The commercial diagnostic landscape for PD-L1 immunohistochemistry (IHC) assays is highly complex. Multiple different companion or complementary diagnostic tests exist for therapeutics targeting the PD-1/PD-L1 pathway, each using a different interpretation to inform therapeutic decision-making. Flagship Biosciences envisions the utilization of Computational Tissue Analysis (cTA®) to develop an approach that could harmonize the interpretation of individual PD-L1 diagnostic tests. Specifically, when a single, continuous cTA-based scoring system is applied across each assay, the assays can be mathematically normalized, harmonizing PD-L1 assay scoring.
In a proof-of-concept study, non–small cell lung cancer (NSCLC) patient samples were stained with the FDA-approved Dako 28-8 and Dako 22C3 tests, as well as the in-house SP142 and E1L3N assays. The cTA platform was used to identify tissue and cell-specific Biofeatures™ and then generate digital scores for PD-L1 test comparison. The performance of the cTA platform in scoring a PD-L1 IHC assay was first examined by comparing the digitally generated PD-L1 scores for the 28-8 assay with (1) manual PD-L1 scores generated by multiple pathologists and (2) an orthogonal reference method (ie, NanoString™). The comparison of manual and digital scores (using cTA) demonstrated that the cTA approach significantly reduced variability in PD-L1 scoring. Additionally, the digitally generated PD-L1 scores showed better correlation to the reference method than did the manual PD-L1 scores.
Following evaluation of the cTA platform performance in scoring the 28-8 PD-L1 assay, the digitally generated scores for each of the 4 PD-L1 assays were compared. The FDA-AACR-ASCO “PD-L1 Blueprint” working group has previously identified similarities and differences between these 4 commercialized assays. Similarly, digital quantification of membrane staining intensity in the tumor compartment using the cTA platform showed that the average intensities of the 22C3 and 28-8 assays were similar, while the SP142 intensity was lower and the E1L3N intensity was higher. The percentage of PD-L1–positive cells identified in each assay was highly correlated across the reference range of PD-L1 expression for each assay. Based on the proof of concept demonstrated in this study, a cTA approach is a method that could potentially enable harmonization of the PD-L1 tests through use of a digital pathology platform.