What are the questions being discussed?

On September 26th, 2024, the U.S. Food and Drug Administration (FDA) convened the Oncologic Drugs Advisory Committee (ODAC). The committee was asked to address several questions regarding use of certain PD-1 inhibitors (i.e., nivolumab, pembrolizumab, or tislelizumab) for the treatment of patients with HER2-negative microsatellite stable gastric/ gastroesophageal junction (GEJ) adenocarcinoma (morning session) and metastatic or unresectable esophageal squamous cell carcinoma (ESCC) (afternoon session).

Specifically, the FDA asked the committee to consider whether PD-L1 is a predictive biomarker in these indications and whether cumulative data support a favorable risk benefit assessment for use of these therapies in patients with PD-L1 expression less than 1.

What are the implications for drug development and patients?

PD-1 inhibitors such as nivolumab, pembrolizumab, and tislelizumab, have been transformative for the treatment of cancer, resulting in improved survival outcomes in many populations, including patients with gastric/ GEJ adenocarcinoma and ESCC. Current labeling for use of these therapies in these settings do not specify the level of PD-L1 expression to select patients most likely to respond to treatment, rather the indications take an “all comer” approach based on the overall populations evaluated in clinical trials. However, cumulative clinical trial data indicate that only a subset of patients with high PD-L1 expression are likely to benefit from these treatments. As a result, certain patients may receive these therapies, which can be expensive and toxic, with low likelihood of significant benefit. Additionally, variable approaches to determining PD-L1 expression and the cutoffs used to define subgroups in primary endpoint analyses in clinical trials complicate efforts to harmonize labeling and define the group of patients most likely to respond to treatment with these therapies. 

What was discussed?

Committee members considered data presentations from multiple clinical trials, FDA’s pooled analysis of data from these trials, and patient and advocate perspectives shared during the open public session of the meeting. After hearing all stakeholder’s perspectives during each session of the meeting, the committee weighed in on the discussion and voting questions posed by the FDA.

Discussion during the morning session focused on the limitations of current approaches to determining PD-L1 expression, both in clinical trials and clinical practice, that can lead to high inter-reporter variability and lack of harmonization in determination of PD-L1 status for patients with gastric/GEJ cancers. Committee members emphasized the need for efforts to harmonize approaches to assess PD-L1 expression and other continuous biomarkers used to define efficacy populations in clinical trials and inform treatment decisions in clinical practice. Despite acknowledgment of the limitations of available trial data, the committee agreed the data presented demonstrated a relative lack of benefit for patients with PD-L1 expression less than one compared to patients with PD-L1 expression ≥1.

ODAC members also noted harms of using these treatments when potential benefit is less certain, such as treatment-related and financial toxicities. Members also acknowledged there have been some success stories for patients with PD-L1 < 1 treated with these therapies, such as those shared during the open public session. Along those lines, the committee agreed on the importance of allowing treatment decisions to remain between patients and their doctors and supported maintaining access to effective treatments. ODAC members called for better alignment and additional data on approaches and cutoffs for determining PD-L1 expression, noting the importance of consistent and harmonized data and guidelines to inform care decisions. At the end of the morning session, Dr. Richard Pazdur, Director of the Oncology Center of Excellence (OCE), asked the participating sponsors to commit to coordinating on uniform biomarker development to ensure future trials are appropriately designed to resolve these challenges.

During the afternoon session, there was similar consensus that there is a clear lack of benefit in the subgroup of patients with ESCC who had PD-L1 expression less than one. In addition, ODAC members were aligned on their concerns around the small sample size of patients included in the data presented and the challenges in determining the efficacy of the treatment based on such limited data. They noted that future trials should be carefully powered to allow for a closer understanding of the subset of patients, including which biomarker negative patients, could potentially derive benefit from these treatments.

For both sessions, the committee generally agreed that cumulative data indicate PD-L1 appears to be a predictive biomarker, with the clearest benefit observed in patients with PD-L1 expression greater than or equal to 10 and modest benefits between 1 and 10. In addition, for the voting questions during both sessions, the majority of ODAC members voted that risk benefit assessment is not favorable for the use of PD-1 inhibitors in these indications in patients with PD-L1 expression less than 1.

What are the outcomes and implications of the meeting?

Though the committee was not asked to determine whether labeling should be refined to specify a specific PD-L1 cut-off for selecting patients for treatment with these therapies, the votes cast during the hearing are likely to influence regulatory decisions regarding labeling for anti-PD-1 antibodies as well as future trial designs and approaches to measuring PD-L1 expression.

Variability in PD-L1 assessment can lead to patients being wrongly categorized as eligible or ineligible for treatment, potentially depriving some patients of effective treatment options or exposing them to ineffective therapies. Differences between assays and antibodies used for PD-L1 testing and variation in PD-L1 scoring systems (e.g., Tumor Proportion Score [TPS] vs. Combined Positive Score [CPS]), can result in inconsistent results across labs and/or pathologists and create confusion in clinical decision-making. In addition, insurance coverage and patient access to treatments may rely on PD-L1 testing and variability in results could lead to discrepancies in treatment access due to differing interpretations of clinical trial data informing clinical guidelines that frequently determine coverage. The ODAC hearing highlighted how lack of alignment on approaches to defining biomarker-based subgroups can complicate efficacy assessments, regulatory decision making, and clinical care, and provides insights to inform an effective path forward. These discussions support the growing body of evidence that improved assay alignment is critical for ensuring patients receive the best treatments.