Functional binding of PD-1 ligands: Overcoming PD-L1 staining limitations to predict therapy response.

Abstract

2581 Background: Immune-checkpoint blockade (ICB) targeting the PD-1/PD-L1 axis has improved outcomes across multiple malignancies, yet response prediction remains unreliable due to fundamental limitations of PD-L1 immunohistochemistry (IHC). As a static expression assay, PD-L1 IHC does not capture functional ligand engagement, does not account for PD-L2 contributions to PD-1 signaling, and is confounded by N-glycosylation that masks epitopes and distorts measurable abundance. These mechanistic gaps undermine biomarker accuracy and impede appropriate patient selection. To overcome these limitations, we established IcAR-PD1, a functional reporter system designed to quantify the true biological availability of PD-L1 and PD-L2 for therapeutic blockade by anti-PD-1/PD-L1 ICBs. Methods: We conducted a double-blinded retrospective study spanning multiple tumor types. IcAR-PD1 scores from patient-derived samples were correlated with clinical outcomes to anti-PD1 therapy and benchmarked against PD-L1 IHC, evaluating predictive accuracy, response probability, and duration-of-response. To define how PD-L1 N-glycosylation modulates inhibitory signaling and drug susceptibility, IcAR-PD1 was combined with a full four-site glycosylation mutation array and validated using primary CD8⁺ T-cell cytotoxicity assays. Results: IcAR-PD1 demonstrated markedly superior predictive power compared with PD-L1 IHC, achieving AUCs of 0.87–0.92 versus 0.55–0.62. High IcAR-PD1 scores aligned with >90% response probability and >3-fold longer median duration-of-response (p<0.01). Incorporation of PD-L2 activity and direct measurement of functional ligand availability explained the improved performance across tumor types. Glycosylation analysis showed that fully glycosylated PD-L1 reduced blockade efficiency, with anti-PD1 treatments markedly more sensitive to glycan status than anti-PDL1. Loss of the N35 site diminished anti-PD1 efficacy by increasing release of functional soluble PD-L1, partially affecting anti-PDL1 activity. The non-glycosylated Nx4 variant displayed enhanced susceptibility to both therapies, with a disproportionately greater improvement for anti-PD1 blockade. Conclusions: IcAR-PD1 resolves the major limitations of PD-L1 IHC by quantifying functional ligand activity, incorporating PD-L2 contribution, and revealing glycosylation-driven resistance mechanisms. These findings support IcAR-PD1 as a precise, mechanism-based biomarker for selecting patients most likely to benefit from anti-PD1 therapy.