• Neuroendocrine
• IMMUNOHISTOCHEMISTRY
• NEOPLASMS

Recently, Pepper et al investigated the expression of the S100 protein in pancreatic neuroendocrine tumours (NETs).1 They found a significantly higher rate of S100 expression (58%) in pancreatic NETs in comparison with NETs from other primary sites (14%) and a high specificity (94%) of moderate to strong (score 2+/3+) expression of S100 protein to indicate a pancreatic origin of a NET. Concurrently but independently of Pepper et al’s study, we observed the expression of the S100 protein in some NETs of the ampullary region, which had not been investigated in the aforementioned study. Consequently, we performed an immunohistochemical analysis of S100 expression in 28 well-differentiated ampullary NETs (23 cases from the major ampulla and 5 cases from the minor papilla/ampulla), which were included in previously published multicentre studies on ampullary NETs2 3 and had available tumour tissue sections. Immunohistochemistry was performed using the same antibody against S100 protein (polyclonal, Dako) and interpretation scoring system proposed by Pepper et al,1 that is, 3+: moderate to strong nuclear/cytoplasmic staining in >50% of cells; 2+: moderate to strong staining in 10–50% of cells or weak staining in >50% of cells; 1+: moderate to strong staining in 5–10% of cells or weak staining in 5–50% of cells; 0: staining in <5% of cells. Expression of S100 by sustentacular or stromal cells was discarded. The vast majority of ampullary NETs (24 cases, 86%) expressed S100 (score ≥1+) and more than half (64%) of ampullary NET cases showed a score 2+ (7 cases, 25%) or 3+ (11 cases, 39%) of S100 protein expression (figure 1). Interestingly, the rates of S100 expression ≥1+, 2+ and 3+ observed in ampullary NETs appear to be even higher than those reported by Pepper et al in pancreatic NETs.1

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

A minor ampulla neuroendocrine tumour showing strong nuclear and cytoplasmic expression of S100 protein in >50% of neoplastic cells (score 3+).

These findings are intriguing for two main reasons. First, they expand the spectrum of S100-positive NETs to include those of the ampullary region, thereby suggesting consideration of these NETs in cases of unknown-origin NETs expressing the S100 protein. Additionally, we recently reported a case of primary gallbladder NET expressing S100 in tumour cells,4 supporting the hypothesis that S100 expression may be a common characteristic of pancreatic, ampullary and biliary NETs, and a relatively specific biomarker of NETs originating from these embryologically related anatomical districts.

Second, in a diagnostic perspective and in particular for small endoscopic biopsies of the ampullary region, the expression of S100 by a neuroendocrine neoplasm can present a diagnostic challenge, potentially leading to an erroneous diagnosis of composite gangliocytoma/neuroma and neuroendocrine tumour (CoGNET). CoGNET, previously referred to as gangliocytic paraganglioma, is a rare mixed epithelial neuroendocrine and neuronal neoplasm with a triphasic morphology, composed of variable proportions of epithelial neuroendocrine cells, ganglion/ganglion-like cells and S100-positive Schwann cells, typically localised in the ampullary/periampullary region. Indeed, both ampullary NETs and the epithelial component of CoGNETs usually express general neuroendocrine markers (such as chromogranin A and synaptophysin) and keratins, and most cases of both tumour types are also positive for somatostatin. Nevertheless, immunostains for SOX10 (consistently reported positive in the Schwannian component of CoGNETs5 and negative in 5 ampullary NETs tested in our hands), as well as for progesterone receptors (reported as positive in the epithelial component of >90% of CoGNETs6 and negative in 10 ampullary NETs tested in this study), may aid in achieving an accurate differential diagnosis between CoGNETs and ampullary NETs, even in superficial biopsies (where the classic triphasic morphology of CoGNETs may not be easily appreciated and S100 immunostaining may be misleading). This distinction is clinically significant because CoGNETs are more indolent neoplasms, with lymph node metastases being less frequent compared with ordinary ampullary NETs.3