Quantcast

Digital pathology for primary diagnosis of screen-detected breast lesions - experimental data, validation and experience from 4 centres.

Research paper by Bethany B Williams, Andrew A Hanby, Rebecca R Millican-Slater, Eldo E Verghese, Anju A Nijhawan, Imogen I Wilson, Justinas J Besusparis, David D Clark, David D Snead, Emad E Rakha, Darren D Treanor

Indexed on: 31 Jan '20Published on: 30 Jan '20Published in: Histopathology



Abstract

The rate of deployment of digital pathology (DP) systems for primary diagnosis in the UK is accelerating. The flexibility and resilience of digital slides versus standard glass slides could be of great benefit in the NHS breast screening programme (NHSBSP). This study aims to document the safety and benefits of DP for preoperative tissue diagnosis of screen detected breast lesions. Concordance data for glass and digital slides of the same cases from 4 sites, were subjected to detailed concordance:discordance analysis. A literature review of DP in the primary diagnosis of breast lesions is presented making this the most comprehensive synthesis of digital breast cancer histopathological diagnostic data to date. Detailed concordance analysis of experimental data from 2 histopathology departments reveals clinical concordance rates for breast biopsies of 96% (216/225) and 99.6% (249/250). Data from direct comparison validation studies in 2 histopathology departments, utilizing the protocol recommended by Royal College of Pathologists found concordance rates for breast histology cases of 99.4% (180/181) and 99.0% (887/896). An intraobserver variation study for glass versus digital slides for difficult cases from the NHSBSP yielded a kappa statistic of 0.80 indicating excellent agreement. Discordances encountered in the studies most frequently concerned discrepancies in grading attributable to mitotic count scoring and identification of weddelite. The experience of 4 histopathology laboratories, and our review of pre-existing literature suggest that DP is safe for the primary diagnosis of NHSBSP breast histology specimens, and does not increase the risk of misclassification. © 2020 John Wiley & Sons Ltd.