Inter-observer agreement in the diagnosis of breast atypical ductal hyperplasia: A systematic review

OBJECTIVE

To review the concordance between pathologists in the diagnosis of atypical ductal hyperplasia of the breast.

METHODS

Systematic review of articles registered in the PubMed database with the terms "pathologists" and "atypical ductal hyperplasia", "atypical ductal hyperplasia" and "interobserver variability", in Embase with "interobserver agreement in atypical ductal hyperplasia" and Lilacs with the terms in Spanish "patólogos" and "hiperplasia ductal atípica", without the use of filters, between 03/16/2022 and 03/26/2022 searching for articles that assess inter-observer agreement in the diagnosis of atypical ductal hyperplasia by using the kappa statistic. Review articles, conference proceedings, and commentaries were excluded.

RESULTS

With the search terms, 507 publications were found. Of these, 491 were excluded from the screening by title and abstract. Of the full reading of the remaining articles, 6 were selected for qualitative analysis. The kappa values ranged from slight agreement, kappa = 0.17, to substantial agreement, 0.69 obtained from pathologists dedicated or experienced in breast pathology and who received a prior tutorial session to review the general criteria for atypia.

CONCLUSIONS

This study reviews the reproducibility in the diagnosis of atypical ductal hyperplasia. A better concordance was obtained with a previous teaching process offered to pathologists dedicated or experienced in breast pathology.

Machine learning-based image analysis for accelerating the diagnosis of complicated preneoplastic and neoplastic ductal lesions in breast biopsy tissues

PURPOSE

Diagnosis of breast preneoplastic and neoplastic lesions is difficult due to their similar morphology in breast biopsy specimens. To diagnose these lesions, pathologists perform immunohistochemical analysis and consult with expert breast pathologists. These additional examinations are time-consuming and expensive. Artificial intelligence (AI)-based image analysis has recently improved, and may help in ordinal pathological diagnosis. Here, we showed the significance of machine learning-based image analysis of breast preneoplastic and neoplastic lesions for facilitating high-throughput diagnosis.

METHODS

Images were obtained from normal mammary glands, hyperplastic lesions, preneoplastic lesions and neoplastic lesions, such as usual ductal hyperplasia (UDH), columnar cell lesion (CCL), ductal carcinoma in situ (DCIS), and DCIS with comedo necrosis (comedo DCIS) in breast biopsy specimens. The original enhanced convoluted neural network (CNN) system was used for analyzing the pathological images.

RESULTS

The AI-based image analysis provided the following area under the curve values (AUC): normal lesion versus DCIS, 0.9902; DCIS versus comedo DCIS, 0.9942; normal lesion versus CCL, 0.9786; and UDH versus DCIS, 1.000. Multiple comparison analysis showed precision and recall scores similar to those of single comparison analysis. Based on the gradient-weighted class activation mapping (Grad-CAM) used to visualize the important regions reflecting the result of CNN analysis, the ratio of stromal tissue in the whole weighted area was significantly higher in UDH and CCL than that in DCIS.

CONCLUSIONS

These analyses may provide a more accurate and rapid pathological diagnosis of patients. Moreover, Grad-CAM identifies uncharted important histological characteristics for newer pathological findings and targets of research for understanding diseases.

[Benign proliferative breast disease with and without atypia]

In the last few years, diagnostics of high-risk breast lesions (atypical ductal hyperplasia [ADH], flat epithelial atypia [FEA], lobular neoplasia: atypical lobular hyperplasia [ALH], lobular carcinoma in situ [LCIS], radial scar [RS], usual ductal hyperplasia [UDH], adenosis, sclerosing adenosis [SA], papillary breast lesions, mucocele-like lesion [MLL]) have increased with the growing number of breast percutaneous biopsies. The management of these lesions is highly conditioned by the enlarged risk of breast cancer combined with either an increased probability of finding cancer after surgery, either a possible malignant transformation (in situ or invasive cancer), or an increased probability of developing cancer on the long range. An overview of the literature reports grade C recommendations concerning the management and follow-up of these lesions: in case of ADH, FEA, ALH, LCIS, RS, MLL with atypia, diagnosed on percutaneous biopsies: surgical excision is recommended; in case of a diagnostic based on vacuum-assisted core biopsy with complete disappearance of radiological signal for FEA or RS without atypia: surgical abstention is a valid alternative approved by multidisciplinary meeting. In case of ALH (incidental finding) associated with benign lesion responsible of radiological signal: abstention may be proposed; in case of UDH, adenosis, MLL without atypia, diagnosed on percutaneous biopsies: the concordance of radiology and histopathology findings must be ensured. No data is available to recommend surgery; in case of non-in sano resection for ADH, FEA, ALH, LCIS (except pleomorphic type), RS, MLL: surgery does not seem to be necessary; in case of previous ADH, ALH, LCIS: a specific follow-up is recommended in accordance with HAS's recommendations. In case of FEA and RS or MLL combined with atypia, little data are yet available to differ the management from others lesions with atypia; in case of UDH, usual sclerosing adenosis, RS without atypia, fibro cystic disease: no specific follow-up is recommended in agreement with HAS's recommendations.

Computational pathology to discriminate benign from malignant intraductal proliferations of the breast

The categorization of intraductal proliferative lesions of the breast based on routine light microscopic examination of histopathologic sections is in many cases challenging, even for experienced pathologists. The development of computational tools to aid pathologists in the characterization of these lesions would have great diagnostic and clinical value. As a first step to address this issue, we evaluated the ability of computational image analysis to accurately classify DCIS and UDH and to stratify nuclear grade within DCIS. Using 116 breast biopsies diagnosed as DCIS or UDH from the Massachusetts General Hospital (MGH), we developed a computational method to extract 392 features corresponding to the mean and standard deviation in nuclear size and shape, intensity, and texture across 8 color channels. We used L1-regularized logistic regression to build classification models to discriminate DCIS from UDH. The top-performing model contained 22 active features and achieved an AUC of 0.95 in cross-validation on the MGH data-set. We applied this model to an external validation set of 51 breast biopsies diagnosed as DCIS or UDH from the Beth Israel Deaconess Medical Center, and the model achieved an AUC of 0.86. The top-performing model contained active features from all color-spaces and from the three classes of features (morphology, intensity, and texture), suggesting the value of each for prediction. We built models to stratify grade within DCIS and obtained strong performance for stratifying low nuclear grade vs. high nuclear grade DCIS (AUC = 0.98 in cross-validation) with only moderate performance for discriminating low nuclear grade vs. intermediate nuclear grade and intermediate nuclear grade vs. high nuclear grade DCIS (AUC = 0.83 and 0.69, respectively). These data show that computational pathology models can robustly discriminate benign from malignant intraductal proliferative lesions of the breast and may aid pathologists in the diagnosis and classification of these lesions.

All atypia diagnosed at stereotactic vacuum-assisted breast biopsy do not need surgical excision

The necessity of excision is debatable when atypia are diagnosed at stereotactic vacuum-assisted breast biopsy (microbiopsy). Among the 287 surgical excisions performed at Institut Bergonié from 1999 to 2009, we selected a case-control study group of 151 excisions; 52 involving all the diagnosed cancers and 99 randomly selected among the 235 excisions without cancer, following atypical microbiopsy (24 flat epithelial atypia; 50 atypical ductal hyperplasia; 14 lobular neoplasia; 63 mixed lesions). Mammographical calcification (type, extension, complete removal) and histological criteria of epithelial atypia (type, number of foci, size/extension), topography and microcalcification extension at microbiopsy were compared according to the presence or absence of cancer at excision. Factors associated with cancer at excision were Breast Imaging Reporting and Data System (BI-RADS5) lesions, large and/or multiple foci of mammographical calcifications, histological type, number, size and extension of atypical foci. Flat epithelial atypia alone was never associated with cancer at excision. BI-RADS5, atypical ductal hyperplasia (alone or predominant) and >3 foci of atypia were identified as independent pejorative factors. There was never any cancer at excision when these pejorative factors were absent (n=31). Presence of one (n=59), two (n=23) or three (n=14) factors was associated with cancer in 24, 15 and 13 cases with an odds ratio=5.8 (95% CI: 3-11.2) for each additional factor. We recommend that mammographical data and histological characteristics be taken into account in the decision-making process after diagnosis of atypia on microbiopsy. With experienced senologists and strict histological criteria, some patients could be spared surgery resulting in significant patient, financial and time advantages.

Atypical ductal hyperplasia: interobserver and intraobserver variability

Interobserver reproducibility in the diagnosis of benign intraductal proliferative lesions has been poor. The aims of the study were to investigate the inter- and intraobserver variability and the impact of the addition of an immunostain for high- and low-molecular weight keratins on the variability. Nine pathologists reviewed 81 cases of breast proliferative lesions in three stages and assigned each of the lesions to one of the following three diagnoses: usual ductal hyperplasia, atypical ductal hyperplasia and ductal carcinoma in situ. Hematoxylin and eosin slides and corresponding slides stained with ADH-5 cocktail (cytokeratins (CK) 5, 14. 7, 18 and p63) by immunohistochemistry were evaluated. Concordance was evaluated at each stage of the study. The interobserver agreement among the nine pathologists for diagnosing the 81 proliferative breast lesions was fair (κ-value=0.34). The intraobserver κ-value ranged from 0.56 to 0.88 (moderate to strong). Complete agreement among nine pathologists was achieved in only nine (11%) cases, at least eight agreed in 20 (25%) cases and seven or more agreed in 38 (47%) cases. Following immunohistochemical stain, a significant improvement in the interobserver concordance (overall κ-value=0.50) was observed (P=0.015). There was a significant reduction in the total number of atypical ductal hyperplasia diagnosis made by nine pathologists after the use of ADH-5 immunostain. Atypical ductal hyperplasia still remains a diagnostic dilemma with wide variation in both inter- and intraobserver reproducibility among pathologists. The addition of an immunohistochemical stain led to a significant improvement in the concordance rate. More importantly, there was an 8% decrease in the number of lesions classified as atypical ductal hyperplasia in favor of usual hyperplasia; in clinical practice, this could lead to a decrease in the number of surgeries carried out for intraductal proliferative lesions.

Differential Diagnosis of Proliferative Breast Lesions

The correct diagnosis of proliferations within the mammary terminal duct-lobular unit has paramount prognostic and therapeutic implications. Occasionally, the differential diagnosis of compact florid hyperplasia, atypical ductal hyperplasia, and low-grade ductal carcinoma in situ can be quite challenging, with seeming morphologic overlap. This article presents s conceptual and practical understanding of these processes and their impact on subsequent cancer risk, with the intention of assisting the practicing pathologist render accurate and clinically relevant diagnoses for this frequently encountered set of mammary epithelial lesions.

Columnar cell lesions of the breast: an update and significance on core biopsy

Columnar cell lesions of the breast refer to the morphological spectrum of alterations of the epithelial lining of variably dilated acini of the terminal duct lobular unit (TDLU), often related to secretions and calcifications. After decades of varied terminologies, the term of 'flat epithelial atypia' by the World Health Organization (WHO) consensus group encompasses the part of the spectrum where columnar cell change or columnar cell hyperplasia acquires low grade cytological atypia, merging with atypical ductal hyperplasia and low grade ductal carcinoma in situ. Its association with low grade invasive carcinoma and lobular neoplasia, whether by proximity to these lesions, or by similar molecular expressions, has prompted greater scrutiny into its clinical significance. Although recent literature attempts to refine the term 'flat epithelial atypia', the applicability of its morphological criteria in routine diagnostic practice remains to be seen, and interobserver variability is highly possible. This poses even greater challenges especially in limited samples of breast tissue, such as in core biopsies, for pre-operative decision-making. The purpose of this review is to elucidate evolving clinical and diagnostically relevant principles that surround and influence the significance of this still controversial entity, especially when discovered on core biopsy in the initial phase of breast diagnosis and management.