81 Inclusion of PD-L1-expressing tumor cells in the combined positive score algorithm yields superior identification of positive specimens around diagnostic cut-offs across multiple indications

BackgroundPD-L1 IHC 22C3 pharmDx uses Tumor Proportion Score (TPS) and Combined Positive Score (CPS) scoring algorithms for the immunohistochemical (IHC) evaluation of PD-L1 protein expression in human cancer tissues; both algorithms include PD-L1 staining tumor cells (TC) in scoring and CPS also includes scoring of PD-L1 staining mononuclear inflammatory cells to aid in the identification of patients for treatment with pembrolizumab (KEYTRUDA®) using indication-specific diagnostic cut-offs. This study evaluated contribution of TC in determining specimen diagnostic status based on the CPS scoring algorithm by looking into four tumor indications approved for use with KEYTRUDA®: gastric or gastroesophageal junction (GEJ) adenocarcinoma (GC/GEJ), urothelial carcinoma (UC), head and neck squamous cell carcinoma (HNSCC), and esophageal squamous cell carcinoma (ESCC). Detection of specimens expressing PD-L1 is significantly dependent on the PD-L1 staining TC component.MethodsA retrospective analysis was done looking at Dako’s internal tumor bank of the mentioned indications that were all stained with PD-L1 IHC 22C3 pharmDx and scored using the TPS, CPS and Quantitative Immune Cell Density (QID) methods described in figure 1. Statistical analysis encompassed looking at the scores generated that met the following criteria: CPS>0, TPS>0 and CPS≠TPS and then evaluating the percentage of those samples that changed from positive to negative diagnostic status upon removal of the TC component from the scoring.ResultsA noticeable downward trend was observed in all four indications in the total number of positives with the removal of the TC component. Table 1 aptly captures this by showing the number of specimens for each indication that had changed from positive to negative around each indication’s diagnostic cut-off(s). The three indications that showed the highest percentages of diagnostic status change were HNSCC (CPS ≥20) with a remarkable 83.3% (130) followed by UC (CPS ≥10) at 46.3% (57) and ESCC (CPS ≥10) at 36.6% (45) of the specimens reclassified as negative.Abstract 81 Figure 1PD-L1 Scoring AlgorithmsThe TPS algorithm (a) is defined as the number of PD-L1 staining tumor cells divided by the total number of viable TC, multiplied by 100. The CPS algorithm (b) includes TC and IC and is defined as the number of PD-L1 staining cells (TC, lymphocytes and macrophages) divided by the total number of viable TC, multiplied by 100. In addition to TPS and CPS, QID (c) was also calculated to quantify the contribution from PD-L1 expressing IC, QID is defined as the CPS minus the TPS.Abstract 81 Table 1Agilent Tumor Bank CPS and QIDConclusionsPD-L1 IHC 22C3 pharmDx (Dako, USA) stains both TC and immune cells. Removal of the PD-L1 staining TC from the CPS algorithm reduces the number of specimens scored as positive for each indication’s respective diagnostic cut-off(s). Scoring only IC reduces the number of specimens scored as positive for each indication’s respective cutoff.