Room: Exhibit Hall | Forum 4
Purpose: To simulate patient-specific prescription selection, based on multivariate dose-response models applied to treatment-plans from patients previously receiving radiotherapy for Stage III Non-Small Cell Lung Cancer (NSCLC).
Methods: Treatment-plans and clinical variables from 162 patients were used to calculate multivariate patient-specific dose-responses for â‰¥ grade 2 acute esophagitis (AE2); radiation pneumonitis (RP2); prescription based tumor BED (TBED, including â€œkickoff-timeâ€? 28 days); and dose-volume metrics after LQ adjustment for comparison with 15 fraction clinical limits. The AE2 and RP2 models were previously selected for relevance and validated against this cohort. Dose-responses were calculated for three baseline protocols (P1:2.75Gy X 24; P2: 2Gy X 30; and P3: 4Gy X 15), by scaling either the number of fractions, or fraction size (scale: 0.5-1.5). For each protocol, all tumor/normal tissue metrics were recorded after the first normal tissue constraint was encountered, and the protocol achieving the highest TBED was selected.
Results: Plans starting from P1, P2, and P3 were chosen in 41 (25%), 17 (11%), and 104 (64%) cases respectively. TBED could be escalated in 35/162 cases (2, 0, and 33 of these starting from P1, P2, and P3, respectively), mostly by increasing the fraction size. P1 was mostly associated with fraction number decrease (ensuring treatment completion before accelerated re-population began). The limiting normal tissue constraint was AE2<45%, RP2<20%, esophagus Dmax<60Gy*Â¹, lung V20<30%, and heart Dmax<60Gy*, in 74, 41, 22, 6, 1, cases respectively, with multiple constraints in 18 cases. Esophagus and heart Dmax were limiting constraints for 30 of the cases where dose-escalation occurred.*when not limiting baseline prescription; Â¹else esophagus Dmax<66Gy.
Conclusion: Multivariate dose-response models are potentially useful tools for assisting physicians in patient-specific selection of treatment prescription.