FDA Approval Summary: Nivolumab for the Adjuvant Treatment of Adults with Completely Resected Esophageal/Gastroesophageal Junction Cancer and Residual Pathologic Disease

In 2021, there will be an estimated 19,260 adults diagnosed with esophageal cancer and 15,530 deaths will occur in the US. In the US, esophageal cancer occurs at a higher rate in men than women (7.3 vs. 1.7 per 100,000) and occurs more frequently in Whites than Blacks, Hispanics, or Asians (4.6, 3.4, 2.5, and 2.3 per 100,000 respectively); the median age at diagnosis is 68 years (1). The global incidence of esophageal cancer varies across geographic regions with the highest incidence occurring in Eastern Asia (predominantly in China), where the highest estimated mortality rates are observed (14.1 per 100,000 men and 4.5 per 100,000 women in East Asia, compared with 5 per 100,000 men and 1 per 100,000 women in North America; ref. 2).

Esophageal squamous cell carcinoma (ESCC; primarily located in the upper or middle esophagus) and esophageal adenocarcinoma (EAC; primarily located in the lower esophagus) comprise the most common histologic subtypes of esophageal cancer. Gastroesophageal junction cancers (GEJC), which are overwhelmingly adenocarcinomas, may be treated as EAC depending on tumor location. While ESCC accounts for 90% of all esophageal cancers worldwide, its incidence has been decreasing, whereas the incidence of adenocarcinoma histology has been increasing in developing countries. The incidence rates of ESCC and EAC are likely due to the prevalence of risk factors: smoking and alcohol for ESCC, and obesity and gastroesophageal reflux disease for EAC (3, 4).

For patients with locally advanced esophageal cancer, standard treatment consists of neoadjuvant chemoradiotherapy followed by resection, which confers an estimated 10-year survival of 38% (5). While the standard of care in the postoperative setting for esophageal cancer treated with neoadjuvant chemoradiotherapy is surveillance, some patients at high risk for recurrence (e.g., positive lymph nodes, positive resection margins) are treated with adjuvant chemotherapy; however, this treatment approach has not been proven effective in randomized trials and is not FDA-approved (6).

Nivolumab (Bristol-Meyers Squibb Company, Princeton, NJ) is a humanized IgG4 monoclonal antibody that binds to programmed death 1 (PD-1) receptor, blocking its interaction with programmed death ligands 1 (PD‐L1) and 2 (PD-L2) and releasing PD-1 pathway–mediated inhibition of the immune response, including the antitumor immune response. Nivolumab is approved for various malignancies, in some cases across several clinical settings for a given tumor (7). Nivolumab has been approved for the treatment of advanced ESCC for patients who are refractory or intolerant to at least one prior fluoropyrimidine- and platinum-based regimen based on the modest improvement in survival demonstrated in the ATTRACTION-3 trial (8).

The FDA granted traditional approval to nivolumab on May 20, 2021 for the adjuvant treatment of completely resected esophageal or gastroesophageal junction cancer (EC/GEJC) with residual pathologic disease in patients who have received neoadjuvant chemoradiotherapy, based on the results of CheckMate 577. This article summarizes the FDA's review of data submitted in the supplemental biologics licensing application and the basis for approval. The investigators’ analyses and interpretation of the data have been previously published (9).

CheckMate 577 (NCT02743494) was a multiregional, randomized, double-blind trial designed to compare disease-free survival (DFS) in patients with completely resected Stage II/III EC/GEJC treated with or without nivolumab who had residual pathologic disease following chemoradiotherapy. Key eligibility criteria included completion of neoadjuvant chemoradiotherapy with a platinum-based chemotherapy followed by complete resection between 4 to 16 weeks prior to randomization with negative margins (R0) and a non-pathologic complete response, and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 1. CheckMate 577 enrolled patients irrespective of baseline tumor cell PD‐L1 status; however, tumor cell PD-L1 status was tested centrally in all patients using the PD-L1 IHC 22–8 pharmDx assay (Agilent/Dako). Patients who had Stage IV resectable disease or those with active autoimmune disease or a medical condition that required immunosuppression were ineligible.

Patients were randomized in a 2:1 ratio to nivolumab 240 mg administered by intravenous infusion every 2 weeks for 16 weeks (Cycles 1–8) followed by 480 mg intravenous infusions every 4 weeks (Cycles 9–17), or placebo. Treatment continued for up to 1 year or until recurrence of disease, unacceptable toxicity, or withdrawal of consent. Randomization was stratified by baseline tumor cell PD-L1 status: (≥ 1% vs. < 1%, indeterminate or non-evaluable), pathologic lymph node status (positive vs. negative), and histology (squamous vs. adenocarcinoma). Disease recurrence/progression was assessed every 12 weeks for 2 years, then every 6 to 12 months between Years 3 and 5. Recurrence was defined as the appearance of one or more new lesions confirmed by cytology/pathology or imaging performed a minimum of 4 weeks from the time of the initial observed recurrence, as assessed by the investigator.

The primary endpoint was DFS, defined as the time between the date of randomization and the date of recurrence or death due to any cause. Overall survival (OS) was a key secondary endpoint. The planned sample size was 760 patients and with 440 DFS events, the study had 91% power to detect a HR of 0.72 using a long-rank test at a two-sided 5% Type I error assuming a median of 21 months in the placebo arm. One interim analysis (IA) at 85% of events and one final analysis were planned for DFS. The efficacy analyses were conducted in the intention-to-treat population, defined as all patients who were randomly assigned to treatment. The log-rank test was used to compare both DFS and OS between study arms, and the Cox proportional hazard model was used to estimate the HR. The Kaplan–Meier product-limit method was applied to summarize DFS and OS. OS was planned to be tested hierarchically if DFS demonstrated superiority in favor of the experimental arm. Two IAs for OS were planned, the first to occur at the same time as the DFS IA. At the time of the DFS IA, the data monitoring committee determined that OS data were not sufficiently mature with 50% information available.

A total of 794 patients were randomized to receive nivolumab (n = 532) or placebo (n = 262); among those, all but two patients in the placebo arm received at least one dose of treatment. Patients were enrolled across multiple regions and countries. Baseline demographics and disease characteristics were mostly balanced (Table 1).

At the first interim DFS analysis (data cutoff, July 3, 2020), 396 DFS events had been observed. With a median follow-up duration of 24.4 months, CheckMate 577 demonstrated a statistically significant improvement in DFS, with a HR of 0.69 [95% confidence interval (CI), 0.56–0.85; two-sided P value = 0.0003]; the estimated median DFS was 22.4 months (95% CI, 16.6–34.0) versus 11.0 months (95% CI, 8.3–14.3) in the nivolumab and placebo arms, respectively (Table 2 and Fig. 1).

The overall treatment effect on DFS was generally consistent across all prespecified exploratory subgroups: age ≥ 65 or < 65 years, sex, ECOG performance status, site of primary tumor, histology, geographic region, etc. Exploratory analyses of subgroups based on baseline tumor characteristics are provided in Table 3.

The safety analysis was based on 792 patients who received at least one dose of nivolumab (n = 532) or placebo (n = 260). The median duration of exposure was 10 months in the nivolumab arm and 9 months in placebo arm.

The risks of nivolumab for this indication appeared consistent with the known safety profile. The most common (≥ 20%) treatment emergent adverse events (TEAE) in the nivolumab arm were fatigue/asthenia (34%), diarrhea (29%), nausea (23%), rash (21%), musculoskeletal pain (21%), and cough (20%); in the placebo arm, TEAEs >20% were fatigue/asthenia (29%), diarrhea (29%), nausea (21%), cough (21%), musculoskeletal pain (20%), and abdominal pain. Twenty percent of patients in each arm experienced a Grade 3 to 4 TEAEs; the most common was fatigue (1.3%) in the nivolumab arm and dysphagia (3.5%) in the placebo arm. Serious adverse events (SAE; i.e., adverse events resulting in death or the need for hospitalization, prolonged hospitalization, or medical intervention to avoid hospitalization) occurred in 30% of patients. The most frequently occurring SAEs (in ≥ 1% of patients) in the nivolumab arm were pneumonia (8%), pneumonitis (4.7%), and dysphagia (1.1%). The most common SAEs in the placebo arm were pneumonia (5.4%), pneumonitis (1.9%), dysphagia (1.9%), pleural effusion (1.5%), diaphragmatic hernia (1.2%), and esophageal stenosis (1.2%).

TEAEs of any grade leading to treatment discontinuation were reported in 68 (13%) of patients in the nivolumab arm, the most common of which were pneumonitis (1.9%), rash (0.6%), and myocarditis (0.6%); TEAEs leading to discontinuation in the placebo arm were reported in 20 (8%) of patients, and the most common was pneumonitis (0.8%).

Immune-related TEAEs (irTEAE) were reported in 160 patients (30%) in the nivolumab arm, and 18 patients (7%) in the placebo arm. The most common irTEAEs (≥ 2% incidence) in patients who received nivolumab were hypothyroidism/thyroiditis (11%), rash (8%), hyperthyroidism (6.6%), pneumonitis (4.7%), diarrhea/colitis (1.9%), and hepatitis (1.1%). The majority (85%) were Grade 1 or Grade 2 in severity.

This approval represents the first approval of an immune checkpoint inhibitor for the adjuvant treatment of completely resected EC/GEJC with residual pathologic disease in patients who have received neoadjuvant chemoradiotherapy. Recent studies show that despite the standard of care for locally advanced EC/GEJC using neoadjuvant chemoradiation followed by resection, the 5-year survival rate is 47% (10). Furthermore, several retrospective cohort reviews indicate that patients who have not achieved pathologic complete response to neoadjuvant chemoradiotherapy are at higher risk of recurrence or progression (11, 12). Residual disease has been used to enrich study populations in other solid tumors, but this is the first time the approach has been used to obtain marketing approval for an indication in EC/GEJC. This strategy may facilitate the identification of high-risk populations, reduce the risk of overtreatment and toxicities, and allow for the use of smaller trials that can be completed more quickly.

While OS is the gold standard endpoint to demonstrate clinical benefit in oncology, a trial with a large sample size and long duration of follow-up is typically needed to evaluate this endpoint. In addition, the analysis of OS results can be confounded by the use of subsequent therapies. For these reasons, in some settings FDA has accepted alternative endpoints to support approval. CheckMate 577 showed a statistically significant and clinically meaningful improvement in DFS in patients receiving adjuvant nivolumab compared with patients receiving placebo. FDA has recognized DFS as an objective and clinically relevant endpoint suitable for traditional approval for solid tumors in the (neo)adjuvant setting (13). At the time of submission, the OS data was immature (50% of events) but given the magnitude of effect on DFS, an acceptable risk profile, and the absence of a detrimental effect on OS, FDA concluded that the applicant had provided substantial evidence of effectiveness and that the benefit:risk assessment is favorable for patients with EC/GEJC with residual pathologic disease following neoadjuvant chemoradiotherapy and complete resection.

Patients with tumor cell PD-L1 expression ≥ 1% and ≥ 5% comprised 16% and 11% of the randomized population, respectively. Similar treatment effects were observed irrespective of tumor cell PD-L1 expression in subgroup analyses. While FDA considers these results to be exploratory as the study was not designed to characterize treatment effect in these subgroups, a similar lack of predictive effect has been observed in other trials in esophageal cancer that measured treatment effects according to tumor cell PD-L1 expression (8, 14). There is accumulating evidence suggesting the limitation of PD-L1 expression as a biomarker based on the use of different methodologies for assessing PD-L1 expression levels, different assays, and the results of exploratory analyses in patients with low expression of PD-L1. PD-L1 combined positive score is an accepted predictor of benefit for patients with esophagogastric cancers (15). Exploration of better biomarkers and/or future trials to assess outcomes specifically in this population will be needed to better characterize the benefit:risk of immune checkpoint inhibitors in esophageal cancer.

The CheckMate 577 trial enrolled 9 participants who reported Black or African American race (1.1%) and 44 participants who reported Hispanic or Latino ethnicity (5.5%), demonstrating an underrepresentation of these populations relative to the US population. The median age of patients in CheckMate 577 was 62 years, slightly younger than the median age of 68 years for all new cases diagnosed in the US each year. Increased efforts are needed to enroll a more diverse study population in esophageal cancer trials to improve the generalizability of study results (16).

The incidence and severity of AEs observed in CheckMate 577 are consistent with the known safety profile of nivolumab. Patients in the nivolumab arm generally experienced similar symptoms to patients in the placebo arm, with the exception of irTEAEs, which were mostly Grade 1 to 2 and managed with dose interruptions and standard treatment with corticosteroids and hormone replacement.

Nivolumab was approved on May 20, 2021 as the first systemic therapy for the adjuvant treatment of completely resected EC/GEJC with residual pathologic disease in patients who have received neoadjuvant chemoradiotherapy. Treatment with nivolumab has a favorable benefit:risk profile, with a clinically meaningful and statistically significant improvement in DFS demonstrated in CheckMate 577. No new safety signals were identified for nivolumab in the trial population, and there was no increased incidence of adverse events with the exception of irTEAEs, which were mostly managed with standard-of-care treatment. Overall, on the basis of the safety profile and duration of treatment, nivolumab appears to be reasonably tolerated with a favorable risk:benefit profile given the demonstrated benefits in patients with EC/GEJC with residual pathologic disease after chemoradiotherapy.

The application was reviewed under various programs designed to expedite the review of applications for patients with cancer, including the Real-Time Oncology program (17) permitting early receipt of datasets prior to the final application submission; Assessment Aid (18), a voluntary review template submitted by the applicant to facilitate FDA's assessment; and Project Orbis (19), a framework in which FDA reviews an application concurrently with international partners, which for this application included Health Canada, Swissmedic, and Australia's Therapeutic Goods Administration. Although collaboration under Project Orbis fostered discussion of issues pertinent to the review of the nivolumab application, the application review was ongoing in Canada, Switzerland, and Australia at the time that FDA approved the application and has subsequently been approved in Canada and Australia.

No disclosures were reported.

The Editor handling the peer review and decision-making process for this article has no relevant employment associations to disclose.