Discussions Swirl on the Role of PD-1 Inhibitors for Gastric/GEJ/ESCC Cancers With 1%-10% PD-L1 Expression

Despite the clear consensus that was reached regarding the unfavorable risk-benefit profile of PD-1 inhibitors in patients with certain PD-L1–negative gastrointestinal cancers in a recent Oncologic Drugs Advisory Committee (ODAC) meeting, the jury is still out regarding the appropriate use of a cutoff, according to Nataliya Uboha, MD, PhD.¹ She added that questions must be addressed before a definitive answer can be given regarding the efficacy of these agents in patients with PD-L1 expression levels ranging from 1% to 10%.

The final votes from two September 26, 2024, FDA ODAC sessions determined that the risk-benefit profile of PD-1 inhibitors was not favorable for the frontline treatment of patients with PD-L1 expression less than 1% who have advanced HER2-negative microsatellite stable gastric/gastroesophageal junction (GEJ) adenocarcinoma or unresectable or metastatic esophageal squamous cell carcinoma (ESCC) (Figure).^1-3

“In my practice, I offer immunotherapy to those with a PD-L1 combined positive score [CPS] of 1% or greater,” Uboha, an associate professor and researcher at the University of Wisconsin School of Medicine and Public Health in Madison, said in an interview with OncologyLive. “I have become less dogmatic about this—when the initial data with nivolumab [Opdivo] came out, my cutoff was 5%. However, I am seeing variability between PD-L1 expression between the primary [and] metastatic sites—[recently], I saw a PD-L1 score of 2.3%, so even with a decimal point. These patients have very serious illnesses, and they need better treatments. In the era of uncertainty, the benefits of the addition of immunotherapy outweigh the risks, especially when the testing for the biomarker is not very reliable.”

Determining whether the use of PD-1 inhibitors in patients with PD-L1 expression ranging from 1% to 10% is optimal remains especially critical as approval is pending for tislelizumab-jsgr (Tevimbra) in combination with platinum-containing chemotherapy in ESCC.² The biologics license application (BLA) was submitted to the FDA on July 18, 2023, and cited data from the phase 3 RATIONALE-306 trial (NCT03783442) showing that the tislelizumab regimen produced a significant improvement in overall survival (OS) compared with placebo and chemotherapy in the overall patient population (n = 649) of the trial.

However, 3-year survival data presented at the 2024 American Society of Clinical Oncology (ASCO) Annual Meeting showed that the survival benefit was greater in patients with a PD-L1 tumor area positivity (TAP) score of 10% or greater vs in those with a TAP score less than 10%.⁴ Among patients with a PD-L1 TAP score of 10% or greater, the median OS was 16.6 months (95% CI, 15.3-23.4) in the tislelizumab plus chemotherapy arm (n = 116) vs 10.0 months (95% CI, 8.6-13.3) in the placebo plus chemotherapy arm (n = 107; HR, 0.70; 95% CI, 0.52-0.95). Among those with a PD-L1 TAP score less than 10%, the median OS was 16.0 months (95% CI, 12.3-19.6) in the tislelizumab arm (n = 151) vs 10.4 months (95% CI, 9.0-13.4) in the placebo arm (n = 168; HR, 0.77; 95% CI, 0.60-0.99).

“The [ODAC] discussion was whether the committee should consider limiting the approval in biomarker-selected patient populations for both nivolumab and pembrolizumab [Keytruda]. Tislelizumab [plus chemotherapy in the frontline setting] is not approved yet, but we’re seeing similar efficacy of these agents across different studies in this patient population,” Uboha said. “I would anticipate that the committee will likely make a decision to approve tislelizumab in similar settings, since we’ve seen similar efficacy, but I also anticipate that there will be a limitation to indications of both nivolumab and pembrolizumab use based on lack of benefit in those with PD-L1–negative tumors.”

Tislelizumab is also pending approval in the locally advanced unresectable or metastatic gastric or GEJ adenocarcinoma setting; the FDA accepted the BLA for tislelizumab plus fluoropyrimidine- and platinum-containing chemotherapy in February 2024 with a target action date expected in December 2024.⁵

Testing for PD-L1: How Accurate Are Assessments and Should Approvals Be Restricted?

Uboha added that the original frontline approval of pembrolizumab plus trastuzumab (Herceptin), fluoropyrimidine, and platinum-containing chemotherapy was for all patients with locally advanced, unresectable or metastatic HER2-positive gastric or GEJ adenocarcinoma. However, the accelerated approval was restricted in November 2023 to include only patients whose tumors have a PD-L1 CPS of 1% or greater, as determined by an FDA-approved test.⁶

“The [ODAC meeting] discussion about patients whose tumors have low PD-L1 expression, in the range of 1% to 10%, was much more nuanced. Primarily, the discussion was geared to how difficult it is to reliably identify the exact score,” Uboha said. “There’s a lot of heterogeneity in expression [and] the pathologists have commented on the heterogeneity [of] expression between the primary [and] metastatic sites. We don’t know what the different treatments, such as radiation, can change [regarding] expression of PD-L1 in these tumors; is expression of PD-L1 dynamic? Looking at patients with a low PD-L1 score and how they respond to immunotherapy treatment was not a primary end point in any of the studies that were discussed.”

Robert Albert Anders, MD, PhD, an associate professor of pathology at the Johns Hopkins University School of Medicine and a pathologist at Johns Hopkins Kimmel Cancer Center, presented the clinical perspective the on PD-L1 testing in clinical practice at the ODAC meeting. In an interview with OncologyLive, he noted that “There are a lot of very detailed portions of PD-L1 testing, [and] the thresholds are presented as all or none. It’s either above a percentage or below a percentage, but there are a lot of caveats that go into PD-L1 testing: The type of specimen, the stain that is used, and how the test is interpreted. I find these things are often overlooked in the discussion of clinical biomarkers.”

Anders added that PD-L1 should be an additional piece of evidence and not the only piece of evidence that determines whether a patient should receive treatment with PD-1 inhibitors. Uboha also noted that the simplification of testing, recommendations, and guidelines is necessary because discordance exists between the prescribing information of a drug, FDA approval, and ASCO as well as NCCN guidelines.

“We need to make it simpler. We need to have or at least attempt to make similar or uniform cutoffs for PD-L1 expression, so that these agents can be used interchangeably, and so that it’s easier for our pathologists and community partners to adopt them in clinical practice,” Uboha said.

“The threshold for positivity is an ongoing problem where we have different PD-L1 tests, and based upon which test is used, the cutoffs are different. Even based upon the tumor type, the cutoffs can be different,” Anders said. “[Therefore], the pathologist who’s scoring PD-L1 is left with almost an impossible task because the way the test is interpreted depends on which test was run and which tumor type it is. That’s not an impossible hurdle, but it’s something we’ve never faced before. The other issue about threshold for positivity is that it’s difficult to reproducibly determine PD-L1 positivity.”

Breaking Down ODAC’s Discussion: Not Enough Benefit for a PD-L1 Score Less Than 1%

Although the benefit of PD-1 agents in patients with PD-L1 scores ranging from 1% to 10% requires further elucidation, according to Uboha, the September 26, 2024, ODAC meeting discussion focused on patients with expression levels less than 1%.^2,3

The FDA noted concerns regarding the lack of benefit observed with PD-1 inhibitors for patients with ESCC who had lower or negative PD-L1 scores (Table 1).² They also detailed concerns for patients with gastric/GEJ adenocarcinoma because of the modest efficacy that has been observed with the agents across patient populations defined by different PD-L1 thresholds (Table 2).³ In both populations, the regulatory agency highlighted the potential for increased toxicities resulting from the addition of these inhibitors.^2,3

“These agents work in many of our patients, but I do believe we need to be more nuanced. We should offer these agents to patients where they are likely to work because they do have toxicities [and] patients can develop immune-mediated toxicities that could negatively affect their quality of life,” Uboha said.

Gastric/GEJ Adenocarcinoma Data

Data from the phase 3 CheckMate 649 (NCT02872116), KEYNOTE-859 (NCT03675737), and RATIONALE-305 (NCT03777657) trials presented in the gastric/GEJ adenocarcinoma space all showed statistically significant primary OS results for patients who were randomly assigned to the PD-L1 agent-containing arm vs the control arm. All patients experienced an OS benefit when treated with a nivolumab plus chemotherapy regimen vs chemotherapy (HR, 0.79; 95% CI, 0.70-0.89), a pembrolizumab plus chemotherapy regimen vs chemotherapy (HR, 0.77; 95% CI, 0.69-0.86), and a tislelizumab plus chemotherapy regimen vs chemotherapy (HR, 0.80; 95% CI, 0.69-0.92).³

However, exploratory results the FDA highlighted showed that the benefits were greatest in patients with a PD-L1 score of 10% or higher treated with the nivolumab regimen vs chemotherapy (HR, 0.65; 95% CI, 0.55-0.78), pembrolizumab regimen vs chemotherapy (HR, 0.64; 95% CI, 0.52-0.77), and tislelizumab regimen vs chemotherapy (HR, 0.57; 95% CI, 0.43-0.76). Respective HRs for patients with PD-L1 expression less than 1% were 0.92 (95% CI, 0.70-1.23), 0.92 (95% CI, 0.73-1.17), and 0.98 (95% CI, 0.64-1.50). Patients with PD-L1 expression less than 10% experienced respective HRs of 0.94 (95% CI, 0.80-1.1), 0.86 (95% CI, 0.75-0.98), and 0.91 (95% CI, 0.77-1.07).

“Everybody agreed that PD-L1 is a predictive biomarker of response,” Uboha said. “Everybody also agreed that it has been a challenging biomarker to use [and] adopt. Across all members, there was strong agreement that patients with a higher PD-L1 score, particularly a PD-L1 CPS score of 10% or greater, derive benefit from the addition of immunotherapy to their treatment. There was also general agreement that those who have no expression of PD-1 in their tumors do not derive much benefit.”

“The problem is the biomarker isn’t particularly perfect,” Anders said. “That makes determining whether an individual will respond [difficult].”

Findings in ESCC

In ESCC, findings from the phase 3 KEYNOTE-590 (NCT03189719), CheckMate 648 (NCT03143153), and RATIONALE-306 trials were discussed. The FDA noted that primary OS data were statistically significant in favor of the anti–PD-L1-containing arm vs chemotherapy alone arm in all the trials, but the point estimates for the treatment effect did not appear to be favorable in patients with PD-L1 expression less than 1% and appeared intermediate in patients with a PD-L1 expression of less than 10%.²

“One distinction that’s important to make is that most ESCCs will express PD-L1, and so the use of anti–PD-1 agents will probably be broader in patients with ESCC,” Uboha noted.

All patients with ESCC experienced an OS benefit when treated with nivolumab plus chemotherapy vs chemotherapy (HR, 0.73; 95% CI, 0.60-0.88), pembrolizumab plus chemotherapy vs chemotherapy (HR, 0.72; 95% CI, 0.59-0.87), and tislelizumab plus chemotherapy vs chemotherapy (HR, 0.68; 95% CI, 0.56-0.82).

FDA OS analyses showed that patients with a PD-L1 expression of 10% or higher experienced respective HRs of 0.62 (95% CI, 0.46-0.84), 0.57 (95% CI, 0.44-0.75), and 0.66 (95% CI, 0.48-0.92) with nivolumab, pembrolizumab, and tislelizumab, respectively. Respective HRs for those with PD-L1 expression less than 1% were 0.93 (95% CI, 0.46-1.91), 1.00 (95% CI, 0.54-1.85), and 1.34 (95% CI, 0.73-2.46). Furthermore, patients with PD-L1 expression of less than 10% experienced respective HRs of 0.77 (95% CI, 0.60-1.01), 0.95 (95% CI, 0.71-1.26), and 0.76 (95% CI, 0.58-0.99).²

Future Directions for Care Amid Paradigm Considerations

Another test will soon be incorporated into practice following the October 18, 2024, approval of zolbetuximab-clzb (Vyloy) plus fluoropyrimidine- and platinum-containing chemotherapy for the frontline treatment of adults with locally advanced unresectable or metastatic HER2-negative gastric or GEJ adenocarcinoma whose tumors are CLDN18.2 positive, as determined by an FDA-approved test.⁷

“I am planning to incorporate reflexive CLDN18.2 testing into our treatment paradigms as well, but this will be applicable only to adenocarcinomas in the advanced setting,” Uboha said.

As targeted agents such as zolbetuximab join the treatment landscape, and the approval of tislelizumab plus chemotherapy for ESCC and gastric cancer indications is pending, clarification has come regarding the optimal approach for patients with a PD-L1 expression of less than 1%.

“The ODAC committee voted to recommend restricting the use of these [anti–PD-1 antibody] agents to patients whose tumors have PD-L1 expression, and I’m hoping that the FDA will further refine the label going forward,” Uboha said. “I’m [also] hoping for more of a unified label across different immunotherapy agents. It will not only make it easier for us to use these agents in clinical practice, but it will potentially simplify future trial designs [in] knowing how to use the appropriate control arm.”

“PD-L1 is a reasonably good biomarker to enrich for patients more likely to respond in a population,” Anders added. “However, the power of its predictive value falls off significantly, almost to being not useful, when applied to an individual patient. Part of that difficulty in being a useful biomarker stems from the inability to uniformly and reproducibly score PD-L1 expression on a given patient.”

Editor’s Note: This interview occurred prior to the approval of zolbetuximab.

References

1. Meeting of the Oncologic Drugs Advisory Committee. FDA. September 26, 2024. Accessed October 21, 2024. https://www.youtube.com/live/ELA3JDqtcFw
2. Immune checkpoint inhibitors in patients with metastatic or unresectable esophageal squamous cell carcinoma. Oncologic Drugs Advisory Committee Meeting. FDA. September 26, 2024. Accessed October 21, 2024. bit.ly/3UjNvS1
3. Immune checkpoint inhibitors in patients with metastatic or unresectable HER2-negative gastric adenocarcinoma. Oncologic Drugs Advisory Committee Meeting. FDA. September 26, 2024. Accessed October 21, 2024. bit.ly/4ferVGm
4. Yoon HH, Kato K, Raymond E, et al. Global, randomized, phase III study of tislelizumab plus chemotherapy versus placebo plus chemotherapy as first-line treatment for advanced/metastatic esophageal squamous cell carcinoma (RATIONALE-306 update): minimum 3-year survival follow-up.J Clin Oncol. 2024;42(suppl 16):4032. doi:10.1200/JCO.2024.42.16_suppl.4032
5. BeiGene’s biologics license application for Tevimbra (tislelizumab) for first-line gastric or gastroesophageal junction cancers accepted by FDA. News release. BeiGene Ltd. February 27, 2024. Accessed October 21, 2024. bit.ly/3YfqoJx
6. FDA amends pembrolizumab’s gastric cancer indication. FDA. Updated November 8, 2023. Accessed October 21, 2024. bit.ly/3YwRX27
7. FDA approves zolbetuximab-clzb with chemotherapy for gastric or gastroesophageal junction adenocarcinoma. FDA. October 18, 2024. Accessed October 21, 2024. bit.ly/3YvZuOE