Comparison of Three Ki-67 Index Quantification Methods and Clinical Significance in Pancreatic Neuroendocrine Tumors
Trynda N Oberg, Jesse S Voss, Christine M Lohse, Tsung-Teh Wu, Thomas C Smyrk, Lizhi Zhang. Mayo Clinic, Rochester, MN
Background: Ki-67 index is essential in the pathological reports for pancreatic neuroendocrine tumors (PNETs). Pathologists usually use three methods to determine Ki-67 index including eyeball estimation (EE), manual counting (MC) or automated quantitation using the Automated Cellular Imaging System III (ACIS) (Dako, Carpinteria, CA). However, there is no sufficient data regarding the concordance among these three methods and their correlations with clinical outcome. The goal of this study was to compare three quantification methods with clinical outcome to determine the best method for clinical practice.
Design: Ki-67 immunostaining was performed on 97 resected PNETs with clinical follow-up information. Three methods of quantification were employed: 1) An average of EE by three pathologists. 2) MC of at least 500 tumor cells within the hot spots. 3) ACIS quantitation by selecting 8-10 hot spot regions (40x tool). Lin's concordance correlation coefficients were used to evaluate agreement among the methods. C indexes from Cox proportional hazards regression models were used to evaluate the ability of each method to predict clinical outcome.
Results: The concordance between the MC and ACIS and between the MC and average EEs were 0.97 and 0.88, respectively. The difference between the MC and ACIS assessments was -0.23, which was not significantly different from zero. But the difference between the MC and EEs was -2.56, which was significantly different from zero (p <0.001). The concordance among the three pathologists' EE was 0.86. More cases were classified as 2010 WHO grade 1 in the EE group due to lower Ki-67 values obtained by this method (65 grade 1 cases by eyeball vs. 31 and 35 by MC and ACIS assessments). There were five grade 3 PNETs and three of them were common by all methods. All three methods were significantly associated with patient survival. The EE had a higher predictive ability for survival compared with the manual and ACIS assessments.
Conclusions: The three quantification methods for Ki-67 index had almost perfect agreement and all of them were significantly associated with clinical outcome. The MC and ACIS assessments were almost equivalent. The EE scores were significantly less than the other two methods and tended to down-grade more tumors to grade 1, but had higher predictive ability for survival. The results suggest the necessity of consensus among pathologists for the method to determine Ki-67 index and proper cutpoint of Ki-67 index for better clinical correlation.
Wednesday, March 21, 2012 9:30 AM
Poster Session V # 251, Wednesday Morning