NEIM-12 PRE- AND POST-TREATMENT RADIOLOGIC FEATURES AND VOLUMETRIC CHANGES LINKED TO TREATMENT RESPONSE OF BRAIN METASTASES AFTER LASER INTERSTITIAL THERMAL THERAPY (LITT)

Abstract

Abstract OBJECTIVE To identify pre-treatment imaging features of brain metastases (BM) associated with lesion-specific progression free survival (PFS-L) after laser interstitial thermal therapy (LITT) and to report longitudinal volumetric changes following the treatment. METHODS Forty-seven BM treated with LITT were retrospectively examined. BM were followed until clinical progression, radiologic progression according to the modified RANO criteria, or censoring. The association between imaging features and PFS-L was assessed with survival analyses. Post-LITT volume shrinkage over time was fit to an exponential decay model. RESULTS Thirteen BM progressed after LITT (median time to progression 3.9 months), seventeen met radiologic criteria for response after LITT (median time to response 12.1 months), and seventeen were stable at time of censoring. Median follow-up duration for censored BM was 18.5 months. The probability of PFS-L at six months (PFS6) was 70.6% for the whole cohort. Pre-LITT enhancing volume <2.5 cc (p=0.0004) and sphericity ≥0.705 (p=0.0026) predicted longer PFS-L (PFS6=86.8%), and BM meeting both cutoffs showed a cumulative benefit (p<0.0001, PFS6=100%). Pre-LITT diffusion imaging did not predict PFS-L, while a subset of lesions (n=7) with highly perfused hotspots had worse PFS-L (p=0.0012). Immediate post-LITT volume was increased compared to pre-LITT (p<0.0001, median increase 80.3%). The percentage of volume increase was not a predictor of PFS-L (p=0.56). In fifteen responding BM (88.2%), the post-LITT volumetric shrinkage over time was described by an exponential decay (R2 ranging 0.92-1.0), and the kinetics of shrinkage was heterogeneous (half-life ranging 0.3-5.4 months). CONCLUSIONS Pre-treatment small size and high sphericity are ideal baseline features for BM considered for LITT treatment and could aid patient selection. Modeling longitudinal volumetric changes of BM responding to LITT is a potential approach to identify BM at risk for progression which deviate from a typical shrinkage pattern.