IMAEs/OESIs (100 days after last dose). AEs were defined as any untoward medical occurrence and were coded according to Medical Dictionary for Regulatory Activities version 23.0 and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0.

Table 10: Definitions of Select AEs, IMAEs, and OESIs in the CheckMate-649 Trial

AE = adverse event; IMAE = immune-mediated adverse event; OESI = other event of significant interest.

Note: Redacted rows have been deleted.

Source: CheckMate-649 Clinical Study Report.¹⁴

Statistical Analysis

Statistical analysis of efficacy outcomes in the CheckMate-649 study is summarized in Table 11. As of the most recent protocol amendment, the final PFS and interim OS analyses were planned to be conducted at a minimum follow-up time of 12 months, and final OS analysis was planned to be conducted at a minimum follow-up of 24 months. At the time of the interim OS analysis, if DMC review had determined that the pre-specified significance levels for all of the objectives in the statistical testing hierarchy were not met, the study would have been terminated.

Table 11: Statistical Analysis of Efficacy End Points in the CheckMate-649 Trial

BICR = blinded independent review committee; CI = confidence interval; CPS = combined positive score; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group performance status; FACT-Ga = functional assessment of cancer therapy – gastric; FACT-G7 = 7-item version of the FACT – general; FOLFOX = oxaliplatin + leucovorin + 5-fluorouracil; GaCS = gastric cancer subscale; HR = hazard ratio; HRQoL = health-related quality of life; KM = Kaplan-Meier; NA = not applicable; OR = odds ratio; ORR = objective response rate; OS = overall survival; PD-L1 = programmed death ligand 1; PFS = progression-free survival; ROW = rest of world; TTSD = time to symptom deterioration; VAS = visual analogue scale; vs. = versus; XELOX = oxaliplatin + capecitabine.

^aFor OS: stratified analysis based on first 420 randomized patients with PD-L1 CPS ≥ 5, and based on first 354 OS events among all randomized patients with PD-L1 CPS ≥ 5. For PFS: stratified analysis based on first 228 PFS events among the first 298 randomized patients with PD-L1 CPS ≥ 5.

Source: CheckMate-649 Clinical Study Report.¹⁴

Type I error was controlled using a hierarchical testing strategy. For the co-primary end points of PFS and OS among patients with PD-L1 CPS of 5 or higher, a 2-sided family-wise error rate of 5% was split between the co-primary outcomes: 2% was allocated to PFS and 3% was allocated to OS. If the OS comparison (nivolumab plus chemotherapy versus chemotherapy) in patients with PD-L1 CPS of 5 or higher was significant, then OS in patients with PD-L1 CPS of 1 or higher and OS in all randomized patients would be sequentially tested at a 2-sided significance level of 1.5% (0.7% each). For analysis of OS in patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, and all randomized patients, the significance levels at the interim and final analyses were obtained using the Lan-DeMets alpha spending function with O’Brien-Fleming type boundary. At the time of the interim analysis, the significance level was based on actual OS events observed and the estimated final number of events; the alpha boundary was 0.016 based on the spending function. At the final analysis, the significance level will be calculated using the number of events, with consideration of the alpha already spent at the interim analysis. For the interim analysis of OS in randomized patients with PD-L1 CPS of 1 or higher and all randomized patients, the significance levels were obtained using the same information fraction as for patients with PD-L1 CPS of 5 or higher.

In the original study design (1:1:1 randomization to nivolumab plus chemotherapy, chemotherapy, or nivolumab plus ipilimumab followed by nivolumab monotherapy), comparisons between the nivolumab plus ipilimumab followed by nivolumab monotherapy and chemotherapy arms were accounted for in the statistical hierarchy and were nested to be tested following the co-primary OS and PFS analyses in patients with PD-L1 CPS of 5 or higher.

Power for Interim and Final Analyses

The final sample size was based on an event-driven analysis (total 2,005 patients). The final PFS analysis and the interim OS analysis was planned at a minimum follow-up of 12 months. The final OS analysis was planned at a minimum follow-up of 24 months. At the database lock of July 10, 2020, the minimum follow-up was 12.1 months (e.g., final PFS and interim OS analysis). Given changes in the study protocol, and assuming that the prevalence of PD-L1 CPS of 5 or higher was 35%, it was estimated that for the July 10, 2020 cut-off (final PFS analysis and interim OS analysis), the primary analysis population would consist of 554 patients with PD-L1 CPS of 5 or higher concurrently randomized to the nivolumab plus chemotherapy and chemotherapy arms. The HR for PFS was modelled as a 2-piece HR with a delayed effect (HR = 1) for the first 3 or 6 months, followed by a constant HR of 0.56 thereafter. With a type I error of 2% at 12 months minimum follow-up, the expected number of PFS events was estimated to be 497 for a 3-month delayed separation of Kaplan-Meier (KM) curves and approximately 99% power, or 506 PFS events for a 6-month delay and approximately 60% power. For OS, the HR was modelled as a 2-piece HR (a delayed effect with a HR of 1 versus chemotherapy for the first 6 months, followed by a constant HR of 0.65 thereafter). At 24 months minimum follow-up (the final analysis), it was expected that 466 events would be observed, providing a power of approximately 85% with a type I error of 3% (2-sided). The actual observed prevalence of PD-L1 CPS of 5 or higher was 60% in the locked database of July 10, 2020 among all randomized patients in the 3 treatment arms. Therefore, the power for PFS and OS analyses was based on 949 patients concurrently randomized to nivolumab plus chemotherapy versus chemotherapy. Using the same PFS model as in the design, with 3 months or 6 months delayed treatment effect, the expected number of PFS events would be 841 and 857, with corresponding power of 99.9% and 84%, respectively. For OS, the expected number of events, using the same model as in the design, was 800 events, providing a power of 97.9%.

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||, data for a February 16, 2021, database lock were provided, reflecting a minimum follow-up of 19.4 months. The planned final analysis of OS (minimum follow-up 24 months) is expected in late 2021 and will include the analysis of the nivolumab plus ipilimumab arm. Power calculations for the February 16, 2021, database lock were not provided.

Protocol Changes

Power considerations for the CheckMate-649 study were affected by amendments to the original study protocol (May 4, 2016) over time, which are summarized in this section. In the original study design, the primary objective was to compare OS in patients with tumour cell PD-L1 ≥ 1% treated with nivolumab plus ipilimumab versus chemotherapy. The planned sample size was 750 patients, based on a piecewise mixture cure rate (with cure defined as long-term survival beyond the study follow-up period) assuming 3 months of delayed separation of curves between nivolumab plus ipilimumab and chemotherapy, a 15% cure rate in the nivolumab plus ipilimumab arm, an HR of 0.7 in non-cured patients, and median OS of 8 months in Europe or North America and 9.6 months in Japan. A total of 240 events in the approximately 300 patients with tumour cell PD-L1 of 1% or more among 750 randomized patients would provide approximately 90% power to detect an average HR of 0.64 with a type I error of 5% (2-sided).

The major changes to the study protocol are as follows:

• In Revised Protocol 02 (Amendment 08; December 7, 2016) the nivolumab plus chemotherapy arm was added, based on positive results of the CA209012³¹ and KEYNOTE-059³² studies. Subsequent randomization was 1:1:1, and the total sample size was increased to 1,349 patients. At the time of this protocol change, 83 patients had been randomized under the original protocol, with 1:1 randomization. The family-wise error rate (5%) was split equally between the 2 comparisons (nivolumab plus ipilimumab versus chemotherapy and nivolumab plus chemotherapy versus chemotherapy).

• In Revised Protocol 04 (Amendment 17, January 5, 2018), PFS and ORR were added as primary end points for the nivolumab plus chemotherapy versus chemotherapy comparison, and the primary analysis population for this comparison was changed to all randomized patients, based on positive results of the ATTRACTION-4 study.¹⁶ The overall significance level of 2.5% allocated to the nivolumab plus chemotherapy versus chemotherapy comparison was split 0.9% for PFS, 1.5% for OS, and 0.1% for ORR.

• In Revised Protocol 05 (Amendment 19, May 29, 2018), the sample size was increased to 1,649 patients to enable more robust analysis across different PD-L1 cut-offs, based on data from the ATTRACTION-4¹⁶ and KEYNOTE-061³³ studies.

• In Revised Protocol 06 (Amendment 20, June 11, 2018), the DMC conducted a pre-planned safety review. Enrolment in the nivolumab plus ipilimumab arm was closed, following DMC review of all 1,003 randomized patients in the 3 treatment arms, due to concerns of the DMC regarding increased early death rates and high toxicity rates.

• In Revised Protocol 07 (Amendment 23, September 14, 2018), the primary analysis population for the nivolumab plus chemotherapy versus chemotherapy comparison was changed again, from all randomized patients to patients with PD-L1 CPS ≥ 5, based on data from the CA209032,³⁴ KEYNOTE-059,³² and KEYNOTE-061³³ studies. OS and PFS were maintained as primary end points, and ORR was moved to a secondary end point. The family-wise error rate of 5% was split 2% for PFS and 3% for OS. The PFS analysis was planned when 228 events occurred in the first 298 patients with PD-L1 CPS ≥ 5, to ensure a minimum follow-up of approximately 8 months. Two interim analyses of OS were planned when 248 and 301 events had occurred in patients with PD-L1 CPS of 5 or higher (the number of patients with PD-L1 CPS ≥ 5 was assumed to be 420).

• In Revised Protocol 08 (Amendment 26, November 15, 2018), the sample size was increased to 2,005 patients, based on pooled blinded monitoring of PD-L1 CPS of 5 or higher prevalence in the CheckMate-649 study. This was necessary to achieve the planned 420 patients with PD-L1 CPS of 5 or higher in the primary PFS and OS analyses.

• In Revised Protocol 09 (Amendment 29, September 16, 2019), the trigger for PFS/OS analyses was changed from event-driven to time-driven, when it was detected that PD-L1 CPS had been significantly underscored for a proportion of study participants, and thus samples had to be rescored. Since enrolment was completed before this amendment, the final sample size (2,005 patients in total) remained based on the original event-driven analysis. The primary analysis population for PFS was updated to patients with PD-L1 CPS of 5 or higher, and the 2 interim OS analyses were reduced to a single interim analysis. Based on data from the KEYNOTE-062 study³⁵ showing delayed separation of the PFS and OS KM curves, the timing of the PFS and OS analyses was changed to a minimum follow-up of 12 months (for the final PFS analysis and the interim OS analysis) and 24 months (for the final OS analysis); these analyses would be based on all events observed at the time of database lock.

Analytical Techniques

For the co-primary end points of OS and PFS per BICR in patients with PD-L1 CPS ≥ 5, the nivolumab plus chemotherapy and chemotherapy arms were compared using a 2-sided stratified log-rank test. The HR comparing treatment arms was calculated from stratified Cox proportional hazards model, with treatment as the sole covariate. Ties were handled using the exact method. Stratification factors for these analyses were region (Asia versus North America versus rest of world), ECOG PS (0 versus 1), chemotherapy regimen (XELOX versus FOLFOX), and tumour cell PD-L1 (≥ 1% versus < 1%, including indeterminate). CIs were adjusted for the corresponding significance level of the HR. The PFS and OS functions were estimated using the KM product limit method; 2-sided 95% CIs for median survival were obtained by log-log transformation. PFS and OS rates at fixed time points and their 95% CIs were derived from KM analysis, using the Greenwood formula variance derivation and on log-log transformation of the survivor function. Analysis of OS and PFS as secondary outcomes in patients with different PD-L1 CPS and analysis of DOR was performed in the same manner as the primary analysis.

ORR and its exact 95% CI was calculated in each treatment group using the Clopper-Pearson method. The difference in ORRs between treatment groups, and the corresponding 95% CI, was calculated using the Cochran-Mantel-Haenszel test, adjusted by stratification factors used for randomization. Stratified ORs, and their 95% CIs, were calculated using the Mantel-Haenszel estimator. Patients who discontinued the study and tumour assessments before objective CR or PR were counted as non-responders. Patient-reported HRQoL data (EQ-5D-3L utility index score and VAS, FACT-Ga total and subscale scores, and FACT-G7 total score) were presented as descriptive and summary statistics. Analysis of HRQoL data was restricted to randomized patients who had an assessment at baseline and at least 1 post-baseline assessment. TTSD was analyzed in the same manner as the primary analyses. A sensitivity analysis was conducted for TTSD, accounting for events (deterioration of symptoms) on or after treatment discontinuation. For HRQoL outcomes (EQ-5D-3L utility index scores and VAS, FACT-Ga total and GaCS score, FACT-G7 scores), missing data were not accounted for, and observed data were presented without imputation. The strategy used for calculation of EQ-5D-3L utility index scores, FACT-G7 scores, and FACT-Ga total and GaCS scores for partially completed questionnaires was not stated.

Sensitivity analyses for the co-primary and hierarchically tested secondary OS and PFS analyses included unstratified analysis, unstratified analysis with stratification factors as covariates, and the originally planned event-driven analysis of Revised Protocol 07. Subgroup analyses of pre-specified subgroups (by region, ECOG PS, chemotherapy regimen, age, sex, race, tumour location at diagnosis, disease stage at diagnosis, locally recurrent/advanced versus metastatic disease, prior cancer surgery, prior radiotherapy, Lauren classification, WHO histologic classification, TNM (tumour, lymph node, metastasis) classification, presence of target lesions, number of organs with baseline lesions, time from diagnosis to randomization, peritoneal metastasis, and liver metastasis) were conducted as per the primary analysis in exploratory fashion. The study was not specifically powered to evaluate outcomes in individual strata.

Analysis Populations

The screened population consisted of all patients concurrently randomized to receive nivolumab plus chemotherapy or chemotherapy, as well as patients enrolled as of the start of the 1:1:1 randomization and not randomized to any of the treatment arms. All randomized patients consisted of the set of patients randomized concurrently to receive nivolumab plus chemotherapy or chemotherapy starting on April 17, 2017. Patients randomized to chemotherapy before the addition of the nivolumab plus chemotherapy arm were not included in the set of randomized patients. All efficacy analyses were conducted in the set of randomized patients. Treated patients (the safety set) consisted of patients concurrently randomized to receive nivolumab plus chemotherapy or chemotherapy who received at least 1 dose of study drug.

Results

Unless noted otherwise, data presented are from the July 10, 2020, database lock.

Patient Disposition

The total number of patients screened in the CheckMate-649 study across all phases was ||||| patients (including the original 1:1 randomization to nivolumab plus ipilimumab versus chemotherapy, 1:1:1 randomization after the nivolumab plus chemotherapy arm was introduced, and the continuing 1:1 randomization to nivolumab plus chemotherapy versus chemotherapy after the nivolumab plus ipilimumab arm was closed). Screening information was provided for the entire study (||||| patients screened). Overall, 2,031 patients were randomized to all 3 arms of the study, and 1,155 (36.3%) were screen failures. The most common reason for screen failure was “patient no longer meets study criteria” (900 patients, 28.2%) because of HER2 positivity (130 patients, 14.4%), no tumour sample available for PD-L1 analysis (130 patients, 14.4%), and ECOG PS of 2 or higher (108 patients, 12.0%).

Patient disposition in the CheckMate-649 trial is summarized in Table 12. In the nivolumab plus chemotherapy and chemotherapy arms, a collective total of 2,687 patients were enrolled. This total represents all concurrently randomized patients to nivolumab plus chemotherapy and to chemotherapy, as well as patients enrolled as of the start of the 1:1:1 randomization and not randomized to any of the treatment arms. Among these 2,687 patients, 1,581 (58.8%) were concurrently randomized to receive either nivolumab plus chemotherapy or chemotherapy (also known as the all-comers population). Among randomized patients, slightly higher proportions received at least 1 dose of protocol therapy in the nivolumab plus chemotherapy arm (|||% of patients with PD-L1 CPS of 5 or higher and 99.1% of all randomized patients) compared with the chemotherapy arm (|||% of patients with PD-L1 CPS ≥ 5 and 96.8% of all randomized patients), primarily due to higher rates of consent withdrawal in the chemotherapy arm. Most patients in the study discontinued treatment (89.3% and 94.9% of all randomized patients in the nivolumab plus chemotherapy and chemotherapy arms, respectively). The most common reasons for discontinuing treatment were disease progression (65.9% and 68.8% of all randomized patients in the nivolumab plus chemotherapy and chemotherapy arms, respectively). Higher proportions of patients in the chemotherapy arm discontinued therapy due to withdrawal of consent or patient request (5.3% and 4.6% of all randomized patients, respectively) compared with the nivolumab plus chemotherapy arm (2.6% and 1.7% of all randomized patients, respectively). Slightly higher proportions of all randomized patients in the nivolumab plus chemotherapy arm (7.7%) discontinued protocol therapy due to toxicity compared with the chemotherapy arm (5.2%). The most common reason for discontinuing the study, other than death, was withdrawal of consent (2.6% in the nivolumab plus chemotherapy arm and 4.7% in the chemotherapy arm)

Table 12: Patient Disposition in the CheckMate-649 Trial

AE = adverse event; chemo = chemotherapy; CPS = combined positive score; nivo = nivolumab; NR = not reported; PD-L1 = programmed death ligand 1; SD = standard deviation.

Note: Results are from the July 10, 2020, database lock.

^aThe enrolled population (N = 2,687) represents all concurrently randomized patients to nivolumab + chemotherapy and chemotherapy as well as patients enrolled as of the start of the 1:1:1 randomization and not randomized to any of the treatment arms. Thus, the frequency of patients enrolled but not randomized does not reflect the actual screen failure rate. The total study enrolment was 3,186 patients.

^bPercentage of patients randomized. Treated patients were randomized patients who received at least one dose of study drug.

^cPercentage of treated patients.

^dPatient status at end of treatment.

^eIncludes patients still on treatment and patients off treatment continuing in the follow-up period.

Source: CheckMate-649 Clinical Study Report.¹⁴

Protocol deviations in the CheckMate-649 trial are summarized in Table 13. Relevant protocol deviations were predefined in the statistical analysis plan, as deviations related to inclusion or exclusion criteria, study conduct, study management, or subject assessment that were programmable and could affect the interpretability of study results. In all randomized patients at baseline, 1.3% and 1.4% of the nivolumab plus chemotherapy and chemotherapy arms, respectively, had a relevant protocol deviation. During the treatment period, 1.0% and 0.5% of the nivolumab plus chemotherapy and chemotherapy arms, respectively, had a relevant protocol deviation.

Table 13: Relevant Protocol Deviations in the CheckMate-649 Trial

Chemo = chemotherapy; CPS = combined positive score; ECOG PS = Eastern Cooperative Oncology Group performance status; Nivo = nivolumab; PD-L1 = programmed death ligand 1.

Note: Results are from the July 10, 2020, database lock.

^aPatients without inoperable, advanced, or metastatic GC or GEJ or distal esophageal carcinoma or without histologically confirmed predominant adenocarcinoma. Of 2 patients with wrong diagnosis, one was diagnosed after randomization and immediately discontinued from the study and one patient’s correct diagnosis (multiple myeloma) was found after the patient discontinued study therapy.

^bPrior adjuvant or neoadjuvant chemotherapy was permitted.

^cNo disease at baseline assessed based on investigator tumour assessments.

^dPalliative radiotherapy was allowed and not counted as a deviation. One patient with prohibited therapy listed in the table received human granulocyte colony stimulating factor which was allowed.

Source: CheckMate-649 Clinical Study Report.¹⁴

Exposure to Study Treatments

Treatment exposure in the CheckMate-649 trial is summarized in Table 14. The mean (SD) duration of nivolumab therapy was similar in patients receiving nivolumab plus FOLFOX and nivolumab plus XELOX (|||| [||||] months and |||| [||||] months, respectively). The mean durations of administration of most chemotherapy components were shorter than that of nivolumab (nivolumab plus FOLFOX: oxaliplatin |||| months, leucovorin |||| months, 5-fluorouracil |||| months, 5-fluorouracil continuous |||| months; nivolumab plus XELOX: oxaliplatin |||| months, capecitabine |||| months). Chemotherapy components were administered for slightly shorter mean durations in the chemotherapy arm compared with the nivolumab plus chemotherapy arm (FOLFOX: oxaliplatin |||| months, leucovorin |||| months, 5-fluorouracil |||| months, 5-fluorouracil continuous |||| months; XELOX: oxaliplatin |||| months, capecitabine |||| months).

Relative dose intensities were calculated using the following formulas. For patients receiving nivolumab plus XELOX, the relative dose intensity of nivolumab was calculated as: 100 × cumulative dose (mg) / ([last dose date of nivolumab 360 mg – start date of nivolumab + 21] × 360/21) + ([last dose date of nivolumab 480 mg – start date of nivolumab 480 mg + 28] × 480/28). For patients receiving XELOX, the relative dose intensity of oxaliplatin was calculated as: 100 × (cumulative dose [mg/m²] / [last oxaliplatin dose date – oxaliplatin start dose date + 21] × 130/21). For patients receiving XELOX, the relative dose intensity of capecitabine was calculated as: 100 × (cumulative dose [mg/m²] / first dose of capecitabine in the last cycle – capecitabine start dose date + 21) × 28,000/21. For patients receiving nivolumab plus FOLFOX, the relative dose intensity of nivolumab was calculated as: 100 × cumulative dose (mg) / ([last dose date of nivolumab 240 mg – start date in the nivolumab monotherapy phase + 14] × 240/14) + ([last dose date of nivolumab 480 mg – start date of nivolumab 480 mg + 28] × 480/28). For patients receiving FOLFOX, the relative dose intensity of oxaliplatin was calculated as: 100 × (cumulative dose [mg/m²] / [last oxaliplatin dose date – oxaliplatin start dose date + 14] × 85/14). For patients receiving FOLFOX, the relative dose intensity of leucovorin was calculated as: 100 × (cumulative dose [mg/m²] / [last leucovorin dose date – leucovorin start dose date + 14] × 400/14). For patients receiving FOLFOX, the relative dose intensity of bolus 5-fluorouracil was calculated as: 100 × (cumulative dose [mg/m²] / [First dose of 5-fluorouracil in the last cycle – 5-fluorouracil start dose date + 14] × 400/14), and that of continuous 5-fluorouracil was calculated as: 100 × (cumulative dose [mg/m²] / [first dose of continuous 5-fluorouracil in the last cycle – continuous 5-fluorouracil start dose date + 14] × 2,400/14).

The relative dose intensity of nivolumab was higher than that of chemotherapy, but the relative dose intensities of chemotherapy components were similar in both treatment groups.

For the 2 database cut-offs included in this report (July 10, 2020, and February 16, 2021), the minimum follow-up time was 12.1 months and 19.4 months, respectively. For the database cut-off of July 10, 2020, the mean (SD) follow-up was |||| (||||) months in the nivolumab plus chemotherapy arm and |||| (||||) months in the chemotherapy arm.

For frequencies and duration of treatment beyond progression in the nivolumab plus chemotherapy arm, see Appendix 5.

Figure 3 shows a KM analysis of time to treatment discontinuation in the CheckMate-649 trial, based on the July 10, 2020, database lock. Median time to treatment discontinuation was 6.75 months (95% CI, 6.11 months to 7.36 months) in the nivolumab plus chemotherapy arm and 4.86 months (95% CI, 4.47 months to 5.29 months) in the chemotherapy arm.

Figure 3: Time to Treatment Discontinuation in the CheckMate-649 Trial

Note: Symbols represent censored observations.

Source: CheckMate-649 Clinical Study Report.¹⁴

Concomitant non–study drug therapies administered in the CheckMate-649 trial are summarized in Table 15. ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Table 15: Concomitant Non–Study Therapies in the CheckMate-649 Trial

|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

^a|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.

^b|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.

Note: Results are from the July 10, 2020 database lock.

Source: CheckMate-649 Clinical Study Report.¹⁴

Anticancer therapies received subsequent to study treatment discontinuation in the CheckMate-649 study are shown in Table 16. Overall, 37.6% of patients in the nivolumab plus chemotherapy arm and 41.2% of patients in the chemotherapy arm received at least 1 subsequent anticancer therapy, which were primarily systemic chemotherapy regimens. A higher proportion of patients in the chemotherapy arm (8.1% of all randomized patients) received subsequent immunotherapy, including anti-PD-1 inhibitors, compared with the nivolumab plus chemotherapy arm (1.5% of all randomized patients).

Table 16: Subsequent Anticancer Therapies in the CheckMate-649 Trial

CTLA-4 = cytotoxic T-lymphocyte-associated protein 4; PD-1 = programmed death 1; PD-L1 = programmed death ligand 1; TS1 = tegafur + gimeracil + oteracil.

Note: Results are from the July 10, 2020, database lock.

^aIncludes palliative radiotherapy.

^bPatients may have received more than 1 subsequent therapy.

^c||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.

Source: CheckMate-649 Clinical Study Report.¹⁴

Efficacy

Only those efficacy outcomes and analyses of subgroups identified in the review protocol are reported in this section. Refer to Appendix 5 for detailed efficacy data. Unless noted otherwise, data presented are from the July 10, 2020, database lock.

Overall Survival

Results for OS in patients with PD-L1 CPS ≥ 5, patients with PD-L1 CPS of 1 or higher, patients with CPS of 10 or higher, and all randomized patients are shown in Table 17, Figure 4, Figure 5, and Figure 6.

At the July 10, 2020, database lock and in the primary analysis population of patients with PD-L1 CPS of 5 or higher, OS events had occurred in 309/473 (65.3%) of patients in the nivolumab plus chemotherapy arm and in 362/482 (75.1%) of patients in the chemotherapy arm. Among patients with PD-L1 CPS ≥ 5, median OS was longer in the nivolumab plus chemotherapy arm (14.39 months; 95% CI, 13.11 to 16.23 months) compared with the chemotherapy arm (11.10 months; 95% CI, 10.02 to 12.09 months) (P < 0.0001). The HR for OS comparing nivolumab plus chemotherapy versus chemotherapy was 0.71 (98.4% CI, 0.59 to 0.86) in favour of nivolumab plus chemotherapy. The 12-month OS rate among patients with PD-L1 CPS ≥ 5 was 57.3% (95% CI, 52.6% to 61.6%) in the nivolumab plus chemotherapy arm and 46.4% (95% CI, 41.8% to 50.8%) in the chemotherapy arm.

At the July 10, 2020, database lock and in the hierarchically tested secondary analysis population of patients with PD-L1 CPS of 1 or higher, OS events had occurred in 434/641 (67.7%) of patients in the nivolumab plus chemotherapy arm and in 492/655 (75.1%) of patients in the chemotherapy arm. Among patients with PD-L1 CPS of 1 or higher, median OS was longer in the nivolumab plus chemotherapy arm (13.96 months; 95% CI, 12.55 to 14.98 months) compared with the chemotherapy arm (11.33 months; 95% CI, 10.64 to 12.25 months; P < 0.0001). The HR for OS comparing nivolumab plus chemotherapy versus chemotherapy was 0.77 (99.3% CI, 0.64 to 0.92) in favour of nivolumab plus chemotherapy. The 12-month OS rate among patients with PD-L1 CPS of 1 or higher was 57.3% (95% CI, ||||% to ||||%) in the nivolumab plus chemotherapy arm and 47.0% (95% CI, |||||||% to |||||||%) in the chemotherapy arm.

At the July 10, 2020, database lock and in the hierarchically tested secondary analysis population of all randomized patients, OS events had occurred in 544 of 789 (68.9%) patients in the nivolumab plus chemotherapy arm and in 591 of 792 (74.6%) patients in the chemotherapy arm. Among all randomized patients, median OS was longer in the nivolumab plus chemotherapy arm (13.83 months; 95% CI, 12.55 to 14.55 months) compared with the chemotherapy arm (11.56 months; 95% CI, 10.87 to 12.48 months; P = 0.0002). The HR for OS comparing nivolumab plus chemotherapy versus chemotherapy was 0.80 (99.3% CI, 0.68 to 0.94) in favour of nivolumab plus chemotherapy. The 12-month OS rate among all randomized patients was 55.0% (95% CI, 51.4% to 58.4%) in the nivolumab plus chemotherapy arm and 47.9% (95% CI, 44.4% to 51.4%) in the chemotherapy arm.

At the July 10, 2020, database lock and in the secondary analysis population of patients with PD-L1 CPS of 10 or higher, OS events had occurred in ||||/|||| (||||%) of patients in the nivolumab plus chemotherapy arm and in ||||/|||| (||||%) of patients in the chemotherapy arm. Among patients with PD-L1 CPS of 10 or higher, median OS was |||| months (95% CI, |||| to |||| months) in the nivolumab plus chemotherapy arm and |||| months (95% CI, |||| to |||| months) in the chemotherapy arm. The HR for OS comparing nivolumab plus chemotherapy versus chemotherapy was |||| (95% CI, |||| to ||||). The 12-month OS rate among patients with PD-L1 CPS of 10 or higher was ||||% (95% CI, ||||% to ||||%) in the nivolumab plus chemotherapy arm and ||||% (95% CI, ||||% to ||||%) in the chemotherapy arm.

Sensitivity analyses of OS as of July 10, 2020, database lock in the primary and hierarchically tested secondary analysis populations showed similar results that were consistent with the main analyses (see Appendix 5). Results were also similar for the more recent database lock of February 16, 2021. KM plots of OS based on data from the February 16, 2021, database lock are shown in Appendix 5.

Figure 4: OS Among Patients With PD-L1 CPS ≥ 5 in the CheckMate-649 Trial

Note: Symbols represent censored observations. HR is from stratified Cox proportional hazards model and P value is from stratified log-rank test.

Source: CheckMate-649 Clinical Study Report.¹⁴

Figure 5: OS Among Patients With PD-L1 CPS ≥ 1 in the CheckMate-649 Trial

Note: Symbols represent censored observations. HR is from stratified Cox proportional hazards model and P value is from stratified log-rank test.

Source: CheckMate-649 Clinical Study Report.¹⁴

Figure 6: OS Among All Randomized Patients in the CheckMate-649 Trial

Note: Symbols represent censored observations. HR is from stratified Cox proportional hazards model and P value is from stratified log-rank test.

Source: CheckMate-649 Clinical Study Report.¹⁴

Subgroup analyses of OS by PD-L1 status, HER2 status, and MSI status were of interest to this review and are shown in Table 18 and Appendix 5. In the primary and hierarchically tested secondary analysis populations (patients with PD-L1 CPS ≥ 5, patients with PD-L1 CPS ≥ 1, and all randomized patients), the HRs from unstratified Cox proportional hazards models comparing nivolumab plus chemotherapy versus chemotherapy were 0.87 (95% CI, 0.71 to 1.06), 0.84 (95% CI, 0.71 to 0.99), and 0.85 (95% CI, 0.73 to 0.99), respectively, in patients with HER2-negative status and 0.53 (95% CI, 0.42 to 0.67), 0.67 (95% CI, 0.55 to 0.82), and 0.71 (95% CI, 0.59 to 0.85), respectively, in patients with unreported HER2 status. In the primary and hierarchically tested secondary analysis populations (patients with PD-L1 CPS ≥ 5, patients with PD-L1 CPS ≥ 1, and all randomized patients), the HRs comparing nivolumab plus chemotherapy versus chemotherapy were 0.33 (95% CI, 0.12 to 0.87), |||| (||||, ||||), and 0.37 (95% CI, 0.16 to 0.87), respectively, in MSI-H patients, and 0.73 (95% CI, 0.62 to 0.85), |||| (||||, ||||), and 0.80 (95% CI, 0.71 to 0.91), respectively, in microsatellite stable (MSS) patients. Among all randomized patients, the HRs comparing nivolumab plus chemotherapy versus chemotherapy favoured nivolumab plus chemotherapy to a greater degree as the PD-L1 CPS cut-off increased. Among randomized patients with PD-L1 CPS less than 10, less than 5, and less than 1, the HRs comparing nivolumab plus chemotherapy versus chemotherapy were 0.94 (95% CI, 0.80 to 1.10), 0.94 (95% CI, 0.78 to 1.13), and 0.92 (95% CI, 0.70 to 1.23), respectively.

Table 18: OS by PD-L1 CPS Among All Randomized Patients in the CheckMate-649 Trial

CI = confidence interval; CPS = combined positive score; HR = hazard ratio; OS = overall survival.

Note: HRs are from unstratified Cox proportional hazards models using data from the July 10, 2020, database lock.

Source: CheckMate-649 Clinical Study Report.¹⁴

Health-Related Quality of Life

EQ-5D-3L utility index scores and EQ VAS scores among all randomized patients in the CheckMate-649 study are shown in Figure 7 and Figure 8. The EQ-5D-3L questionnaire completion rate during the treatment period for all randomized patients was 90% or more at baseline and 80% or more at most subsequent assessments for which at least 10 patients were eligible to respond. Among patients who had discontinued study treatment but remained in the study, completion rates ranged from ||||% to ||||% in the nivolumab plus chemotherapy arm and from ||||% to ||||% in the chemotherapy arm at follow-up visits || and || after treatment discontinuation. Among patients who had discontinued study treatment but remained in the study, completion rates ranged from ||||% to ||||% in the nivolumab plus chemotherapy arm and from ||||% to ||||% in the chemotherapy arm during survival follow-up at assessments for which |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

Figure 7: Mean Changes in EQ-5D-3L Utility Index Score From Baseline Among All Randomized Patients in the CheckMate-649 Trial

Note: Utility index scores range from 0 (no utility) to 1 (perfect utility). Error bars represent standard error of the mean. Horizontal lines indicate the sponsor-defined MID of 0.08 points. Only time points with a minimum of 5 patient responses are plotted.

Source: CheckMate-649 Clinical Study Report.¹⁴

Figure 8: Mean Changes in EQ-5D-3L Self-Rated Health Status EQ VAS From Baseline Among All Randomized Patients in the CheckMate-649 Trial

This figure has been redacted at the sponsor’s request.

Note: EQ VAS scores range from 0 (worst imaginable health) to 100 (best imaginable health). Error bars represent standard error of the mean. Horizontal lines indicate the sponsor-defined MID of 7 points. Only time points with a minimum of 5 patient responses are plotted.

Source: CheckMate-649 Clinical Study Report.¹⁴

FACT-Ga total scores and GaCS scores among all randomized patients in the CheckMate-649 study are shown in Figure 9. The FACT-Ga questionnaire completion rate during the treatment period for all randomized patients was 90% or more at baseline and 80% or more at most subsequent assessments for which at least 10 patients were eligible to respond. Among patients who had discontinued study treatment but remained in the study, completion rates for the GaCS and FACT-G7 ranged from 45.6% to 78.6% in the nivolumab plus chemotherapy arm and from 31.3% to 55.6% in the chemotherapy arm at assessments for which at least 10 patients were eligible to respond. Patients in both arms had similar FACT-Ga total scores at baseline (mean = 126.6, SD = 28.3, in the nivolumab plus chemotherapy arm and mean = 126.8, SD = 26.8, in the chemotherapy arm). Patients in both arms had similar GaCS scores at baseline (mean = 49.6, SD = 15, in the nivolumab plus chemotherapy arm and mean = 50.0, SD = 15.0, in the chemotherapy arm). FACT-Ga total scores and GaCS scores were numerically higher (improved) at post-baseline assessments compared with the baseline assessment until 121 weeks and 127 weeks post-baseline, respectively, in both treatment groups.

Figure 9: Mean Changes in FACT-Ga Total and GaCS Scores From Baseline Among All Randomized Patients in the CheckMate-649 Trial

This figure has been redacted at the sponsor’s request.

Note: FACT-Ga total scores range from 0 to 184, with higher values indicating better HRQoL. GaCS scores range from 0 to 76, with higher values indicating better HRQoL. Error bars represent standard error of the mean. Horizontal lines indicate the sponsor-defined MID of 8.2 points for the FACT-Ga GaCS score). Only time points with a minimum of 5 patient responses are plotted.

Source: CheckMate-649 Clinical Study Report.¹⁴

Progression-Free Survival

Results for PFS per BICR in patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, patients with CPS of 10 or higher, and all randomized patients are shown in Table 19.

At the July 10, 2020, database lock and in the primary analysis population of patients with PD-L1 CPS of 5 or higher, PFS events had occurred in 328 of 473 (69.3%) patients in the nivolumab plus chemotherapy arm and in 350 of 482 (72.6%) patients in the chemotherapy arm (refer to Figure 10 for KM curves). Among patients with PD-L1 CPS of 5 or higher, median PFS was longer in the nivolumab plus chemotherapy arm (7.69 months; 95% CI, 7.03 to 9.17 months) compared with the chemotherapy arm (6.05 months; 95% CI, 5.55 to 6.90 months; P < 0.0001). The HR for PFS comparing nivolumab plus chemotherapy versus chemotherapy was 0.68 (98% CI, 0.56 to 0.81) in favour of nivolumab plus chemotherapy. The 12-month PFS rate among patients with PD-L1 CPS of 5 or higher was |||% (95% CI, |||% to |||%) in the nivolumab plus chemotherapy arm and |||% (95% CI, |||% to |||%) in the chemotherapy arm.

At the July 10, 2020, database lock and in the secondary analysis population of patients with PD-L1 CPS of 1 or higher, PFS events had occurred in 454 of 641 (70.8%) patients in the nivolumab plus chemotherapy arm and in 472 of 655 (72.1%) patients in the chemotherapy arm. Among patients with PD-L1 CPS of 1 or higher, median PFS was 7.49 months (95% CI, 7.03 to 8.41 months) in the nivolumab plus chemotherapy arm and 6.90 months (95% CI, 6.08 to 7.03 months) in the chemotherapy arm. The HR for PFS comparing nivolumab plus chemotherapy versus chemotherapy was 0.74 (95% CI, 0.65 to 0.85). The 12-month PFS rate among patients with PD-L1 CPS of 1 or higher was |||% (95% CI, |||% to |||%) in the nivolumab plus chemotherapy arm and |||% (95% CI, |||% to |||%) in the chemotherapy arm.

At the July 10, 2020, database lock and in the secondary analysis population of all randomized patients, PFS events had occurred in 559 of 789 (70.8%) of patients in the nivolumab plus chemotherapy arm and in 557 of 792 (70.3%) of patients in the chemotherapy arm. Among all randomized patients, median PFS was 7.66 months (95% CI, 7.10 to 8.54) months in the nivolumab plus chemotherapy arm and 6.93 months (95% CI, 6.60 to 7.13 months) in the chemotherapy arm. The HR for PFS comparing nivolumab plus chemotherapy versus chemotherapy was 0.77 (95% CI, 0.68 to 0.87). The 12-month PFS rate among all randomized patients was |||% (95% CI, |||% to |||%) in the nivolumab plus chemotherapy arm and |||% (95% CI, |||% to |||%) in the chemotherapy arm.

At the July 10, 2020, database lock and in the secondary analysis population of patients with PD-L1 CPS of 10 or higher, PFS events had occurred in 252 of 375 (67.2%) patients in the nivolumab plus chemotherapy arm and in 289 of 393 (73.5%) patients in the chemotherapy arm. Among patients with PD-L1 CPS of 10 or higher, median PFS was 8.31 months (95% CI, 6.97 to 9.69 months) in the nivolumab plus chemotherapy arm and 5.78 months (95% CI, 5.45 to 6.87 months) in the chemotherapy arm. The HR for PFS comparing nivolumab plus chemotherapy versus chemotherapy was 0.63 (95% CI, 0.53 to 0.75). The 12-month PFS rate among patients with PD-L1 CPS of 10 or higher was |||% (95% CI, |||% to |||%) in the nivolumab plus chemotherapy arm and |||% (95% CI, |||% to |||%) in the chemotherapy arm.

Sensitivity analyses of PFS as of July 10, 2020, database lock in the primary analysis populations showed similar results (Appendix 5). Results were also similar for PFS analysis using the more recent database lock of February 16, 2021. KM plots of PFS based on data from the February 16, 2021, database lock are shown in Appendix 5.

Subgroup analyses of PFS by PD-L1 status, HER2 status, and MSI status were of interest to this review and are shown in Appendix 5. In the primary analysis population (patients with PD-L1 CPS ≥ 5), the HR for PFS comparing nivolumab plus chemotherapy versus chemotherapy was 0.78 (95% CI, 0.64 to 0.96) in patients with HER2-negative status and 0.57 (95% CI, 0.45 to 0.72) in patients with unreported HER2 status. In the primary analysis population (patients with PD-L1 CPS ≥ 5), the HR for PFS comparing nivolumab plus chemotherapy versus chemotherapy was 0.32 (95% CI, 0.12 to 0.85) in MSI-H patients and 0.70 (95% CI, 0.59 to 0.82) in MSS patients. Among all randomized patients, the unstratified HRs for PFS comparing nivolumab plus chemotherapy versus chemotherapy favoured nivolumab plus chemotherapy to a greater degree as the PD-L1 CPS cut-off increased. Among randomized patients with PD-L1 CPS less than 10, less than 5, and less than 1, the HRs for PFS comparing nivolumab plus chemotherapy versus chemotherapy were ||| (95% CI, ||| to |||), ||| (95% CI, ||| to |||), and ||| (95% CI, ||| to |||), respectively.

Table 19: PFS by BICR Among Patients With PD-L1 CPS ≥ 10, PD-L1 CPS ≥ 5, PD-L1 CPS ≥ 1, and All Randomized Patients in the CheckMate-649 Trial

Chemo = chemotherapy; CI = confidence interval; CPS = combined positive score; HR = hazard ratio; Nivo = nivolumab; PD-L1 = programmed death ligand 1; PFS = progression-free survival.

^aAnalysis based on database lock of July 10, 2020.

^bAnalysis based on database lock of February 16, 2021.

^cBased on Kaplan-Meier estimates.

^dStratified Cox proportional hazards model; HR for nivolumab + chemotherapy compared with chemotherapy. Stratification factors were region, ECOG PS, tumour cell PD-L1, and chemotherapy regimen.

^eTwo-sided P value from stratified log-rank test.

Source: CheckMate-649 Clinical Study Report.¹⁴

Figure 10: PFS Among Patients With PD-L1 CPS ≥ 5 in the CheckMate-649 Trial

Note: Symbols represent censored observations. HR is from stratified Cox proportional hazards model and P value is from stratified log-rank test.

Source: CheckMate-649 Clinical Study Report.¹⁴

Objective Response Rate

ORR (best overall response of PR or CR) per BICR in patients with PD-L1 CPS of 10 or higher, patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, and all randomized patients is shown in Table 20. Among patients with PD-L1 CPS of 10 or higher, the difference in ORR comparing nivolumab plus chemotherapy versus chemotherapy was |||% (95% CI, |||% to |||%) and the OR for response was ||| (95% CI, ||| to |||). Among patients with PD-L1 CPS of 5 or higher, the difference in ORR comparing nivolumab plus chemotherapy versus chemotherapy was |||% (95% CI, |||% to |||%) and the OR for response was ||| (95% CI, ||| to |||). Among patients with PD-L1 CPS of 1 or higher, the difference in ORR comparing nivolumab plus chemotherapy versus chemotherapy was |||% (95% CI, |||% to |||%) and the OR for response was ||| (95% CI, ||| to |||). Among all randomized patients, the difference in ORR comparing nivolumab plus chemotherapy versus chemotherapy was |||% (95% CI, |||% to |||%) and the OR for response was ||| (95% CI, ||| to |||).

Table 20: ORR per BICR Among Patients With PD-L1 CPS ≥ 10, PD-L1 CPS ≥ 5, PD-L1 CPS ≥ 1, and All Randomized Patients in the CheckMate-649 Trial

BICR = blinded independent central review; CI = confidence interval; CPS = combined positive score; CR = complete response; NA = not applicable; NR = not reported; OR = odds ratio; ORR = objective response rate; PD-L1 = programmed death ligand 1; PR = partial response; SD = stable disease; UTD = unable to determine.

^aConfirmed CR or PR by RECIST 1.1.

^bStratified adjusted difference in response rate (nivolumab + chemotherapy vs. chemotherapy) based on DerSimonian and Laird method of weighting.

^cStratified by region, ECOG PS, tumour cell PD-L1, and chemotherapy regimen.

^dStratified adjusted odds ratio (nivolumab + chemotherapy vs. chemotherapy) using Mantel-Haenszel method.

^eTwo-sided P value from stratified Cochran-Mantel-Haenszel test. Note that this P value is not adjusted for multiplicity.

Source: CheckMate-649 Clinical Study Report.¹⁴

Duration of Response

DOR per BICR in patients with PD-L1 CPS of 10 or higher, patients with PD-L1 CPS of 5 or higher, patients with PD-L1 CPS of 1 or higher, and all randomized patients is shown in Table 21. Among patients with PD-L1 CPS of 10 or higher, the median DOR was ||| months (95% CI, ||| to ||| months) in the nivolumab plus chemotherapy arm and ||| months (95% CI, ||| to ||| months) in the chemotherapy arm. Among patients with PD-L1 CPS of 5 or higher, the median DOR was 9.49 months (95% CI, 7.98 to 11.37 months) in the nivolumab plus chemotherapy arm and 6.97 months (95% CI, 5.65 to 7.85 months) in the chemotherapy arm. Among patients with PD-L1 CPS of 1 or higher, the median DOR was ||| months (95% CI, ||| to ||| months) in the nivolumab plus chemotherapy arm and ||| months (95% CI, ||| to ||| months) in the chemotherapy arm. Among all randomized patients, the median DOR was 8.51 months (95% CI, 7.23 to 9.92 months) in the nivolumab plus chemotherapy arm and 6.93 months (95% CI, 5.82 to 7.16 months) in the chemotherapy arm.

Table 21: DOR per BICR Among Patients With Measurable Disease and PD-L1 CPS ≥ 10, PD-L1 CPS ≥ 5, PD-L1 CPS ≥ 1, and All Randomized Patients in the CheckMate-649 Trial

BICR = blinded independent central review; chemo = chemotherapy; CI = confidence interval; CPS = combined positive score; DOR = duration of response; nivo = nivolumab; PD-L1 = programmed death ligand 1.

Note: + indicates a censored value. Results are from the July 10, 2020, database lock.

^aBased on Kaplan-Meier estimates.

Source: CheckMate-649 Clinical Study Report.¹⁴

Symptom Severity

TTSD, defined as time from randomization to a decline of 8.2 points from baseline in GaCS score, is shown for all randomized patients in Table 22 and Figure 11. Median TTSD was not reached in the nivolumab plus chemotherapy arm. The median TTSD in the chemotherapy arm was 21.03 months (95% CI lower bound 12.45 months). The HR for comparison of the nivolumab plus chemotherapy arm versus the chemotherapy arm was 0.77 (95% CI, 0.63 to 0.95).

Table 22: TTSD Among All Randomized Patients in the CheckMate-649 Trial

CI = confidence interval; HR = hazard ratio; NA = not applicable; NR = not reached; TTSD = time to symptom deterioration.

Note: Results are from the July 10, 2020, database lock.

^aBased on Kaplan-Meier estimates.

^bStratified Cox proportional hazards model; HR for nivolumab + chemotherapy compared with chemotherapy. Stratification factors were region, ECOG PS, tumour cell PD-L1, and chemotherapy regimen.

Source: CheckMate-649 Clinical Study Report.¹⁴

Figure 11: TTSD Among All Randomized Patients in the CheckMate-649 Trial

TTSD = time to symptom deterioration.

Note: Symbols represent censored observations. HR is from stratified Cox proportional hazards model.

Source: CheckMate-649 Clinical Study Report.¹⁴

Harms

Only those harms identified in the review protocol are reported below. Refer to Table 23 for detailed harms data.

Adverse Events

Almost all patients in the nivolumab plus chemotherapy arm (99.2%) and the chemotherapy arm (98.0%) experienced at least 1 AE. In the nivolumab plus chemotherapy arm, 69.1% of patients experienced at least 1 AE of Grade 3 or higher, while in the chemotherapy arm, 59.5% of patients experienced at least 1 AE of Grade 3 or higher. Common AEs in both arms were nausea, vomiting, diarrhea, peripheral neuropathy, neutropenia, and anemia.

Serious Adverse Events

SAEs occurred in 54.1% (Grade ≥ 3, 35.9%) of patients receiving nivolumab plus chemotherapy and 43.7% (Grade ≥ 3, 29.9%) of patients receiving chemotherapy. SAEs occurring in at least 2% of either treatment group were vomiting, diarrhea, dysphagia, malignant neoplasm progression, pneumonia, anemia, febrile neutropenia, pneumonitis, and pyrexia. Aside from malignant neoplasm progression, each of these SAEs occurred in less than 4% of either treatment group.

Withdrawals Due to Adverse Events

WDAEs occurred when AEs led to discontinuation of at least 1 study drug, regardless of whether the patient continued with study treatment. Overall, 47.4% of patients in the nivolumab plus chemotherapy arm and 32.7% of patients in the chemotherapy arm experienced WDAEs. Common WDAEs included progressive disease, peripheral neuropathy, and diarrhea. Note that only 7.7% of patients in the nivolumab plus chemotherapy arm and 5.2% of patients in the chemotherapy arm discontinued study treatment due to unacceptable toxicity (Table 12).

Mortality

During the study period, 68.8% of patients receiving nivolumab plus chemotherapy and 74.6% of patients receiving chemotherapy died. Deaths attributed to study drug toxicity by investigators occurred in 1.5% of patients in the nivolumab plus chemotherapy arm and 0.5% of patients in the chemotherapy arm.

Notable Harms

Refer to Table 10 for definitions of select AEs, IMAEs, and OESIs, which were protocol-defined to capture the expected toxicity profile of nivolumab. Select AEs in all categories, and most IMAEs and OESIs, occurred more frequently in the nivolumab plus chemotherapy arm than in the chemotherapy arm. Select AEs affecting the gastrointestinal system (40.3% in the nivolumab plus chemotherapy arm and 33.9% in the chemotherapy arm), select AEs affecting the hepatic system (34.1% and 24.3%), select AEs affecting the skin (33.5% and 17.9%), select AEs affecting the endocrine system (15.0% and 1.8%), and hypersensitivity/infusion reactions (15.1% and 5.9%) were the most common select AEs in the nivolumab plus chemotherapy arm. Hypothyroidism/thyroiditis (9.5% in the nivolumab plus chemotherapy arm and 0.8% in the chemotherapy arm), rash (6.5% and 0.5%), pneumonitis (4.2% and 0%), diarrhea/colitis (3.3% and 0%), hyperthyroidism (2.9% and 0.3%), and hepatitis (2.4% and 0%) were the most common IMAEs in the nivolumab plus chemotherapy arm.

Table 23: Summary of Harms in the CheckMate-649 Trial