{
  "abstract": "Background Lack of functional biomarkers to classify melanoma patient responses to immune checkpoint blockade (ICB) complicates the prediction of which patients will benefit most from ICB. 1 2 While PD1/PD-L1 interactions are promising biomarker candidates,3 4 previous studies have relied on gene or protein expression assays that only infer interaction and do not accurately reflect PD1/PDL1 functional status. We have developed a novel spatial imaging platform, FuncO:TiME,5 to directly quantify PD1/PD-L1 interactions and map them onto single-cell PD1 expression across melanoma tumor-immune microenvironments (TiMEs), aiming to identify distinct interaction profiles among ICB treatment responders and non-responders.Methods FuncO:TiME was performed on melanoma patient formalin-fixed paraffin embedded (FFPE) samples collected pre- and post-treatment as examples of complete, partial and non-responses to standard neoadjuvant ICB therapy. 5 FuncO:TiME’s computational pipeline integrates data acquired from two spatial imaging platforms; functional oncology mapping (FuncOmap6) and COMET sequential immunofluorescence (seqIF7). FuncOmap quantifies and maps 30x106 PD1/PD-L1 interactions on a per-pixel basis which are overlaid onto single cell PD1 expression across the TiME on COMET seqIF images. FuncO:TiME also generates global and region-specific violin plots highlighting distribution of PD1/PD-L1 interactions pre- and post-ICB between complete, partial and non-responder patients.Results Global violin plots confirmed that complete, partial and non-responders to ICB therapy had unique shifts in the distributions of whole tissue PD1/PD-L1 interactions between pre- and post-treatment tissues ( figure 1). Post-treatment PD1/PD-L1 interactions were significantly higher in non-responder tissue but did not correlate with TiME PD1 expression. Region-specific FuncO:TiME images and corresponding violin plots (figure 2) highlight tissue areas of higher PD1 expression yet no observable PD1/PD-L1 interactions and low PD1 expression with high PD1/PD-L1 interactions, honing the point that PD1 expression does not reflect its functional state.Conclusions FuncO:TiME closes critical gaps in functional biomarker discovery for immunotherapies and highlights the need to consider PD1/PD-L1 interactions in the context of specific TiME cell sources. The high-resolution spatial quantification of checkpoint interactions by FuncO:TiME has the potential to revolutionize ICB use in melanomas and can be applied broadly across other cancers and immunotherapies.References Li H, van der Merwe PA, Sivakumar S. Biomarkers of response to PD-1 pathway blockade. Br J Cancer. 2022;126(12):1663–1675. doi:10.1038/s41416-022-01743-4Wagner E, Larijani B, Kirane AR. Predictive biomarkers for immune checkpoint inhibitor therapy in advanced melanomas. Surg Oncol Clin N Am. 2025;34(3):437–451. doi:10.1016/j.soc.2025.01.006Sánchez-Magraner L, Gumuzio J, Miles J, et al. Functional engagement of the PD-1/PD-L1 complex but not PD-l1 expression is highly predictive of patient response to immunotherapy in non-small-cell lung cancer. J Clin Oncol. 2023;41(14):2561­­­­­­­­­­­­-2570. doi:10.1200/JCO.22.01748Kirane AR, Lee D, Lowe M, et al. Toward functional biomarkers of response to neoadjuvant oncolytic virus in stage II melanoma: immune-förster resonance energy transfer and the dynamic tumor immune microenvironment. JCO Oncol Adv. 2025;(2):e2400049. doi:10.1200/OA-24-00049Legg S, Wagner E, Applebee CJ, Kirane AR, Padget J, Larijani B. Functional spatial mapping of the tumour immune microenvironment in advanced melanoma patients. Published online June 1, 2025:2025.05.29.656855. doi:10.1101/2025.05.29.656855Safrygina E, Applebee C, McIntyre A, Padget J, Larijani B. Spatial functional mapping of hypoxia inducible factor heterodimerisation and immune checkpoint regulators in clear cell renal cell carcinoma. BJC Rep. 2024;2(1):1-11. doi:10.1038/s44276-023-00033-7Rivest F, Eroglu D, Pelz B, et al. Fully automated sequential immunofluorescence (seqIF) for hyperplex spatial proteomics. Sci Rep. 2023;13:16994. doi:10.1038/s41598-023-43435-wEthics Approval All human materials were collected and processed according to Stanford University Institutional Review Board (IRB: 65607).Abstract 103 Figure 1Overview of FuncO:TiME process and global violin plots highlighting shifts in pre- and post-treatment PD1/PD-L1 interactions between complete, partial and non-responders to immune checkpoint blockade therapy. A) FuncO:TIME incorporates FuncOmap and COMET seqIF imaging data to simultaneously address the spatial arrangement of cell types in the TIME, their PD1/PD-L1 expression and which cells show quantifiable PD1/PD-L1 interactions. B) Global violin plots were generated to highlight shifts in the distribution of 30x105 PD1/PD-L1 interactions pre- versus post-treatment in patients 27, 34 and 38 as examples of partial, non- and complete responders to ICB therapy respectively. PD1/PD-L1 interactions are quantified as Förster resonance energy transfer (FRET) efficiency (Ef) values, where Ef above 4% is considered as interactive based on the Förster radius of chromophores used to label PD1 and PD-L1. Statistical differences between pre- and post-treatment Ef values were calculated using non-parametric Mann-Whitney U testing on 1000 randomly sampled data points (dark lines) from each of the 30x10 datapoints acquired per tissue sample. Significant differences between pre- and post-treatment PD1/PD-L1 interactions were observed for all patient response categories, however post-treatment interactions were significantly higher for patient 34 compared to patients 27 and 38, which is reflective of this patient‘s non-response to ICB therapyAbstract 103 Figure 2Mapping of PD1/PD-L1 interactions TiME sources of PD1 expression using FuncO:TIME: Melanoma FFPE samples collected from patient 27 pre- and post-treatment with ICB were profiled using COMET seqIF and FuncOmap. Top row: COMET images were inspected to find regions of PD1 expression not impacted by autofluorescence (orange squares). Middle row: The resolution of regions within the orange squares were increased using Deep Zoom software and pattern matching algorithms were applied to find the corresponding region on FuncOmap images. Bottom row: Composite Funco:TiME images showing PD1/PD-L1 interactions (Ef values, from FuncOmap) overlaid onto TiME PD1 expression from COMET seqIF images. Per-pixel Ef values are colored according to the adjacent scale (0-50%, where Ef values 0-4% are indicative of no PD1/PD-L1 interaction due to the Förster radius between chromophore pairs during FRET-based assays). Corresponding violin plots show the distribution of all Ef values in the localized regions. Note the pre-treatment FuncO:TIME image & violin plot shows a region with seemingly low PD1 expression yet high distributions of Ef values. Whereas the post-treatment FuncO:TIME image & violin plot shows the inverse: a region with seemingly high PD1 expression yet the majority of PD1/PD-L1 Ef values are 5% and below, indicating little to no PD1/PD-L1 interactions in this region. Top right: Overview of FuncO:TIME computational image processing workflow [2]",
  "authors": [
    {
      "affiliations": [
        "Stanford University School of Medicine, Stanford, CA, USA",
        "Stanford University, Palo Alto, CA, USA"
      ],
      "name": "Emma Wagner"
    },
    {
      "affiliations": [
        "University of Bath, Bath, Bath, UK",
        "Cell Biophysics Laboratory, University of Bath, Bath, Bath, UK"
      ],
      "name": "Samuel Legg"
    },
    {
      "affiliations": [
        "Stanford University School of Medicine, Stanford, CA, USA"
      ],
      "name": "Yichen Zhang"
    },
    {
      "affiliations": [
        "Cell Biophysics Laboratory, University of Bath, Bath, Bath, UK"
      ],
      "name": "Chris Applebee"
    },
    {
      "affiliations": [
        "Stanford University School of Medicine, Stanford, CA, USA"
      ],
      "name": "Allison Betof Warner"
    },
    {
      "affiliations": [
        "University of Bath, Bath, Bath, UK"
      ],
      "name": "Julian Padget"
    },
    {
      "affiliations": [
        "University of Bath, Bath, Bath, UK"
      ],
      "name": "Banafshe Larijani"
    },
    {
      "affiliations": [
        "Stanford University School of Medicine, Stanford, CA, USA"
      ],
      "name": "Amanda Kirane"
    }
  ],
  "title": "103 Functional spatial mapping of the tumour immune microenvironment in advanced melanoma patients",
  "uid": "840b88ee-15c9-5e8f-9e5b-438b13d2b4b7"
}