Augmented interpretation of HER2, ER, and PR in breast cancer by artificial intelligence analyzer: enhancing interobserver agreement through a reader study of 201 cases

Abstract

Background: Accurate classification of breast cancer molecular subtypes is crucial in determining treatment strategies and predicting clinical outcomes. This classification largely depends on the assessment of human epidermal growth factor receptor 2 (HER2), estrogen receptor (ER), and progesterone receptor (PR) status. However, variability in interpretation among pathologists pose challenges to the accuracy of this classification. This study evaluates the role of artificial intelligence (AI) in enhancing the consistency of these evaluations.

Methods: AI-powered HER2 and ER/PR analyzers, consisting of cell and tissue models, were developed using 1,259 HER2, 744 ER, and 466 PR-stained immunohistochemistry (IHC) whole-slide images of breast cancer. External validation cohort comprising HER2, ER, and PR IHCs of 201 breast cancer cases were analyzed with these AI-powered analyzers. Three board-certified pathologists independently assessed these cases without AI annotation. Then, cases with differing interpretations between pathologists and the AI analyzer were revisited with AI assistance, focusing on evaluating the influence of AI assistance on the concordance among pathologists during the revised evaluation compared to the initial assessment.

Results: Reevaluation was required in 61 (30.3%), 42 (20.9%), and 80 (39.8%) of HER2, in 15 (7.5%), 17 (8.5%), and 11 (5.5%) of ER, and in 26 (12.9%), 24 (11.9%), and 28 (13.9%) of PR evaluations by the pathologists, respectively. Compared to initial interpretations, the assistance of AI led to a notable increase in the agreement among three pathologists on the status of HER2 (from 49.3 to 74.1%, p < 0.001), ER (from 93.0 to 96.5%, p = 0.096), and PR (from 84.6 to 91.5%, p = 0.006). This improvement was especially evident in cases of HER2 2+ and 1+, where the concordance significantly increased from 46.2 to 68.4% and from 26.5 to 70.7%, respectively. Consequently, a refinement in the classification of breast cancer molecular subtypes (from 58.2 to 78.6%, p < 0.001) was achieved with AI assistance.

Conclusions: This study underscores the significant role of AI analyzers in improving pathologists' concordance in the classification of breast cancer molecular subtypes.

Keywords: Artificial intelligence (AI); Breast cancer; Concordance; Digital pathology; Estrogen receptor (ER); Human epidermal growth factor receptor 2 (HER2); Progesterone receptor (PR); Whole-slide image (WSI).

Fig. 1

A schematic flow of the reader study (AI: artificial intelligence, ER: estrogen receptor, HER2: human epidermal growth factor receptor 2, PR: progesterone receptor)

Fig. 2

Concordance among pathologists in HER2 (human epidermal growth factor receptor 2) dataset (A), ER (estrogen receptor) dataset (B), and PR (progesterone receptor) dataset (C). Concordance between the consensus of pathologists and the AI (artificial intelligence) analyzer in HER2 dataset (D), ER dataset (E), and PR dataset (F)

Fig. 3

A Proportion of revisited and revised cases by artificial intelligence (AI) analyzer in Pathologist 1 (P1), Pathologist 2 (P2), and Pathologist 3 (P3). Initial and revised pathologists’ consensus of HER2 (human epidermal growth factor receptor 2) in All cases (B) or revisited cases only (C). Initial and revised pathologists’ consensus of ER (estrogen receptor) in All cases (D) or revisited cases only (E). Initial and revised pathologists’ consensus of PR (progesterone receptor) in All cases (F) or revisited cases only (G)

Fig. 4

A Initial and revised pathologists’ consensus of subtype. B Initial and revised concordance rates of subtype among pathologists (ER: estrogen receptor, HER2: human epidermal growth factor receptor 2, PR: progesterone receptor)

Fig. 5

Rate of revising by pathologists according to the artificial intelligence (AI) analyzer’s results, when one, two, or all three pathologists revisited (ER: estrogen receptor, HER2: human epidermal growth factor receptor 2, PR: progesterone receptor)

Fig. 6

A Carcinoma in situ areas containing HER2 (human epidermal growth factor receptor 2)-stained cells were classified as cancer area (CA) by HER2 analyzer (bar: 200 μm). B ER (estrogen receptor)/PR (progesterone receptor) analyzer recognized the inked area at the margin of the tissue as a CA with ER-positive tumor cells (bar left: 5 mm, right: 200 μm). C ER/PR analyzer barely caught the CA through the entire area of PR-stained slide (bar left: 5 mm, right: 100 μm). D In a PR-stained slide, pathologists focused the visible clusters of positive tumor cells at low magnification and classified them as low positive all together even after revision. In contrast, ER/PR analyzer counted all the tumor cells and interpreted them as negative (bar left: 5 mm, right: 100 μm)