Abstract

High Content Single Cell Analysis Using the Programmable Bio-Nano-Chip System: New Tools for Cancer Diagnosis

John McDevitt, Chair, Department of Biomaterials, New York University College of Dentistry Bioengineering Institute

Over the past few decades, use of biomarkers has become increasingly intrinsic to practice of medicine and clinical decision-making. Diagnosis and management of oral cancer is a promising area whereby biomarker driven testing has potential to provide significant impact on patient care. Oral cancer is sixth most common cancer worldwide and has been marked by high morbidity and poor survival rates with little over the past few decades. Beyond prevention, early detection is the most crucial determinant for successful treatment and survival of oral cancer. This talk will feature details related to a new ‘cytology-on-a-chip’ platform capable of high-content single-cell measurements. This methodology permits concurrent analysis of molecular biomarker expression and cellular/nuclear morphology using over 200 fluorescence intensity and shape parameters for each region of interest extracted from multi-spectral fluorescence images. Molecular biomarkers: EGFR, avß6, CD147, ß-catenin, MCM2, and Ki67 were selected based on their capacity, through prior immunohistochemistry studies, to distinguish stages of disease progression towards oral cancer. Measurement time to complete this chip-based image analysis is approximately 20 minutes vs. about 1-3 days for gold standard pathology exam. This new clinical decision tool has been developed and validated in context of major clinical study involving 714 prospectively recruited patients. These efforts have led to collection of data across 6 diagnostic categories and assembly of one of largest well-qualified cytology database (confirmed by tissue biopsy) ever collected for prospectively recruited potentially malignant oral lesions. The application of statistical machine learning algorithms exploiting this large database has led to development of robust classification models with validated and stable parameters. High sensitivity and high specificity adjunctive diagnostic aids have been developed through these efforts.