Room: Exhibit Hall | Forum 5
Purpose: Achieving desirable dose to target is difficult in pancreas tumors due to its location and close proximity of dose limiting OARs. We investigated the geometric connections between tumor and OARs by analyzing their nearest distances and volume sectors around the tumor in terms of OAR filling. GTV coverage was assessed relative to this data.
Methods: 21 patients (185 fractions) from our institution undergoing adaptive MR-IGRT for pancreatic cancer were included. All OAR portions falling within a 3 cm ring of the PTV were re-contoured in real-time using daily MR images. The previous treatment’s plan was applied to the new anatomy and unmet OAR dose constraints or potential improved coverage facilitated an adaptive replan. Plan optimization is more heavily weighted toward OAR constraints and the most common reason for adaptation is unmet OAR constraints. In this work, nearest distances between OARs and tumor were tracked throughout treatment, and OAR volumes falling within 1, 2, and 3 cm of the tumor were calculated.
Results: Target coverage is not explained by the nearest distance between OARs and tumor. 182 out of the 185 fractions involved at least one OAR abutting the GTV, yet target coverages spanned a wide range of values. When the space around the GTV is further divided into discrete units, OAR filling within each volume sector correlates with coverage in that sector. Overall, the best predictor of eventual coverage is the percentage of the space within 1 cm of the GTV that is occupied by OARs.
Conclusion: This work shows that a more complete picture involving the distribution of OAR volumes around the target better explains expected coverage as compared to OAR nearest distances. Notably, for SBRT patients the percentage of space within 1 cm of the GTV that is occupied by OARs strongly correlates with eventual target coverage.