FDG-PET associations with pathological response and survival with neoadjuvant immunotherapy for melanoma.

Abstract

9523 Background: This study sought to describe the changes seen with FDG-PET following neoadjuvant immunotherapy in melanoma patients (pts), and explore associations with pathological response and recurrence-free survival (RFS). Methods: Two prospective clinical trial cohorts of stage III pts with RECIST measurable nodal melanoma were included; 1) NeoTrio, 13 pts received 2 doses of preoperative pembrolizumab alone, and 2) NeoPele, 20 pts received 2 doses of pembrolizumab and 5 weeks of lenvatinib preoperatively. All pts underwent baseline and week 6 (preoperative) FDG-PET assessments and all had surgery. PET responses were evaluated based on the modified EORTC criteria. In addition to the standard categories Complete Metabolic Response (CMR), Partial Metabolic Response (PMR), Stable Metabolic Disease (SMD), and Progressive Metabolic Disease (PMD), a novel category was established, near-CMR, where the maximum standardized uptake value (SUVmax) decreased by more than 90%. PET responses were determined while blinded to the outcome data. Pathological response was determined as per INMC criteria. Results: 33 pts were included, 67% male, median age 65 years, 48% BRAF mutant. Pathological response, PET response and correlations are shown in the Table. All 8 (24%) pts achieving CMR or near-CMR obtained major pathologic response (MPR), while the 14 pts (42%) with PMR had variable pathological outcomes (57% MPR, 21% pPR, 21% pNR). For the 6 pts with PMD, 5 (83%) had pNR and the one pts who achieved pCR recurred with brain metastases. After a median 29.9 mo follow-up (95% CI 27.0-33.1), PET response associated with RFS; those with CMR/near CMR had 100% 24mo RFS (nil recurrences to date) while those with PMR or SMD had inferior survival (79% and 80% 24mo RFS), and those with PMD the worst outcomes (17% 24mo RFS, all pts except one have recurred) (p=0.0029). The predictive value of PET and pathology for RFS was assessed using the Cox Proportional Hazards model. 12- and 24-month RFS AUCs of PET and pathology response were similar at 81.5% (95% CI: 67.7-95.3) vs. 80.9% (95% CI: 67.8-93.9), and 83.3% (95% CI: 69.1-97.5%) vs. 80.8% (95% CI: 66.6-94.9), respectively. Conclusions: FDG-PET demonstrates utility in predicting pathological response and survival with neoadjuvant immunotherapy in melanoma. PET may identify pts who are not going to be pathological responders and who have the worst survival outcomes, enabling a potential switch of neoadjuvant systemic therapy. [Table: see text]