Cumulative Target Volume Dose and Locoregional Failure in Pancreatic Cancer Patients With Treated With Ablative Stereotactic MR-Guided Adaptive Radiation Therapy (SMART)

Abstract

<h3>Purpose/Objective(s)</h3> Emerging data suggest that ablative stereotactic magnetic resonance-guided adaptive radiation therapy (SMART) for inoperable pancreas cancer (PCa) achieves favorable local control (LC) and overall survival (OS) compared to non-ablative dose. In the current SMART workflow, each adaptive fraction is reoptimized without accounting for previously delivered dose. Cumulative target volume (TV) dose from all SMART fractions has not previously been evaluated with respect to clinical outcomes. <h3>Materials/Methods</h3> We evaluated 44 inoperable PCa patients who were prescribed 50 Gy in 5 daily fractions on a 0.35T-MR LINAC. 33 patients (75%) were treated with elective nodal irradiation, typically including the celiac axis and superior mesenteric artery. PTV margin was 3 mm. On-table adaptive replanning was performed to primarily satisfy gastrointestinal (GI) organ at risk (OAR) constraints, and secondarily maximize TV coverage. Cumulative dose for all fractions was summed onto a TRUFI MR fractional scan (i.e., fraction 5). A rigid transformation was performed for fractions 1 to 4 scans onto the fraction 5 frame of reference (FOR) based on alignment of gross disease. Each fractional dose was transformed onto fraction 5 FOR and summed cumulatively for all 5 fractions. Locoregional failure (LRF) was delineated on follow-up imaging by a diagnostic radiologist with PCa expertise. Cumulative TV dose metrics (D80, D90, mean, max) were evaluated with respect to LRF. Diagnostic scans were rigidly registered to the fraction 5 FOR based on recurrence region to investigate recurrence characteristics and target dose metrics. A Spearman correlation was used to investigate LRF and (1) ENI vs. no ENI, (2) TV coverage metrics, (3) planning TV (PTV) volume. <h3>Results</h3> The median PTV volume for all patients, with ENI, and without ENI was 125.9 cc (range: 20.6-356.8), 137.3 cc (range: 72.8-356.8, and 55.4 (range: 20.6-280.15), respectively. LRF occurred in 12 patients (27.3%); the median LRF volume was 10.9 cc (range: 1.1-145.5cc). The median Dice similarity coefficient taken to characterize LRF overlap with PTV was 0.05 (range: 0-0.45). Moderate negative correlation coefficients were noted between recurrence to ENI [-0.59 (<i>P</i> < 0.001)], recurrence to PTV volume [-0.41 (<i>P</i> < 0.001)] and recurrence to max PTV dose [–0.47 (<i>P</i> < 0.001)]. No correlation was observed for cumulative D80, D90, mean dose to recurrence. <h3>Conclusion</h3> To our knowledge, this is the first analysis of cumulative TV dose from 5-fraction SMART for PCa. While use of ENI is controversial, moderate and statistically significant correlations were noted between LRF and ENI, in addition to PTV volume and maximum dose. Additional investigation is warranted to better understand whether specific dose thresholds exist for gross disease and electively treated volumes with respect to long-term LC and OS.