Proposed histopathologic grading system derived from a study of KIT and CK19 expression in pancreatic endocrine neoplasm.

Predicting the biologic behavior of pancreatic endocrine neoplasm in the absence of local invasion or metastasis is difficult. We recently proposed an immunohistochemical classification system based on a combination of KIT and CK19 expression: low risk (KIT-/CK19-), intermediate risk (KIT-/CK19+), and high risk (KIT+/CK19+). In the current study, we sought histopathologic features that correlated with the immunohistochemical categories and that could be used in predicting tumor behavior. We assessed histologic findings of 97 pancreatic endocrine neoplasm that had been classified into 3 risk groups based on KIT/CK19 expression. Clinicopathologic associations with prognosis were evaluated using Cox proportional hazards regression models. Tumor size, mitoses, infiltrative border, extrapancreatic extension, perineural invasion, and presence of amyloid were significantly different among the 3 risk groups, and tumor necrosis approached statistical significance as well (P = .089). A scoring system using mitoses, necrosis, and tumor border was developed as follows: mitosis: 0 (0/50 high-power field), 1 (1-3/50 high-power field), and 2 (≥4/50 high-power field); necrosis: 0 (absent) and 1 (present); tumor border: 0 (noninfiltrating) and 1 (infiltrating), giving a possible histology score of 0 to 4. Although there was an overall difference in disease-specific survival among the 4 histology scores (P < .001), there was not a statistically significant difference between patients with scores of 0 and 1 or between patients with scores of 3 and 4. Therefore, a 3-tied grading system was developed by combining score 0 and 1 tumors as grade 1, score 2 tumors as grade 2, and score 3 and 4 tumors as grade 3. There was a significant difference in survival, tumor metastasis, tumor recurrence, and functioning status among the 3 grades. By analyzing histopathologic features in KIT/CK19 risk groups, we developed a 3-tiered histopathologic grading system using reproducible parameters (mitoses, tumor necrosis, and infiltrative border) that are easy to apply in practice.