Gastrointestinal stromal tumors (GISTs) are very rare. Although they are the most common mesenchymal tumors of the alimentary tract they only account for 1–2% of malignant gastrointestinal tumors [1]. The most common locations of GISTs are the stomach (50–60%), small intestine (30–35%), and colorectum (5%) [1]. Extragastrointestinal GISTs (E-GISTs) are usually located in the mesentery, retroperitoneum, omentum, and pancreas [1],[2]. The most common anatomic location of pancreatic GISTs (P-GISTs) is the head [1]. Preoperative diagnosis of P-GISTs is difficult because almost half show a heterogeneous appearance with a cystic component in radiological studies and their hypervascular behavior makes differentiation with pancreatic neuroendocrine tumors (PNET) difficult [1],[3]. The treatment of P-GISTs depends mainly on the anatomical location in the pancreas, with extended resections and lymphadenectomy being unnecessary. We present a case localized in the area of the pancreatoduodenal groove treated with local resection and the literature is reviewed to analyze the prognostic influence of the surgical technique performed to treat GISTs of the pancreatic head and uncinate process.

A 50-year-old woman, with no history of interest, reported weight loss and occasional abdominal pain in the right upper quadrant. Laboratory test parameters were normal and ultrasound showed a solid hypoechogenic lesion of 30 × 35 × 31 mm, in the hepatic hilum, which could be an adenopathy or pancreatic head tumor. A contrast-enhanced computed tomography (CT) scan confirmed the presence of a tumor measuring 39 × 61 × 35 mm at the level of the pancreatoduodenal groove, with a heterogeneous attenuation pattern and enhancement in the arterial phase with areas of central necrosis. The rest of the pancreas was normal without focal lesions or dilation of the main duct. There were no radiological criteria for locoregional lymph node involvement. There were also two enhancing liver lesions in arterial phase in segments IVa (7 mm) and VII (26 mm), isodense with the liver parenchyma in portal phase, suspicious of secondary deposits of a hypervascular lesion, all compatible with metastatic PNET. In magnetic resonance imaging (MRI) the findings were similar, although in SPECT the possibility of benign lesions (focal nodular hyperplasia) cannot be ruled out. A core needle biopsy of the segment VII lesion was performed without showing malignancy. Upper GI endoscopy showed erosive gastritis and a suggestive image of extrinsic compression at the duodenal bulb. There were no signs of malignancy in the biopsies performed on the gastric body, antrum, and duodenum.

Endoscopic ultrasound showed a well-defined, hypervascularized, heterogeneous, and hypoechogenic lesion of 4 cm in diameter, suggestive of PNET. Fine needle aspiration (FNA) cytology was performed where scarce cellularity was observed with neuroendocrine cells without cytological atypia.

Elective surgery was performed, and a pseudo-encapsulated tumor was found close to the pancreatoduodenal groove at the second duodenal portion, approximately 4 cm in diameter, with an exophytic component. In the rest of the abdominal exploration, no peritoneal implants or palpable liver lesions were observed. Intraoperative ultrasound showed that the lesion was separated from the Wirsung and confirmed the existence of three liver lesions of 2.7 cm in segment VII, 1.5 cm in segment Iva, and another of 1 cm in segment I. Radioablation of the three lesions is performed, followed by Kocher’s maneuver, cholecystectomy, and introduction of a Fogarty catheter. Through an anterior duodenotomy it is confirmed that the papilla is distal to the lesion. Because of these findings and with the suspicion of a PNET, it is decided to perform a local resection including pancreatic parenchyma and section of the duodenum at 2 cm from the pylorus with wide safety margins and closure of the distal duodenum with endostapler, preserving the papilla. The intervention is completed with a Roux-en-Y duodenojejunostomy, without local lymphadenectomy. The macroscopic histological study shows a tumor of 3.5 × 2.8 × 2 cm with pancreatic and duodenal tissue. Microscopically the findings are compatible with an encapsulated P-GIST, with spindle cells (Figure 1), without necrosis, 4 mitosis/50 high power fields (HPF) (low risk), without the involvement of borders neither of four peripancreatic nodes examined (pT2, N0, AJCC 8th edition). The lesion shows positivity for the immunohistochemical technique of CD 117 (Ckit), actin, and DOG1 and negativity for desmin, S100, HMB45, EMA, CD34, CD31, chromogranin, and synaptophysin. Ki65 is less than 2%.

The postoperative period includes a prolonged hospital stay due to a duodenal stump fistula that required surgical reintervention (Grade IIIb of Clavien’s classification). The patient did not receive adjuvant treatment with imatinib and after 11 months of follow-up is asymptomatic with no signs of recurrence or liver lesions on the control CT scan.

Gastrointestinal stromal tumors are rare tumors and account for only 1–2% of malignant gastrointestinal tumors [1]. The most frequent locations are the stomach (50–60%), small intestine (30–35%), and colorectum (5%) with only 4% being duodenal GISTs (D-GISTs) [1],[4]. The main prognostic factors are the size, location, surgical resection margins, intraoperative tumor rupture, and mitotic index [1],[3],[5],[7]. Extragastrointestinal GISTs only represent 5–10% of GISTs, although they have a worse prognosis than gastric GISTs (G-GISTs) [2],[6]. The treatment of choice for any potentially resectable GIST ≥2 cm without liver metastases is surgery, and there is currently a consensus on the role of imatinib in adjuvant therapy in patients with higher risk of relapse [3],[7],[8],[9].

Pancreatic GISTs are exceptional, with a small number of published cases, short postoperative follow-ups, and no conclusive evidence on their biological behavior or prognostic differences according to the anatomical location in the pancreas. Liu et al. have conducted the only comparative study between G-GISTs and P-GISTs, being P-GISTs larger and more malignant [1]. The disease-free interval (DFS) has been lower than in G-GISTs, although there has been no difference in disease-related survival [1].

In a review of the literature we found, including our case, a total of 51 P-GIST. The most frequent location has been the pancreatic head in 22 cases (43.1%) (three tumors in the uncinate process and two in the head-body). Of these, we have analyzed 19 GISTs of the pancreatic head and uncinate process after excluding two cases in which the surgical treatment was not specified and another because it was mainly body involvement with extension to the head (Table 1) [10],[11],[12]. There were no differences in sex (10 women and 9 men), the mean age was 50.2 years (31–70) and the mean size of the tumor was 6.8 cm (15–2.4). Most pancreatic head GISTs are symptomatic. Abdominal pain is the most frequent symptom (68.4%) and 36.8% patients have anemia. Despite the large size of the lesions, no patient presented obstructive jaundice. Preoperative diagnosis is difficult to make and only in three cases (15.7%) was the diagnosis of P-GIST confirmed by a FNA [13],[14],[15]. The preoperative diagnosis was cystic tumor in five patients (26.3%) and PNET in four other cases (21%). Most were spindle cell type (94.4%) and 77% showed an intermediate or high-risk category. Pancreatoduodenectomy was performed in 14 patients (10 Whipple and 4 with pyloric preservation). In five cases (26.3%) the surgical technique was more limited: three enucleations and two wedge local resections, with one pancreatico-jejunostomy and a duodenojejunostomy in our case. Eight patients received adjuvant treatment with imatinib, all of them after PD. Four patients (all with high-risk GIST) developed liver metastases: three after PD and imatinib, and one after local resection without adjuvant therapy.

Numerous pathologies have been described at the pancreatoduodenal groove. Preoperatively it is difficult to differentiate a P-GIST from a PNET or other cystic tumors of the pancreas. Endoscopic ultrasound is the best diagnostic technique to determine the exact location by differentiating between a pancreatic or duodenal tumor adjacent to the pancreas [1]. In P-GISTs the surgical treatment will depend on the location, although extensive resections and lymph node dissection are unnecessary. Pancreatoduodenectomy with or without pyloric preservation is the most commonly used technique in pancreatic head GISTs (73.6%), while local resection is more frequent in D-GISTs (62–70.8%) [4],[16],[17]. In D-GISTs, PD has been used in larger lesions (76% in lesions ≥5 cm) and located in the second duodenal portion (64.3–81%), although it has had higher rates of recurrence than local resection because it has been performed on more biologically malignant tumors [4],[16],[17]. Although PD is a safe technique in the treatment of pancreatic head tumors, it is not exempt from significant morbidity and mortality [16], [18]. Local resections of P-GISTs are less surgical time consuming, allow preservation of pancreatic function and have fewer postoperative complications [4],[7],[16]. The main determining factors for local resection are the general condition of the patient, preoperative diagnosis of PNET, invasion of adjacent structures, size and location [3],[16]. The margins of resection are not clearly established, but it is recommended that they be 1–2 cm [4],[16]. Intraoperative ultrasound must ensure that the lesion is sufficiently separated from the Wirsung before indicating an enucleation. When the GIST is located at the level of the pancreatoduodenal groove the choice of technique is controversial [4].

Of the 19 GISTs in the pancreatic head reviewed, the DFS has been analyzed using the Kaplan–Meier method and the log-rank test (Mantel-Cox), considering a statistically significant p < 0.05. We found no statistically significant differences in DFS when analyzing sex (p = 0.07), age (> or or < 5 cm, p = 0.58) and risk grade (p = 0.08). Shen et al. [16] reported a worse prognosis in the group of male patients operated by D-GIST and Lopes et al. reported a significantly higher incidence of androgen receptors in E-GISTs [19]. In our review all recurrences have been in male patients with high-grade GISTs. However, there has been no statistical significance when analyzing both variables, probably because of the small number of cases.

Given the impossibility of knowing the biological behavior of the tumor preoperatively, the main factors for deciding the extension of the surgical resection in GISTs of the pancreatic head are the size of the tumor and its anatomical relationship with the duodenum and the papilla. In all lesions larger than 5 cm or close to the ampulla of Vater, PD should be performed, and pyloric preservation is recommended [2]. In tumors of the uncinate process [4]. In tumors among 3–5 cm at the first duodenal portion it would be advisable to associate an antrectomy. If there is sufficient margin, the duodenum near the pylorus can be sectioned, with stapling of the duodenum proximal to the ampulla of Vater, and a Roux-en-Y duodenojejunostomy performed. In those GISTs with involvement of the third duodenal portion, the options for reconstruction after a local resection are a duodeno-duodenostomy or a Roux-en-Y duodeno-jejunostomy [17].

In conclusion, the surgical technique does not influence the survival of pancreatic head GISTs. Local resection is oncologically safe, and the preoperative decision depends on the size and location of the tumor. The postoperative histological study will determine the risk of recurrence and the need for adjuvant treatment with imatinib.