Intratumoral heterogeneity of Ki67 index and its impact on the diagnosis and prognostication of pancreatic neuroendocrine tumors: hot-spot count differs significantly from random-area count

Abstract

The pivotal role of Ki67 in grading pancreatic neuroendocrine tumors (PanNETs) is well recognized and firmly established in current WHO guidelines. Intratumoral heterogeneity is a well-known phenomenon, and it has also been documented for Ki67; however, the data on the magnitude of the impact of this heterogeneity on the final grade in primary NETs is relatively limited. In this study, Ki67 labeling index (KI) was calculated by using the manual count of camera-captured image method in 91 PanNETs both in hot-spots (KI(h)) as well as 6 different random-areas (KI(r)), each counted in a minimum of 2000 cells. The process was repeated for 29 samples from metastatic foci. Mean KI of the hot-spots was more than twofold higher than that of the random (5 vs 2.1%), with a low kappa agreement (0.2770). This changed the final grade in almost half of the cases (42/91; p < 0.001). Random counting missed all 4 G3 NETs as well as 6/7 of the Grade 2b cases (those with Ki67 10-<= 20%). On the other hand, in only 4 cases, the difference in KI was > 10%. Similar heterogeneity was also observed in the 29 metastatic tumors analyzed with the final grade differing in 55% based on KI(h) versus KI(r). KI(h) had a stronger correlation with signs of aggressiveness including metastasis and tumor size and also trended with perineural invasion in tumors > 5 cm. The intratumoral heterogeneity in Ki67 in pancreatic NETs lead to a change in final grade in 46% of cases by hot-spot versus random count. This underscores the importance of both making a reliable Ki67 count and providing a numerical index in addition to the final grade, and interpretation of the results case-by-case basis for management purposes, rather than the rigid grade-based approach. This study also supports the usage of hot-spot rather than random count as the grade parameter. Reporting Ki67 in cytology and small biopsy specimens should be supplemented by a comment highlighting that the final grade may change when the entire tumor is evaluated but that the difference in the nominal count will seldom be higher than 10%.