Can multiplexing Immunohistochemistry for PD-L1, CD68, and CD3 improve scoring for “Keytruda®” Indication?

Abstract

Abstract Introduction/Objective FDA-approved companion diagnostics immunohistochemistry scoring for PD-L1 is essential for Keytruda ® treatment in patients with NSCLC. The challenge with PD-L1 testing alone is false positive PD-L1 expression score due to alveolar macrophage infiltration. Multiplexing with PD-L1, CD68, and CD3 could help alleviate this issue. By viewing PD-L1 expression on tumor cells, CD68 expression on macrophages, and the presence of CD3+ T cells on a single slide, pathologists will be able to see the presence of an immune microenvironment (high PD-L1 expression with abundant macrophages) and the potential for response to other types of Immunotherapies (presence of T cells). Methods/Case Report Positive NSCLC FFPE blocks were sectioned and stained using Agilent Dako Omnis. Two combinations of multiplex staining were prepared for each block by staining PD-L1 (brown) alone first followed by CD3 and CD68. First slide: CD3(Vina Green), CD68(Magenta); Second slide: CD3(Magenta), CD68(Vina Green). To prevent wash out, dehydration and clearing occurred with two changes of 100% alcohol and three changes of xylene for manual cover slipping. Results (if a Case Study enter NA) Manual scoring of PD-L1 alone vs two different multiplex combinations was compared to Agilent Proscia digital AI scoring using a preliminary study set (9 positives, 2 negatives). CD68-green and CD3-magenta showed better discrimination than CD68-magenta and CD3-green at a glance. This could be due to the staining intensity and nonspecific background staining of CD68-magenta. Manual scoring was higher compared to digital scoring, which could be due to the human error of positive bias. Digital scoring was significantly lower for PD-L1 alone vs both multiplex stains. The digital scoring for the multiplex stains showed a higher scoring percentage by removing false positive macrophages. Conclusion Reflex scoring workflow is being developed to focus on difficult cases using the multiplex triple stain, i.e. clear negative and clear >50% with 3+ intensity. More cases will be collected and studied to further investigate the improved PD-L1 scoring with digital imaging assisted by artificial intelligence and to compare the multiplex color combinations.