Expression of CD24, A Stem Cell Marker, in Pancreatic and Small Intestinal Neuroendocrine Tumors

Abstract

Objectives

CD24 has been considered as a normal and cancer stem cell marker. Potential intestinal stem cells weakly express CD24. In the pancreas, CD24 is a possible cancer stem cell marker for ductal adenocarcinoma.

Methods

Expression of CD24 in intestinal and pancreatic neuroendocrine tumors (NETs) was examined. Immunohistochemistry was performed on benign duodenum, ileum mucosa and pancreas, as well as primary duodenal NETs, primary and metastatic ileal and pancreatic NETs.

Results

Scattered CD24 positive cells were noted in the duodenal and ileal crypts, most of which showed strong subnuclear labeling pattern. Similar expression was observed in 41 of 43 (95%) primary ileal NETs, but only in 4 of 26 (15%) duodenal NETs (p<0.01). In addition, metastatic ileal NETs retained CD24 expression. Pancreatic islets did not express CD24, and only rare cells had subnuclear labeling of CD24 in the pancreatic ducts. Unlike ileal NETs, only 5 of 92 (5%) pancreatic NETs expressed CD24 in the subnuclear compartment (p<0.01). All 5 NETs showed a unique morphology with prominent stromal fibrosis.

Conclusion

CD24 expression was frequent in primary and metastatic midgut NETs, but rare in pancreatic and duodenal NETs. Expression of CD24 in ileal NETs may have future diagnostic and therapeutic implications.

INTRODUCTION

Well differentiated neuroendocrine tumors (NETs) of the gastroenteropancreatic system compromise approximately two-thirds of total NETs with an annual incidence rate of 3–5 cases per 100,000 persons^1–4. Based on embryonic source of the organ, they are classified into 3 groups: foregut, midgut and hindgut. Foregut NETs include those arising in the stomach, pancreas, duodenum and proximal jejunum. Midgut NETs are always located in the distal jejunum and ileum. Although rare, small intestinal and pancreatic NETs are the second most common malignant neoplasms in the small intestine and in the pancreas, respectively^{5, 6}. In addition, both tumors are frequently associated with liver metastasis. Most patients die of hepatic failure due to extensive liver involvement by metastases ^7–9.

Our understanding of gastroenteropancreatic NETs (GEP-NETs) has grown considerably in recent years. The pathogenesis of foregut NETs is distinct from that of midgut NETs. For example, MEN1 mutations are frequently detected in duodenal and pancreatic NETs, but they are rare in midgut NETs. Nevertheless, all NETs are thought to originate from a variety of neuroendocrine cell types or progenitor cells¹⁰. There are more than 10 types of neuroendocrine cells in the gut. Midgut NETs arise from enterochromaffin cells, which are the major neuroendocrine cell type of the small intestine³. In the duodenum, gastrin-producing and somatostatin-producing cell hyperplasia precedes the development of gastrinoma and somatostatinoma, respectively⁴. Pancreatic NETs were thought to derive from islet cells; however, recent studies suggest that they may arise from multipotent pancreatic stem/progenitor cells residing within the ductal epithelium¹¹.

CD24, a heat stable antigen, has been recognized as both a normal and a malignant stem cell biomarker^{12, 13}. Intestinal stem cells weakly express CD24¹⁴ while neighboring cells such as Paneth and neuroendocrine cells in the small intestine express higher amounts of CD24 (^{12, 14–17}. In the pancreas, CD24/PDX1 positive progenitor cells were shown to differentiate into insulin-producing cells¹⁸. In addition, CD24 may be one marker of cancer stem cells for intestinal and pancreatic ductal adenocarcinomas^{7, 12, 19–22}. Furthermore, CD24 expression was associated with a poor prognosis and chemotherapy resistance in pancreatic cancer^{23, 24}

Despite extensive studies on CD24 as a stem cell marker, its expression in intestinal and pancreatic NETs remains unknown. In this study, we wanted to determine CD24 expression in pancreatic and small intestinal NETs.

MATERIALS AND METHODS

CD24 expression was immunohistochemically assessed in 5 tissue microarrays (TMAs) containing 92 pancreatic NETs (including 6 cases with both primary and liver metastasis) and 32 midgut (jejunal/ileal) NETs (including 7 with both primary and liver metastasis). Immunohistochemical labeling for CD24 was also performed on 26 duodenal NETs, either from endoscopic mucosal resection or Whipple resection. In order to further examine whether CD24 expression was preserved in liver metastasis, additional 11 jejunal/ileal cases with both resected primary and metastatic tumor were included in the study. In addition, 13 liver metastases from patients with pancreatic NETs (6 wedge resections and 7 biopsies) were immunohistochemically labeled with CD24. Among the 13 cases, 7 had primary tumor in the TMAs, 1 had both resected primary and metastasis, and 5 had liver tumor only. This study was approved by Vanderbilt Institutional Review Board.

Four µM unstained slides from the TMAs and formalin fixed paraffin embedded resection/biopsy specimens were deparaffinized by routine methods. For antigen retrieval, the sections were heated to 105°C for 20 minutes in a pH-9.0 EDTA buffer, and then allowed to cool to room temperature. After the retrieval, the tissue sections were quenched with 3% H₂O₂ in sodium azide for 5 minutes at room temperature. Anti-CD24 antibody at 1:100 dilution (ab118070, abcam^®, Cambridge, MA ) was then incubated with the tissue sections, followed by antibody localization using the Dako Envision⁺ HRP-labeled polymer (DAKO, Carpinteria, CA ). Staining was visualized by 5 minute incubation with diaminobenzidine.

Tumors with greater than 5% of the tumor cells demonstrating subnuclear concentrated cytoplasmic CD24 labeling were considered to be positive. The immunohistochemical stains were read by two pathologists (SNS and CS).

RESULTS

Expression of CD24 in the Small Intestine and Small Intestinal NETs

Consistent with observations by others¹⁴, we found CD24 expression by normal small intestine epithelial cells. Scattered CD24 positive cells were seen in normal intestinal crypts, but rare in the intestinal villi (Figure 1A). Several labeling patterns were found, including weak membrane, supranuclear, subnuclear, and diffuse cytoplasmic staining; however, a majority of the CD24-positive cells displayed strong subnuclear cytoplasmic labeling (Figure 1B).

Image 1.

CD24 (ab118070, abcam^®, Cambridge, MA) expresssion in normal small intestine epithelium. A, CD24 expression is concentrated in the crypt epithelium (×40). B, CD24 positive cells demonstrate strong subnuclear cytoplasmic labeling (×200; arrows: subnuclear CD24 labelling).

Next, we immunohistochemically labeled small intestinal NETs with anti-CD24 antibodies. Forty one of 43 (95%) midgut NETs showed moderate to strong CD24 expression. The CD24 labeling was characteristically located in the subnuclear compartment (Figure 2A), the staining pattern was similar to that observed in some CD24-positve cells in normal small intestine, suggesting that these CD24 positive cells in normal intestinal epithelium are initiating cells for midgut NETs. It appeared that the liver metastases retained CD24 expression (Figure 2B). Only 1 of 18 liver metastases (6%) had negative CD24 labeling, in this case the corresponding primary tumor also lacked CD24 expression. Primary tumors and their respective metastases showed CD24 positivity in more than 50% tumor cells, with moderate to strong subnuclear labeling. This pattern was observed in all but two of the cases. The liver metastases demonstrated CD24 labeling that was always stronger at the periphery than in the center of the lesion.

Image 2.

Expression of CD24 in midgut neuroendocrine tumors. A, Primary midgut NET with CD24 expression (×100), inset, higher magnification demonstrates subnuclear CD24 labelling (×400). B, A liver lesion from the same patient showing CD24 expression (×100). C, CD24 expression in the intestinal mucosa (×40). D, CD24 expression in the inflamed mucosa from the same patient shown in C (×40; arrow: NETs)

The adjacent benign intestinal mucosa was also evaluated for CD24 expression. With the exception of one case, there was no evident increase in CD24 expressing cells. In this one case, there was a clinical history of Crohn’s disease, and the NET appeared to arise in a background of inflammatory bowel disease. The adjacent mucosa showed prominent increase in CD24 positive cells; however, the unaffected normal mucosa showed only scattered CD24 positive cells (Figure 2C–2D).

CD24 expression in duodenal NETs was also examined. Twenty six duodenal NETs included 14 from duodenal bulb, 4 from second portion, 5 from the ampulla/papilla, 2 from distal duodenum, and 1 from an unknown segment. Although the duodenal crypts had CD24 positive cells, duodenal NETs infrequently expressed CD24 when compared to midgut NETs (p<0.01). Only 4 of the 26 (15%) tumors expressed CD24 (Figure 3A–3B). Of the 4 tumors, 2 were from the ampulla, and 2 from the duodenal bulb. There were no morphological differences between CD24 positive tumors and those that lacked expression

Image 3.

Expression of CD24 in duodenal neuroendocrine tumors. A, Duodenal NETs with no CD24 expression (×100). B, Duodenal NET with subnuclear CD24 expression (×100). Note: some cells in the background crypts expressed CD24 (A).

Expression of CD24 in the Pancreas and Pancreatic NETs

CD24-positive cells were not observed in the islets of Langerhans. Previous studies reported frequent CD24 expression in pancreatic ductal adenocarcinoma; however, CD24 positive cells were rare in benign pancreatic ducts^{23, 25}. Immunohistochemical labeling 20 pancreatic sections only detected 2 cells in the pancreatic ducts showing subnuclear CD24 expression (Figure 4A).

Image 4.

Expression of CD24 in the pancreas and pancreatic neuroendocrine tumors. A, CD24 positive (subnuclear labeling) cell in a large pancreatic duct (×200). B, Pancreatic NET showing no subnuclear CD24 labeling (×200). C, A hematoxylin and eosin stain showing a pancreatic NET with prominent stromal fibrosis (×40). D, the same tumor shown in C with CD24 expression (×200).

Only 5 of 92 (5%) pancreatic NETs demonstrated moderate to strong CD24 expression. The labeling was also predominantly located in the subnuclear region of the tumor cells as seen in the small intestine NETs. A group of pancreatic NETs (10%, 9/93) displayed weak and diffuse cytoplasmic staining (Figure 4B). When we examined benign acinar and islet cells in some cases, we also found a similar pattern of weak cytoplasm staining. Therefore, this pattern of labeling was considered to be non-specific. None of the positive cases had corresponding liver metastasis. However, 18 liver metastases, including 8 with CD24 negative primary tumor, showed no CD24 expression.

The histology of the 5 CD24 positive pancreatic NETs was further examined. Interestingly, these tumors all shared unique morphologic features, including dense fibrosis, small nests/tubules, and an infiltrative growth pattern. In addition, they frequently surrounded a large pancreatic duct, causing pancreatic duct stenosis and chronic pancreatitis in the surrounding pancreas (Figure 4C–D). This variant of pancreatic NET has been described in previous studies. They shared some features of midgut NETs including serotonin and CDX2 expression^26–28, raising the possibility of a common precursor for pancreatic and midgut NETs.

DISCUSSION

CD24 expression has been studied in small intestinal crypts and pancreatic ductal adenocarcinomas. It has been identified as a marker of intestinal stem cells. CD44⁺CD24^loCD166⁺ cells were identified as putative intestinal stem cells in both mouse and human, and single CD44⁺CD24^loCD166⁺ cells were shown to give rise to enteroids containing multiple intestinal epithelial lineages, including neuroendocrine cells, paneth cells and goblet cells¹⁴. CD44⁺CD24^hiCD166⁺ cells isolated from mouse intestine were found to be predominantly non-proliferative secretory cells displaying markers of enteroendocrine, Paneth and goblet cells.. The mature neuroendocrine cells derived from the CD44⁺CD24^loCD166⁺ cell were found to strongly express CD24 in mouse models. In humans, scattered CD24 positive cells are present in both duodenal and ileal epithelium. Some have been identified as neuroendocrine cells by previous studies¹⁷. Interestingly, we observed a similar pattern of CD24 expression in a majority of midgut NETs, however this was not a frequent observation in duodenal NETs (95% vs. 14%). Midgut NETs demonstrated a predominantly strong subnuclear CD24 labeling pattern. This unique immunohistochemical labeling was also observed in liver metastases from the primary CD24 positive midgut NETs (94%). These data suggest that CD24 expressing cells in the intestinal crypts are the precursor cells of most intestinal NETs.

In all but one case, the adjacent mucosa showed no increase in CD24 expressing cells; however, this case in a patient with inflammatory bowel disease showed prominent CD24 positive cells. These data confirm that most small intestinal NETs are not initiated by way of the hyperplasia and dysplasia pathway, and further suggest that the tumorigenetic mechanisms may be different between small intestinal NETs arising in normal mucosa and those arising in inflamed mucosa.

The cell of origin of pancreatic NETs is unclear. A small subpopulation of human PDX-1 and nestin expressing pancreatic stem/progenitor cells isolated from adult human pancreatic duct were shown to be capable of differentiating into insulin-, glucagon and somatostatin-positive cells in vitro in the presence of specific growth factors²⁵. In addition, the human CXCR4-positive cells expressing different stem cell markers isolated from the islet-depleted pancreas were able to form islet-like structures ²⁹. These observations suggest that adult stem/progenitor cells residing in adult pancreatic duct give rise to pancreatic endocrine cells, and genetic and/or epigenetic alterations in these cells may potentially transform them into pancreatic NETs.

In adult mice, CD24 negative pancreatic ductal cells differentiated into insulin-secreting cells³⁰; however, CD24 expression in human adult pancreatic ducts had not been explored previously. We noted that CD24 expressing cells were extremely rare in the pancreatic duct. In addition, islet cells do not express CD24. These findings may help to explain why a vast majority of pancreatic NETs (95%) did not demonstrate any CD24 expression. Consequently, a lack of CD24 expression was also noted in all the liver metastases from the pancreatic NETs.

In particular, those pancreatic NETs demonstrating CD24 expression were morphologically distinct from CD24 negative pancreatic NETs. While the CD24 negative pancreatic NETs were hypercellular with minimal stromal fibrosis, CD24 positive pancreatic NETs showed histomorphologic overlap with small intestine NETs. These tumors were often adjacent to large ducts, and demonstrated dense fibrotic bands with infiltrative nests and tubules of neoplastic cells. Although, it remains unclear if the CD24 positive cells in the pancreas are indeed neuroendocrine cells, the possibility that these tumors may derive from the rare CD24 positive cells seen in the pancreatic duct should be considered. Their rarity precludes a definitive determination of origin by immunofluorescence double staining.

CD24 expression can be either cytoplasmic or membranous in malignancies. In pancreatic ductal adenocarcinoma, CD24 expression has been associated with poor prognosis in patients although the effect was small and was lost when analyzing CD44; CD24; Epcam triple positive cells ^{23, 24}. Cytoplasmic CD24 expression was associated with shortened patient survival in colorectal cancers³¹. All CD24 expressing NETs in this series had cytoplasmic labeling in the subnuclear compartment. Recently, intracellular CD24 expression was shown to inhibit the specific endoribonuclease activity of G3BP, a phosphorylation-dependent endoribonuclease³². In addition, the study showed that intracellular CD24 inhibits cell invasion by posttranscriptional regulation of BART through interaction with G3BP. However, whether cytoplasmic expression of CD24 in small intestine NETs confers an indolent clinical course requires further study.

Immunohistochemical studies for CD24 expression might have potential clinical implications. Ninety five percent of midgut NETs expressed CD24, whereas only 5% of pancreatic and 15% of duodenal NETs had CD24 expression. Importantly, metastatic midgut NETs retained subnuclear CD24 expression. Thus, CD24 proves to be a useful biomarker in determining the origin of metastatic NETs.

CONCLUSION

The incidence of pancreatic and small intestinal neuroendocrine tumors has consistently risen over the past decade. Currently, in North America and Europe in terms of gastrointestinal malignancies, it is second in prevalence only to colorectal carcinoma. The heterogeneous nature of these tumors along with their propensity for treatment resistance via alternative growth pathways has made molecular-targeted therapy an essential tool in treatment. We propose CD24 as a novel marker for identifying primary midgut neurendocrine tumors as well as a subset of primary pancreatic neuroendocrine tumors with a unique histomorphologic pattern of disease. This distinction is important, as therapeutic options depend on location of the primary tumor. CD24 has potential as a small intestinal cancer stem cell marker. In particular CD24 reactive NETs could be prospective candidates for both anti-cancer stem cell and anti-VEGF/angiogenic therapeutic regimens³³.

Table 1.

Expression of CD24 in Duodenal, Midgut and Pancreatic Neuroendocrine Tumors

	Duodenal NET (n=26)	Midgut NET		Pancreatic NET
		Primary (n=43)	Liver metastasis (n=18)	Primary (n=92)	Liver metastasis (n=18)
CD24+	4 (15%)	41 (95%)	17 (94%)	5 (5%)	0 (0%)

NET: neuroendocrine tumor