NIRA-01 GENERATING CHARACTERISTIC RESPONSE CURVES FOR BRAIN METASTASES TREATED WITH SRS AND LITT

Abstract Brain metastases are significant sources of morbidity and mortality in patients with cancer. Potential minimally invasive treatment modalities include stereotactic radiosurgery (SRS) and laser interstitial thermal therapy (LITT), but comprehensive understanding of response profiles to such treatments is lacking. Our objective is to generate characteristic response curves quantifying post-contrast lesion enhancement leading up to and following SRS/LITT. Towards this end, we have established a computational pipeline that performs multi-modal registration of CT-simulation and longitudinal MRI studies, allowing long-term tracking of lesion responses pre- and post-treatment. As proof-of-concept, this pipeline was applied to three exemplar patients to generate characteristic response curves for treatment scenarios of varying complexity. Patient 1 received SRS to a single lesion that exhibited a complete response but recurred locally. Patient 2 received SRS to a single lesion that exhibited a partial response before rapidly progressing. Subsequent LITT re-established partial control with continued lesion involution at last follow-up. Patient 3 received SRS to six lesions that exhibited a mixture of complete and partial responses, two of which show signs of local recurrence in long-term follow-up. These proof-of-concept studies yielded characteristic response curves that model clinical course and response. With further optimization, we are applying this framework to 1500+ patients with 5000+ lesions to establish a comprehensive portfolio of characteristic response curves. This portfolio of lesion behavior will be integrated to construct a 4D brain map summarizing responses in both space and time following SRS and LITT. We envision construction of this atlas will enable evidence-based selection of optimal treatment modalities as well as provide accurate predictions of response and recurrence as a function of treatment type and lesion site.