Click here to


Are you sure ?

Yes, do it No, cancel

Modeling Local Versus Distant Tumor Recurrence in Non-Small Cell Lung Cancer Patients Receiving Combined Chemoradiotherapy and Molecularly Targeted Drugs

D McClatchy*, H Paganetti , H Willers , C Grassberger , Massachusetts General Hospital and Harvard Medical School, Boston, MA


(Wednesday, 7/17/2019) 7:30 AM - 9:30 AM

Room: 225BCD

Purpose: There is a great clinical interest in integrating targeted agents such as tyrosine kinase inhibitors (TKI) with definitive chemoradiotherapy (CRT) for locally-advanced (LA) non-small cell lung cancers (NSCLC) exhibiting targetable mutations in EGFR (epidermal growth factor receptor) or other oncogenes. We developed a novel evolutionary and radiobiological model of local (LR) and distant tumor recurrence (DR) following CRT with TKI induction and maintenance for quantitatively optimizing the administration of targeted agents in this setting.

Methods: The tumor progression model simulates a clinical trial by drawing patients from probability distributions and models Gompertzian tumor volume trajectories using a system of non-linear differential equations capturing fundamental principles of evolutionary drug resistance and radiobiology. LR and DR were predicted to create Kaplan Meier (K-M) curves, which were then calibrated to published single-institutional studies in EGFR-mutant versus wild-type populations (calibration dataset). Model-predicted progression free survival (PFS) was independently evaluated against clinical trial outcomes. Finally, multimodal treatment regimens were explored.

Results: The calibrated model reproduced observed LR and DR dynamics, and independently predicted the results of recent CRT trials in LA-NSCLC (RTOG0617/PROCLAIM, predicted vs. reported median/5-yr. PFS of 12-mo./13% vs. 10.6-mo./14%) and a TKI trial for EGFR-mutant stage IV NSCLC (EURTRAC, predicted vs. reported median/2-yr. PFS of 7-mo./14% vs. 9.7-mo./11%). Further investigations showed that predicted PFS is most sensitive to the growth rate of the tumor (5-yr. ΔPFS -16% for growth above median), more so than pre-existing targeted resistance. Unexpectedly, shorter TKI induction periods predicted significantly higher PFS by targeting early drug resistant disease (5-yr. ΔPFS 6% for 2 wk.-12 wk. induction).

Conclusion: Our tumor progression model accurately captures the LR and DR dynamics in EGFR-mutant and wild-type NSCLC populations treated with sequential TKI-CRT and can be used to optimize and individualize TKI use in future clinical trials of multimodality therapy for oncogene-driven LA-NSCLC.


Modeling, Targeted Radiotherapy, Tumor Control


TH- Dataset analysis/biomathematics: biostatistics and clinical trial design

Contact Email