PROACTING: predicting pathological complete response to neoadjuvant chemotherapy in breast cancer from routine diagnostic histopathology biopsies with deep learning.

BACKGROUND: Invasive breast cancer patients are increasingly being treated with neoadjuvant chemotherapy; however, only a fraction of the patients respond to it completely. To prevent overtreatment, there is an urgent need for biomarkers to predict treatment response before administering the therapy. METHODS: In this retrospective study, we developed hypothesis-driven interpretable biomarkers based on deep learning, to predict the pathological complete response (pCR, i.e., the absence of tumor cells in the surgical resection specimens) to neoadjuvant chemotherapy solely using digital pathology H&E images of pre-treatment breast biopsies. Our approach consists of two steps: First, we use deep learning to characterize aspects of the tumor micro-environment by detecting mitoses and segmenting tissue into several morphology compartments including tumor, lymphocytes and stroma. Second, we derive computational biomarkers from the segmentation and detection output to encode slide-level relationships of components of the tumor microenvironment, such as tumor and mitoses, stroma, and tumor infiltrating lymphocytes (TILs). RESULTS: We developed and evaluated our method on slides from n = 721 patients from three European medical centers with triple-negative and Luminal B breast cancers and performed external independent validation on n = 126 patients from a public dataset. We report the predictive value of the investigated biomarkers for predicting pCR with areas under the receiver operating characteristic curve between 0.66 and 0.88 across the tested cohorts. CONCLUSION: The proposed computational biomarkers predict pCR, but will require more evaluation and finetuning for clinical application. Our results further corroborate the potential role of deep learning to automate TILs quantification, and their predictive value in breast cancer neoadjuvant treatment planning, along with automated mitoses quantification. We made our method publicly available to extract segmentation-based biomarkers for research purposes.

Association of DCIS size and margin status with risk of developing breast cancer post-treatment: multinational, pooled cohort study.

OBJECTIVE: To examine the association between size and margin status of ductal carcinoma in situ (DCIS) and risk of developing ipsilateral invasive breast cancer and ipsilateral DCIS after treatment, and stage and subtype of ipsilateral invasive breast cancer. DESIGN: Multinational, pooled cohort study. SETTING: Four large international cohorts. PARTICIPANTS: Patient level data on 47 695 women with a diagnosis of pure, primary DCIS between 1999 and 2017 in the Netherlands, UK, and US who underwent surgery, either breast conserving or mastectomy, often followed by radiotherapy or endocrine treatment, or both. MAIN OUTCOME MEASURES: The main outcomes were 10 year cumulative incidence of ipsilateral invasive breast cancer and ipsilateral DCIS estimated in relation to DCIS size and margin status, and adjusted hazard ratios and 95% confidence intervals, estimated using multivariable Cox proportional hazards analyses with multiple imputed data RESULTS: The 10 year cumulative incidence of ipsilateral invasive breast cancer was 3.2%. In women who underwent breast conserving surgery with or without radiotherapy, only adjusted risks for ipsilateral DCIS were significantly increased for larger DCIS (20-49 mm) compared with DCIS <20 mm (hazard ratio 1.38, 95% confidence interval 1.11 to 1.72). Risks for both ipsilateral invasive breast cancer and ipsilateral DCIS were significantly higher with involved compared with clear margins (invasive breast cancer 1.40, 1.07 to 1.83; DCIS 1.39, 1.04 to 1.87). Use of adjuvant endocrine treatment was not significantly associated with a lower risk of ipsilateral invasive breast cancer compared to treatment with breast conserving surgery only (0.86, 0.62 to 1.21). In women who received breast conserving treatment with or without radiotherapy, higher DCIS grade was not significantly associated with ipsilateral invasive breast cancer, only with a higher risk of ipsilateral DCIS (grade 1: 1.42, 1.08 to 1.87; grade 3: 2.17, 1.66 to 2.83). Higher age at diagnosis was associated with lower risk (per year) of ipsilateral DCIS (0.98, 0.97 to 0.99) but not ipsilateral invasive breast cancer (1.00, 0.99 to 1.00). Women with large DCIS (≥50 mm) more often developed stage III and IV ipsilateral invasive breast cancer compared to women with DCIS <20 mm. No such association was found between involved margins and higher stage of ipsilateral invasive breast cancer. Associations between larger DCIS and hormone receptor negative and human epidermal growth factor receptor 2 positive ipsilateral invasive breast cancer and involved margins and hormone receptor negative ipsilateral invasive breast cancer were found. CONCLUSIONS: The association of DCIS size and margin status with ipsilateral invasive breast cancer and ipsilateral DCIS was small. When these two factors were added to other known risk factors in multivariable models, clinicopathological risk factors alone were found to be limited in discriminating between low and high risk DCIS.