Manjiri Bakre, Charusheila Ramkumar, Chetana Basavaraj, Arun Kumar, Lekshmi Madhav, Chandra Prakash, and Prathima R
OncoStem Diagnostics Pvt. Ltd., India
Posters & Accepted Abstracts: J Mol Biomark Diagn
Assessment of �risk of cancer recurrence� in ER+ breast cancer patients based on clinical parameters and biomarkers is insufficient leading to over-treatment with chemotherapy. OncotypeDx, Mammaprint are useful in limited sets of node negative (-) patients, but they are largely prognostic with limited chemo-predictivity and are prohibitively expensive in India and SE Asia. A cost-effective �predictive� test which will: i) accurately estimate the �risk of recurrence� for ii) a �broader� (node - and +) set of patients is urgently required. Using a retrospective training cohort of 330 patients, we developed a Morphometric Immunohistochemistry based test comprising 5 biomarkers plus three clinical parameters to arrive at �CanAssist Score� using a Machine Learning Statistical algorithm. The CanAssist Score stratifies patients into �low or high� risk for recurrence. Analytical validation experiments performed to assess critical IHC variables confirmed the robustness of the test. Initial �test validation� on 130 cases was 93% specific. Extended clinical validation on an additional 500 pre and post-menopausal cases shows NPV of 95% and specificity of 80%. Majority patients in �low risk� group had Stage 2, Grade 2/3 disease over Grade 1. In a head-to-head pilot study with Oncotype Dx test �CanAssist-Breast� had superior NPV and specificity. Importantly CanAssist-Breast correctly re-stratified many recurred cases as �high risk� which were stratified as �low risk� by Oncotype Dx and thus were not treated with chemotherapy. In conclusion, we have developed a highly specific, low-cost test to predict risk of recurrence and prevent overtreatment in patients with early stage breast cancer.
Molecular Biomarkers & Diagnosis received 2054 citations as per Google Scholar report