QS5. Functional Oncology Mapping (FuncOmap): checkpoint-interaction maps as functional biomarkers to guide neoadjuvant immunotherapy in melanoma

Neoadjuvant immune checkpoint blockade (ICB) is increasingly used in melanoma, yet biomarkers such as PD-L1 expression poorly predict response. Surgeons need tools that identify non-responders early to enable timely surgical rescue and avoid delays from ineffective therapy.


Methods:

FuncOmap, the first automated platform for functional digital spatial profiling uses immune Förster Resonance Energy Transfer (iFRET) as a “chemical ruler” to directly quantify millions of checkpoint interactions (e.g., PD-1:PD-L1) in tumors at < 10 nm distances. High Ef indicates strong checkpoint interaction; low Ef indicates weak interaction. We upgraded FuncOmap to integrate cyclic multiplex immunofluorescence (mIF) immune maps, generating functional immune maps of the tumor microenvironment. Paired pre- and post-treatment melanoma specimens from patients receiving neoadjuvant ICB were analyzed; FFPE compatibility was maintained. We use supervised (tumor associated macrophages, M1-like vs M2-like) and unsupervised (all PD-1/PD-L1 cells) machine learning to distinguish critical patterns of checkpoint interactions in the tumor-immune microenvironment.


Results: In the first 10 patients analyzed (40 total in training set), loss of PD-1:PD-L1 interactions was associated with favorable pathologic response, whereas gainwas observed with non-response. Importantly, FuncOmap distinguished patients with similar partial pathologic responses, where loss of interaction predicted superior survival compared with gain of interaction. Integrated cell-specific analyses showed immunostimulatory tumor-associated macrophages were enriched in pre-ICB samples of responders within regions of high interaction. These findings demonstrate that checkpoint-interaction change patterns stratify outcomes more effectively than PD-L1 scores.


Conclusions:

FuncOmap establishes a new class of functional immune biomarkers by quantifying checkpoint interactions in situ. Preliminary results suggest predictive capacity superior to PD-L1. In completion of our training data, we aim to i) develop biomarker-driven decision-making. including identifying opportunity for early surgical rescue and tailored operations for responders and ii) further.. FuncOmap is also optimized for additional checkpoint and targeted receptor:ligand pairs (e.g., CTLA4:CD80, LAG3:MHCII, AXL:Gas6), broadening translational potential.