ESTRO 2024 - Abstract Book

S1669

Clinical - Lung

ESTRO 2024

aims to develop and evaluate a framework for personalized adaptive radiation treatment (PART) to reduce the risk of cardiac adverse events (AEs) among locally advanced NSCLC patients using a Bayesian continuous learning risk prediction model and a Bayesian adaptive monitoring procedure.

Material/Methods:

As shown in Figure 1, there are four major components in this trial: 1) multidimensional personalized normal complication tissue probability (NTCP) model to guide heart dose constraints; 2) sequential learning from the previous cohorts of treated patients to refine the model; 3) continuous adaptation of the target risk level for further improvement, and 4) go/no-go monitoring of the realized cardiac risk resulting from the PART approach. There are two co-primary endpoints: elevation of high-sensitivity cardiac Troponin T (hs-cTnT) during radiation [5] and grade 2 or higher cardiovascular AEs within 12 months post-completion of chemoradiotherapy. Building upon the historical data from the protons vs. photons randomized trial for locally advanced NSCLC,[6] we have developed a Bayesian personalized NTCP model for cardiac risk prediction using clinical, biological, and dosimetry factors to guide personalized dosimetric constraints. Through the trial, the effectiveness of the PART plan in reducing cardiac toxicity is continuously monitored based on an optimal Bayesian design.

Results: