Abstract 6904: Model-derived progression-free survival (mPFS): A better metric for patient-derived xenograft studies

Abstract Patient-derived xenografts (PDX) are a powerful assay system for translational prediction of therapeutic potential for experimental therapeutics. The primary readout from PDX models is tumor shrinkage, measured by the ratio of treated/control tumor volume. This readout corresponds loosely to the Objective Response Rate (ORR) metric used in clinical trials, in that it is a readout of drug effect on the sensitive subclones within a tumor. In prior work, we have shown that there is a paradoxical relationship between short-term tumor response and progression-free survival (PFS). While ORR is the primary measure utilized in the Phase II to III transition in clinical development, approval hinges on survival metrics, and the disconnect between ORR and PFS creates a potential risk of late-stage failures for experimental anticancer therapeutics. Thus, it is of interest to understand whether there are alternative approaches to analyzing PDX data that can enable the assessment of therapeutic potential in terms of a projected PFS. In this simulation study, we will examine the accuracy of progression-free survival (PFS) estimates obtained directly from observation during a PDX study and compare their accuracy with the accuracy of PFS estimates derived using a mathematical model-based approach that relies on a kinetic model of tumor subclone growth (model-based PFS, or mPFS) fitted directly to PDX tumor growth and rebound kinetics under treatment. We generated our test dataset using a mathematical model of tumor kinetic growth to simulate xenograft data using fitted parameters derived from literature, adding in experimental error consistent with prior experimental measurement-derived variability. We used the simulated raw data to calculate an observed PFS as well as mPFS derived from fitting the tumor clonal kinetic model to the raw data in a blinded fashion. Our results show that the empirical PFS from PDX models is subject to overestimation, as a function of the observation time interval. On the other hand, mPFS is significantly less biased and more accurate, but is subject to some logistical constraints on sampling. Our work demonstrates the practical translational utility of a novel analysis method (mPFS) in deriving estimates of progression-free survival from PDX model data. Citation Format: Andrew Chen, Madison Stoddard, Lin Yuan, Debra Van Egeren, Arijit Chakravarty, Dean Bottino. Model-derived progression-free survival (mPFS): A better metric for patient-derived xenograft studies abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts) ; 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84 (6Suppl): Abstract nr 6904.