Clinicopathological Correlates of Hormone Expression-Based Subtypes of Non-Functioning Duodenal/Ampullary Neuroendocrine Tumors: A Multicenter Study of 151 Cases

Abstract

Duodenal neuroendocrine tumors (Duo-NETs) may arise in the ampullary and non-ampullary duodenum. Non-functioning Duo-NETs (NF-Duo-NETs), which account for most cases, may express various hormones. Previous studies have suggested that hormone production might be associated with biological aggressiveness. Current treatment protocols are based on functionality, tumor size, and location, but small NF-Duo-NETs may also have metastatic potential. We aimed to investigate whether tumor cell subtyping, based on hormone expression, could provide further insights into NF-Duo-NET biological behavior. We analyzed the clinico-pathological correlates of hormone expression in a multicenter series of 151 NF-Duo-NETs, subdividing tumors into five subtypes: gastrin-producing G-cell NETs (Gas-NETs), somatostatin-producing D-cell NETs (Som-NETs), serotonin-producing enterochromaffin-cell NETs (Ser-NETs), plurihormonal NETs, and gastrin-, somatostatin-, and serotonin-negative NETs (GSSN-NETs). Som-NETs were the most frequent (31%), followed by plurihormonal NETs (26%), Gas-NETs (24%), GSSN-NETs (13%), and Ser-NETs (4%). Som-NETs and GSSN-NETs were more commonly located in the ampullary region and showed significantly larger size, more frequent lymphatic and/or vascular invasion, and higher pT, pN, and American Joint Committee on Cancer (AJCC-9th edition) stages compared to Gas-NETs, which were often (77%) diagnosed at AJCC stage I. Ampullary Som-NETs showed a more invasive and metastatic potential compared to non-ampullary Som-NETs, while, among plurihormonal NETs, the predominantly expressed hormone influenced tumor biological features, with gastrin-predominant NETs showing less invasive potential. At logistic regression, both tumor cell subtype and tumor size were independently associated with aggressiveness (pT3, pN1, or pM1 stage at diagnosis). Hormonal expression profiling may be clinically relevant in NF-Duo-NETs, independently of tumor size.

Keywords: Cell subtyping; Gastrin; Null-cell tumors; Plurihormonal tumors; Somatostatin.

Fig. 1

Histological images of a Gas-NET, a Som-NET, and a Ser-NET. A-B) Gas-NET showing a trabecular architecture and diffuse positivity for gastrin (A, hematoxylin–eosin; B, gastrin immunostaining). C-D) Som-NET featuring a tubular architecture and extensive positivity for somatostatin (C, hematoxylin–eosin; D, somatostatin immunostaining). E–F) Ser-NET exhibiting a nested structure and diffuse positivity for serotonin (E, hematoxylin–eosin; F, serotonin immunostaining)

Fig. 2

Histological images of a plurihormonal non-functioning duodenal neuroendocrine tumor. The tumor shows trabecular architecture (A, hematoxylin–eosin). The neoplasm exhibits expression of both gastrin (in most tumor cells; B, gastrin immunostaining), and serotonin (in 5–10% of tumor cells; C, serotonin immunostaining). Rare somatostatin-positive cells are also seen (D, somatostatin immunostaining)

Fig. 3

Histological images of a gastrin-, somatostatin-, and serotonin-negative neuroendocrine tumor (GSSN-NET) of the ampullary region. The low-power view (A) highlights a local lymph node metastasis (arrowhead) and infiltration of pancreatic parenchyma (arrow) (A, hematoxylin–eosin). The tumor shows a well-differentiated neuroendocrine morphology with a predominantly trabecular architecture (B), minor areas with a nested pattern and occasional nuclear pleomorphism (B-C, hematoxylin–eosin). The neoplasm shows diffuse expression of synaptophysin (not shown) and focal positivity for chromogranin A (D, chromogranin A immunostaining). Tumor cells are negative for serotonin (E, serotonin immunostaining), gastrin (F, gastrin immunostaining) and somatostatin (G, somatostatin immunostaining), with positive internal controls in the duodenal mucosa

Fig. 4

Forrest plot displaying OR (dots) and 95% CI (whiskers) estimated from a logistic model for aggressive tumor as a function of tumor cell subtype and tumor size (mm). Model Chi2 138.44, p < 0.001. Subtype: p < 0.001; Tumor size: p < 0.001. Model discrimination: AUC ROC = 0.85

Fig. 5

Kaplan–Meier survival estimates of patients with NF-Duo-NETs by tumor cell subtype