How Many Needle Core Biopsies Are Needed to Comfortably Predict the Histologic Grade of Metastatic Well-Differentiated Neuroendocrine Tumors to the Liver?
Zhaohai Yang, Laura H Tang, David S Klimstra. Penn State Milton S. Hershey Medical Center, Hershey, PA; Memorial Sloan-Kettering Cancer Center, New York, NY
Background: As determined by Ki67 labeling index (especially from the highest labeling region), the histologic grade of metastatic well-differentiated neuroendocrine tumors (NETs) correlates with patient survival. Substantial intratumoral heterogeneity in Ki67 labeling index exists, resulting in different WHO grades (G1 vs G2). It is of practical importance to determine the optimal number of random needle biopsy cores needed to sample the area with the highest grade.
Design: The question was addressed in a simulated fashion using virtual core biopsies on resection specimens. Immunohistochemical staining for Ki67 was performed on 45 resected metastatic NETs to the liver. The scanned whole slide images were segmented into multiple subsections to simulate needle core biopsies, and the Ki67 labeling index for each subsection was determined using digital image analysis. The final grade of each case was based on the highest labeling region. In heterogeneous cases (containing both G1 and G2 regions) the probability of obtaining that final grade was expressed as the ratio of the number of cores containing the final grade over total number of cores, while in homogeneous cases the probability was 100% by definition.
Results: The cohort consisted of 24 homogeneous and 21 heterogeneous cases. Percentage of cases for which the final grade was correctly predicted on virtual core biopsies increased with the number of cores in a logistic fashion. The percentages of correctly predicted cases in the heterogeneous group (P1) were 27%, 40%, 48%, 53%, and 57% for 1, 2, 3, 4 and 5 cores, respectively. The percentages in the whole cohort (P2) increased to 66%, 72%, 76%, 78% and 80% for the respective number of cores due to inclusion of homogeneous cases. A 90% accuracy required 31 cores in the heterogeneous group, and 17 cores for the whole cohort.
Conclusions: A high degree of accuracy in predicting histologic grade of NETs requires multiple core biopsies particularly in cases exhibiting intratumoral heterogeneity. At least five cores are needed to achieve reasonably acceptable accuracy and may be clinically feasible.
Monday, March 19, 2012 1:00 PM
Platform Session: Section E, Monday Afternoon