Room: Karl Dean Ballroom A1
Purpose: Radiation-induced outcome modeling has been studied extensively to predict treatment outcomes. However, to our knowledge, such models have not been incorporated into the treatment planning process. We investigate mathematical modeling of 2-year survival and applying it prospectively in SBRT planning for central and ultra-central lung tumors.
Methods: In a multi-institutional study, we analyzed follow-up data from 58 consecutive central lung SBRT patients (27 females, 31 males, 48-94 years old, median tumor size 2.0 cm, 3-8 fractions with 7.5-20 Gy/fraction). At 36mo median follow-up, 29% patients died within 2 years post-SBRT. Using multivariate logistic regression, we formulated a 2-year survival probability model. Predictor variables included age, sex, ECOG performance status (PS), prior lung cancer, prescribed radiation dose and fractionation (as equivalent 4-fx dose), tumor size, body mass index (BMI), and dosimetric parameters. As proof of concept, we integrated our model along with clinical dose-volume constraints in the IMRT planning objective function for one central lung SBRT case.
Results: Based on our 58-patient cohort, the model indicated that age, tumor size, prescribed dose, ECOG PS, sex, prior lung cancer, BMI, lung V20 and max heart dose were significant predictors of 2-year survival (P<0.05). By incorporating the survival model in IMRT optimization, max dose to heart was reduced by 39% improving 2-year survival by 5.4% at the cost of increased max dose to spinal cord - 29.8Gy, still meeting the clinical constraint of 30Gy.
Conclusion: Current radiotherapy planning uses aggregated, population-based dose constraints and fractionation, which are then personalized on a case-by-case basis according to institutional or individual experience. In contrast, our proof-of-concept study demonstrates an objective, systematic framework of prospectively integrating and personalizing outcome models to maximize 2-year survival. Our proposed framework therefore represents a significant potential shift in the current treatment planning paradigm.
Not Applicable / None Entered.