Diagnostic agreement in the evaluation of image-guided breast core needle biopsies - Results from a randomized clinical

Background: Image-guided core needle biopsies (CNBs) are commonly used as the initial sampling method for nonpalpable, mammographically detected breast lesions. Although prior studies have shown that this procedure is a highly sensitive and accurate method for the detection of breast cancer, the level of diagnostic agreement between pathologists in the analysis of CNB has not been previously studied in detail. Methods: To address this, we reviewed the pathologic findings in 2004 CNB from patients enrolled in the Radiologic Diagnostic Oncology Group 5 study, a randomized, multicenter trial designed to determine the role of CNB and fine needle aspiration biopsy in the evaluation of nonpalpable breast lesions. Slides of CNB specimens were initially diagnosed by pathologists at the 22 participating institutions (local diagnosis) and were then sent to the study pathologists for central review (central diagnosis). Local and central diagnoses were compared. Results: Overall, the central diagnosis and local diagnosis were concordant in 1925 cases (96%), indicating an excellent level of agreement by kappa statistic analysis (kappa = 0.90; 95% confidence interval 0.88-0.92). The level of agreement between local and central pathologists did not vary with the image guidance system (stereotactic mammography vs. ultrasound) or with the mammographic findings (soft tissue density vs. microcalcifications). The level of diagnostic agreement observed for CNB was comparable to that observed among 596 open surgical biopsies obtained from patients in this study and subjected to central pathology review (93% agreement; kappa = 0.89, 95% confidence interval 0.86-0.92). Conclusions: The level of diagnostic agreement in interpretation of breast CNB is extremely high among pathologists and is comparable to that seen for open surgical biopsy.