Whole-Slide Telepathology Grading of Prostate Cancer in Biopsies Using International Society of Urological Pathology Criteria: Interobserver and Intraobserver Concordance and Implications for Active Surveillance.

Abstract

Subjectivity of prostate cancer (PCa) grading affects biomarker investigations, assessments of imaging, and treatment decisions, especially for active surveillance. We conducted the first telepathology concordance study of PCa needle biopsy grading using contemporary International Society of Urologic Pathology guidelines. Whole slide images (n = 120) of PCa biopsy cases across grade groups (GGs) were examined by 11 urologic pathologists. Forty cases were blindly regraded, oversampling cases with lower agreement on first review. Pathologists provided GG, percentage and type of pattern Gleason 4, and tumor length. Consensus for a case was defined as GG agreement by ≥ 7 pathologists. GG agreement among and within pathologists was measured using model-based weighted Kappa statistics (κ_w). Consensus was achieved on 86/120 (72%) cases. Of 34 cases not achieving consensus, most were GG2 vs GG1 (41%) or GG2 vs GG3 (38%). The inter-rater κ_w was 0.34 (95% CI 0.28, 0.39) indicating fair agreement. Forty regraded cases had an intra-rater κ_w of 0.49 (95% CI 0.46, 0.53; moderate agreement). Sixty-four percent of the repeated grades agreed. Of the repeated ratings that did not agree, 38% involved GG1 vs GG2. Of the GG1 vs GG2 paired repeated ratings, 77% reported 1% to 10% pattern 4, and indicated common observation was "poorly formed glands." Inter-rater and intra-rater variability in GG is significant, particularly discerning GG1 from GG2 with ≤ 10% grade 4. These observations have profound implications for treatment decisions (surveillance vs definitive therapy) and studies of new biomarkers and imaging of prostate cancer.