A Disease—Focused Quality Improvement Process Is an Effective Tool for Improvement of Inter—Observer Reproducibility in HER2 Immunohistochemistry Scoring for Biopsies from Gastric and Esophageal Adenocarcinomas

Abstract

Introduction: Over—expression of the HER2 protein is found in 15—20% of patients with gastric and esophageal adenocarcinomas (GEAC). The Clinical “ToGA” (Trastuzumab® in Gastric cancer) trial demonstrated a clear survival benefit of adding Trastuzumab® to the treatment of patients with advanced GEAC and HER2 over—expression. To ensure uniformity of HER2 testing by immunohistochemistry (IHC) and in—situ hybridization (ISH), the College of American Pathologists issued guidelines for precise grading of HER2 IHC; however, inter—observer IHC semi—quantitative scoring variability tends to be high. This study evaluated the utility of the Inform Diagnostics “Disease—Focused Project” (DFP) quality program in improving inter—observer reproducibility of HER2 IHC scoring. Methods: The IDEA (Inform Diagnostics Electronic Archives) database was searched for cases with GEAC that underwent HER2 IHC testing in 2016, and 70 consecutive cases were reviewed to identify the most common challenges for scoring, e.g., borderline staining, marked heterogeneous staining, intense luminal staining, intense cytoplasmic staining, and crush artifact. Ten H&E and HER2 IHC—stained slides from cases representative of these challenges were then selected and circulated among 24 GI pathologists for blinded scoring of HER2 IHC staining on a scale from 0 to 3. After staining scores were submitted, formal review of pertinent literature and group microscope sessions, highlighting common difficulties in HER2 IHC assessment, were performed. After a delay of 2 weeks, participants were asked to rescore HER2 staining on the original 10 cases, applying knowledge gained from DFP. Pre— and post—DFP inter—observer reproducibility of HER2 IHC scoring was assessed by calculating Fleiss ? values. Results: Inter—observer reproducibility of HER IHC scoring measured by Fleiss ? values of pre— and post—DFP was 0.30 (95% confidence interval 0.27—0.33) and 0.67, (95% confidence interval 0.64—0.7) respectively (p = 0.007). The pre—DFP correlation kappa value (0.30) is considered “fair” while post—DFP value (0.67) is considered “substantial,” 2 categories higher. Conclusion: Disease—Focused Projects at Inform Diagnostics have been shown to effectively reduce variability in various applications of anatomic pathology. This DFP represents an effective tool to improve accuracy in HER2 IHC scoring, which is essential for selecting GEAC patients who would benefit from adding Trastuzumab® (Herceptin) to their therapy.