The FDA’s Oncologic Drug Advisory Committee voted against the use of checkpoint inhibitors in esophageal cancer with PD-L1 expression less than 1.
In an 11 to 1 vote, the FDA’s Oncologic Drug Advisory Committee (ODAC) voted that the risk-benefit assessment is not favorable for the use of checkpoint inhibitors (CPIs) in patients with metastatic or unresectable esophageal squamous cell carcinoma with PD-L1 expression less than 1.1
The ODAC discussed this question through data supporting the current approvals of nivolumab (Opdivo) with chemotherapy or ipilimumab (Yervoy) and pembrolizumab (Keytruda) plus chemotherapy, as well as the application of tislelizumab-jsgr (Tevimbra) plus chemotherapy, for this indication.
“My vote was no,” said ODAC member Hanna Sanoff, MD, MPH, University of North Carolina at Chapel Hill. “Given the completely overlapping survival curves and even some suggestion of no benefit at all, even potential harm, I did not see enough evidence. Despite the small sample sizes, I think the fact that effective treatment is statistically modified by PD-L1 [expression] is pretty compelling.”
Sandra Casak, MD, acting clinical team leader of gastrointestinal malignancies, Division of Oncology 3, Office of Oncologic Diseases, presented the introductory remarks for the FDA. Casak explained that the approvals of pembrolizumab and nivolumab were granted agnostic of PD-L1 expression.
She summarized that the data for pembrolizumab, nivolumab, and tislelizumab were marginal or not favorable in patients with PD-L1 expression under 1 and moderate in patients with intermediate PD-L1 expression under 10. The data did not support CPI use in patients with a PD-L1 expression under 1, with higher magnitudes of benefits observed with expression over 1.
“We now have results across 3 trials suggesting lack of efficacy in [patients with] PD–L1-negative [tumors]. To summarize, improvement in survival with the addition of a checkpoint inhibitor was greatest in patients with higher PD-L1 expression, 10 or more, in all 3 trials.”
The notable risks of CPIs in all patients, regardless of PD-L1 expression, are immune-mediated adverse effects (IMAEs). Casak noted IMAEs could become chronic and often manifested as endocrine, lung, neurologic, or arthritic conditions. Steroids, the most common line of treatment for these AEs, could compromise patient quality of life.
Geetika Srivastava, MD, MSPH, clinical reviewer, Division of Oncology 3, Office of Oncologic Diseases, expanded on the FDA’s perspective. In KEYNOTE-590 (NCT03189719), which supported the approval of pembrolizumab in this indication, the overall survival (OS) HR for patients with esophageal cancer and a combined positive score (CPS) less than 1 was 1.00, suggesting no benefit. Comparatively, the HR for patients with a CPS greater than 10 was 0.57. CheckMate 648 (NCT03143153), which supported the approval of nivolumab, showed an OS HR of 0.93 in patients with a CPS less than 1 and 0.62 for patients with a CPS greater than 10. In Rationale-306 (NCT03430843), supporting the application of tislelizumab, the HR for a tumor area positivity (TAP) score less than 1 was 1.34 and 0.66 with a TAP greater than 10.
Srivastava concluded that the OS benefits seen in the 3 trials appeared to be driven by PD–L1-high subgroups.
Peter C. Enzinger, MD, gastrointestinal oncologist at Dana-Farber Cancer Institute, discussed the high unmet needs and challenges for patients with metastatic esophageal cancer. The considerations for patients in need of first-line treatment included therapeutic urgency and timing of biomarker testing, adverse effect profiles, and long-term survival.
“Some of the reasons [for these challenges] are as follows. There can be different assays and antibody clones used by various organizations that are not FDA approved, which may result in staining variability. Unlike a clinical trial, in the real world, patients may not have a sample of sufficient quality for PD-L1 testing. Finally, there is significant inconsistency in pathologists training in cut point interpretation,” explained Enzinger.
“Treatment recommendations, in my opinion, should be as simple as possible for this disease,” said Ronan J. Kelly, MBBCh, MBA, FASCO, Baylor University Medical Center. Kelly explained that restricting CPI to patients with a CPS score greater than 1 would exclude 11% of patients who are PD–L1-negative, while restricting the CPS score to greater than 10 would exclude 48% of patients.
“To reduce discomfort and improve their calorific intake, we often give [patients] palliative radiation to improve their ability to eat. Many of these patients struggle to even swallow their own saliva. When we offer them radiation, what we do is we upregulate PD-L1… We are changing the PD L1 status of that patient; therefore, spatial and temporal heterogeneity may be even more problematic in this disease where radiation is the norm,” Kelly explained.
On March 22, 2021, the
A total of 548 patients with esophageal cancer and known PD-L1 status were evaluated. The HR for OS was 0.72 (95% CI, 0.59-0.87; P <.0001). The HR was 0.69 (range, 0.56-0.85) in patients with a CPS greater than 1 and 0.57 (range, 0.44-0.75) with a CPS greater than 10.
Pooja Bhagia, MD, executive director of global clinical development, late-stage oncology at Merck, presented efficacy and safety data from the study.1,2 The median OS in the esophageal cancer cohort was 12.6 months (range, 10.2-14.3) in the pembrolizumab arm vs 9.8 months (range, 8.6-11.1) in the placebo arm (HR, 0.72; 0.60-0.88; P =.0006). The OS HR in patients with a CPS less than 1 was 1.00 (range, 0.54-1.85), 0.82 (range, 0.54-1.24) with a CPS between 1 and 5, 1.03 (range, 0.61-1.75) with a CPS between 5 and 10, and 0.57 (range, 0.44-0.75) with a CPS greater than 10.
Regarding safety, “there is no biological rationale to suggest that the safety profile of pembrolizumab would change based on PD-L1 expression,” explained Bhagia.
Bhagia concluded that the magnitude of benefit increased with higher levels of PD-L1 expression, and efficacy trends in the CPS less than 1 subgroup favored the combination over chemotherapy alone.
On May 27, 2022,
Dana Walker, MD, MSCE, vice president and global program lead at Bristol Myers Squibb, presented the data from CheckMate 648. The median OS for nivolumab and chemotherapy was 15.4 months (95% CI, 11.9-19.5) vs 9.1 months (range, 7.7-10.0) with chemotherapy (HR, 0.54; 99.5% CI, 0.37-0.80; P <.0001).
Looking at an analysis of CPS subgroups, in contrast to what was seen in gastric cancer where higher CPS scores derived more benefit from the combination, there was no evidence of increased benefit with a higher CPS score in esophageal cancer. Similar was seen in patients who were treated with the chemotherapy-free option of nivolumab and ipilimumab. Further, 90% of patients had a positive CPS score.
Walker also noted that, similar to KEYNOTE-590, there was no difference in safety profiles based on PD-L1 status.1
On July 18, 2023, the
Mark Lanasa, MD, PhD, senior vice president and chief medical officer of solid tumors at BeiGene, tislelizumab’s sponsor, presented the study’s findings.1,4 The median OS was 17.2 months (95% CI, 15.8-20.1) with tislelizumab vs 10.6 months (range, 9.3-12.1) with placebo (HR, 0.66; 0.54-0.80; 1-sided P <.0001). Lanasa explained that there was no meaningful differentiation in treatment benefit above or below a TAP of 10. Therefore, he proposed that a cut-off value of less than 1 is most appropriate in this population.
Members of the ODAC expressed concern about sample sizes that were discussed in the studies, which presented statistical difficulties decision-making.1 The availability of pooled analyses, which were not previously available at the trial level, helped to make conclusions; however, Committee members expressed discomfort with making a decision based on the current data.
“Because this is such a dynamic field…it is when we do the pooled analyses that we see these higher order effects… I think we, as a field, have to be prepared for the fact that if there are other drugs approved in the same setting for a specific cancer type that do show these higher order effects where a specific biomarker cut-off does not meet the mark, I think we just have to be prepared for that. If we have these high-quality analyses, that may change the way we treat our patients,” said ODAC member Neil Vasan, MD, PhD, Columbia University Medical Center.
“We see in the evidence, of course, is that in the high-expressing PD-L1 tumors we see significant benefit. At [PD-L1 expression] 1 to 10, there is some modest activity but plausible activity. Less than 1, we have the same story of lack of overall survival benefit,” said Christopher Lieu, MD, University of Colorado and acting ODAC chairperson.
Lieu supported the idea of standardizing PD-L1 testing to make it simpler for practitioners as well as ensure “the right testing gets done on the right patients.”
For the voting question, 1 member voted yes and 11 voted no. One member, Ravi Madan, MD, National Cancer Institute, abstained from voting.
“I am not seeing clear benefit, but I am not seeing robust enough numbers that I feel I am comfortable making a determination,” said Madan.
“My vote was no. My honest answer to that question—is the risk-benefit assessment favorable? No. Despite the small numbers, I think this is the best data set we are going to get,” said ODAC member Heidi McKean, Avera Cancer Institute, Sioux Falls, South Dakota.
