AIM: To provide long-term survival results of operable duodenal gastrointestinal stromal tumors (DGISTs) in a tertiary center in China.

METHODS: In this retrospective study, the pathological data of 28 patients with DGISTs who had been treated surgically at the Second Department of General Surgery, Sir Run Run Shaw Hospital (SRRSH) from June 1998 to December 2006 were reviewed. All pathological slides were examined by a single pathologist to confirm the diagnosis. In patients whose diagnosis was not confirmed by immunohistochemistry at the time of resection, representative paraffin blocks were reassembled, and sections were studied using antibodies against CD117 (c-kit), CD34, smooth muscle actin (SMA), vimentin, S-100, actin (HHF35), and desmin. Operative procedures were classified as wedge resection (WR, local resection with pure closure, without duodenal transection or anastomosis), segmental resection [SR, duodenal transection with Roux-Y or Billroth II gastrojejunostomy (G-J), end-to-end duodenoduodenostomy (D-D), end-to-end or end-to-side duodenojejunostomy (D-J)], and pancreaticoduodenectomy (PD, Whipple operation with pancreatojejunostomy). R0 resection was pursued in all cases, and at least R1 resection was achieved. Regional lymphadenectomy was not performed. Clinical manifestations, surgery, medical treatment and follow-up data were retrospectively analyzed. Related studies in the literature were reviewed.

RESULTS: There were 12 males and 16 females patients, with a median age of 53 years (20-76 years). Their major complaints were “gastrointestinal bleeding” (57.2%) and “nonspecific discomfort” (32.1%). About 14.3%, 60.7%, 17.9%, and 7.1% of the tumors originated in the first to fourth portion, respectively, with a median size of 5.8 cm (1.6-20 cm). Treatment was by WR in 5 cases (17.9%), SR in 13 cases (46.4%), and by PD in 10 cases (35.7%). The morbidity and mortality rates were 35.7% and 3.6%, respectively. The median post-operative stay was 14.5 d (5-47 d). During a follow-up of 61 (23-164) mo, the 2-year and 5-year relapse-free survival was 83.3% and 50%, respectively. Eighty-four related articles were reviewed.

CONCLUSION: Surgeons can choose to perform limited resection or PD for operable DGISTs if clear surgical margins are achieved. Comprehensive treatment is necessary.

Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal neoplasms of the gastrointestinal tract, although the annual incidence rates reported worldwide are less than 20 per million, and only about 5000 new cases are diagnosed annually in the United States[1-3]. GIST is a primary gastrointestinal disease that can arise anywhere along the digestive tract in adults. The stomach (60%) and jejunoileum (30%) are the most common primary sites, and only a small number of cases have been reported in the colorectum (< 5%), esophagus and appendix (< 1%)[4,5]. In addition, duodenal lesions represent approximately 5% of GISTs.

All GISTs harbor some malignant potential, although only 10%-30% are clinically malignant. In the past decade following Fletcher’s report[6], primary GISTs are not classified as “benign” or “malignant”, but are stratified by the probability of recurrence after complete resection into very low, low, intermediate, and high risk on the basis of their size and mitotic rate. Subsequently, Miettinen et al[7] suggested that the anatomical origin may be another independent factor for risk stratification, indicating that DGISTs share maximal risks with rectal GISTs compared with those of the stomach and jejunoileum.

Although DGISTs are relatively rare, they account for nearly 30% of all primary tumors of the duodenum, and the vast majority of patients present with gastrointestinal bleeding[4]. With regard to treatment, DGISTs often pose difficult surgical problems, due to the complex anatomical relationship around the duodenum, i.e., unlike the stomach or other intestinal segments where complete excision with wide margins are relatively straightforward procedures, wide resection of DGISTs will almost always entail a pancreaticoduodenectomy (PD), which is massively invasive and technically challenging[8,9].

In recent years, a limited resection (LR) of DGISTs demonstrated a comparable effect to PD in selected cases[10]. However, the optimal surgical approach (LR or PD) for DGISTs is largely unknown, as all the available evidence has been derived from small retrospective series[11]. In addition, scholars have gradually recognized the complexities of DGISTs, and these tumors have been classified separately from other small intestine GISTs into an independent category[12]. Also, a number of papers on DGISTs have been released[8-10,13-16]. Nevertheless, more experience with long-term oncological observations is required, especially for surgeons. This article aims to provide a single center experience of operable DGIST cases in China, and an update on the clinical management of DGISTs.

GISTs are a family of tumors thought to arise from the interstitial cells of Cajal in the gastrointestinal tract. Recently, the putative stem and progenitor cells for GISTs have been identified[1]. Most GISTs have oncogenic mutations in either KIT or platelet-derived growth factor receptor-α (PDGFRA), and there is substantial evidence that these mutations are pathogenetic for the initiation of GISTs. Histopathologically, GISTs are usually well circumscribed and surrounded by a pseudocapsule, ranging in size from millimeters to 40 cm, with a median size between 5 cm and 8 cm, while large GISTs often show cystic degeneration or central necrosis[18]. Microscopically, GISTs are defined as morphologically spindle cell, epitheloid, or occasionally pleomorphic, mesenchymal tumors, usually (approximately 95%) express the KIT protein and often (up to 90%) harbor mutations of a gene that encodes for a type III receptor tyrosine kinase (either KIT or PDGFRA)[19].

Surgery is the mainstay of treatment for localized, resectable GISTs. The tumor should be removed en-bloc with its pseudocapsule to yield an adequate resection margin. The optimal width of the tumor-free margin has not been defined, and it is unclear if re-resection is beneficial for positive microscopic surgical margins (R1), especially as the free radial margin is the one that is positive in most instances and there is no additional tissue to be removed[20]. Lymphadenectomy is not warranted unless there is gross nodal involvement. In cases of unresectable or marginally resectable disease, adjuvant tyrosine kinase inhibitor (TKI) therapy should be considered. Following surgical resection, GISTs often recur locally, spread diffusely throughout the serosal surfaces of the abdomen and/or metastasize to the liver. Advanced disease is associated with metastases to distant sites, including the lung and bone. Prior to the advent of TKI therapeutics, the prognosis for advanced GISTs was poor owing to their inherent resistance to both chemotherapy and radiation therapy[1].

DGISTs share the above-mentioned factors, but have individuality. DGISTS are unique entities, not only due to their anatomical location, but also their clinical manifestations, particularly difficult surgical decisions and poor prognosis. This is why DGISTs have attracted the authors’ interests as well as the attention of the French Sarcoma Group (GSF-GETO). In the 47^th Annual Meeting of the American Society of Clinical Oncology held in Chicago in June 2011, Duffaud et al[12] at GSF-GETO12 retrospectively analyzed 66 resectable DGIST patients with a median tumor size of 6 cm (1.5-31 cm), 29 of whom underwent WR, 23 SR, and 14 PD. During a median follow-up of 36 (1-168) mo, their 4-year OS and RFS rates were 89% and 58%, respectively. Duffaud’s report is the largest cohort study in the surgical rather than pathological field, thus has current significance.

The clinical presentations of DGISTs are highly variable according to their size and the existence of mucosal ulceration, but not tumor site[8,9]. The most common clinical presentation is reported to be gastrointestinal bleeding or abdominal pain. Interestingly in this series, the authors found that large tumors (> 5 cm) caused less gastrointestinal bleeding (P < 0.05); this can be rationalized by the different phenotype of DGISTs, “submucoscal/ulcerous type” or “serosal/massive type”, most small tumors were the former type and the majority of large tumors were the latter type[21]. More than 60% of DGISTs are located in the descending section, however, the reason for this is unclear.

Preoperatively, a variety of alternative examinations can be adopted, among which CT and MRI seem to be the best imaging modalities for assessment of the primary lesion and detection of metastases, whereas EUS is the optimum non-invasive tool for the clinical diagnosis[22,23]. Furthermore, EUS-guided biopsy has been established for the pathological diagnosis, although the sensitivity of DGIST samples obtained by EUS-guided biopsy is unsatisfactory compared with stomach GISTs (37.5% vs 84.4%)[24]. Recently, CT- or US-guided biopsy has been abandoned for resectable GISTs, due to the risk of pseudocapsule rupture and tumor spillage in the peritoneal cavity[4]. By integrating all the diagnostics, 71.4% of patients were accurately or probably diagnosed as having a DGIST. Moreover, the sensibility of preoperative diagnosis was correlated to Fletcher’s or Miettinen’s risk (P < 0.01) and initial hemoglobin levels (P < 0.05); that is, the tumor with malignant behavior has a tendency to be diagnosed as an abdominal/retroperitoneal tumor or a pancreatic cancer rather than a DGIST, whereas decreased hemoglobin potentially raises suspicion of gastrointestinal diseases, and possibly DGISTs.

The optimal surgical approach (LR or PD) for DGISTs is controversial. Goh et al[10] suggested that LR is associated with a shorter operation time, a similar complication rate, and a comparable disease-specific survival; Duffaud et al[12] concluded that LR rather than PD should be pursued to preserve optimal pancreas function for a better quality of life. According to the relatively few patients with long-term follow-up in this report, the findings support the views above. Although the DRS following PD seemed poor (P < 0.05), after adjusting the covariates, PD tended to be performed in patients with high risk (P < 0.05), and the results proved that LR and PD had a similar impact on RFS and DRS (P > 0.05), thus, both surgical approach lead to a similar oncological prognosis if clear surgical margins are achieved[4,14]. Patients undergoing LR and PD showed similar overall morbidities (44.4% vs 20.0%; P > 0.05) with the exception of DGE which was more frequent in the SR group (46.2%), and prolonged postoperative stay (P > 0.05).

Regardless of the pros and cons outlined above, essential factors which influence whether LR or PD is chosen are tumor size, location, proximity to the duodenal papilla, and technical feasibility[25-28]. In general, WR with primary closure can be performed for small lesions if the resulting lumen is adequate and the Vater ampulla can be preserved, even by laparoscopy or combined laparoendoscopic surgery[29,30]. Occasionally, the antimesenteric defect after WR can be closed by Roux-en-Y duodenojejunostomy, whereas the mesenteric defect inside the “C” loop of the pancreas head can by repaired by translocation of the distal common bile duct as a patch[31,32]. SR with gastrojejunostomy or end-to-side/-end duodenojejunostomy may be performed for larger tumors located in the D1, D3 and D4. The side[D]-to-end[J] duodenojejunostomy is not recommended due to possible duodenal leakage and stump stasis (stump syndrome)[33]. Some scholars advocate resection and anastomosis even for lesions close to the papilla by performing the anastomosis just below the ampulla, which has been achieved by performing a lateromedial anastomosis opposite the papilla or by performing papilloplasty with a temporary stent catheter inserted into the papilla to avoid possible postoperative stenosis[34,35]. PD is only indicated when the tumor is located in the D2 and involves the papilla, pancreas, or if the tumor is large with high malignant potential and has involved the adjoining organs[8]. PD combined with major hepatectomy for a DGIST with localized liver metastases has also been reported[36]. PD can provide a wider tumor clearance, but reconstruction is difficult and there is an increased risk of long-term anastomotic stenosis, as both the pancreatic and common bile ducts are likely to be smaller in diameter[37]. In this context, the author introduced the binding pancreaticojejunostomy, which resolved these problems, improved the anastomotic operability, and decreased postoperative complications[38,39]. In addition, for ES-accessible mini tumors (less than 1.2 cm) in the D1, D2, and proximal D3, EUS-assisted band ligation is also feasible, although the necessity is debated[40].

Beside complete resection, pharmacological treatments are necessary[41]. Glivec^® (imatinib), a TKI, is now widely prescribed in the United States and Western countries for high-risk local GIST patients as adjuvant therapy after surgery and in metastatic GIST patients as first-line treatment[2,20,42]. However, in China, only 8.1% of urban patients (Table 3, 7.1% of present cases) take imatinib, although the Glivec^® International Patient Assistance Program (GIPAP) was officially started in September 2003 by the China Charity Foundation, the Tumor Drug Department of Novartis, and the Max Foundation, with the objective of providing free medicines to patients who needed treatment. GIPAP has relatively strict eligibility criteria, and the annual Glivec^® cost is far beyond the economic realities in China, therefore only a few urban patients benefit from this donation program, let alone rural populations. Accordingly, neoadjuvant imatinib therapy is impractical in China, despite the fact that it is believed to allow LR in patients with locally advanced DGISTs[12].

In terms of the survival analyses, Duffaud et al[12] indicated that only mitotic rate predicted RFS; whereas the present study showed that not only mitotic rate, but also tumor size and various combinations of these two parameters in addition to Gold’s point, Fletcher’s risk and Miettinen’s risk predict RFS with similar statistical powers (P < 0.001). Moreover, preoperative diagnosis of a duodenal tumor is also a positive factor for better patient survival, however, it is not an independent parameter. Gold et al[17] of the Memorial Sloan-Kettering Cancer Center have developed a nomogram to predict the probability of 2- and 5-year RFS for resected GISTs patients, and tested it in patients from the Spanish Group for Research on Sarcomas and the Mayo Clinic. With regard to the present cohort, Table 4 shows that the actual 2-year RFS rate was similar to the predicted value (83.3% vs 83%), but the actual 5-year RFS rate was lower than the predicted value (50% vs 70%), this may be due to the more malignant behavior of DGISTs compared with other small intestinal GISTs.

The limitations of this study include its retrospective design, small sample size, single center experience, and lack of adjuvant therapy. As analyses with small numbers of patients sometimes give misleading results, readers should be careful in evaluating these findings. However, based on a comprehensive literature review, it is necessary to strengthen these results. Future prospective studies enrolling larger numbers of patients and/or multiple medical centers are required.