Abstract 1218P
Background
Durable survival benefits of immune checkpoint inhibitors are often manifested by survival plateaus borne by potential long-term survivors (LTS). This study explored the survival heterogeneity and the underlying fraction of LTS among previously treated aESCC patients receiving NIVO in the phase III ATTRACTION-3 study.
Methods
Mixture cure models were applied to model both overall survival (OS) and progression-free survival (PFS) outcomes using patient-level data from the trial with minimum 36-months of follow-up. Patients are classified in two mutually exclusive latent subgroups as LTS and non-LTS. The survival trend of LTS was assumed to follow that of a hypothetical cohort within the general population having the same distribution of age, sex, nationality and smoking status as the trial population. The survival function of LTS was modelled using data obtained from the Human Mortality Database. Time-to-event outcomes for non-LTS were modeled by parametric survival functions that were estimated alongside the corresponding proportion of LTS via maximum likelihood methods. Candidate models were evaluated based on statistical goodness-of-fit and visual comparison to Kaplan-Meier curves reported from the trial.
Results
According to best-fitting models, estimated proportion of LTS was 10.4% [95% CI :6.2%-17.1%] from the OS data and 4.4% [95%CI: 2.1%-8.8%] from the PFS data, where 20-year restricted mean OS and PFS times were 31.4 and 12.4 months, respectively. For both OS and PFS outcomes, ranges of estimated proportions of LTS were narrow across all candidate models: 8.1%-10.9% from OS analysis, and 2.3%-4.9% from PFS analysis. Restricted mean OS and PFS times measured within the trial follow-up were 12.5 and 4.1 months for the non-LTS, and 13.2 and 5.0 months for the control arm of the trial, respectively.
Conclusions
A modest proportion of patients treated with NIVO in the ATTRACTION-3 study can be identified as having no risk of disease-related mortality or progression. Comparable survival outcomes between the non-LTS and the control arm of the study can explain the significant survival benefit of NIVO over taxane-based chemotherapy with the underlying presence of LTS.
Clinical trial identification
ONO-4538-24/BMS CA209473, NCT02569242.
Editorial acknowledgement
Legal entity responsible for the study
Authors on behalf of Bristol-Myers-Squibb.
Funding
Bristol-Myers-Squibb.
Disclosure
J.A. Ajani: Financial Interests, Personal, Advisory Role: BMS, Merck, Taiho, AZ, Daiichi, Novartis, Amgen, Gilead, Astellas, Zymeworks, Beigene, Innovent, OncLive, Geneos, Arcus, Servier; Financial Interests, Personal, Research Grant: Leap, BMS, Astellas, Amgen, Taiho, DeltaFly, Zymeworks, Prolinx, Gilead, Daiichi, LaNova, Macrogenics; Financial Interests, Institutional, Affiliate: ASCO. D.J. Sharpe: Financial Interests, Institutional, Full or part-time Employment: Parexel International; Financial Interests, Institutional, Other, Received research grants/funds from BMS to conduct this research: Bristol-Myers Squibb. T. De: Financial Interests, Institutional, Full or part-time Employment: Parexel; Financial Interests, Institutional, Other, Received research grants/funds from BMS to conduct this research: Bristol-Myers Squibb. I. Kim: Financial Interests, Personal, Full or part-time Employment: Bristol Myers Squibb (BMS); Financial Interests, Personal, Stocks/Shares: Bristol-Myers Squibb (BMS); Non-Financial Interests, Personal, Proprietary Information: Bristol-Myers Squibb (BMS). J.A. Gricar: Financial Interests, Personal, Full or part-time Employment: bristol-myers squibb; Financial Interests, Personal, Stocks/Shares: bristol-myers squibb. M. Kurt: Financial Interests, Personal, Full or part-time Employment, Employee of Bristol-Myers Squibb since June-2018: Bristol-Myers Squibb; Financial Interests, Personal, Stocks/Shares, Owns restricted shares of Bristol Myers Squibb since 2019: Bristol Myers Squibb; Non-Financial Interests, Project Lead: Bristol-Myers Squibb.