NRK-ABMIL: Subtle Metastatic Deposits Detection for Predicting Lymph Node Metastasis in Breast Cancer Whole-Slide Images.

The early diagnosis of lymph node metastasis in breast cancer is essential for enhancing treatment outcomes and overall prognosis. Unfortunately, pathologists often fail to identify small or subtle metastatic deposits, leading them to rely on cytokeratin stains for improved detection, although this approach is not without its flaws. To address the need for early detection, multiple-instance learning (MIL) has emerged as the preferred deep learning method for automatic tumor detection on whole slide images (WSIs). However, existing methods often fail to identify some small lesions due to insufficient attention to small regions. Attention-based multiple-instance learning (ABMIL)-based methods can be particularly problematic because they may focus too much on normal regions, leaving insufficient attention for small-tumor lesions. In this paper, we propose a new ABMIL-based model called normal representative keyset ABMIL (NRK-ABMIL), which addresseses this issue by adjusting the attention mechanism to give more attention to lesions. To accomplish this, the NRK-ABMIL creates an optimal keyset of normal patch embeddings called the normal representative keyset (NRK). The NRK roughly represents the underlying distribution of all normal patch embeddings and is used to modify the attention mechanism of the ABMIL. We evaluated NRK-ABMIL on the publicly available Camelyon16 and Camelyon17 datasets and found that it outperformed existing state-of-the-art methods in accurately identifying small tumor lesions that may spread over a few patches. Additionally, the NRK-ABMIL also performed exceptionally well in identifying medium/large tumor lesions.

BCR-Net: A deep learning framework to predict breast cancer recurrence from histopathology images.

Breast cancer is the most common malignancy in women, with over 40,000 deaths annually in the United States alone. Clinicians often rely on the breast cancer recurrence score, Oncotype DX (ODX), for risk stratification of breast cancer patients, by using ODX as a guide for personalized therapy. However, ODX and similar gene assays are expensive, time-consuming, and tissue destructive. Therefore, developing an AI-based ODX prediction model that identifies patients who will benefit from chemotherapy in the same way that ODX does would give a low-cost alternative to the genomic test. To overcome this problem, we developed a deep learning framework, Breast Cancer Recurrence Network (BCR-Net), which automatically predicts ODX recurrence risk from histopathology slides. Our proposed framework has two steps. First, it intelligently samples discriminative features from whole-slide histopathology images of breast cancer patients. Then, it automatically weights all features through a multiple instance learning model to predict the recurrence score at the slide level. On a dataset of H&E and Ki67 breast cancer resection whole slides images (WSIs) from 99 anonymized patients, the proposed framework achieved an overall AUC of 0.775 (68.9% and 71.1% accuracies for low and high risk) on H&E WSIs and overall AUC of 0.811 (80.8% and 79.2% accuracies for low and high risk) on Ki67 WSIs of breast cancer patients. Our findings provide strong evidence for automatically risk-stratify patients with a high degree of confidence. Our experiments reveal that the BCR-Net outperforms the state-of-the-art WSI classification models. Moreover, BCR-Net is highly efficient with low computational needs, making it practical to deploy in limited computational settings.

Residual breast tissue after robot-assisted nipple sparing mastectomy.

INTRODUCTION: While the long-term oncologic safety of robot-assisted nipple sparing mastectomy (RNSM) remains to be elucidated, histologically detected residual breast tissue (RBT) can be a surrogate for oncologically sound mastectomy. The objective of this study is to determine the presence of RBT after RNSM. METHODS: Between August 2019-January 2020, we completed 5 cadaveric RNSMs. Full thickness biopsies from the mastectomy skin flap were obtained from predefined locations radially around the mastectomy skin envelop and nipple areolar complex to histologically evaluate for RBT. RESULTS: The first case was not technically feasible due to inability to obtain adequate insufflation. Five mastectomy flaps were analyzable. The average mastectomy flap thickness was 2.3 mm (range 2-3 mm) and the average specimen weight was 382.72 g (range 146.9-558.3 g). Of 70 total biopsies, RBT was detected in 11 (15.7%) biopsies. Most common location for RBT was in the nipple-areolar complex, with no RBT detected from the peripheral skin flaps. CONCLUSIONS: In this cadaveric study, RNSM is feasible leaving minimal RBT on the mastectomy flap. The most common location for RBT is in the periareolar location consistent with previous published findings after open NSM. Clinical studies are underway to evaluate the safety of RNSM.

Impact of the 2018 ASCO/CAP HER2 Guideline Focused Update.

OBJECTIVES: The 2018 American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) human epidermal growth factor receptor 2 (HER2) guideline focused update revises the HER2 scoring criteria. We evaluated the impact on HER2 rates in breast carcinoma diagnosed at our center. METHODS: In a retrospective series of breast core biopsies with invasive carcinoma diagnosed between 2014 and 2017 (n = 1,350), HER2 status was classified according to 2013 and 2018 ASCO/CAP guidelines and changes in HER2 status identified. RESULTS: The 2018 guidelines reclassified the HER2 status of 6% of patients. Most changed from HER2 equivocal status (equivocal by immunohistochemistry and fluorescence in situ hybridization under the 2013 guidelines) to HER2-negative status (2018 guidelines). The HER2-positive rate decreased by 0.4%. CONCLUSIONS: The 2018 guidelines decrease the rate of HER2 equivocal and positive breast cancer and reduce repeat HER2 testing on excision specimens. Approximately 0.4% of patients will become newly ineligible for anti-HER2 therapy.

A combination of GATA3 and SOX10 is useful for the diagnosis of metastatic triple-negative breast cancer.

In metastatic breast cancer (MBC), it can be difficult to establish the origin if the primary tumor is triple negative or if there is a loss of biomarker expression. SOX10 expression has been reported in primary triple-negative breast cancer but is poorly studied in metastatic lesions. In this study, the diagnostic utility of a panel of SOX10, GATA3, and androgen receptor (AR) in MBC negative for estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 was evaluated and compared with the expression of these markers in the matched primary breast cancer. In a series of 57 triple-negative MBCs, 82% were positive for GATA3, 58% for SOX10, and 25% for AR. Nearly all MBCs (95%) were positive for either GATA3 or SOX10, with 46% dual positive and 5% of cases negative for both markers. Most GATA3-negative MBC cases were SOX10 positive (70%). AR expression was only seen in GATA3-positive MBC (25%) and was significantly more frequent in SOX10-negative MBC (48%) versus SOX10-positive MBC (9%; P = .001). Concordance for GATA3, SOX10, and AR between the primary and metastasis was 89%, 88%, and 80%, respectively. Although GATA3 is a more sensitive lineage marker than SOX10 in MBC, SOX10 is a useful adjunct because it is positive in most GATA3-negative breast metastases. Using both GATA3 and SOX10 is recommended for confirming breast as the site of origin in metastases that lack estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 expression, whereas the addition of AR is not helpful.

HER2 equivocal breast cancer that is positive by alternative probe HER2 FISH are classified as HER2 negative by Oncotype DX.

In breast cancer, human epidermal growth factor receptor 2 (HER2) status is determined by immunohistochemistry (IHC) and/or in situ hybridization. Oncotype DX also reports HER2 status by an rt-PCR-based assay. Assay concordance between IHC and fluorescent in situ hybridization (FISH) (including alternative probe HER2 FISH) vs Oncotype DX HER2 rt-PCR has not been described in the post-2013 American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) HER2 Guideline revision setting. We performed a retrospective review of HER2 equivocal invasive breast carcinoma from 2014 to 2016 with the Oncotype DX HER2 result. Fifteen patients with HER2 equivocal invasive breast cancer had Oncotype DX performed. Of these, 13 underwent alternative probe HER2 FISH yielding 4 negative, 6 equivocal and 3 positive results. All 15 cases were classified as HER2 negative by Oncotype DX rt-PCR, including the three cases which were positive by alternative probe HER2 FISH, yielding a discordance rate for Oncotype DX rt-PCR HER2 of 20% (3/15). All three patients with HER2-positive breast cancer on the basis of alternative probe HER2 FISH received anti-HER2-targeted therapy. Treatment decisions based on HER2 status should utilize the IHC/FISH result as Oncotype DX results may incorrectly disqualify some patients from being eligible for anti-HER2 therapy based on the current 2013 ASCO/CAP HER2 Guidelines.