Assessing PD-L1 without a biopsy and through PD-L1 PET imaging with 18F-BMS-986229.

2578 Background: Programmed death ligand-1 (PD-L1) is usually determined by immunohistochemistry (IHC). Determining PD-L1 status by whole body, non-invasive PD-L1 PET imaging with 18F-BMS-986229 tracer infusion may be a way to circumvent limitations to PD-L1 IHC such as poor tissue availability; discrepant PD-L1 IHC results among different tumors within the same patient; and inability to measure PD-L1 longitudinally given challenges with repeated biopsies. Methods: Within the prospective ADAPT-IT trial (NCT03122522) testing an abbreviated course of nivolumab (nivo) + ipilimumab (ipi) in patients with unresectable stage III or IV melanoma, we investigated PD-L1 PET imaging in a 10-patient expansion cohort at baseline and after 6 weeks of treatment. Maximum standard uptake value (SUV max) and mean of all SUV max for lesions for PD-L1 PET were calculated on a whole patient level for each target/ non target lesion. Standard treatment response was determined by RECIST v1.1 at weeks 6 and 12. PD-L1 IHC was scored using the E1L3N antibody by tumor proportion score (TPS), Immune Cell Score (ICS), and a Combined Positive Score (CPS). Comparisons between PD-L1 PET parameters and response were determined by complete or partial response (CR/PR); stable disease (SD), and progressive disease (PD) at respective imaging timepoints. Absolute changes in PD-L1 SUV mean and max were calculated between scans. The correlation between PD-L1 PET and IHC was estimated using Spearman’s coefficient correlation. Results: Five patients (50%) had a PR to treatment at week 6; one of these PR became a CR at week 12 and one PR was not assessable at week 12. All RECIST responders at week 6 (n = 5) had baseline PD-L1 SUV mean ≥3.00, and all progressors at week 6 (n = 3) had baseline PD-L1 SUV mean ≤2.60 (Table). A similar trend was observed when assessing response at week 12 and when considering baseline PD-L1 SUV max. Patients had changes in their PD-L1 SUV levels between baseline and week 6. Baseline PD-L1 SUV mean (inclusive of all lesions within a patient) correlated with PD-L1 IHC by TPS, ICS, and CPS (r = 0.64, 0.5, 0.47, respectively). No patients had side effects from 18F-BMS-986229 tracer infusion. Conclusions: The signal of PD-L1 positivity by PET imaging with 18F-BMS-986229 at baseline appears associated with early efficacy from nivo + ipi in this small cohort and may offer additional information than PD-L1 IHC. The ability to assess PD-L1 on a whole patient level at multiple timepoints on treatment may have future implications in how best to sequence and combine immunotherapies but further study in larger patient cohorts is needed. Clinical trial information: NCT03122522. [Table: see text]