Author(s)

  • Jéssica Guedes (Presenting Author) | Clinical Protein Science and Imaging | Lund University - European Cancer Moonshot Lund Center, 221 84, Lund, Sweden
  • Nicole Woldmar | Clinical Protein Science and Imaging | Lund University - European Cancer Moonshot Lund Center, 221 84, Lund, Sweden
  • András Kriston | Peter Horvath Laboratory | Biological Research Centre, 6726, Szeged, Hungary
  • Ede Migh | Peter Horvath Laboratory | Biological Research Centre, 6726, Szeged, Hungary
  • Ferenc Kovacsv | Peter Horvath Laboratory | Biological Research Centre, 6726, Szeged, Hungary
  • Henriett Oskolás | Clinical Protein Science and Imaging | Lund University - European Cancer Moonshot Lund Center, 221 84, Lund, Sweden
  • István Németh | Marys Cancer Laboratory | Department of Dermatology and Allergology, 6720, Szeged, Hungary
  • Peter Horvath | Peter Horvath Laboratory | Biological Research Centre, 6726, Szeged, Hungary
  • Gyulai Rolland | Department of Dermatology and Allergology | Department of Dermatology and Allergology, 6720, Szeged, Hungary
  • Esters Baltas | Department of Dermatology and Allergology | Department of Dermatology and Allergology, 6720, Szeged, Hungary
  • Jeovanis Gil | Clinical Protein Science and Imaging | Lund University - European Cancer Moonshot Lund Center, 221 84, Lund, Sweden
  • György Marko-Varga | Clinical Protein Science and Imaging | Lund University - European Cancer Moonshot Lund Center, 221 84, Lund, Sweden

Abstract

Despite advances in cancer therapy, treating late-stage cases remains challenging, mainly due to resistance and non-responsiveness treatment. Comprehensive profiling of individual tumors can identify actionable targets, enabling the selection of the most effective therapy for each patient. In this study, we developed an integrative workflow to support treatment decision-making for late-stage cancer patients by combining AI-driven morphological analysis, laser microdissection of tumor and stromal regions, including specific subclones, and in-depth molecular profiling through gene mutation panels and spatial proteomics. Applying this workflow to multiple metastatic patients uncovered novel therapeutic opportunities. The AI-based analyses identified distinct cell subpopulations within primary and metastatic lesions, while spatial proteomics revealed significant inter-tumor heterogeneity, highlighting markers of treatment resistance and upregulation of cancer-associated pathways, including PI3K/AKT/MTOR, glycolytic signaling, and mitochondrial metabolism, which were identified as drivers of cancer progression in the analyzed patients. Our findings support alternative treatments based on individual molecular profiles, showing that combining molecular and morphological characterization enables personalized care. This approach sets the ground for omics analysis inside clinical surroundings, enhances treatment response prediction and has the potential to extend patient survival.