Phase Ib/II study of Pembrolizumab with Lanreotide depot for advanced, progressive Gastroenteropancreatic neuroendocrine tumors (PLANET)

While performing a study of immune checkpoint blockade with the anti-PD-1 antibody pembrolizumab combined with the somatostatin analogue (SSA) lanreotide in patients with low- and intermediate-grade gastroenteropancreatic neuroendocrine tumors (GEP-NETs), we studied whether there were any immune correlates of response to the anti-PD-1 therapy that could guide future attempts to integrate immunotherapy into the treatment of NETs. Patients with grade 1 and 2 GEP-NETs who had progressed on a prior SSA received lanreotide 90 mg subcutaneously and pembrolizumab 200 mg intravenously every 3 weeks until progression or intolerable toxicity. Objective response rate (ORR) at any time in the study, clinical benefit rate (CBR, defined as stable disease or better), progression-free survival (PFS), and overall survival (OS) were measured. Changes in T cell subsets in peripheral blood before and during therapy were analyzed by multiparameter mass cytometry (CyTOF). Archived tissue samples were analyzed for PD-L1 expression and TIL infiltration. Twenty-two (22) patients (GI/pancreatic 14/8, median Ki67 7% [IQR 4, 10%], median 1.5 prior systemic therapies [range 1–4]) were enrolled. Among the GI-NETs, there was one partial response, the CBR was 50%, the median PFS was 8.5 months, and the median OS was 32.7 months. No responses were seen in pancreatic NETs, which had 0% CBR, a PFS of 2.7 months, and an OS of 23.9 months. Of the 16 analyzable tumors, 6 had detectable PD-L1 expression and 15 had detectable TILs. Neither TILs nor PD-L1 expression correlated with ORR or CBR. However, clinical benefit (SD or better) was associated with peripheral blood on-treatment effector memory T cell activation and progressive disease was associated with baseline peripheral blood regulatory T cell (Treg) activation. We conclude that immune checkpoint blockade had low activity in unselected patients with grade 1 and 2 GEP-NETs. Further study of strategies to reduce Treg activation or enhance effector memory activation during immunotherapy is warranted.