New Study Confirms High Inter-Observer Agreement Using the WATS3D Biopsy for Evaluation of Esophageal Pre-Cancer

SUFFERN, NY--(Marketwired - May 4, 2014) - CDx Diagnostics today announced results from a study showing high inter-observer agreement rates in the diagnosis of Barrett's esophagus and dysplasia among pathologists analyzing samples from the WATS3D biopsy. The study showed that WATS3D computer-assisted pathology analysis results in significantly better pathologist inter-observer agreement rates compared to standard manual microscopic histopathology assessment of Barrett's esophagus and dysplasia. The results were presented during an oral session at Digestive Disease Week® 2014 (DDW) on May 3, 2014 in Chicago.

"This study has potentially important implications for clinical care because we know that histopathologic diagnosis of Barrett's esophagus-associated dysplasia has traditionally suffered from poor inter-observer agreement, even among experienced GI pathologists," said lead author Prashanth Vennalaganti, MD of the Division of Gastroenterology and Hepatology, University of Kansas Medical Center, Kansas City, MO. "These high kappa values for the WATS3D biopsy provide powerful validation for the clinical value of this procedure. Providing a more accurate, consistent assessment of esophageal pre-cancer will allow gastroenterologists to make more confident decisions about care strategies for their patients."

The WATS3D biopsy collects a wide area, disaggregated tissue specimen of the entire thickness of the suspect epithelium. This unique tissue specimen is then subjected to specialized, computer-assisted 3-dimensional analysis which selects the 200 "highest ranking" cells from each patient for presentation to a pathologist. The high sensitivity of WATS3D is due to the large tissue area sampled, and the proprietary computer imaging system that is based on an algorithm developed as part of the U.S. Strategic Defense Initiative missile defense program.

In standard manual microscopic inspection of a biopsy specimen the first cells evaluated by the pathologist from each slide are selected randomly. An initial impression of the degree of pathology on the case can then be formed based on these randomly selected cells and this can lead to "confirmation bias" against modifying this initial impression during the remainder of the pathologist's analysis.

In contrast, during WATS3D analysis the pathologist's initial impression of the case is not based on evaluation of a few randomly selected cells but on a subset of 200 mathematically pre-selected cells which are rank ordered for potential abnormality, eliminating "confirmation bias" as a source of inter-observer variability.