No abstract

BackgroundCanAssist-Breast is an immunohistochemistry based test that predicts risk of distant recurrence in early-stage hormone receptor positive breast cancer patients within first five years of diagnosis. Immunohistochemistry gradings for 5 biomarkers (CD44, ABCC4, ABCC11, N-Cadherin and pan-Cadherins) and 3 clinical parameters (tumor size, tumor grade and node status) of 298 patient cohort were used to develop a machine learning based statistical algorithm. The algorithm generates a risk score based on which patients are stratified into two groups, low- or high-risk for recurrence. The aim of the current study is to demonstrate the analytical performance with respect to repeatability and reproducibility of CanAssist-Breast.MethodsAll potential sources of variation in CanAssist-Breast testing involving operator, run and observer that could affect the immunohistochemistry performance were tested using appropriate statistical analysis methods for each of the CanAssist-Breast biomarkers using a total 309 samples. The cumulative effect of these variations in the immunohistochemistry gradings on the generation of CanAssist-Breast risk score and risk category were also evaluated. Intra-class Correlation Coefficient, Bland Altman plots and pair-wise agreement were performed to establish concordance on IHC gradings, risk score and risk categorization respectively.ResultsCanAssist-Breast test exhibited high levels of concordance on immunohistochemistry gradings for all biomarkers with Intra-class Correlation Coefficient of ≥0.75 across all reproducibility and repeatability experiments. Bland-Altman plots demonstrated that agreement on risk scores between the comparators was within acceptable limits. We also observed > 90% agreement on risk categorization (low- or high-risk) across all variables tested.ConclusionsThe extensive analytical validation data for the CanAssist-Breast test, evaluating immunohistochemistry performance, risk score generation and risk categorization showed excellent agreement across variables, demonstrating that the test is robust.Electronic supplementary materialThe online version of this article (10.1186/s12885-019-5443-5) contains supplementary material, which is available to authorized users.

Accurate recurrence risk assessment in hormone receptor positive, HER2/neu negative breast cancer is critical to plan precise therapy. CanAssist Breast (CAB) assesses recurrence risk based on tumor biology using artificial intelligence-based approach. We report CAB risk assessment correlating with disease outcomes in multiple clinically high- and low-risk subgroups. In this retrospective cohort of 925 patients [median age-54 (22–86)] CAB had hazard ratio (HR) of 3 (1.83–5.21) and 2.5 (1.45–4.29), P = 0.0009) in univariate and multivariate analysis. CAB's HR in sub-groups with the other determinants of outcome, T2 (HR: 2.79 (1.49–5.25), P = 0.0001); age [< 50 (HR: 3.14 (1.39–7), P = 0.0008)]. Besides application in node-negative patients, CAB's HR was 2.45 (1.34–4.47), P = 0.0023) in node-positive patients. In clinically low-risk patients (N0 tumors up to 5 cms) (HR: 2.48 (0.79–7.8), P = 0.03) and with luminal-A characteristics (HR: 4.54 (1–19.75), P = 0.004), CAB identified >16% as high-risk with recurrence rates of up to 12%. In clinically high-risk patients (T2N1 tumors (HR: 2.65 (1.31–5.36), P = 0.003; low-risk DMFS: 92.66 ± 1.88) and in women with luminal-B characteristics (HR: 3.24; (1.69–6.22), P < 0.0001; low-risk DMFS: 93.34 ± 1.34)), CAB identified >64% as low-risk. Thus, CAB prognostication was significant in women with clinically low- and high-risk disease. The data imply the use of CAB for providing helpful information to stratify tumors based on biology incorporated with clinical features for Indian patients, which can be extrapolated to regions with similarly characterized patients, South-East Asia.

Background CanAssist Breast (CAB) is a prognostic test for early stage hormone receptor‐positive (HR+), human epidermal growth factor receptor 2 negative (HER2−) breast cancer patients, validated on Indian and Caucasian patients. The 21‐gene signature Oncotype DX (ODX) is the most widely used commercially available breast cancer prognostic test. In the current study, risk stratification of CAB is compared with that done with ODX along with the respective outcomes of these patients. Methods A cohort of 109 early stage breast cancer patients who had previously taken the ODX test were retested with CAB, and the results respectively compared with old cut‐offs of ODX as well as cut‐offs suggested by TAILORx, a prospective randomized trial of ODX. Distant metastasis‐free survival after 5 years was taken as the end point. Results CanAssist Breast stratified 83.5% of the cohort into low‐risk and 16.5% into high‐risk. With the TAILORx cut‐offs, ODX stratified the cohort into 89.9% low‐risk and 10.1% into high‐risk. The low, intermediate, and high‐risk groups with ODX old cut‐offs were 62.4%, 31.2%, and 6.4%, respectively. The overall concordance of CAB with ODX using both cut‐offs is 75%‐76%, with ~82%‐83% concordance in the low‐risk category of these tests. The NPV of the low‐risk category of CAB was 93.4%, and of ODX with TAILORx cut‐offs was 91.8% and 89.7% with old cut‐offs. Conclusions Compared to the concordance reported for other tests, CAB shows high concordance with ODX, and in addition shows comparable performance in the patient outcomes in this cohort. CAB is thus an excellent and cost‐effective alternative to ODX.

541 Background: Treatment decisions for early stage HR+/HER2neu- breast cancer patients in the West routinely depend on prognostic tests that predict risk of recurrence. However, such tests are rarely used in Asia due to prohibitive costs and lack of validation data on Asian patients. Chemotherapy is thus often a default treatment leading to physiological and financial toxicity. To address these, we have developed CanAssist Breast (CAB) as an affordable IHC-based prognostic test, retrospectively validated on ~1400 patients, 63% South Asians and rest Caucasians. To date CAB has been prescribed by 180+ doctors across 30 cities in India for ~600 patients in clinics, enabling personalized treatment decisions. Methods: Primary surgical FFPE blocks and clinical follow-up data were obtained from hospitals. GraphPad Prism and MedCalc were respectively used for Kaplan-Meier survival analyses and Cox logistic regression to calculate hazard ratios. Results: The median age of diagnosis in the validation cohort was 56 years, 63% patients with stage II disease and 60% node negative tumors. Distant Metastasis Free Survival (DMFS) in the low-risk category of the validation cohort was 95%, and 84% in high-risk (P < 0.0001). Similar results were obtained with the Caucasian subgroup, as also with the chemotherapy-naive subgroup (30% of the cohort), demonstrating that risk stratification by CAB is unaffected by race or chemotherapy. Next, the performance of CAB was compared with Oncotype DX (ODX). 83% patients stratified as low risk by ODX (RS 0-25) in a sub-cohort of 109 were also stratified as low-risk by CAB. To assess the impact of CAB in treatment decision making, we assessed the data of 589 patients who have undergone CAB testing so far, 288 were identified as low-risk. 93% of these CAB low-risk patients were not given chemotherapy, demonstrating the clinical impact of CAB. Conclusions: CAB is the first test of its kind to be retrospectively validated in Asia. It shows high concordance with ODX in risk stratification of patients. CAB has been in clinical practice in India and near-India markets for 2 years and is helping clinicians and patients in making affordable treatment decisions.