Use of Whole Slide Digital Images for Determination of Ki-67 (MIB-1) Labeling Index in Infiltrating Astrocytomas: Comparison of Current Counting Methods with an Image-Based Algorithm
M Bonham, R Monroe, R Allen, M Almaula, T Tihan. University of California, San Francisco, CA; Bioimagene, Sunnyvalle, CA
Background: While mitotic rates as well as proliferation indices have been reported as significantly different among different grades of infiltrating astrocytomas, a reproducible cut-off value has not emerged. The goal of this study was to determine whether more reproducible algorithms using whole slide digital images (WSDI) can define better cut-off values for different grades of infiltrating astrocytomas.
Design: Whole tissue sections from 35 infiltrating astrocytomas grades II through IV were evaluated to determine the best slides/blocks for Mib1 immunohistochemistry. 1000 cells were manually counted for MIB1 positivity. The same slides were then used for WSDI. The slides were scanned using the iScanTM automated whole slide scanning system and were viewed using the ImageViewerTM. Single fields of view (FOV) from each WSDI measuring ∼420,000 µ2 that represented the highest density of Ki-67 positive nuclei were selected for image analysis. Three different FOVs were analyzed in a subset of slides to determine the variability in each case. The total Ki-67 labeling index as well as average surface area of positive and negative tumor cells were calculated in MatLabTM.
Results: We analyzed labeling indices in 13 grade II, 6 grade III and 16 grade IV astrocytomas using both methods. The mean labeling indices for manual and automated counts were 2.4 and 2.6 for grade II tumors, 10.3 and 12.4 for grade III and 15.8 and 20.9 for grade IV tumors, respectively. The correlation coefficients between manual and automated counts were 0.79 for grade II, 0.86 for grade III, and 0.79 for grade IV astrocytomas. The overall correlation coefficient for the entire series was 0.87. In addition, the mean nuclear area for positive cells was 92.2 µ2 for grade II, 102.2 µ2 for grade III and 86.8 µ2 for grade IV astrocytomas. Analysis of three different FOVs per slide yielded a correlation coefficient of 0.87 when compared to analysis of a single separate FOV.
Conclusions: Our analysis yielded a similarly increasing labeling index for astrocytoma grades II through IV using both methods. Additional information such as the mean size of tumor nuclei was obtained for WSDI. Both counting methods still suffer from a sampling bias due to subjective selection of the area to be counted. Developing optimal algorithms based on tissue and tumor type may be more reproducible in predicting the value of proliferation indices in CNS neoplasms.
Monday, March 22, 2010 1:00 PM
Poster Session II # 252, Monday Afternoon