Diagnostic concordance of breast pathologists: lessons from the National Health Service Breast Screening Programme Pathology External Quality Assurance Scheme.

Aims: Previous concordance studies examining accuracy of breast diagnosis by pathologists, typically targeting difficult, histologically challenging breast lesions using artificial and restrictive environments, have reported aberrantly high levels of diagnostic discordance. The results of these studies may be misinterpreted by non-pathologists and raise concerns relating to routine practice. This study aims to assess the diagnostic agreement among UK breast pathologists.

Methods: Two hundred and forty consecutive breast lesions, submitted by participants from their routine practice, included in the UK National Health Service Breast Screening Programme (NHSBSP) breast pathology EQA scheme during the last 10 years were reviewed. An average of approximately 600 participants viewed each case. Data on diagnostic categories (benign, atypical, in-situ malignant and invasive malignant) were collected. In this study, benign and atypical diagnoses were grouped together.

Results: The overall diagnostic agreement level was in the almost perfect range. Thirty-five cases (14.6%) showed diagnostic concordance of ≤95%. Reasons for discordance included one or more of: (1) scheme methodology limitations such as: (i) miscoding of certain lesions (e.g. phyllodes tumours and lobular neoplasia) (n = 7) and (ii) variable representation of the index lesion on glass slides (n = 18); and (2) diagnostically challenging cases that may be interpreted more easily using immunohistochemistry (n = 28). These latter included benign and malignant papillary lesions (n = 12), complex sclerosing lesions (n = 7), intraductal epithelial proliferative lesions (n = 6) and an unusual special tumour type (n = 1). Further review identified pathologists' misinterpretation in 13 cases (5.4%), with an average discordance rate of only 4.2%.

Conclusions: The performance of breast pathologists is high. Exclusion of the effect of the scheme methodology limitations highlights further the high performance rate and identifies true diagnostically challenging entities. These difficult cases may benefit from additional diagnostic work-up and second opinions.