Pancreaticoduodenectomy versus local resection in the treatment of gastrointestinal stromal tumors of the duodenum

Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal (GI) tract. They can occur in the stomach (45% to 65%), small intestine (15% to 25%), colon and rectum (5% to 10%) or esophagus (5% to 10%)[1]. The clinical manifestations of GISTs are variable and render accurate diagnosis challenging. The current diagnosis of GIST is based on histologic and immunohistochemical criteria, the most important of which is the expression of the receptor tyrosine kinase KIT (CD117, c-Kit)[1, 2]. Duodenal GISTs are a very rare presentation, accounting for approximately 20% of tumors in the small intestine and 1% to 4% of all GISTs[3].

Complete surgical resection remains the best option for the treatment of GISTs, although imatinib mesylate, a tyrosine kinase inhibitor, is effective for GISTs. Unlike gastric GISTs, which can be adequately treated by wedge resection instead of formal gastrectomy, the optimal surgical procedures for duodenal GISTs have not been well characterized in the surgical literature[3‐9]. Some authors advocate radical procedures like pancreaticoduodenectomy (PD), whereas others support conservative procedures such as segmental duodenectomy and local or wedge resection (LR) based on their biology: since they are encapsulated tumors, GISTs do not widely infiltrate at the microscopic level and rarely metastasize to lymph nodes. In this study, we retrospectively reviewed the clinicopathologic characteristics of duodenal GISTs in our institution and compared the outcomes of patients undergoing PD versus LR with the main objective of determining if LR is a viable treatment option for these tumors.

A total of 48 patients who underwent surgical resection for a duodenal GIST were retrospectively reviewed from January 2006 to January 2012 at the First Affiliated Hospital, Zhejiang University School of Medicine. The preoperative diagnosis of a duodenal GIST was made through computed tomography or endoscopy. Furthermore, the precise diagnosis of GIST was made based on the standard histologic criteria. Complete tumor removal, either LR or PD, was the treatment in each case. In general, following LR the duodenal defect was closed primarily when possible or with a Roux-en-Y duodenojejunostomy when primary closure was not possible. Routine lymphadenectomy was not performed. The following characteristics were collected for each patient: age, sex, presenting symptoms, location of primary tumor, pathologic features including tumor size, mitotic count, immunohistochemistry, operative method, complications and the most recent follow-up information. Data on the use of preoperative or postoperative therapies including imatinib and chemotherapy were also recorded.

The primary study end point was disease-free survival (DFS), which was defined as the time from surgery to GIST recurrence. Patients who did not have evidence of local recurrence or metastasis at the last follow-up and those who died from causes unrelated to GIST were excluded from the DFS analysis. Furthermore, we analyzed prognostic factors, including age, sex (male versus female), clinical presentation (asymptomatic versus symptomatic), risk classification (high versus intermediate, low and very low) and operative method (PD versus LR).

Duodenal GISTs are a rare tumor entity, accounting for 4% of GISTs[10]. The clinical manifestations of duodenal GISTs are variable depending on tumor size and the existence of mucosal ulceration. As in previous reports, in our study, gastrointestinal bleeding was the most common clinical presentation of a duodenal GIST, followed by incidental finding, abdominal discomfort and abdominal pain. In general, duodenal GISTs most frequently involve the second portion of the duodenum, followed by the third portion, fourth portion and first portion. In the current study, we also noted that the second portion of the duodenum (35.4%) was most commonly involved, while we also found a high incidence of lesions in the first portion (22.9%). Furthermore, the median size of duodenal GISTs in our series was 4.7 cm. Similarly, most authors report a smaller size of tumor (4.0 to 5.0 cm) for duodenal GISTs compared with gastric and small bowel GISTs[6, 8, 11]. Interestingly, we noted that patients who ultimately required PD were more likely to present with a larger tumor and more commonly presented with a tumor in the second portion of the duodenum compared with LR. So we suggest that a tumor in the second portion of the duodenum, and which involves the papilla, pancreas or the duodenal bulb, or if the tumor is large with a high malignant potential, then PD is mandated since this type of tumor usually cannot be locally resected.

Surgical resection with a clear margin is the best option for the treatment of GISTs[7‐9, 12]. The size of the surgical margin along the segment of the digestive tract is not formally defined; however, there is little submucosal spread in GIST and clear margins of 1 or 2 cm are recommended. Theoretically, the choice of surgical procedure for a duodenal GIST mainly depends on its size, location and proximity to the duodenal papilla. However, the optimal surgical procedure for a duodenal GIST is currently unknown. Various surgical procedures have been reported, such as pancreaticoduodenectomy (PD) and local resection (LR), including wedge or segmental resection. Some authors advocate radical procedures like PD when the tumor is located at the medial wall of the second portion of the duodenum and involves the ampulla of Vater, or if the tumor has involved the adjoining organs[8, 9, 13]. Whereas, others support LR based on the tumor’s biology; since they are encapsulated tumors, GISTs do not widely infiltrate at the microscopic level and rarely metastasize to lymph nodes. As reported, LR can be used if the resection can achieve histologically clear margins (R0) and can preserve more of the pancreas parenchyma, duodenum and common bile duct without the cost of increased disease recurrence[7, 14‐16]. LR can be beneficial for patients because it contributes to a better quality of life and does not involve the excessive resection associated with PD. However, the main concern regarding LR is the risk from the margins involved and hence the theoretical increased risk of local recurrence[7].

Whilst there are several case reports and series about the clinicopathological features and frequency of duodenal GISTs, only a few reports compare the oncological long-term outcome of LR versus PD. The largest series of patients with duodenal GISTs (n = 156) indicated that prognosis was associated with tumor grade. In this review, 84 patients underwent LR (15 enucleations, 48 segmental resections and 21 wedge resections, 80%) and 21 (20%) underwent PD, and the retrospective analysis did not address the effect of operative method on disease recurrence after the operation[5]. Very recently, Johnston et al. retrospectively reviewed 96 patients at five institutions, and there were 58 LRs and 38 PDs. They concluded that factors associated with a worse recurrence-free survival included tumor size, mitotic count, being AJCC stage III disease and an NIH high risk classification rather than surgical approach[9]. Tien et al. analyzed nine patients who underwent PD and sixteen who underwent LR. They also showed that type of operation was not correlated to operative risk and disease recurrence[7]. Of thirteen patients reported by Beham et al., eight underwent segmental duodenectomy and five underwent pylorus-preserving PD. They suggested that the operative procedure and type of mutation (the KIT and PDGFRA receptor tyrosine kinase genes and comparative genomic hybridization) did not correlate with long-term survival[17]. In the present study, we noted that LR did not seem to have an adverse impact on the longer-term outcomes of patients. Multivariable analysis showed that operative method was not associated with DFS. The only significant factor associated with a worse DFS was an NIH high risk classification. Furthermore, PD was significantly associated with a longer operation time and a longer hospital stay compared to LR. Collectively, these data indicate that LR might be a reliable and curative option for a significant subset of patients with duodenal GIST, which is in accordance with the report of El-Gendi[18].

In the current study, the 1- and 3-year DFS was 100% and 88%, respectively. This is similar to previous reports, which showed that the DFS rate at 1 to 3 years after a complete surgical resection, for all kinds of resection, varies from 86% to 100%[8, 13, 16, 18]. Prognosis and recurrence are mainly dependent on tumor biology, which is determined by size and mitotic index (Fletcher scale)[19]. In our study, 75% of patients with a duodenal GIST had a low mitotic count, which is consistent with the findings reported by others (72% to 78%) and is lower than for gastric and small bowel tumors (where a mitotic count >5/50 HPF has been found in more than 30% of cases)[6‐9, 18]. The data from the study by Miettinen et al. also indicated that the prognosis for localized duodenal GISTs was poor with a high risk of relapse (>50%)[5]. Due to the earlier presentation, smaller tumors and a lower NIH risk classification, patients with a duodenal GIST may belong in a better prognostic category than other small bowel GISTs.

Imatinib mesylate, a tyrosine kinase inhibitor, plays a key role in the management of GISTs. It can be used in neoadjuvant therapy, adjuvant therapy and to treat tumor recurrence[20]. The role of neoadjuvant imatinib mesylate in the management of GISTs is currently the subject of an ongoing multicenter trial. In our study, 9 of 48 patients received postoperative adjuvant imatinib and there was no local recurrence or metastasis.

Because the recurrence of a duodenal GIST is correlated to tumor biology rather than type of operation, LR with clear surgical margins is a reliable and curative option for duodenal GISTs, and is compatible with long-term DFS. PD should be reserved for patients where LR is not technically feasible because of the involvement of the papilla of Vater.