Differentiation of small bowel and pancreatic neuroendocrine tumors by gene-expression profiling

Neuroendocrine tumors (NETs) are rare neoplasms whose incidence has increased five-fold over the past three decades1. These tumors are derived from enterochromaffin and islet cells, and may secrete a variety of polypeptides, which may cause symptoms and can be useful for diagnosis, detection, and treatment of these tumors2. Two of the most common sites of these malignancies involving the digestive system are the small bowel (SBNETs) and pancreas (PNETs), and approximately 58% of patients with SBNETs and 72% of those with PNETs present with metastatic disease3, primarily to the liver. It has been estimated that as many as 10–20% of patients with liver metastases present with an unknown primary site4, and in these cases, being able to determine the primary site has important implications for surgical approach and therapeutic measures. The preferred treatment for SBNETs and PNETs is surgical resection, even in the face of metastatic disease5. Medical therapy for metastatic NETs includes the use of somatostatin analogs, which helps to control hormonal symptoms as well as improve time to progression6. Sandostatin is an 8 amino acid analog of somatostatin which mediates its effects by binding to the type 2 somatostatin receptor (SSTR2, a G-protein coupled receptor [GPCR]) on the cell surface of NETs. Sandostatin has also been useful in localization of tumors when labeled with Indium, but its success in locating primary GI sites of NETs in patients presenting with liver metastases has been limited4. Chemotherapy has also had limited response rates in NETs, with improved survival rates in those with pancreatic tumor sites. Treatment of pancreatic NETs (PNETs) has included combinations of streptozotocin and 5-fluorouracil and/or doxorubicin, with response rates of approximately 40%7. However, in non-pancreatic NETs, the response rate is only 25%7. More recently, Sunitinib and Everolimus have been demonstrated to lead to improved disease free survival for patients with advanced PNETs8, 9. Even patients with advanced disease may have prolonged survival, with median survivals of 56 and 24 months in metastatic SBNETs and PNETs, respectively1. These statistics could potentially be further improved by uncovering new targets, such as additional cell surface receptors (besides SSTR2) which could be exploited for diagnosis, imaging, and therapy. Further understanding of these NETs could come from analysis of their expression patterns, which might then be used to categorize the genetic profiles of different primary sites. This would hold promise for determining the site of origin from biopsies of liver metastases, and to identify targets for future therapy. To date, the limited gene expression studies performed in SBNETs and PNETs have not identified consistent and useful patterns of over and underexpressed genes10–14, and therefore the objective of this study was to examine matched tumor and normal tissues from these sites, then determine whether these could be used to predict the primary sites from liver metastases.