Endosonography in decision making and management of gastrointestinal endocrine tumors

Abstract

Objective: gastroenteropancreatic (GEP) neuroendocrine tumors, suspected on clinical basis, are often difficult to localize. We report our experience with endoscopic ultrasonography (EUS) in the preoperative localization of pancreatic endocrine tumors (PETs), compared to other imaging modalities, and in staging and following up carcinoid tumors (CTs) of the gastrointestinal (GI) wall. Methods: 50 patients (20 males; mean age 54 years), 39 with suspected PETs and 11 with GI CTs underwent EUS (Olympus GF-UM2 or GF-UM3). EUS data could be compared with resected specimens in 25 out of the 39 PETs and five out of the 11 CTs. Results: in the PETs group 42 tumors (35<20 mm) were removed: 23 in the pancreas, eight in the duodenum, and 11 in the lymph nodes. EUS correctly localized 20 out of the 23 (87%) pancreatic tumors, included 11 out of the 12 (91.6%) insulinomas, three out of the eight (37.5%) duodenal gastrinomas, and ten out of the 11 (90.9%) metastatic lymph nodes. Furthermore EUS accurately evaluated the depth of parietal invasion of CTs in three out of four patients (75%) (two after and one prior to endoscopic resection). In three patients EUS was confirmed as normal on resected specimens (two pancreas and one stomach). In the PETs group, a correct localization was obtained by ultrasonography (US) only in 17.4% of cases, by computed tomography (CT) in 30.4%, by magnetic resonance imaging (MRI) in 25%, by angiography in 26.6%, and by somatostatin receptor scintigraphy in 15.4%. Conclusion: EUS must be considered the first-intention method in localizing PETs and is helpful in decision making and management of GEP endocrine tumors.

Introduction

Gastroenteropancreatic (GEP) neuroendocrine tumors are rare neoplasias. They can be subdivided into carcinoid tumors (CTs) of the gastrointestinal (GI) wall and pancreatic endocrine tumors (PETs) (including islet cell endocrine neoplasia of the duodeno-pancreatic area, pancreatic carcinoid, otherwise called well-differentiated endocrine carcinoma, and small-cell carcinoma, also called poorly differentiated endocrine carcinoma, according to Capella et al., 1994). Several particular features make these lesions distinguishible from other GI tumors, i.e. unique aspects of pathogenesis, histology, diagnosis, localization, and treatment (Lloyd, 1990, Jensen and Norton, 1998).

Carcinoid tumors arise from enterochromaffin cells and are mostly located in the appendix, rectum, and small intestine. Unusual localizations are the esophagus, the bile duct, and the ampulla of Vater. They are often small in size: 80% of the rectal lesions are less than 1 cm and only 5% more than 2 cm. The risk of metastases increases with the size of the lesion (Yoshikane et al., 1993, Kvols, 1998). Gastric carcinoids need to be considered as a distinct disease, apart from the other GI carcinoids previously described, because they arise from the enterochromaffin-like cells (ECL), mostly observed in hypergastrinemic conditions such as atrophic gastritis associated with hypochlorhydia, pernicious anemia, or Zollinger–Ellison syndrome (Maton et al., 1990, Modlin et al., 1995).

Endoscopic ultrasonography (EUS) allows the visualization of GI CTs, as small as 2 mm in diameter, in the esophago-gastric, duodenal, and colo-rectal wall. They usually present as round or egg-shaped lesions with hypoechoic or intermediate homogeneous echopattern, well demarcated and mainly located in the second and/or third wall layer (mucosa and/or submucosa) (Palazzo and Roseau, 1992, Yoshikane et al., 1993), sometimes with a transmural invasion of the GI wall, usually accessible to histological diagnosis by means of endoscopic biopsies.

EUS is used for the staging of gastric and colo-rectal CTs by accurately determining their size, the depth of parietal invasion, and the local nodal involvement, with an overall accuracy of 75% for the last two features (Yoshikane et al., 1993), so providing information of prognostic value and allowing the physician to plan therapeutic management. EUS can be of value also in patient follow-up and in evaluating if endoscopic resection has been radical (De Angelis et al., 1997).

The islet cell tumors of the pancreas entail different problems in terms of diagnosis, localization and therapy. Diagnostic suspicion is based on clinical and functional signs and is confirmed by means of specific biochemical parameters. PETs still represent a difficult challenge for diagnostic imaging tecniques for several reasons: small size, deep location in the abdomen, and extrapancreatic or multiple localizations (Stefanini et al., 1974, Lloyd, 1990, Jensen and Norton, 1998).

Nevertheless, once the diagnosis is proved, if surgery is considered in order to cure the disease, at least in patients without metastases, a correct preoperative localization and staging is mandatory to optimize operative management, reduce duration, and complexity of the operation and improve surgical results (Creutzefeldt, 1995, Eriksson et al., 1995, Guillausseau et al., 1995, Matuchansky et al., 1995, Orbuch et al., 1995).

Different preoperative imaging modalities are currently available. These tecniques have become increasingly complex with the introduction of more sophisticated biotechnological procedures (Modlin and Tang, 1997), but increased sensitivity does not necessarily mean increased clinical usefulness. Depending upon their size, PETs can be preoperatively localized by standard imaging studies, such as ultrasonography (US), computed tomography (CT), magnetic resonance imaging (MRI), also with short-time inversion recovery sequences (STIR-MRI) and angiography at best in 40–70% of all cases (Krudy et al., 1984, Broughan et al., 1986, Galiber et al., 1988, Frucht et al., 1989, Päivansalo et al., 1989, Pisegna et al., 1993, Guillausseau et al., 1995, Orbuch et al., 1995, Modlin and Tang, 1997, Jensen and Norton, 1998). Other promising methods increasingly used are somatostatin receptor scintigraphy (SRS) (Cadiot et al., 1996, Gibril et al., 1996b, Zimmer et al., 1996, Termanini et al., 1997, Krausz et al., 1998) and selective intra-arterial injection of secretin or calcium (SAIS/SAIC) during angiography with epatic venous sampling (Imamura et al., 1987, Doppman et al., 1991, Thom et al., 1992, Doppman et al., 1995, Gibril et al., 1996a). Both these techniques have demonstrated good sensitivity, even if a wide range is reported in the literature from 10 to 100% (depending on type, site and size of the tumor) (Imamura et al., 1987, Doppman et al., 1991, Kwekkeboom et al., 1995, Strader et al., 1995, Cadiot et al., 1996, Gibril et al., 1996a, Gibril et al., 1996b, Zimmer et al., 1996, Termanini et al., 1997, Jensen and Norton, 1998), but they are rarely available and very expensive (Kwekkeboom et al., 1996, Bansal et al., 1999). SAIS and SAIC are also time-consuming and carry the same complication rate (0.6–2%) as reported for angiography (Lilly et al., 1990, Strader et al., 1995). Portal venous sampling, despite high sensitivity (46–96%), is time-consuming, uncomfortable, requires considerable expertise by the radiologist and is associated with too many risks (complication rate about 20%) (Miller et al., 1992, Strader et al., 1995). Dual phase helical CT has the potential to improve the detection of small PETs and their hepatic metastases (Van Hoe et al., 1995, Chung et al., 1997, King et al., 1998), but till now only a small number of patients have been evaluated and we lack prospective comparative studies between spiral CT and other imaging modalities in the setting of the PETs. It may be impossible to locate pancreatic neuroendocrine tumors, 10 mm or less in diameter, by intraoperative palpation alone (Norton et al., 1990, Norton, 1995). Intraoperative US has been able to detect 88% of insulinomas and 91% of pancreatic gastrinomas (Klotter et al., 1987, Galiber et al., 1988, Norton et al., 1988, Norton et al., 1990, Norton, 1995), but only 30% of duodenal gastrinomas (Norton et al., 1988, Norton, 1995, Sugg et al., 1993). So if the precise location of an endocrine tumor is preoperatively known, surgery can be made considerably easier and surgical morbidity may be reduced.

At the moment EUS is the most accurate diagnostic technique available for pancreatic imaging and can provide high-resolution images of pancreatic parenchyma and surrounding structures (Rösch et al., 1992, Ruszniewski et al., 1995, Fukuda and Hirata, 1995, Savides et al., 1996, Sahai et al., 1998), therefore it is higly accurate for the diagnosis of small pancreatic carcinomas (Yasuda et al., 1988, Rösch et al., 1991). In several studies EUS has shown high sensitivity (57–100%) and specificity (88–95%) in localizing PETs (Dancygier and Classen, 1986, Lux et al., 1986, Hayashi et al., 1989, Lightdale et al., 1991, Glover et al., 1992, Palazzo et al., 1992, Rösch et al., 1992, Thompson et al., 1994, Zimmer et al., 1994, Zimmer et al., 1996, Bansal et al., 1995, Ruszniewski et al., 1995, Shumacher et al., 1996, Proye et al., 1998).

Diagnostic EUS criteria for PETs are the evidence of single or multiple well demarcated lesions (Fig. 1), with an echo-poor, homogeneous pattern (Fig. 2, Fig. 3). Rarely echo-rich (Fig. 4), inhomogeneous, poorly outlined tumors have been observed, sometimes with cystic areas, calcifications and/or a central area slightly more echogenic than the rest of the lesion. Lymph nodes need to be carefully looked for. A round shape, well defined boundaries and a hypoechoic or isoechoic pattern compared with the primary tumor are considered to be strongly suggestive for tumoral involvement.

Unfortunately the sensitivity clearly decreases for lesions located in the pancreatic tail (37% in one study) (Shumacher et al., 1996) and for duodenal wall tumors (less than 50%) (Thompson et al., 1994, Ruszniewski et al., 1995). The combination of conventional endoscopy and EUS provided correct localization of duodenal gastrinomas in 60% of cases (Ruszniewski et al., 1995). Intraoperative endoscopic transillumination of the duodenum improves the detection of duodenal wall gastrinomas (Frucht et al., 1990), but the medial wall of the duodenum cannot be explored (Sugg et al., 1993, Norton, 1995). Notwithstanding these problems, EUS has emerged as an accurate means for preoperative localization of PETs and is now considered the first-intention method for detection of small PETs by several authors (Lightdale et al., 1991, Glover et al., 1992, Palazzo et al., 1992, Rösch et al., 1992, Thompson et al., 1994, Zimmer et al., 1994, Bansal et al., 1995, Ruszniewski et al., 1995).

We report our experience in EUS preoperative localization of PETs comparing endosonographic results with those of other imaging modalities and with histopathological findings. Furthermore we try to answer the question whether the information provided by EUS make a difference in clinical decision making and management of GI endocrine tumors.