Tissue Markers
Historically, NET classification has been based primarily on morphological appearance of the tumor (size, extent of spread, histologic classification, genetic background, and assessment of proliferative markers [Ki-67 and mitotic index]). This approach has limitations as results are not always reproducible and may even be inaccurate in certain cases. Indeed, the histopathological assessment of NETs is, in general, regarded as a particularly challenging pathological exercise.
In other tumors, gene expression profiles have proven advantageous in refining approaches to classification and prognosis e.g. astrocytoma grading (>90% accuracy), prostatic carcinomas (74-80% accuracy), breast and colon cancer prognosis (70-90%). We have examined NET neoplasia in an attempt to help clarify the notoriously ambiguous and amaranthine morphology of this disease. distinct transcript expression profiles of 9 potential marker genes that were phenotypically relevant to primary and metastatic NETs (adhesion, migration, proliferation, apoptosis, metastasis, and hormone secretion), could be used for classification, albeit with suboptimal accuracy rates <80%. Expanding this to a multi-gene panel (21 marker genes, sixteen of which were preferentially expressed in NETs), we could differentiate and predict tumor grades and types with 100% accuracy. Gene expression profiling and supervised machine learning provides a bias-free method for classification of primary and metastatic SI NETs. The techniques we have utilized may also have further applications in the accurate delineation of molecular signaling networks involved in NET initiation, local development, and invasion.