PCN7 Analysis of Individual-Level Association of Progression-Free Survival (PFS) and Time to Treatment Failure (TTF) with Overall Survival (OS) for Previously Untreated Advanced Renal Cell Carcinoma Patients in the CheckMate 214 Trial


      To determine the association between overall survival (OS) and two other endpoints, progression-free survival (PFS) and time to treatment failure (TTF), for patients with previously untreated advanced renal cell carcinoma. We used individual patient data from the intent-to-treat population (n=1096) in the CheckMate 214 study.


      Exploratory landmark analyses assessed whether early progression status and earlier switch to next line of therapy by clinically meaningful timepoints predict subsequent survival. In parallel, correlation analyses using copula (Clayton, Plackett, and Hougaard) functions and Poisson modeling explored individual-level associations using the 42-month follow-up data, where the strength of association was measured by Kendall’s τ. For both landmark and correlation analyses, experiments examined the sensitivity of the results using data from the intermediate/poor-risk patients only and 30-month follow-up.


      In landmark analyses, patients who progressed by 3, 6, and 12 months were at higher risk of death versus those who did not progress by corresponding landmark timepoints, with hazard ratios (HR) adjusted for age, gender, Karnofsky performance status, and IMDC risk score being 3.18 (95% CI: 2.57-3.93), 3.70 (95% CI: 3.01-4.53), and 3.71 (95% CI: 2.88-4.78), respectively. Similarly, with respect to TTF status by 3, 6, and 12 months, HRs were 3.39 (95% CI: 2.55-4.51), 2.93 (95% CI: 2.39-3.60), and 3.12 (95% CI: 2.52-3.87), respectively. TTF yielded slightly higher Kendall’s than PFS (0.50 vs 0.48) in nearly all copula and Poisson models. These findings were similar for the intent-to-treat population at the 30-month follow-up and for the intermediate/poor-risk patients.


      Patients in the CheckMate 214 trial who progressed or changed medication early on were at substantially higher risk of death than those who had not. Both intermediate endpoints showed moderate correlation to OS. Future research can analyze the association between the two pairs of endpoints with alternative measures and/or longer-term follow-up data.