The combination of soluble forms of PD-1 and PD-L1 as a predictive marker of PD-1 blockade in patients with advanced cancers: a multicenter retrospective study.

Introduction: The clinical relevance of soluble forms of programmed cell death-1 (sPD-1) and programmed cell death-ligand 1 (sPD-L1) remains unclear. We here investigated the relation between the efficacy of PD-1 blockade and pretreatment plasma levels of sPD-1 and sPD-L1 across a broad range of cancer types.

Methods: We retrospectively analyzed clinical data from 171 patients with advanced solid tumors who received nivolumab or pembrolizumab monotherapy regardless of treatment line. The concentrations of sPD-1 and sPD-L1 were measured with a fully automated immunoassay (HISCL system).

Results: The study subjects comprised patients with head and neck cancer ( = 50), urothelial cancer ( = 42), renal cell cancer ( = 37), gastric cancer ( = 20), esophageal cancer ( = 10), malignant pleural mesothelioma ( = 6), or microsatellite instability-high tumors ( = 6). High or low levels of sPD-1 or sPD-L1 were not significantly associated with progression-free survival (PFS) or overall survival (OS) for PD-1 blockade in the entire study population. Comparison of treatment outcomes according to combinations of high or low sPD-1 and sPD-L1 levels, however, revealed that patients with low sPD-1 and high sPD-L1 concentrations had a significantly poorer PFS (HR of 1.79 [95% CI, 1.13-2.83], = 0.01) and a tendency toward poorer OS (HR of 1.70 [95% CI, 0.99-2.91], = 0.05) compared with all other patients.

Conclusion: Our findings suggest that the combination of low sPD-1 and high sPD-L1 levels is a potential negative biomarker for PD-1 blockade therapy.