POSTER: Building Predictive Models Of Clinical Outcomes In Immune Oncology Using Computational Tissue Analysis Scorecards


Several studies have shown that the location and expression of infiltrating immune cells in patient tumors can better identify which patients are more likely to respond to anti- PD-1/PD-L1 therapy. In particular, immunohistochemistry- based studies have shown that the spatial location of PD-L1 expression has particular biological relevance. Here, we use Flagship’s digital pathology platform (cTA®) to investigate IHC based PD-L1 and CD8 staining patterns in Non-Small Cell Lung (NSCLC) tissue biopsies. The cTA platform creates thousands of per-cell Biofeatures™ derived from the scanned images of the IHC stained tissue, and applies Artificial Intelligence (AI) to the data to score endpoints for patient and cohort classification. In this approach, each tissue’s IO landscape is represented using an “IO Scorecard“, which summarizes the IHC biomarker data and captures a comprehensive analysis of the tissue sample. The scorecard models can be used to monitor changes before and after drug treatment and/or create predictive models for patient response outcomes. In this study, NSCLC samples were sectioned and stained using either Dako 22C3 or SP263 PD-L1 IHC assays. Serial sections of each tissue specimen were also stained for CD8 expression. The cTA process detected all cells, assigned them to the tumor or TME compartments, and recorded the Biofeatures™ data which characterized PD-L1 or CD8 staining in the Tumor or TME compartments. The method was validated by its ability to reproduce pathologist scoring for PD-L1 and CD8. The AI Scorecard approach demonstrated that certain PD-L1 staining Biofeatures™ may also predict the CD8 status of a tumor, suggesting that additional CD8 staining may not be necessary to understand important expression patterns pertaining to cytotoxic T-cells.


Biofeatures generated by Flagship’s cTA provide contextual information about biomarker staining and expression that can be used to predict clinical outcomes and endpoints.

  • Flagship AI separates All Cells, Tumor, and Stroma (TME) in whole tissue sections to quantify biomarker heterogeneity and give context to the tumor microenvironment
  • Quantified CD8 expression in NSCLC samples showed higher correlation with SP263 PD-L1 expression on TME-associated cells (R2= 0.6144). PD-L1 22C3 did not correlate as highly with CD8 expression alone (R2 All Cells = 0.3754)
  • Biofeature profiles of each individual cell describe different phenotype presences in tissues. PD-L1 staining of individual cell types in CD8high and CD8low tissues shows larger fluctuations in SP263 staining in certain cell subtypes based on CD8 expression levels
  • Predictive outcome models can be built using the Flagship IO Scorecard approach, which takes in to account hundreds of data points from each cell in a tissue
  • Flagship IO Scorecard-derived endpoints built a model to predict CD8+ infiltration using only PD-L1 staining in NSCLC tissues. IO Scorecard models increase CD8+ infiltration predictive accuracy of SP263 (R2= 0.9012) and 22C3 (R2=0.8833) over tumor/TME specific biomarker quantification methods
Share This Post: