351 - Multi-Institutional Comparison of Ablative 5-Fraction MR-Guided Online Adapted vs. 15/25-Fraction CT-Guided Offline Adapted Radiation Therapy for Locally Advanced Pancreas Cancer

M. D. Chuong¹, E. OReilly², R. Herrera³, M. Zinovoy⁴, K. E. Mittauer⁵, M. Rubens⁶, P. B. Romesser⁴, A. J. Wu⁴, N. Bassiri-Gharb⁵, C. Hajj⁴, J. J. Cuaron², A. Ucar⁷, F. de Zarraga⁷, S. Aparo⁷, W. Lu², A. Gutierrez⁵, C. H. Crane⁴, and M. Reyngold⁴; ¹Miami Cancer Institute, Miami, FL, ²Memorial Sloan Kettering Cancer Center, New York, NY, ³Florida International University Herbert Wertheim College of Medicine, Miami, FL, ⁴Department of Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, NY, ⁵Department of Radiation Oncology, Miami Cancer Institute, Baptist Health South Florida, Miami, FL, ⁶Department of Biostatistics, Miami Cancer Institute, Baptist Health South Florida, Miami, FL, ⁷Department of Medical Oncology, Miami Cancer Institute, Baptist Health South Florida, Miami, FL

Purpose/Objective(s): Ablative radiation therapy (A-RT) defined as a prescribed dose of ~100 Gy BED₁₀ can achieve durable freedom from local failure (FFLF) and may improve overall survival (OS) versus non-ablative RT for locally advanced pancreas cancer (LAPC). The optimal technique for delivering A-RT is uncertain. Materials/

Methods: We retrospectively compared A-RT outcomes among LAPC patients treated at 2 U.S. institutions with different A-RT techniques excluding those who had surgery at any time. Institution (Inst) A used stereotactic magnetic resonance (MR)-guided online adaptive radiation therapy (SMART) with automatic beam gating on a 0.35T MR-Linac prescribing 5 fx (50 Gy, 95.6%; 45 Gy 4.4%). Inst B used a hypofractionated ablative RT (HART) delivered on a CBCT-guided linac with respiratory gating and selective offline replanning, prescribing 25 fx (75 Gy, 80.8%) or 15 fx (67.5 Gy, 19.2%). Elective coverage including the celiac trunk and SMA was routine at both Inst A and B. Clinical characteristics were compared using Chi square or Fisher’s exact test. FFLF as per RECIST, freedom from distant failure (FFDF), and overall survival (OS) were estimated from RT with the Kaplan-Meier method and log-rank test. Cox proportional hazards regression model was used for univariate (UVA) and multivariate (MVA) analysis.
Results: 211 patients (n=91, SMART; n=120, HART) were evaluated. Similar baseline characteristics (SMART vs. HART) included maximum tumor size (3.7 vs. 3.8 cm), median pre-RT CA19-9 level (55 vs. 70 U/mL), induction FOLFIRINOX or gemcitabine/nab-paclitaxel use (82.4% vs. 85.8%), and median induction chemo duration (3.9 vs. 3.7 months). Significant differences (SMART vs. HART) included median age (71 vs. 68 years; p=0.004), ECOG performance status 0 (41.8% vs. 90.0%; p<.001), head tumor (80.2 vs. 62.5%; p=.005), and N+ (25.3% vs. 42.5%; p<.001). SMART plans were hotter (e.g., GTV V120% 14.5 vs. 0.0 cc; p<.001) with higher target coverage (e.g., median GTV D90 scaled 101.1% vs. 84.5%; p<.001). Median follow-up from A-RT was 12.0 vs. 17.5 months (p<.001). 2-yr FFLF, FFDF, and OS for SMART vs. HART were 85.6% vs. 69.0% (p<.001), 30.3% vs. 26.5% (p=.513), and 31.0% vs. 35.3% (p=.056). On MVA, worse FFLF was associated with 15-25 vs. 5 fractions (HR 4.982, 95% CI 1.105-22.454) and induction non-FOLFIRINOX vs. FOLFIRINOX (HR 2.010 (95% CI 1.023-3.949) while worse OS was associated with higher GTV V120% (HR 1.016, 95% CI 1.001-1.031). Acute grade >3 toxicity was similar (3.3% vs. 5.8%; p=.390) while less late grade >3 toxicity occurred after SMART (2.2% vs. 9.2%; p=0.037).
Conclusion: Favorable 2-yr FFLF and OS were achieved after induction chemo and A-RT for LAPC regardless of delivery technique. Higher FFLF after SMART may be related to higher GTV coverage, albeit this was not significant on MVA. Despite this, OS was similar potentially due to more unfavorable baseline characteristics among the SMART cohort. Lower rates of late grade >3 toxicity may have been facilitated by online adaptive RT.