Evaluation of Automated Tissue Microarray Scoring by Quantitative Imaging Cytometry
Electronic Poster - Award to Submit Full Manuscript on Behalf of APIII to Archives of Pathology and Lab MedicineKenneth J Craddock MD; University Health Network, University of Toronto; Farshid Siadat MD; University Health Network, University of Toronto; William Geddie MD; University Health Network, University of Toronto; Suzanne K Lau MSc; University of Toronto; Arezou A Ghazani MSc; Ontario Cancer Institute;
Content:
Several new platforms capable of assessing chromogenic and fluorescent staining are available for high throughput semi-automated analysis of tissue microarrays. The iCys? Imaging Cytometer (Compucyte) is unique in its use of laser excitation, multiple detectors, and analysis software, providing capability for accurate spectral deconvolution of both chromogenic and fluorescent detection systems. We assessed the capability of this instrumentation for quantitative measurement of protein expression in a tissue microarray (TMA), by comparing pathologist scoring to the instruments automated assessment, using Quantum dots, a conventional fluorophore, and a chromogenic dye.
Technology:
The iCys? Imaging Cytometer and its accompanying iCys software (Compucyte, Cambridge, MA) were used to scan and analyze a tissue microarray stained with EGFR monoclonal antibody using either streptavidin-conjugated Qdot 655 (Invitrogen, Eugene, OR), Alexa 647, or NovaRed peroxidase detection methods.
Design:
A tissue microarray was constructed from xenografts derived from 9 lung carcinoma cell lines, selected for their differential expression of epithelial growth factor receptor (EGFR). High-resolution scanning fields (pixel size 0.25?m2) were placed within cores defined by low-resolution scanning, and light loss (argon 488 and HeNe 605) or emitted fluorescence recorded. Measurements were filtered by the software to isolate positive staining colocalized to nuclei. The ability of each detection method to differentiate the protein expression of individual cell lines was evaluated using two-tailed t-tests, and compared to manual scoring by two pathologists using the HSCORE method.
Results:
By automated assessment, Quantum dots and Alexa showed a small improvement over NovaRed chromagen in distinguishing cell lines by their protein expression. There was an intensity-dependent effect; Quantum dots and Alexa performed best at distinguishing cell lines with high staining intensities, while NovaRed more effectively separated the cell lines with low staining intensities. The manual pathologist scoring using HSCOREs performed the best overall, with a modest improvement over the Qdot and Alexa automated scoring. The main improvement seen in the manual scoring was in tissue cores showing a high proportion of necrosis.
Conclusion:
The automated scoring of the TMAs provided a very good estimation of the EGFR expression in the cell lines as compared to manual scoring and RNA measurement. The Qdot and Alexa fluorescent detection systems performed comparably, showing an improved dynamic range in the high staining intensities. The main area identified for improvement of the automated TMA scoring was in distinguishing viable tumour from necrotic areas.
