Noninvasive assessment of programmed-death ligand-1 (PD-L1) in esophagogastric (EG) cancer using 18F-BMS-986229 PET.

Abstract

409 Background: PD-1 inhibitors are now standard in advanced EG cancer, with approval often based on PD-L1 CPS. However, single site biopsies are inadequate due to disease heterogeneity, dynamic changes over time, and variability due to operator, choice of assay, and tumor content. Better methods of non-invasive, comprehensive PD-L1 evaluation are needed. Methods: This is a prospective, pilot PET study of the positron-emitting agent 18F-BMS-986229, a macrocyclic peptide with high affinity for PD-L1. Patients were administered 18F-BMS-986229 intravenously and underwent whole body PET/CT 60 minutes later. The primary endpoint was safety and feasibility. Results: Ten patients underwent PD-L1 PET imaging. All had adenocarcinoma, 70% had metastatic disease, 20% had locally advanced, unresectable tumors, and 10% had resectable tumors. 30% of patients received prior treatment, including 2 with a PD-1 inhibitor. Median PD-L1 CPS was 10 (IQR 5-20). No patients experienced an adverse event associated with 18F-BMS-986229. Although PD-L1 CPS ≥1 was required for inclusion, 2 patients who had PD-L1 CPS ≥1 on initial biopsy, had a subsequent biopsy closer in time, but still prior to PD-L1 PET imaging, with PD-L1 CPS <1. In 8 of 10 patients, the biopsy site was evaluable by PD-L1 PET imaging. 5 of the 6 biopsy sites that were PD-L1 CPS ≥1, were PD-L1 PET avid; both biopsy sites that were PD-L1 CPS <1 were appropriately non-avid. In sum, 88% of patients had radiographic and pathologic concordance. The Spearman rank correlation coefficients (rs) between PD-L1 CPS and PD-L1 PET visualization score (scored 1 to 5, defined by accumulation greater than adjacent background) was 0.64, and 0.61 between PD-L1 CPS and PD-L1 PET SUV. Patients with a higher PD-L1 PET visualization score had a higher PD-L1 CPS: those with a maximum visualization score of 5 (n=4), had a median PD-L1 CPS of 22.5 (range 5-70), those with a maximum score of 4 (n=2) had PD-L1 CPS of 5 and 20, and those without any avid lesions (score of 3 or lower) had a median PD-L1 CPS of 5 (range 0-10). 71% of patients with PD-L1 PET avid lesions, also had non-avid sites of disease, 80% of whom had a more than 2-fold difference in SUV between the most and least avid lesions. 7 patients had untreated, advanced disease, 6 of whom were treated with PD-1 inhibitors as frontline therapy. 3 had PD-L1 PET avid lesions: 2 achieved a complete response (CR), the third achieved a partial response (PR), and progression free survival (PFS) was 32, 28, and 22 months respectively. Among the 3 patients with non-avid PD-L1 PET imaging who received PD-1 inhibitors as frontline therapy, 1 achieved a CR, 1 achieved a PR, and 1 had progressive disease, with PFS of 14.5, 10, and 2 months, respectively. Conclusions: We report the first use of PD-L1 PET imaging in patients with EG cancer and demonstrate its potential use as a companion diagnostic tool to aid in identifying patients who may benefit from PD-1 therapy. Clinical trial information: NCT04161781 .