Gastrointestinal stromal tumour of the duodenum: single institution experience

Abstract

Background

Primary gastrointestinal stromal tumours (GISTs) of the duodenum are rare. The aim of this study was to review the surgical management of GISTs in this anatomically complex region.

Methods

Retrospective review from January 1999 to August 2011 of patients with primary GISTs of the duodenum.

Results

Forty-one patients underwent resection of duodenal GISTs. All operations were performed with intent to cure with negative margins of resection. The most common location of origin was the second portion of the duodenum. Local excision (n = 19), segmental resection with primary anastomosis (n = 11) and a pancreatoduodenectomy (n = 11) were performed. Two patients underwent an ampullectomy with local excision. Peri-operative mortality and overall morbidity were 0 and 12, respectively. Patients with high-risk GISTs (P = 0.008) and those who underwent a pancreatoduodenectomy (P = 0.021) were at a greater risk for morbidity. The median follow-up was 18 months. Eight patients developed recurrence. High-risk GISTs and neoplasms with ulceration had the greatest risk for recurrence (P = 0.017, P = 0.029 respectively). The actuarial 3- and 5-year survivals were 85% and 74%, respectively.

Conclusion

The choice and type of resection depends on the proximity to the ampulla of Vater, involvement of adjacent organs and the ability to obtain negative margins. The morbidity depends on the type of procedure for GIST.

Introduction

Gastrointestinal stromal tumours (GISTs) are the most common mesenchymal neoplasms of the gastrointestinal tract and can arise anywhere from the oesophagus to the anus.^1,2 The most common locations of origin for GISTs are the stomach (60%–70%), small intestine (25%–35%), oesophagus (2%–3%), and rarely in the colon, rectum or appendix (collectively 5%).³ Duodenal GISTs account for 6%–21% of all resected GISTs of the small bowel.⁴ The operative management and outcomes of duodenal GISTs in the literature are limited to small case series or case reports.^1–3,5–10 The aim of the present study was to address the presentation, management and surgical outcomes of 41 patients with this gastrointestinal neoplasm occurring in a rather rare and anatomically complex location.

Patients and methods

After approval by the institutional review board, a retrospective review of the medical records of all patients treated operatively for duodenal GIST at the Mayo Clinic in Rochester, MN from January 1999 to August 2011 was performed. GISTs incidentally discovered by the pathologist in the duodenum resected for other pathology were excluded. Clinical presentation, diagnostic evaluation, type of operative procedure, pathological and morphological GIST characteristics, pathological risk category (low, intermediate and high)¹¹ and treatment outcomes were collected during a detailed chart review. The tumour was classified as low, intermediate or high risk of malignant potential based on the established criteria of size and mitotic activity per 50 high power field (HPF). The neoplasm was low risk if size was no larger than 2 cm and <5 mitosis/50 HPF, intermediate risk if size was 2–5 cm with >5 mitosis/50 HPF or >10 cm and no more than 5 mitosis/50 HPF, and high risk if size was >5 cm and >5 mitosis/50 HPF.¹¹

All operations were performed with intent to cure and negative margins of resection. The margin also included the cut ends of the pancreatic and bile duct in patients who underwent an ampullectomy when necessary to achieve complete resection. The follow-up protocol at this institution currently includes a history, physical examination and imaging every 6 months for 5 years and then yearly afterwards for low-risk GISTs and every 3–4 months for 3 years, then every 6 months until 5 years and annually thereafter for high-risk patients including those on tyrosine kinase inhibitors. Follow-up was until the most recent date of a clinic visit. Follow-up data were available for all patients until death or as of August 2011 with a median follow-up of 18 months (range: 12 days to 12 years). The last date of follow-up also included meaningful clinical correspondence and imaging studies from physicians of patients who chose to have their follow-up closer to their place of residence because this institution is a tertiary referral institution.

Surgical complications were classified according to the Clavien-Dindo Classification where Grade I is any deviation from the normal post-operative course, Grade II is those requiring pharmacological treatment, Grade III are those that require surgical, endoscopic or radiological intervention with or without general anaesthesia, Grade IV are those with a life-threatening complication with single- or multi-organ dysfunction and Grade V is complication resulting in death.¹²

Medians and ranges are used to express continuous data, unless specified otherwise. Logistic regression models were used for binary outcomes; P < 0.05 was considered statistically significant. Kaplan–Meier methodology and Cox regression analysis were used to analyse survival outcomes. Statistical analyses were performed using the JMP Statistical package (JMP software, Cary, NC, USA).

Results

Forty-one patients with duodenal GISTs underwent resection at this institution over a 12-year period. There were 22 men and 19 women with median age of 59 years (range: 39–79) at the time of operation. Eleven patients presented with symptoms of fatigue and 11 patients had a duodenal GIST recognized incidentally on imaging, such as computed tomography or endoscopic ultrasonography, or at the time of an operation for another indication. The median duration of symptoms was 30 days (range: 1 day to 36 months). A total of 18 patients had a pre-operative evidence of a clinically important haemorrhage attributed to the duodenal GIST, 14 of whom required a blood transfusion. The mean blood haemoglobin concentration initiating transfusion was 6.8 ± 0.4 g/dl (range 5–9.2 g/dl) and the median transfusion requirement for patients with pre-operative GI bleeding was 4 units (range 1–10 units). Pre-operative diagnostic studies varied and included an upper endoscopy in 32 patients, computed tomography in 30 patients and upper GI series and magnetic resonance imaging in 1 patient each. The second portion of the duodenum (n = 26) was the most common location. The median size was 4 cm (range 0.7–17 cm). Eighteen patients underwent an endoscopic ultrasonography in addition to the upper endoscopy, 14 of whom also had a fine needle aspiration (FNA) biopsy of the mass. Seven of the 14 FNA biopsy results were negative for GIST. The median size of the tumour for true positive and false negative pathology results on FNA was 3.2 cm (range: 1–4.5) and 3.2 cm (range: 1.3–4.3), respectively. Ulceration (n = 18) of the tumour and submucosal location (n = 18) were the most common gross pathological features of GIST, followed by exophytic growth (n = 5). Classification of the neoplasm by risk of recurrence included low risk (n = 27), intermediate risk (n = 5) and high risk (n = 9) GISTs.¹¹ GISTs in the third portion of the duodenum and those with ulceration were more likely to present with a pre-operative haemorrhage (P = 0.028 and P < 0.001, respectively). The clinical and pathological characteristics of GIST by location in the portion of duodenum are listed in Table 1.

Table 1.

Charasteristics of GIST per location in duodenum

Location (n)	D 1 (3)	D 2 (26)	D 3 (8)	D 4 (4)
Age in yearsa	53 (52–58)	58 (39–81)	62 (39–81)	60 (50–81)
Size in cma	5.3 (1–10)	4.5 (0.5–17)	3.3 (1.2–6.5)	6.2 (5–8)
Risk classf	3	26	8	4
Low	2	17	6	2
Intermediate	0	2	2	1
High	1	7	0	1
Operation
Local resection	2	14b	3	0
Pancreatoduodenectomy	0	11	0	0
Segmental resection	1d	1c	5c	4
Morbiditye,f	0	9	1	2
I	0	1	0	0
II	0	1	0	1
IIIa	0	5	1	0
IIIb	0	0	0	1
Iva	0	1	0	0
Presenting symptoms/signsf
Incidental	1	4	3	3
Fatigue	0	8	2	1
Abdominal pain	1	6	2	0
Melena/blood in stools	0	7	1	0
Early satiety	1	1	0	0

^aMean (range).

^bAmpullectomy in 2 of 14 patients.

^cSphincteroplasty in 1 patient in the D2 and D3 group.

^dEn block antro-duodenal resection, Bilroth II reconstruction.

^eMorbidity per Clavien-Dindo classification.

^fNumber of patients.

All patients in this series underwent resection and all resections were performed with intent to cure. All margins of resection were confirmed as negative by intra-operative frozen section histopathology. Local excision (n = 19), segmental resection with primary anastomosis (n = 11) and pancreatoduodenectomy (n = 11) were performed. Five of the 41 patients underwent laparoscopic resection, including 1 laparoscopic pancreatoduodenectomy. The median size of the tumours for local excision, segmental resection and pancreatoduodenectomy were 2.5 cm (range: 0.5–5.1), 5.4 cm (range: 1.2–10) and 7 cm (range: 2–17), respectively. The size of the GIST was a predictor for local excision vs. segmental resection in the first portion of the duodenum (P < 0.001). One en-bloc resection of the first portion of the duodenum with an antrectomy was included under the category of segmental resection (Table 1). Two patients who underwent local excision also had an ampullectomy. Pancreatico-biliary to duodenal continuity after a septectomy was performed with pancreatico-biliary stenting in each of these patients. One additional patient had a sphincteroplasty during a segmental resection extended immediately adjacent to the ampulla. Two patients had hepatic metastases (6 and 2 metastases each) resected at the time of the initial operation.

There was no peri-operative mortality. The overall morbidity rate was 29% [Clavien-Dindo class I (n = 2), II (n = 2), IIIa (n = 6), IIIb (n = 1) and IVa (n = 1)] with severe morbidity (Clavien-Dindo class ≥ III) in 20%. Severe morbidity included fascial dehiscence in 1 patient, a pancreatic fistula in 3 (all after pancreatoduodenectomy), enteric leak in 2 patients requiring percutaneous placement of drains and anastomotic site bleeding requiring endoscopic intervention in 2. Patients with high-risk histological characteristics of the GIST are more likely to have complications when compared with patients in the intermediate- or low-risk category (OR 8.8: CI 1.7–55.2; P = 0.008). Also, patients who underwent a pancreatoduodenectomy were more likely to develop complications than those who underwent a local excision or a segmental resection (OR 14.8: CI 2.5–131.3 P = 0.002). One patient who underwent a local excision of the tumour with sphincteroplasty developed a benign, distal common bile duct stricture requiring a Roux-en-Y hepaticojejunostomy. Three patients in the ampullectomy and sphincteroplasty group did not have any complications.

The median follow-up was 18 months (range: 12 days to 12 years). Two patients received neoadjuvant and 7 patients received adjuvant therapy with Gleevec® (Imatinib; Novartis Pharmaceutical Corporation, Basel, Switzerland). There have been seven deaths during the follow-up period; four were related to recurrent GISTs, whereas three were unrelated to the disease. There have been eight recurrences of the GIST, involving six patients in the high-risk and one each in intermediate- and low-risk groups. Three of the eight patients with recurrence GISTs underwent resection of the recurrence, including a metastectomy of a pelvic recurrence necessitating a hysterectomy/bilateral salpingo-oophorectomy combined with recto-sigmoidectomy, resection of the remainder of duodenum with Roux-en-Y reconstruction for local duodenal recurrence and one with both a local duodenal recurrence and hepatic metastasis had a local excision and sub-segmental hepatic resections. Six of the seven patients with recurrent disease underwent some form of tumour-directed therapy, including targeted inhibitor therapy in six, surgical resection in three and radiofrequency ablation of a hepatic recurrence in one. Most patients had more than one form of therapy for recurrence. Intervention was discontinued in one patient who could not tolerate imatinib therapy and whose performance status precluded other interventions. Patients with ulceration had the greatest risk of recurrence compared with submucosal or exophytic growth (3-year recurrence-free survival was 43% vs. 100%; P = 0.031). Those in the pathological high-risk category were also at an increased risk of recurrence (3-year recurrence-free survival was 29%, 67% and 90% for the high-, intermediate- and low-risk pathological category, respectively, P = 0.017).

The median time to recurrence was 18 months (range: 7 months to 11 years). There were 15 and 11 patients evaluated at 3- and 5-years, respectively. The 3- and 5-year overall survival was 85% and 74%, respectively (Fig. 1a). The 3- and 5-year disease-free survival was 80% (Fig. 1b). Survival outcomes were not influenced by type of operation or pathological risk group (P = 0.22 and P = 0.76, respectively). Recurrent disease was associated with decreased survival (RR 6; 95% CI 1.26–32.66; P = 0.02). Recurrence accounted for all four disease-related deaths.

Figure 1.

(a) Kaplan–Meier curve for overall survival (in months). (b) Kaplan–Meier curve for recurrence-free survival (in months)

Discussion

GISTs are believed to originate from the pluripotent mesenchymal stem cells destined to differentiate into the interstitial cells of Cajal and are characterized by expression of the transmembrane receptor tyrosine kinase c-KIT and the CD 117 antigen.¹⁰ GISTs have a predilection for the upper gastrointestinal tract and when sought have been found as incidental lesions in up to 10% of pathological specimens resected for other indications.¹¹ An incidental pre-operative diagnosis of duodenal GIST accounted for 27% of patients in our study (n = 11); a similar number of patients had symptoms suggestive of fatigue secondary to chronic blood loss anaemia. The risk of recurrence for small bowel GISTs is classified pathologically as low, intermediate or high based on the number of mitosis per high power field plus the size of the tumour.¹³ The greatest chance of disease-free survival is based on the ability to achieve pathologically negative margins during the initial operation with clinically localized disease.⁹ Comparing size and the mitotic index in various locations in the gastrointestinal tract, duodenal GISTs appear to have the greatest risk of recurrence.¹⁴ Given the complex anatomy in the region of the duodenum, resection of a duodenal GIST mandates a carefully planned approach depending on the location in the duodenum and whether the GIST is limited to the anti-mesenteric vs. the mesenteric border.

Currently the treatment of choice for GIST, regardless of origin, is local excision with histologically negative margins.¹¹ The low frequency of local recurrence after limited resections and the local growth characteristics of GIST do not dictate extended resections. A GIST arising from the duodenum is more problematic because of adjacent anatomy. Although large GISTs more frequently dictate resection of adjacent structures, a small GIST may arise at organ interfaces or potentiate a desmoplastic reaction which dictates resection of adjacent structures. This finding appears to be more frequent for duodenal GISTs. Our findings highlight this issue. GISTs arising in the second and third portion of the duodenum do not always mandate a pancreatoduodenectomy; indeed, only 11 of the 26 GISTs in these locations were treated by pancreatoduodenectomy. The median size of GIST in this location was 7 cm, and 4 of the 11 patients had a severe desmoplastic reaction. The goal of resection is to obtain a histopathologically negative margin. Although size was a significant factor for selection of operative approach with GISTs arising in the first portion of the duodenum, proximity to the ampulla of Vater was the most frequent factor affecting operative approach in the second portion of the duodenum. Negative margins are often achieved with local excision or segmental resection while preserving the distal bile duct and pancreatic duct. Involvement of the ampulla may dictate an ampullectomy with limited resections or pancreatoduodenectomy. Biliary sphincteroplasty with or without septotomy may also be advantageous to ensure free flow of bile and pancreatic secretion if resection margins extend to the ampulla. The morbidity of an ampullectomy was less than that of a pancreatoduodenectomy (P = 0.018). For a smaller tumour that does not cause a severe desmoplastic reaction around the pancreatic head within which malignant extension cannot be excluded, a less invasive but technically challenging pancreas-sparing duodenectomy may be considered.¹⁵

Non-periampullary GISTs or those arising on the anti-mesenteric border of the duodenum can be addressed by local excision or a segmental duodenectomy depending on the circumferential extension (Table 1). The defect in the duodenum is usually closed in a transverse fashion to ensure lumen patency. Another option for large defects that can cause luminal narrowing after local resection is an on lay Roux-en-Y jejuno-duodenostomy.⁷ In the D1, D3, and D4 segments of the duodenum, a larger neoplasm may be resected with a segmental duodenal resection with end-to-end anastomosis. A lymphadenectomy is not carried out routinely, because GISTs do not routinely spread to the lymph nodes except in the rare patient with Carney's triad.^16,17 Large tumours and those with a severe desmoplastic reaction involving adjacent organs may need an en bloc resection to ensure negative margins which is the most important factor in recurrence-free and progression-free survival.⁹ Local excision or segmental resections with anastomosis or Roux reconstructions have low morbidity compared with the more extensive pancreatoduodenectomy which may prove necessary to achieve negative margins. In the setting of duodenal GIST, laparoscopic excision with sound oncological principles is feasible. Long-term outcomes, however, depend on histology, size of the primary GIST and the presence of metastatic disease at the time of the initial operation.⁹ Although these neoplasms may shrink with neoadjuvant therapy and improved recurrence-free survival has been demonstrated with adjuvant imatinib therapy,¹⁸ the number of patients undergoing such treatment in the present study was too few to draw any meaningful conclusions. Another limiting factor was the retrospective nature of the study and a lack of randomization given the low incidence of GISTs in this anatomic location.

Conclusion

The duodenum is an uncommon site for GISTs. When present, symptoms are usually non-specific and can include fatigue from occult anaemia or abdominal pain. Asymptomatic duodenal GISTs also occur. Management is complicated by the location of the GIST in the second portion of the duodenum in relation to the pancreaticobiliary confluences. Sound oncological principles of obtaining negative margins should guide operative management. The operative procedure is dictated by proximity to or involvement of either the distal biliary tract or extension into the pancreatic head and can include local excision, segmental resection, transduodenal resection with ampullectomy or sphincteroplasty and a pancreatoduodenectomy. Our data suggest that there may be a role for operative intervention for recurrence in some patients. Whether adjuvant therapy for patients with duodenal GISTs differs from that of other gastrointestinal GISTs could not be determined from this study.

Conflict of interest

None declared.