Abstract 1072P
Background
Several studies showed the inaccuracy of classical surrogate endpoints, such as progression-free survival (PFS) and objective response rate (ORR), in predicting overall survival (OS) in trials of ICI. Patients that respond to immunotherapy have longer median duration of response (mDR) than those that achieve an ORR with standard CT. We aimed to explore the validity of new surrogate endpoints that take into account ORR and mDR in the context of combination ICI trials.
Methods
Systematic review of randomized controlled trials (RCT) investigating anti-PD1/PD-L1 drugs plus CT versus CT alone (published until May 2020). We performed both (i) arm-level analysis to evaluate median overall survival (mOS) predictors; and (ii) comparison-level analysis for OS hazard ratio (HR). Linear regression models weighted by trial size were fitted and adjusted R 2 was used to quantify OS prediction.
Results
A total of 11 RCT involving 6,675 patients met the inclusion criteria (5 non-small cell lung, 2 small-cell lung, and one bladder, breast, gastric and head & neck cancer each). All RCT were conducted in first-line setting, 55% with anti-PD-L1 and 45% with anti-PD1 combinations. mOS ranged from 10 to 25 months across trials and HR for OS ranged from 0.49 to 0.85. In the arm-level analysis, the best mOS prediction was obtained with a new endpoint that combines ORR and mDR (mDORR = ORR*mDR), with an adjusted R 2 = 0.71 (Table). In the comparison-level analysis, the best predictor for OS HR was again mDORR ratio (ORR odds ratio * mDR ratio) with an adjusted R 2 = 0.55. The classical PFS HR showed a weaker association with OS HR (adjusted R 2 = 0.38). Table: 1072P
| Median OS (arm-level analysis) | Adjusted R2 |
| Surrogates | |
| ORR * mDR | 0.71 |
| 1-year PFS | 0.66 |
| mDR | 0.58 |
| ORR | 0.19 |
| OS HR (comparison-level analysis) | |
| ORR odds-ratio * mDR ratio | 0.55 |
| ORR odds-ratio | 0.54 |
| PFS HR | 0.38 |
| mDR ratio | 0.01 |
Conclusions
The new surrogate endpoint ORR*mDR is a promising predictor of mOS in trials of anti-PD1/PD-L1 plus CT. There is moderate association with OS HR, but the new surrogate is still more accurate than PFS for this purpose. mDORR may be used for estimating the potential efficacy of ICI combinations in early trials and making a decision to proceed to larger RCT.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
G. Villacampa Javierre: Advisory/Consultancy: AstraZeneca; Speaker Bureau/Expert testimony: Merck Sharp & Dohme. R. Dienstmann: Advisory/Consultancy, Speaker Bureau/Expert testimony: Roche; Speaker Bureau/Expert testimony, Research grant/Funding (self): Merck; Speaker Bureau/Expert testimony: Ipsen; Speaker Bureau/Expert testimony: Amgen; Speaker Bureau/Expert testimony: Sanofi; Speaker Bureau/Expert testimony: Servier Laboratories; Advisory/Consultancy: Boehringer Ingelheim; Research grant/Funding (self): Pierre Fabre. All other authors have declared no conflicts of interest.