Digital Imaging in Breast Cancer: An Ode to Microscopy
Kinan Drak Alsibai, Sandy Azoulay, Odile Languille-Mimoune, Bruno Poulet, Anne Barres, Camilo Adem. Institut de Pathologie de Paris, Paris, France
Background: Digital image analysis is currently available using strong electronic algorithm in order to evaluate biomarkers (ER, PR, Her2 and Ki67) in breast pathology. Our objective was to compare this technology with the routine manual appreciation of the mentioned breast biomarkers.
Design: Breast cancer cases were prospectively collected over a time period. Pathologists reported breast panel as they did routinely according to the ASCO/CAP recommendation. After their report signed out, all slides were scanned using Ventana iScan Coreo. Digital slides were evaluated by the mean of Virtuoso (software, Ventana Roche). All values were compared and discrepancies noted.
Results: Complete date was available for 59 cases (23 biopsies, 36 surgical specimens). There was no discordant result in ER evaluation (8 ER-; 51 ER+) using both methods. Six cases (10%) were scored differently for PR (three considered negative or positive by either method). Digital evaluation for Her2 reported more score 3 than manual scoring (12 cases versus 9; 20% versus 15%), more score 2 (17% versus 12%), but less scores 0/1 (63% versus 73%). Interestingly, SISH was available and showed gene amplification for 2 discordant cases reported manually score 2, and by image analysis score 3. No Her2 amplification was seen by SISH in another case scored 2 by both methods. Since there is no consensus on the threshold for Ki67 classifying highly proliferative tumor versus low proliferative tumor, three different analyses were done: using 14% cutoff showed 11 discordant cases, while using 20% cutoff showed 8 discordant cases, and 30% cutoff showing 3 discordant cases (5%).
Conclusions: Manual evaluation of breast cancer biomarkers remains the gold standard as it can be used by all pathologists. It is a cost effective method, as well as time saving. Digital imaging may help in training pathologists, or when applied on low proliferative tumor. Recommendations from consensus conference might emphasize using different approaches: routine microscopy as 1st screening, and in some instances depending on the final advocated cutoff either independent reading by two pathologists or the use of digital imaging.
Wednesday, March 6, 2013 9:30 AM
Poster Session V # 265, Wednesday Morning