Tumour grade is one of the prognostic factors in pancreatic ductal adenocarcinoma, but its value is controversial. In this study, the predictive value and the reproducibility of the WHO grading system were reconsidered and the possibility of supplementing it with the immunohistochemically assessed proliferative activity was investigated. Seventy resected ductal adenocarcinomas of the head of the pancreas were evaluated. A total of 60 HPF fields on two to four sections per tumour were screened for glandular differentiation, mucin production, mitosis, and nuclear atypia by two observers with different degrees of experience. Each criterion was scored and the grade was calculated from the mean value of all single scores. Corresponding slides were immunohistochemically stained with the proliferation marker Ki-S5. The percentage of positive nuclei was assessed and a proliferation index (PI) assigned (<10%=1; 10-50%=2; >50%=3). Multivariate analysis (Cox regression) identified grade and R stage as the most significant factors for predicting survival. The PI determined on the basis of Ki-S5 staining did not prove to be an independent prognostic factor. In 30 of 70 carcinomas, it correlated with the tumour grade. Within a given tumour grade, the cases with the least favourable prognosis could be distinguished on the basis of their PI. The inter-observer variability was considerable, with the main differences occurring in the group of G1 tumours. According to the refined WHO criteria, the histopathological grade of pancreatic ductal carcinoma is an important independent prognostic factor, but reproducibility depends on the expertise of the observer. Criteria that relate to cellular and structural differentiation seem to be more predictive than those related to proliferation.
Copyright 2000 John Wiley & Sons, Ltd.