Evaluation of prognostic indicators using validated canine insulinoma tissue microarrays

Tissue microarray (TMA) technology allows analysis of multiple tumour samples simultaneously on a single slide. The aim of the present study was to develop and assess a TMA containing 32 primary canine insulinomas and 13 insulinoma metastases. The results of histopathological and immunohistochemical analyses of triplicate core biopsies were compared with those of individual tissue sections using weighted κ statistics. Inter-observer agreement of TMA immunohistochemistry scores were assessed for chromogranin A (CgA), insulin, growth hormone (GH), growth hormone receptor (GHR) and Ki67 index, as well as the prognostic utility of clinicopathological, histopathological and immunohistochemical criteria. There was substantial agreement of scores for histopathological parameters (κ = 0.64-0.70) and a substantial to near-perfect agreement for homogenous immunohistochemical parameters (κ = 0.69-1.00). Except for GH, which demonstrated heterogeneous staining, there was good to excellent inter-observer agreement for all other immunohistochemical staining scores (intra-class correlation coefficients: 0.70-1.00). On univariate analysis, the presence of nuclear atypia was significantly predictive of disease-free intervals (DFIs) for canine insulinoma, while tumour size, TNM stage, necrosis and Ki67 index were significant in terms of prognosis, with respect to both DFI and survival time. On multivariate analysis, tumour size and Ki67 index retained predictive power for survival time, as did tumour size for DFI. This study confirms the applicability of TMA technology for evaluation of canine insulinoma.