{
  "abstract": "Background Immune checkpoint inhibitors (ICIs) have transformed the oncology treatment landscape, yet effective treatments for rare cancers remain a clinical unmet need. The scarcity and heterogeneity of rare cancers makes widespread progress challenging; a generalizable biomarker to predict response and improve patient selection is critically needed. The immune profile score (IPS) is a DNA- and RNA-based molecular signature validated as a prognostic biomarker in a large cohort of >1,500 advanced solid cancer patients. Here we evaluate IPS in a cohort of patients with rare, advanced, solid cancers who received off-label ICI therapy.Methods From a multimodal real-world database, patients with an advanced solid cancer diagnosis defined as rare by the FDA (<200,000 patients diagnosed) who received any ICI-containing regimen off-label in the first or second line of therapy were eligible (patients with high TMB (>10mut/Mb) or microsatellite instability excluded). IPS was calculated as a continuous score and patients were further categorized into IPS-high(H) and IPS-low(L) using a previously determined threshold. Cox proportional hazards models were fit to demonstrate prognostic utility for overall survival.Results Among 90 evaluable patients with a median follow up of 20 months, 26 tumor types were represented, where carcinosarcoma (n=19, 21%), pancreatic ductal adenocarcinoma (n=17, 19%) and gynecological clear cell carcinoma (n=7, 7.8%) were most common. 16 (17.8%) patients were classified as IPS-high. Only 7 cancer types had representation from both IPS-H and IPS-L. IPS-H patients demonstrated significantly longer survival compared to IPS-L patients (HR=0.26, 95% CI= [0.09-0.73]). IPS remained significant when restricted to cancer types with representation from both IPS-H and IPS-L (HR=0.18, [0.04-0.79]). Additionally, we evaluated the prognostic utility of IPS as a continuous variable. The HR (for a 50 unit increase, HR=0.15 [0.05-0.46]) was similar in magnitude and direction to the categorical representation of IPS-H versus L. IPS was consistently prognostic across all subgroups (demographics, regimen group) and clinically relevant confounders (liver metastases, lung metastases, PD-L1 gene expression, and cancer type).Conclusions Our results support that IPS is a generalizable pan-cancer biomarker that can accurately stratify ICI treatment outcomes. Importantly, IPS may support label expansion of ICIs to rare cancer types without current approvals, representing an area where more effective therapies are urgently needed. To our knowledge we are the first to report a pan-cancer biomarker for ICI in rare cancers.",
  "authors": [
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Michelle Ting-Lin"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Alia D Zander"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Rossin Erbe"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Yan Liu"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Ailin Jin"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Seung W Hyun"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Ben Terdich"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Kyle A Beauchamp"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Timothy Taxter"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Victoria L Chiou"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Michelle M Stein"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Kate Sasser"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA",
        "University of California San Diego, Moores Cancer Center, La Jolla, CA, USA"
      ],
      "name": "Ezra Cohen"
    },
    {
      "affiliations": [
        "Sylvester Comprehensive Cancer Center, University of Miami Miller School of Medicine, Miami, FL, USA"
      ],
      "name": "Anna Lasorella"
    },
    {
      "affiliations": [
        "Lee Schwartzberg Research Center at the West Cancer Center and Research Institute, Memphis, TN, USA"
      ],
      "name": "Gregory A Vidal"
    },
    {
      "affiliations": [
        "Vanderbilt University Medical Center, Nashville, TN, USA"
      ],
      "name": "Scott Haake"
    },
    {
      "affiliations": [
        "Sylvester Comprehensive Cancer Center, University of Miami Miller School of Medicine, Miami, FL, USA"
      ],
      "name": "Antonio Iavarone"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA"
      ],
      "name": "Chithra Sangli"
    },
    {
      "affiliations": [
        "Tempus AI, Inc., Chicago, IL, USA",
        "Northwestern University Feinberg School of Medicine, Chicago, IL, USA"
      ],
      "name": "Halla Nimeiri"
    },
    {
      "affiliations": [
        "Northwestern University Feinberg School of Medicine, Chicago, IL, USA"
      ],
      "name": "Aparna Kalyan"
    }
  ],
  "title": "157 A novel multi-omic algorithm to predict real-world outcomes among patients with rare, advanced, solid cancers treated with off-label immune checkpoint inhibitors",
  "uid": "2c802e37-7327-5c03-b2c1-3766aec9111d"
}