PathAI Composite PD-L1 and AI-powered CD8+ Topology Biomarker May Improve Prediction of Immuno-Oncology Treatment Response in Patients with Melanoma at the Society for Immunotherapy of Cancer Meeting 2021
Boston, Massachusetts – PathAI, a global provider of AI-powered technology applied to pathology announced that results from a recent exploratory biomarker analysis on digitized slides to apply AI-predicted CD8 topology assessment of patients with advanced melanoma will be presented at the annual Meeting of Society for Immunotherapy of Cancer, November 10-14 2021.
The results will be shared in two posters, Lee et al. The utility of AI-powered spatial classification of intratumoral CD8+ immune-cell topology in predicting overall survival in patients with melanoma as part of the CheckMate 067 clinical trial, and Glass et al. Machine Learning Models Can Quantify CD8 Positivity in Melanoma Clinical Trial Samples.
PathAI developed Machine Learning (ML) models to quantify CD8+ T cells in digitized whole slide images (WSI) of melanoma patient samples, and validated their performance against a consensus of pathologists on a held-out data set. These models were deployed on CD8-stained biopsy samples collected from patients with previously untreated advanced melanoma in the CheckMate 067 clinical trial. The AI-based predictions of CD8 positivity were used by ML models created by Bristol Myers Squibb to categorize each WSI by spatial pattern and density of CD8+ T cell infiltration (CD8 topology) as desert (deficient in CD8+ T cells), excluded (CD8+ T cells at the tumor boundaries and surrounding stroma), or inflamed (CD8+ T cells within the tumor parenchyma).
PD-L1 stained tumor cell positivity scores of either PD-L1 <1% or PD-L1 >1%, collected previously during the CheckMate 067 clinical trial, were combined with the CD8+ spatial phenotype scores to create a composite biomarker. Correlations between overall survival of patients in CheckMate 067 and CD8+ spatial phenotype alone, or the composite biomarker identified a subpopulation of patients with PD-L1 expression < 1%, and a CD8+-excluded tumor spatial pattern that benefited significantly from the drug combination treatment (nivolumab and ipilimumab) compared with PD-L1 negative patients with a CD8+-inflamed tumor spatial pattern (P = 0.002). No difference in survival between excluded and inflamed phenotypes was observed for patients treated with monotherapy (nivolumab alone) (P = 0.41), nor with any patients with PD-L1 >1%.
The data presented here suggest that in addition to PD-L1 positive patients, PD-L1 negative CheckMate 067 patients who were also CD8+ in specific subregions of tumors had overall survival benefit when treated with ipilimumab plus nivolumab combinations compared to those that were treated with ipilimumab monotherapy.
Together, these publications highlight the promise of ML-based CD8+ phenotype scoring to reveal populations of patients that might have the greatest benefit from existing treatments.
PathAI is a leading provider of AI-powered research tools and services for pathology. PathAI’s platform promises substantial improvements to the accuracy of diagnosis and the efficacy of treatment of diseases like cancer, leveraging modern approaches in machine and deep learning. Based in Boston, PathAI works with leading life sciences companies and researchers to advance precision medicine. To learn more, visit pathai.com.