S. Cui1, J. Zeng2, D. S. Hippe3, J. Fu1, F. Yaseen4, Y. He1, J. Kang5, C. Grassberger1, R. Rengan2, and S. R. Bowen1,3; 1Department of Radiation Oncology, University of Washington/ Fred Hutchinson Cancer Center, Seattle, WA, 2Department of Radiation Oncology, University of Washington - Fred Hutchinson Cancer Center, Seattle, WA, 3University of Washington, Department of Radiology, Seattle, WA, 4Department of Bioinformatics and Medical Education, University of Washington, Seattle, WA, 5University of Washington School of Medicine, Fred Hutch Cancer Center, Department of Radiation Oncology, Seattle, WA
Purpose/Objective(s): Personalized treatment concepts include functional lung avoidance and response-adaptive escalation radiotherapy (FLARE-RT). We report mature outcomes and patterns of failure of the FLARE-RT phase II trial, which evaluated multimodal FDG-PET/CT and SPECT/CT imaging for biologically adaptive therapy in patients with unresectable non-small cell lung cancer (NSCLC). Materials/
Methods: This study included 49 patients with AJCCv7 stage IIB-IIIB NSCLC who enrolled on FLARE-RT (NCT02773238) and initiated chemoradiation to 60Gy in 30fx. Select patients prospectively classified as week 3 (24 Gy nominal) mid-RT non-responders on FDG-PET received a concomitant integral radiation boost of 14Gy in 15fx to residual areas of metabolically active disease. Boosts were spatially conformed to and non-uniformly scaled by mid-RT PET uptake distributions, resulting in peak tumor doses of >90Gy. Overall survival (OS, primary endpoint) and progression-free survival (PFS) were estimated by the Kaplan-Meier method. Locoregional progression (LRP) and distant metastasis (DM) cumulative incidence rates were estimated by the Aalen-Johansen method, with death and the other mode of progression as competing risks. Significant predictors for each treatment failure mode were identified through univariable Fine-Gray regression. Results: Over a median follow-up of 52.3 months, 1-year and 2-year OS were 81.6% (95%CI: 71.5% - 93.2%) and 54.2% (CI: 41.8% -70.4%), respectively; 1-year and 2-year PFS were 53.1% (CI: 40.9% - 69.0%) and 40.5% (CI: 28.8% - 57.0%), respectively. The cumulative incidence rates for 1-year LRP and DM were 12.2% (CI: 4.9%-23.2%) and 34.7% (CI: 21.7% - 48.1%), respectively; 2-year LRP and DM were 18.5% (7.4% - 29.6%) and 43.1% (CI: 28.9%-57.2%), respectively. Higher pre-RT PET total lesion glycolysis (TLG) correlated with an increased risk of LRP (HR per 1-SD increase 1.87 [CI: 1.41-2.49], p=0.01), whereas higher mid-RT PET metabolic tumor volume (MTV) was potentially linked to increased risk of DM (HR per 1-SD increase 1.60 [CI: 0.97-2.62], p=0.06). 25 patients treated with durvalumab exhibited a lower risk of DM (1-year DM 20.0% vs. 54.2%, p=0.04) with equivocal LRP rates (1-year LRP 20.0% vs. 8.3%, p=0.26). Conclusion: Biological image-guided FLARE-RT achieved high locoregional control with consistent survival outcomes relative to historical controls (RTOG0617, RTOG1106, PACIFIC). Patients with baseline bulky metabolic disease burden at risk of local failure may benefit from upfront and adaptive dose escalation, while those with residual disease at risk of distant failure may require intensified systemic therapy. Our experience can inform next-generation biomarker-guided and risk-adaptive therapy trial designs for unresectable NSCLC.
