(PRWEB) November 22, 2010
The concordance to manual scorings by expert assessors was higher for the HER2-CONNECT(TM) image analysis software than for any previously reported HER2 IHC image analysis method, and operators save time due to easy outlining of the tissue regions to be analysed.
The method was developed independently of reagent manufacturers and validated in collaboration with Nordic Immunohistochemical Quality Control (NordiQC), an internationally recognized organization for laboratory proficiency testing.
Significant savings in operator time: As opposed to other commercially available image analysis algorithms for HER2 IHC, the HER2-CONNECT(TM) algorithm does not require meticulous outlining of tumor areas (see illustration). The method automatically discards stromal cells, detects membranes, and quantifies the connectivity of detected membranes.
Connectivity is a continuous end point: Membrane connectivity is a continuous measurement that easily translates into classical scores of 0, 1+, 2+, and 3+. This makes the method more descriptive and well suited for biomarker research and pre-clinical drug development.
Says Dr. Mogens Vyberg, director of NordiQC: 'The most direct interpretation of the ASCO/CAP guideline recommendations for HER2 IHC testing requires analysis of the invasive tumor cells individually in order to determine the ratio of tumor cells with complete membrane staining and to determine the intensity of the membrane staining. This interpretation may sound straightforward, but in fact there are serious limitations causing unacceptable inter-observer variations. The automated identification of connectivity of stained membranes represents a practical interpretation of the ASCO/CAP guidelines. And it shows a lot of promise in addressing the practical challenges that we are currently confronted with, when deploying digital pathology in the clinic.'
High concordance, sensitivity and specificity: The method was subjected to blinded validation studies in collaboration with NordiQC. The study was conducted by five expert pathologists on tissue sections from breast cancer biopsies. The percentage agreement achieved by the HER2-CONNECT(TM) method was 92.3% (Cohen’s kappa: 0.86). Specificity wrt FISH: 99.2%, Sensitivity wrt FISH: 100%. These results and further data were published at the Pathology Visions conference in San Diego on 24-27 Oct 2010.
Reagent independence: The underlying algorithmic principles of the HER2-CONNECT(TM) are robust towards the visual differences seen when using reagents from various manufacturers. The method was validated for the two leading assays: HercepTest (from Dako) and Pathway HER2 (from Ventana/Roche).
Computationally efficient: Researchers can work with HER2-CONNECT(TM) in two different ways: 1) Connectivity and score are immediately computed based on operator defined regions of interest, and 2) Batch processing: Full sections and/or Tissue Micro Arrays can be subjected to analysis in batch processing mode. Computational efficiency makes both approaches possible and feasible on standard desktop computers.
Slide scanners: The method was validated for the NanoZoomer from Hamamatsu. Corresponding studies for other leading scanners are on their way. Preliminary results suggest that the method is also scanner independent.
The HER2-CONNECT(TM) software is for research use only, including clinical research, and wil run on Windows based computers.
# # #