155P - Artificial Intelligence (AI) - powered human epidermal growth factor receptor-2 (HER2) and tumor-infiltrating lymphocytes (TIL) analysis for HER2-positive early breast cancer patients treated with HER2-targeted neoadjuvant chemotherapy (NAC)

Background

HER2 is an important predictive biomarker of HER2-targeted therapies for breast cancer. Reliable HER2 testing is essential for clinical practice, but immunohistochemistry (IHC) staining results are susceptible to considerable interobserver variations.

Methods

Whole-slide images (WSI) of pretreatment HER2 IHC- and H&E-stained slides, and clinicopathological data were obtained from stage I – III, HER2-positive breast cancer patients treated with NAC including anti-HER2 agents at Samsung Medical Center, in Seoul, Korea, between 2009 and 2017. Lunit SCOPE HER2 is an AI-powered HER2 IHC analyzer, developed from 1,133 HER2 IHC stained WSI of breast cancer, annotated by 113 board-certified pathologists. It classifies tumor cells by HER2 staining intensity (H0, H1, H2 or H3) and quantifies the cells in each staining category. Also, it calculates AI HER2 IHC categories in accordance with the latest ASCO/CAP HER2 IHC evaluation algorithm. Lunit SCOPE IO is a separate AI-powered WSI analyzer that identifies and quantifies TIL within the cancer epithelium or stroma from H&E-stained WSI.

Results

In a total of 254 patients, the AI-scored HER2 categories were 1+ in 3 (1.2%), 2+ in 12 (4.7%), and 3+ in 239 (91.3%). The overall concordance of HER2 categories between AI and pathologists was 93.7%. Pathologic complete response (pCR, ypT0N0 or ypTisN0) was observed in 117 (46.1%) cases. The patients in pCR had a significantly higher proportion of tumor cells in H3 intensity category by AI (84.7% vs. 76.7%, p = 0.010 in all patients; 87.7% vs. 81.0%, p = 0.019 in HER2 IHC 3+ by pathologists), and significantly higher densities of stromal and intratumoral TILs (Table). Table: 155P

Conclusions

AI-powered automated HER2 scoring and TIL analysis can provide additional information for response prediction in HER2-positive early breast cancer patients treated with NAC.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Lunit Inc.

Funding

Lunit Inc.

Disclosure

Y. Lim: Financial Interests, Personal, Full or part-time Employment: Lunit Inc., IMBdx. S.I. Cho: Financial Interests, Personal, Full or part-time Employment: Lunit Inc.; Financial Interests, Personal, Stocks/Shares: Lunit Inc. S. Kim, G. Park, S. Song, H. Song, S. Park, W. Jung, K. Paeng: Financial Interests, Personal, Full or part-time Employment: Lunit Inc. M. Ma: Financial Interests, Personal, Full or part-time Employment: Lunit; Financial Interests, Personal, Stocks/Shares: Lunit. C. Ock: Financial Interests, Personal, Full or part-time Employment: Lunit Inc.; Financial Interests, Personal, Invited Speaker: Ybiologics; Financial Interests, Personal, Stocks/Shares: Lunit Inc., Ybiologics. All other authors have declared no conflicts of interest.