{
  "abstract": "Background Head and neck cancer remains the 6th most common cancer globally, with patients often diagnosed in the advanced setting. With increasing use of targeted therapies and immunotherapy earlier in the treatment paradigm, there is growing interest in comprehensive molecular profiling of the tumor microenvironment (TME). Spatial proteomics technologies now enable high-plex characterization of TME biology at single-cell resolution, offering valuable insights into clinical outcomes. However, analysis of high-plex spatial data, especially using unsupervised methods across whole-slide images and large patient cohorts, remains computationally complex and methodologically fragmented.Methods We present Celtix, a fully integrated platform developed by Tixomics that streamlines data quality control, cell segmentation, phenotyping, and spatial analysis for high multiplex immunofluorescence imaging data. Designed for scalability and cross-cohort reproducibility, Celtix enables whole-slide, single-cell resolution analysis across large datasets. To demonstrate its utility, we applied Celtix to a cohort of head and neck squamous cell carcinoma (HNSCC) patients undergoing immunotherapy, analyzing over 8 million cells. A comprehensive 52-antibody panel targeting tumor, immune, vascular, and metabolic markers was used to characterize the TME in a spatially resolved, high-dimensional manner.Results The Celtix platform, powered by Tixomics’ analytical framework, demonstrated robust performance in handling high-dimensional spatial proteomics data, delivering an end-to-end pipeline from data ingestion to spatial insight generation. Using samples from 10 HNSCC patients (5 responders, 5 non-responders), we identified 18 distinct cell phenotypes through unsupervised clustering (figure 1). Responders showed enrichment of B cells, particularly plasma cells, and elevated proportions of helper T cells, neutrophils, and macrophages. Non-responders exhibited increased distance between proliferating B cells and plasma cells, indicative of impaired B cell maturation and disrupted tertiary lymphoid structure (TLS) organization. They also showed reduced distance between tumor cells and regulatory T cells (Tregs), consistent with enhanced local immunosuppression and physical exclusion of cytotoxic lymphocytes. The distance between Tregs and tumor cells was greatest in partial responders, intermediate in progressive disease, and shortest in stable disease (figure 2), suggesting that Treg proximity to tumor may contribute to local immunosuppression and resistance.Conclusions Celtix, developed by Tixomics, provides a scalable and reproducible platform for high-plex spatial proteomic analysis, enabling comprehensive profiling of the TME at single-cell, whole-slide resolution. Its application to HNSCC demonstrates its utility in uncovering spatial biomarkers of immunotherapy response. Future work will expand to region-of-interest (ROI) comparisons, integrate metabolic marker insights, and include larger cohorts to support statistical validation and biomarker discovery.Abstract 1081 Figure 1Cell phenotyping heatmap. Cell phenotyping heatmap showing main cell types and key markers that differentiate themAbstract 1081 Figure 2Distances to tumor cells. Comparisons between RECIST for tumor cells distances to immune cells",
  "authors": [
    {
      "affiliations": [
        "Tixomics, Los Angeles, CA, USA"
      ],
      "name": "Bassem Ben Cheikh"
    },
    {
      "affiliations": [
        "Wesley Research Institute, Auchenflower, QLD, Australia"
      ],
      "name": "Meg Donovan"
    },
    {
      "affiliations": [
        "Frazer Institute, Faculty of Health, Medicine and Behavioural Sciences, The University of Queensland, Woolloongabba, QLD, Australia"
      ],
      "name": "James Monkman"
    },
    {
      "affiliations": [
        "University of Queensland, Brisbane, QLD, Australia"
      ],
      "name": "Arutha Kulasinghe"
    }
  ],
  "title": "1081 Celtix: an end-to-end spatial analysis platform for high-multiplex immunofluorescence with application to head and neck cancer",
  "uid": "c81c0246-ee55-5b60-9a75-55bc186b83b8"
}