{
  "abstract": "Background Immune checkpoint inhibitors (ICIs) have transformed the treatment for Mismatch repair-deficient (MMRd) endometrial cancer (EC), but a substantial proportion of patients fail to respond. Predictive biomarkers to guide patient selection are still lacking. Prior work from our group identified collagen remodeling within the extracellular matrix (ECM) of the tumor microenvironment (TME) as a feature associated with ICI resistance in MMRd EC. 1 2 We aimed to develop a clinically applicable collagen-based score to predict ICI resistance.Methods Pre-ICI FFPE samples from advanced/recurrent MMRd EC patients treated with ICIs at Gustave Roussy Institute (2016–2024) were retrospectively collected. Patients were classified as ICI responders (ICI-Rs: CR, PR, or SD ≥12 months) or Non-responders (ICI-NRs: PD or SD <12 months). Collagen organization was assessed using polarization-resolved Second Harmonic Generation (pSHG) microscopy in tumor center and periphery (up to two regions of interest [ROIs] per area). Quantitative metrics included collagen density, and entropy and circular variance to describe collagen fiber spatial organization. A machine-learning pipeline was implemented, including single-feature assessment using MaxStat, feature selection using a regularized logistic regression multivariate model, and construction of a two-feature score via unpenalized logistic regression. The model was validated through bootstrap analysis to derive a clinically applicable predictive score.Results Collagen density and spatial organization were analyzed in the tumor center and periphery of samples from 28 MMRd EC patients (17 ICI-Rs, 11 ICI-NRs). Taken individually, the most accurate predictors of ICI-NR were high circular variance (AUC = 0.833), and high entropy of collagen fibril organization in the periphery (AUC = 0.827), two features associated with disorganized collagen deposition. While increased collagen fibril entropy in the tumor center showed lower individual predictive capacity (AUC = 0.780), it consistently emerged as a strong predictor in a multivariable model alongside high collagen density in the tumor center (importance coefficients = 0.61 and 0.65, respectively). A final composite model using these two features achieved high accuracy (AUC = 0.917; specificity = 95.1%; sensitivity = 83.1%) for ICI resistance. The algorithm is patented and supports future clinical application.Conclusions A composite score combining increased collagen density and collagen fibril spatial disorganization within the ECM of the tumor center accurately predicted ICI resistance in MMRd endometrial cancer. Validation in an independent cohort is planned to confirm its predictive capacity. These data suggest that dense and disorganized ECM collagen may contribute to ICI-resistance, and support investigating strategies modulating ECM stiffness to revert ICI-resistance.References Grau Béjar JF, Zeng Q, Mehnert M, Colomba E, Genestie C, Le Formal A, et al. Exploring the tumor microenvironment (TME) in patients with mismatch repair-deficient (MMRd) endometrial cancer (EC) undergoing immune checkpoint inhibitor (ICI) therapy: proteomic insights into immune resistance mechanisms. J Clin Oncol. 2024 Jun;42(16_suppl):5609–5609.Grau JF, Aimé C, Mehnert M, Yaniz-Galende E, Genestie C, Formal AL, et al. Abstract 5149: Extracellular matrix remodeling in metastatic mismatch repair-deficient endometrial cancer: implications for immune checkpoint inhibitor response prediction. Cancer Res. 2024 Mar;84(6_Supplement):5149–5149.",
  "authors": [
    {
      "affiliations": [
        "Gynecologic Cancer Translational Research Laboratory, INSERM Unit 981, Institut Gustave Roussy, Villejuif, France"
      ],
      "name": "Juan Francisco Grau Bejar"
    },
    {
      "affiliations": [
        "Chimie Physique et Chimie du Vivant (CPCV), Département de Chimie, École Normale Supérieure, PSL University, Sorbonne Université, CNRS, Paris, France"
      ],
      "name": "Carole Aimé"
    },
    {
      "affiliations": [
        "Laboratory for Optics and Biosciences, École Polytechnique, CNRS, INSERM, Institut Polytechnique de Paris, Palaiseau, France"
      ],
      "name": "Vaky Abdelsayed"
    },
    {
      "affiliations": [
        "Else Kroener Fresenius Center for Digital Health, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology, Dresden, Germany"
      ],
      "name": "Qinghe Zeng"
    },
    {
      "affiliations": [
        "Tumor Genetics Department, Institut Gustave Roussy, Villejuif, France"
      ],
      "name": "Étienne Rouleau"
    },
    {
      "affiliations": [
        "Department of Pathology, Institut Gustave Roussy, Villejuif, France"
      ],
      "name": "Catherine Genestie"
    },
    {
      "affiliations": [
        "Laboratory for Optics and Biosciences, École Polytechnique, CNRS, INSERM, Institut Polytechnique de Paris, Palaiseau, France"
      ],
      "name": "Marie-Claire Schanne-Klein"
    },
    {
      "affiliations": [
        "Gynecologic Cancer Translational Research Laboratory, INSERM Unit 981, Institut Gustave Roussy, Villejuif, France"
      ],
      "name": "Alexandra Leary"
    }
  ],
  "title": "80 Composite score combining collagen density and spatial organization predicts immune checkpoint inhibitor resistance in mismatch repair-deficient endometrial cancer",
  "uid": "d71d502c-100a-53c0-a816-fff578cc7fb3"
}