Heterogeneity of tumor prognostic markers: a reproducibility study applied to liver metastases of pancreatic endocrine tumors

Liver biopsy of metastatic pancreatic endocrine tumors allows confirmation of the diagnosis and assessment of prognosis. However, sampling variability is a potential limitation. Our aim was to use the tissue microarray technique to assess the heterogeneity of three prognostic markers, ie, MIB-1 proliferation index, microvascular density and somatostatin receptor type 2, inside single or between synchronous or metachronous liver metastases of pancreatic endocrine tumors. Tissue microarrays were constructed, which included core biopsies taken from surgically resected liver metastases in 29 patients. MIB-1, microvascular density and somatostatin receptor type 2 were evaluated after immunostaining. The heterogeneity was highlighted by the calculation of the reproducibility of the values of two cores randomly selected among all the cores studied. For quantitative variables, it was assessed by the intraclass correlation coefficient and by a Bland-Altman approach. For qualitative variables, observed agreement and weighted kappa were given. A total of 184 liver metastases were analyzed. For MIB-1, the intraclass correlation coefficients were 0.63, 0.69 and 0.67 and for microvascular density, the intraclass correlation coefficients were 0.48, 0.60 and 0.00, respectively, in single, synchronous and metachronous liver metastases. The variability increased for higher mean values of microvascular density. For somatostatin receptor type 2, the observed agreements were 91% (kappa = 0.81), 69% (kappa = 0.49) and 79% (kappa = 0.68) in single, synchronous and metachronous liver metastases, respectively. In conclusion, tissue microarray analysis identifies heterogeneity of protein expression in pancreatic endocrine metastases, which depends on the marker tested. The reproducibility is better for MIB-1 and somatostatin receptor type 2 than for microvascular density. Sampling variability should be taken into consideration as a potential limitation to the assessment of prognostic and therapeutic markers in biopsy samples from metastatic pancreatic endocrine tumors.