ABSTRACT
CD8+ tumor-infiltrating T cells can be regarded as one of the most relevant predictive biomarkers in immune-oncology. Highly infiltrated tumors, referred to as inflamed (clinically "hot"), show the most favorable response to immune checkpoint inhibitors in contrast to tumors with a scarce immune infiltrate called immune desert or excluded (clinically "cold"). Nevertheless, quantitative and reproducible methods examining their prevalence within tumors are lacking. We therefore established a computational diagnostic algorithm to quantitatively measure spatial densities of tumor-infiltrating CD8+ T cells by digital pathology within the three known tumor compartments as recommended by the International Immuno-Oncology Biomarker Working Group in 116 prospective metastatic melanomas of the Swiss Tumor Profiler cohort. Workflow robustness was confirmed in 33 samples of an independent retrospective validation cohort. The introduction of the intratumoral tumor center compartment proved to be most relevant for establishing an immune diagnosis in metastatic disease, independent of metastatic site. Cut-off values for reproducible classification were defined and successfully assigned densities into the respective immune diagnostic category in the validation cohort with high sensitivity, specificity, and precision. We provide a robust diagnostic algorithm based on intratumoral and stromal CD8+ T-cell densities in the tumor center compartment that translates spatial densities of tumor-infiltrating CD8+ T cells into the clinically relevant immune diagnostic categories "inflamed", "excluded", and "desert". The consideration of the intratumoral tumor center compartment allows immune phenotyping in the clinically highly relevant setting of metastatic lesions, even if the invasive margin compartment is not captured in biopsy material.
Subject(s)
CD8-Positive T-Lymphocytes , Image Processing, Computer-Assisted , Immunophenotyping/methods , Melanoma/pathology , Adult , Aged , Aged, 80 and over , Deep Learning , Female , Humans , Male , Melanoma/immunology , Middle AgedABSTRACT
Future Medicine: Digital Pathology Abstract. Pathology is facing a paradigm shift. Digitization enables highly efficient, networked diagnostics and the simplified exchange of expert knowledge. Algorithms for image analysis and artificial intelligence have the potential to further increase the quality of diagnostics in pathology. Structured electronic reporting enables the continuous development of digital diagnostics and improves the communication between clinical disciplines. Here we identify and discuss the main trends that will shape digital pathology.
