Editor’s Note: To advance the standard of care for gastrointestinal cancers in China and share the latest developments and insights, the "2024 CSCO Workshop on Standardized Diagnosis and Treatment of Gastrointestinal Cancers," co-hosted by the Chinese Society of Clinical Oncology (CSCO), the Beijing Xisike Clinical Oncology Research Foundation, and the Hangzhou Oriental Clinical Oncology Research Center, took place in Shanghai from August 17–18, 2024.

At the workshop, Dr. Jun Zhou from Shanghai GoBroad Cancer Hospital China Pharmaceutical University delivered a compelling keynote presentation titled “No Perfect Study Design, Only Continuous Refinement: The Evolution of the CheckMate649 Study.” Professor Zhou discussed the rationale behind 29 key protocol revisions during the study’s development, shedding light on how these changes shaped the trial’s outcomes. In an exclusive interview with Oncology Frontier, he shared his insights on how these adjustments can serve as a guide for designing more effective immunotherapy trials in the future.

1. Reflecting on the Lecture: A Closer Look at the CheckMate649 Study

Oncology Frontier: Professor Zhou, your presentation titled “No Perfect Study Design, Only Continuous Refinement” resonated strongly at this year’s conference. Could you summarize the main points of your lecture?

Dr. Jun Zhou: The CheckMate649 study has provided groundbreaking data on the use of first-line immunotherapy combined with chemotherapy for advanced gastric cancer. What many don’t realize, however, is that it took 29 protocol revisions over four years to finalize the study’s design. Among these, four major adjustments stand out:

1. Expanding Study Cohorts: Initially, the study focused on comparing a dual-immunotherapy regimen (nivolumab with ipilimumab) to a chemotherapy regimen (oxaliplatin and fluoropyrimidine). This design aimed to assess the feasibility of chemotherapy-free treatments. However, findings from the CheckMate012 study in lung cancer showed that combining nivolumab with chemotherapy significantly improved overall survival compared to immunotherapy alone. Based on this evidence, a third cohort combining nivolumab with chemotherapy was added to CheckMate649, evolving it into a three-arm trial.
2. Refining Primary Endpoints: The study initially focused solely on overall survival (OS) as its primary endpoint. However, data from the ATTRACTION-4 study demonstrated that combining nivolumab with chemotherapy achieved impressive objective response rates (ORR) of 66.7%–70.6%, alongside significant progression-free survival (PFS) benefits. In response, ORR and PFS were added as co-primary endpoints. However, since the FDA did not permit ORR as a primary endpoint, it was later reassigned as a secondary endpoint.
3. Optimizing Biomarkers: The study initially used tumor proportion score (TPS) as a biomarker to predict treatment response, reflecting its established role in lung cancer immunotherapy. However, subsequent analyses from the ATTRACTION-2 and ATTRACTION-4 studies, as well as data from the KEYNOTE series, revealed that TPS was not a reliable predictor of outcomes in gastric cancer. Instead, the combined positive score (CPS) proved more effective, particularly for patients with CPS ≥10. This led to CPS replacing TPS as the primary biomarker in CheckMate649.
4. Defining Treatment Duration and Follow-Up: The optimal duration of nivolumab treatment had not been clearly established. Data from earlier studies, such as CA209-003 and CheckMate153, suggested that two years of treatment could maximize survival benefits while balancing adverse events. Based on this, the treatment duration in CheckMate649 was capped at two years. Additionally, follow-up intervals were defined at 12, 24, and beyond, informed by survival trends seen in studies like KEYNOTE-062.

At the four-year follow-up, CheckMate649 demonstrated that nivolumab combined with chemotherapy provided durable survival benefits for Chinese patients, with a median OS (mOS) of 14.3 months. In patients with CPS ≥5, mOS reached 15.5 months, with an ORR of 68% and a median duration of response (DOR) of 12.5 months. For patients achieving complete or partial remission (CR/PR), mOS extended to 21.5 months—nearly two years. These results underscore the importance of CPS ≥5 as a key predictor of benefit from this treatment combination.

2. The Impact of Protocol Refinements on Study Outcomes

Oncology Frontier: The numerous adjustments to the protocol clearly contributed to the study’s success. In your opinion, how significant were these refinements, and what lessons can be drawn for future trials?

Dr. Jun Zhou: The CheckMate649 study is a prime example of how dynamic trial design can maximize both scientific rigor and clinical relevance. Its 29 revisions were not arbitrary but grounded in strong evidence from other studies, highlighting the value of continuous learning and adaptation during a trial.

One of the key takeaways is the critical role of identifying the right patient population. This requires a robust foundation in both basic and clinical research, as well as close collaboration between investigators. Accurate targeting ensures that the trial produces reliable, actionable results that can directly benefit patients.

This study also reinforces the need for systematic thinking in clinical research. Investigators should actively monitor advancements in the field, adjust protocols dynamically, and address potential biases promptly. This approach not only ensures scientific validity but also enhances the trial’s ability to reflect real-world outcomes.

3. Recommendations for Future Immunotherapy Research

Oncology Frontier: Based on your experience with CheckMate649, what are your recommendations for designing future immunotherapy trials?

Dr. Jun Zhou:

1. Set Clear Objectives and Adapt Endpoints: Defining the trial’s purpose is fundamental, whether it’s improving OS, PFS, or both. While endpoints should be informed by past research, they should remain flexible enough to evolve during the study.
2. Select Effective Biomarkers: Precise biomarkers are crucial for identifying the optimal beneficiary population. Trials that focus on clearly defined subgroups are more likely to produce reliable and clinically meaningful results. Overly complex designs, even if successful, can face challenges in broader clinical adoption.
3. Foster Collaboration: Regular investigator meetings that bring together clinicians and basic scientists are vital. Collaborative discussions help identify and resolve issues early, ensuring trials are both scientifically robust and clinically applicable.

Ultimately, the success of any trial lies in its ability to strike a balance between scientific innovation and real-world practicality.

About Dr. Jun Zhou:

• Chief Physician and Graduate Supervisor
• Director of Oncology, Shanghai GoBroad Cancer Hospital China Pharmaceutical University
• Recipient of the “People’s Distinguished Physician” Award
• Member and Secretary of the CSCO Gastric Cancer Expert Committee
• Active member of various CSCO committees, including those on biliary tract tumors, antitumor drug safety, radiotherapy and interventional therapy, and smart healthcare