Automated Tumour Recognition and Digital Pathology Scoring Unravels New Role for PD-L1 in Predicting Good Outcome in ER-/HER2+ Breast Cancer

Figure 2

(a) Areas of tumour cores targeted for use in subsequent tissue microarrays. Where sufficient material was available, three cores of tissue were taken from three areas, (1) enriched with tumour epithelium (green), (2) enriched with tumour stroma, (<50% tumour epithelium) (black), and (3) enriched for lymphoid infiltrates within the tumour (red). (b) (i) Representative cores ranging from absent to high PD-L1 protein expression assessed by IHC using SP142, as indicated. Both raw immunohistochemistry images and QuPath cell detection masks are shown. A magnified region is shown in an exploded view. The key displays the QuPath cell classifier on the mask images. (ii) Displays PD-L1 expression identified by RNAscope. Cores ranging from absent to high PD-L1 RNAScope expression, as indicated. A magnified region is shown in an exploded view.

(a)

(b)