- Email: [email protected]
EUS-guided pancreatic duct intervention: outcomes of a single tertiary-care referral center experience
ORIGINAL ARTICLE: Clinical Endoscopy
EUS-guided pancreatic duct intervention: outcomes of a single tertiary-care referral center experience Larissa L. Fujii, MD,1 Mark D. Topazian, MD,1 Barham K. Abu Dayyeh, MD,1 Todd H. Baron, MD,1 Suresh T. Chari, MD,1 Michael B. Farnell, MD,2 Ferga C. Gleeson, MD,1 Christopher J. Gostout, MD,1 Michael L. Kendrick, MD,2 Randall K. Pearson, MD,1 Bret T. Petersen, MD,1 Mark J. Truty, MD,2 Santhi S. Vege, MD,1 Michael J. Levy, MD1 Rochester, Minnesota, USA
Background: EUS can provide access to the main pancreatic duct (MPD) for therapeutic intervention. The longterm clinical success of EUS-guided MPD interventions is unknown. Objective: To determine technical and clinical success rates, predictors of success, and long-term outcomes of EUS-guided MPD intervention. Design: Retrospective, single-center study. Setting: Tertiary-care referral center. Patients: Forty-five patients. Intervention: EUS-guided MPD stent retrieval or placement. Main Outcome Measurements: Technical and clinical success rates, adverse events, and long-term clinical outcomes. Results: Among the 45 patients, 37 had undergone failed ERCP, and 29 had surgically altered anatomy. Median follow-up after initial EUS-guided intervention was 23 months. Two patients underwent EUS for stent removal, and EUS-guided MPD stent placement was attempted in 43 patients. Technical success was achieved in 32 of 43 patients (74%) with antegrade (n Z 18) or retrograde (n Z 14) stent insertion. Serious adverse events occurred in 3 patients (6%). Patients underwent a median of 2 (range 1-6) follow-up procedures for revision or removal of stents, without adverse events. Complete symptom resolution occurred in 24 of 29 patients (83%) while stents were in place, including all 6 with nondilated ducts. Stents were removed in 23 patients, who were then followed for an additional median of 32 months; 4 patients had recurrent symptoms. Among the 11 failed cases, most had persistent symptoms or required surgery. Limitations: Retrospective study design, individualized patient management. Conclusion: EUS-guided MPD intervention is feasible and safe, with long-term clinical success in the majority of patients. EUS provides important treatment options, particularly in patients who would otherwise undergo surgery. (Gastrointest Endosc 2013;78:854-64.)
Endoscopic retrograde pancreatography (ERP) is the least-invasive method to access the main pancreatic duct (MPD) and provide therapeutic intervention. Historically,
failed ERP necessitated repeat endoscopic attempts and/or percutaneous radiologic or surgical intervention. EUSguided pancreatic interventions are a new alternative to
Abbreviations: ERP, endoscopic retrograde pancreatography; MPD, main pancreatic duct.
Received March 18, 2013. Accepted May 9, 2013.
DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
Current affiliations: Division of Gastroenterology and Hepatology (1), Division of Gastroenterologic and General Surgery, Mayo Clinic Rochester, Rochester, Minnesota (2).
Copyright ª 2013 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2013.05.016
Reprint requests: Michael J. Levy, MD, Mayo Clinic, Division of Gastroenterology and Hepatology, 200 1st St SW, Rochester, MN 55905.
854 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
Take-home Message EUS-guided pancreatic duct access and stent placement often can provide drainage safely, with good long-term clinical outcomes in selected patients. Prior endoscopic retrograde pancreatogram during the same procedure was associated with failed EUS-guided stent placement, reflecting the importance of determining early which patients would benefit from EUS guidance. Patients with symptoms suggestive of a pancreatic origin but who have normal pancreatogram results during EUS also may benefit from stent placement.
Figure 1. EUS-guided pancreatogram from the gastric body reveals partial pancreatic duct obstruction due to a pancreaticojejunal anastomotic stricture.
ERP, particularly when ERP has failed or is unlikely to succeed (as in surgically altered anatomy). The proximity of the gut lumen to the MPD and the unparalleled image quality and continuous real-time imaging of EUS allow precise targeting of the MPD. Although ductal access is typically easily achieved by using FNA techniques, subsequent interventions may be challenging. Limited data have been published regarding this therapeutic modality, particularly long-term patient outcomes. We assessed technical and clinical outcomes and predictors of success among patients undergoing EUS-guided MPD intervention at our institution.
30 days. Each EUS was performed by 1 of 2 endosonographers (M.J.L. or M.D.T.), both of whom have performed more than 10,000 upper GI tract EUS examinations. Technical success was defined as complete if the intent of the procedure was accomplished; otherwise it was defined as failed. Patient outcomes were monitored for at least 12 months after initial intervention. Patients who had less than 12 months of follow-up within our facility were contacted via telephone to determine clinical outcome. The 3 patients who underwent stent placement and could not be contacted were excluded from the analysis on long-term analysis. All clinical, laboratory, radiologic, and surgical findings after EUS were reviewed. The last available clinical follow-up was used to assess patients’ responses to interventions. Patients were deemed to have complete responses to therapy only if it was clearly documented that all clinical symptoms fully resolved after the procedures. Patients were considered to have partial responses when there was improvement in severity and/or frequency of presenting symptoms. Patients who experienced no benefit were categorized as having no response to endoscopic therapy.
MATERIALS AND METHODS Statistical analysis A prospectively maintained EUS database was reviewed to identify all patients who underwent attempted EUSguided MPD intervention at Mayo Clinic, Rochester, Minnesota. The Institutional Review Board granted approval for retrospective data retrieval and patient contact. Written informed consent was obtained before each procedure. Data collected pertained to all aspects of the clinical, laboratory, and radiologic findings. All endoscopic, interventional radiologic, and surgical procedures performed before and after the attempted EUS-guided MPD access were reviewed. Pancreatograms obtained after initial MPD access (but before guidewire placement) were categorized during the procedure as showing complete obstruction (no contrast material drainage to the small bowel lumen), partial obstruction (slow or poor drainage of contrast material to the small bowel), or no definite obstruction (good drainage of contrast material to the small bowel). Adverse events possibly related to the procedure were identified within
Continuous variables are reported as either a mean and standard deviation (SD) or median and range. Means were reported unless the SD was larger than the mean itself, in which case medians were reported. The t test was used for continuous variables, and a Pearson chi-square analysis was used for categorical variables. A P value of ! .05 was considered statistically significant.
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 855
Technique A therapeutic channel curvilinear array echoendoscope (CF-UCT140-AL5, GF-UC140P-AL5, or GF -UC160P-AT8A; Olympus America, Inc, Center Valley, Pa) by using carbon dioxide insufflation was used to access the MPD. The route was selected based on the site of pathology, to traverse the shortest distance between the transducer and the MPD, to avoid intervening structures, and to provide a needle trajectory to facilitate subsequent track dilatation and stent placement. A 22-gauge or 25-gauge FNA needle (Cook
EUS-guided pancreatic duct intervention
Fujii et al
Figure 2. EUS-guided antegrade stent placement. A, A guidewire is advanced through the needle into the pancreatic duct. B, The track between the gastric body to the pancreatic duct is balloon dilated. C, Antegrade transluminal stent placement. The pancreaticoenteric anastomosis could not be traversed. D, During a subsequent examination, the anastomotic stricture was traversed, and an antegrade transanastomotic stent was placed.
Medical, Bloomington, Ind) was typically used for the initial pancreatogram. When MPD drainage was attempted, a 19gauge FNA needle (Cook Medical) was used for duct access (Fig. 1) followed by guidewire placement via the needle. Guidewires were either 0.035 or 0.025 inches in diameter and included straight and angled hydrophilic wires (Glidewire, Jagwire; Boston Scientific, Inc, Natick, Mass). Attempts were made to advance the guidewire in an antegrade fashion across the site of MPD obstruction to the gut lumen, via the papilla or pancreaticoenteral anastomosis (in postoperative cases) under EUS and fluoroscopic guidance. Pancreatic stent placement was then performed over the guidewire. Stent placement was categorized as (1) retrograde (sometimes referred to as a rendezvous procedure) from the papilla and/or anastomosis into the MPD via another endoscope or (2) antegrade placement across the proximal gut wall into the MPD via the echoendoscope. Antegrade stent placement was further subdivided based on whether the stent crossed the site of ductal obstruction and papilla and/or anastomosis (transpapillary/transanastomotic) or not (transluminal). For retrograde stent placement, the guidewire was left in place, and the echoendoscope was withdrawn. Either
a side-viewing duodenoscope (TJF-160VF; Olympus America) in patients with native anatomy or a pediatric or adult colonoscope (PCF-Q180AL or CF-H180AL; Olympus America) in patients with postoperative anatomy was advanced to the pancreatic-enteric junction. The transpancreatic guidewire was grasped with a snare or biopsy cable and withdrawn through the endoscope, or a second guidewire was passed retrograde alongside the first wire into the MPD, sometimes after sphincterotomy. A stent was then deployed in retrograde fashion, from the small bowel through the papilla and/or anastomosis and into the MPD. For antegrade stent placement (Fig. 2), transmural track dilatation was performed by using hydrostatic balloons (Titan Biliary Dilation Balloon; Cook Medical or Hurricaine Biliary Dilation Balloon; Boston Scientific), tapered catheters (Sohendra; Cook Medical), and/or cannulas (Proforma; ConMed Endoscopic Technologies, Utica, NY). When these devices failed to traverse the transmural track to the pancreatic duct, a wire-guided needle-knife catheter was used (Boston Scientific), initially without electrosurgical current, and if needed, with short bursts of electrosurgical current at standard sphincterotomy current, to access the MPD. If the guidewire did not initially cross the site of
856 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
Figure 3. Overview of the technical success and adverse events for each procedure type. MPD, main pancreatic duct; TS, technical success.
pancreatic duct obstruction, further attempts were made after accessing the MPD with a catheter via the transmural track. Both straight and double-pigtail stents were used (Cook Medical). Typically, the proximal end of the stent was left in the lumen of the proximal GI tract, facilitating subsequent stent exchange, upsizing, or removal. Additional side holes were often created within the intraductal portion of the stent to facilitate drainage.
RESULTS Forty-five patients (mean age 57 y [SD 16 y], 24 men, mean body mass index 23.5 kg/m2 [SD 3.6]) underwent 49 EUS examinations in an attempt to access the MPD under EUS guidance between January 2006 and December 2012 (initial examinations) as well as 59 subsequent examinations to exchange or remove stents (follow-up examinations). The number of total procedures, successful procedures, and adverse events are outlined in Figure 3. Forty-four of the initial 45 examinations were performed with patients under general anesthesia. Intravenous antibiotics (fluoroquinolones or piperacillin-tazobactam) were administered prophylactically in 30 procedures. The median MPD caliber at the site of attempted entry based on EUS or by CT if not measured during EUS was 4.4 mm (range 0.8-10 mm). www.giejournal.org
The clinical and preprocedure structural diagnoses are listed in Table 1. Eighteen patients had more than 1 clinical indication. There was no difference in age, sex, and mean MPD size in patients with and without recurrent acute pancreatitis (the most common clinical symptom). Previous pancreatic surgery had been performed in 29 patients (64%) at 19.9 months (range 3.3-150.6 months) before EUS. Types of prior pancreatic surgery and altered anatomy are listed in Appendix 1, available online at www.giejournal.org. EUS-guided MPD intervention was performed after failed ERP in 37 patients (82%), among whom 21 patients had surgically altered anatomy, and 16 had normal anatomy. Twenty-four ERPs had been attempted on a previous day, whereas 15 ERPs were attempted immediately before EUS-guided intervention. Three of the 37 patients underwent at least 2 separate failed ERP procedures. Reasons for failed ERP are outlined in Appendix 2, available online at www.giejournal.org. Two patients were diagnosed with recurrent pancreatic cancer after EUS-guided MPD intervention.
Attempted stent retrieval Two patients underwent EUS for attempted retrieval of a retained MPD anastomotic stent placed during pancreaticoduodenectomy performed 6 and 1.5 years previously. Both patients underwent EUS for abdominal pain, at which Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 857
EUS-guided pancreatic duct intervention
Fujii et al
In the remaining 43 patients, the procedural intent was to place an MPD stent. Pancreatogram before the attempted stent placement showed complete obstruction in 13 patients, partial obstruction in 11, and no definite obstruction in 13 patients. The remaining 6 patients had either a disconnected duct (n Z 5) or the MPD was not identified (n Z 1). MPD access was achieved by using the gastric body for the majority of cases (41 patients, 95%). One patient had a failed attempt at accessing the MPD from the duodenal
bulb, whereas the remaining patient had a successful EUSguided pancreaticobulbostomy via the retrograde route. Complete technical success with stent placement was achieved in 32 patients (74%). Stent placement was possible during the index examination in 30 patients (94%), whereas the other 2 patients required 1 and 2 additional attempts, respectively. The stents were deployed by using a retrograde or antegrade technique in 14 and 18 patients, respectively; of the 18 antegrade stent placements, the stent crossed the site of pancreatic duct obstruction and the papilla and/or anastomosis in 13 and did not in 5. Among these 5, a stent could be advanced across the site of ductal obstruction in 2 patients during a follow-up procedure (Fig. 2D). The mean procedure duration was 162 minutes (SD 37), 125 minutes (SD 66), and 130 minutes (SD 19) for the retrograde and antegrade methods, respectively; retrograde stent placement was significantly lengthier than either antegrade approach (P Z .05). This is likely explained by a significantly lower mean procedure time if 1 endoscopist performed the entire examination as compared with when separate physicians performed the EUS and ERP portions (109 37 minutes vs 171 44 minutes, respectively; P ! .001). In addition, the retrograde technique was less commonly used in postoperative patients (84% vs 35% antegrade and retrograde techniques, respectively; P Z .006) but more commonly used after failed ERP attempts (100% of patients undergoing the retrograde technique had a prior ERP compared with 61% of patients who had an antegrade approach; P Z .008). Although it was possible to access, inject contrast material, and insert a guidewire into the MPD in 42 patients (98%), failed stent placement (n Z 11) occurred because of the inability to advance the guidewire into the MPD (n Z 1) or through the anastomosis or papilla (n Z 8), inability to adequately dilate the transmural track (n Z 1), or loss of the guidewire during ERP (n Z 1). A needleknife was used to facilitate transgastric ductal access in 6 patients when other efforts failed, allowing transgastric access and examination completion in all 6 patients. When the procedure was successful, the track was dilated to a mean of 5.5 mm (SD 1.5). Baseline characteristics between the patients with and without successful stent placement are shown in Table 2. There was a statistically greater chance of failure when EUS access for stenting was attempted on the same day as a failed ERP. One (n Z 30) or 2 stents (n Z 2) were placed during the initial examination. Straight plastic and double-pigtail plastic and metal stents were used in 17, 11, and 1 case, respectively (3 unknown). The plastic stent diameters were 10F (n Z 1), 7F (n Z 28), 5F (n Z 3), and 3F (n Z 1) with mean stent length of 9 cm (SD 3.3). The 1 covered metal stent used was 10 mm in diameter and 80 mm in length. In 12 patients, stents (11 placed retrograde and 1 placed antegrade) were removed during the subsequent procedure
858 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
TABLE 1. Indications for attempted EUS-guided intervention in 52 patients Indications
No.
Clinical (n Z 79; 18 had more than 1 indication) Recurrent acute pancreatitis
30
Chronic pancreatitis
8
Pancreatic duct leak
3
Retained surgically placed stent
2
Abdominal pain and dilated MPD
23
Weight loss and dilated MPD
8
Steatorrhea and dilated MPD
5
Structural (n Z 45) Benign anastomotic stricture
25
Benign non-anastomotic stricture
5
Malignant non-anastomotic stricture
1
Retained MPD stent
2
Disconnected pancreatic duct
5
Pancreatic duct leak
3
Symptomatic pancreas divisum
3
Papillary stenosis
1
MPD, main pancreatic duct.
time the retained stents were identified and believed responsible for the patients’ pain. EUS transductal displacement into the jejunum was successful by using a 19-gauge FNA needle and guidewire advancement to dislodge the stent in the patient with chronic pancreatitis with an MPD diameter of 1.5 mm. The total procedure duration was 160 minutes (44 minutes of fluoroscopy). After stent removal, the patient’s pain intensity improved but persisted. EUS-guided stent retrieval was unsuccessful in the other patient with a MPD diameter of 1.4 mm. The total procedure duration was 96 minutes (33 minutes of fluoroscopy). The patient’s abdominal pain persisted until his death from recurrent pancreatic cancer 7 months later.
Attempted stent placement
Fujii et al
EUS-guided pancreatic duct intervention
TABLE 2. Patient characteristics relative to technical success of stent placement Stenting successful (n [ 32)
Stenting failed (n [ 11)
P value
57.8 (16.1)
52.7 (18.6)
.39
18
4
.25
23.5 (3.7)
23.6 (3.7)
.96
Prior pancreatic surgery
20
7
.95
Prior ERP attempt
25
11
.09
ERP immediately before EUS
7
7
.04
Fellow present
18
5
.54
Dilated MPD (R3 mm)
3
6
.35
4.9 (2.4)
3.9 (3.2)
.43
Complete MPD obstruction by pancreatogram
5
8
.14
R5 EUS features of chronic pancreatitis
11
6
.24
Duration, total time, mean (SD), min
142 (51)
154 (54)
.49
Duration, EUS mean (SD), min
89 (45)
97 (20)
.68
Duration, fluoroscopy, mean (SD), min
40 (23)
34 (20)
.43
Age, average (SD), y Male, no. Body mass index, mean (SD), kg/m2
Maximum MPD caliber, mean (SD), mm
SD, Standard deviation; ERP, endoscopic retrograde pancreatography; MPD, main pancreatic duct.
after a median of 3 months (range 1-42). In 1 patient who underwent prior pancreaticoduodenectomy, stent retrieval was attempted at 4 months, but the pancreaticojejunal anastomosis could not be reached by using a single-balloon endoscope. It was later removed by using EUS-guidance 46 months after insertion. Of the remaining 11 patients, 5 failed stent replacement (3 because of the inability to identify the papilla and 2 because of the inability to recannulate the MPD), 4 had normal pancreatogram results, and 2 were removed without attempted re-evaluation of the MPD. Two patients did not undergo repeat endoscopy because of their poor prognosis related to underlying malignant disease. The remaining 18 patients (2 retrograde and 16 antegrade stent placements) underwent a median of 2 (range 1-6) additional endoscopic procedures, with stent replacement or upsizing, with the first follow-up examination occurring at a median of 46 days (range 13-77) after the initial procedure. The maximum number of stents placed was 1, 2, 3, 4, and 5 stents in 2, 7, 6, 2, and 1 patients, respectively. In general, these patients underwent stent placement until maximum resolution of pancreatic duct stenosis was obtained. Ten patients had their stents endoscopically removed at a mean of 9 months (range 2.5-39), 1 had her stents removed during completion pancreatectomy, 2 died from underlying disease with stents in place, and 5 remain with stents in place at a median of 42.6 months (range 13-69.5) after initial placement. Detailed follow-up was available for 9 of the 11 patients in whom stent placement failed, with a median follow-up
of 31 months (range 3-80 months). Five patients (45%) required surgery for ongoing symptoms, including lateral pancreaticojejunostomy (n Z 2), revision of pancreaticojejunal (n Z 2), and partial pancreatectomy (n Z 1). After surgery, all but 1 patient had complete resolution of symptoms. Of the 6 patients who have not undergone surgery after EUS, 2 are asymptomatic at 15 and 30 months without intervention, 2 experience ongoing recurrent acute pancreatitis that has been medically managed, and 2 have insufficient follow-up. Complete follow-up was available until death or at least 12 months in 29 of 32 (91%) successfully stented patients. Six died of underlying malignancy or unrelated causes at 11 months (range 2-54 months); of these, 4 patients had complete pancreatic symptom resolution until death. In the remaining 23 patients, complete clinical success with full symptom resolution occurred in 16 patients (69.6%) at the time of last follow-up (median 37 months [range 12-72 months]). Complete symptom resolution while the stent was in place was seen in 24 patients (82.8%). Stents were removed in 23 patients after a median of 4 months (range 1-47 months); the patients were then followed for a median of 32 months (range 15.5-50.5 months), with symptom recurrence in 4 patients after a median of 14 months (range 5-32 months). Seven living patients (30.4%) with complete follow-up had partial clinical success with symptom recurrence at a median of 7 months (range 2-45 months). Of these, 3 had their stents previously removed, whereas 4 had their stents in place at the
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 859
EUS-guided pancreatic duct intervention
Fujii et al
TABLE 3. Patient characteristics relative to long-term clinical success after stent placement* Partial clinical success (n [ 9)
Complete clinical success (n [ 20)
58.8 (12.2)
58.1 (18.7)
.92
5
11
.98
22.2 (4.2)
24.5 (3.3)
.12
Fellow present during first procedure
5
7
.30
Postoperative anatomy
8
11
.08
Structural indication: benign anastomotic stricture
8
10
.046*
Clinical indication: recurrent acute pancreatitis
3
10
.40
Clinical indication: abdominal pain
3
7
.93
Prior ERP attempt
5
17
.09
ERP immediately before EUS
1
4
.87
Dilated MPD (R3 mm)
9
14
.065
5.4 (1.8)
4.4 (2.6)
.33
Complete MPD obstruction by pancreatogram
2
6
.61
Antegrade stent insertion
6
10
.40
Transpapillary/transanastomotic stent placement during first procedure
8
16
.56
Double-pigtail stent placed during first procedure
4
7
.87
Stent length during first procedure (cm)
10.8 (3.3)
8.1 (2.6)
.03*
R5 EUS features of chronic pancreatitis
3
5
.64
Stent exchanged
6
11
.49
Stent eventually removed
6
15
.82
Age, mean (SD), y Male, no. 2
Body mass index, mean (SD), kg/m
Maximum MPD caliber mean (SD), mm
P value
SD, Standard deviation; ERP, endoscopic retrograde pancreatography; MPD, main pancreatic duct. *Includes 29 patients in whom long-term follow-up (until death or O12 months) was available.
time of symptom recurrence. None required surgical or percutaneous intervention. Nineteen patients had a repeat CT scan after stent placement at a median of 335 days (range 30-1393 days) after the initial procedure. The mean MPD size decreased from 5.9 mm (SD 2.3) before the procedure to 4.5 mm (SD 2.2) after stent placement, for a mean difference of 1.4 mm (SD 1.6). Four patients had a slight increase in MPD size ranging between 0.2 and 0.8 mm, which may be related to differences in measurements rather than an actual increase. There was no difference in mean MPD size before and after stent placement or mean difference in MPD size in patients who had complete versus partial and/or no clinical success. Patient and procedural characteristics relative to longterm clinical success are detailed in Table 3. Shorter stent lengths and any structural indication other than benign anastomotic stricture were associated with complete clinical success in univariate analysis. Notably, all 6
patients with normal MPD diameter (defined as diameter !3 mm) had complete responses to stent placement (P Z .065 compared with patients with dilated ducts). The indications for intervention in these patients were recurrent acute pancreatitis (n Z 2), pancreatic duct leakage (n Z 2), abdominal pain (n Z 1), and weight loss (n Z 1) in the setting of surgically altered anatomy (n Z 2) or pancreas divisum (n Z 1). The initial pancreatogram in patients with successful stent placement revealed complete obstruction in 8 patients, partial obstruction in 10 patients, no definite obstruction in 7 patients, and 4 patients with a disconnected pancreatic duct. There was no correlation between pancreatogram findings and likelihood of clinical success (P Z .26 for the comparison between those with partial or complete obstruction versus no definite obstruction). Six of the 7 patients (85.7%) with absence of obstruction on pancreatogram had complete symptom resolution, which was not statistically significant (P Z .26 compared with those with partial or complete
860 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
obstruction). Patients having prior pancreatic surgery experienced a nonsignificant trend toward a lower likelihood of clinical success.
EUS-guided pancreatic duct intervention
The concept of combining ERP with EUS technology led to the first report in 1995 of EUS-guided pancreatography for a patient requiring MPD stone removal after pancreaticoduodenectomy.1 Since that time, there has been increased use and evolution in EUS-guided MPD techniques, particularly in patients in whom previous ERP attempts were unsuccessful. ERP may be particularly challenging after pancreatic head resection because of the inability to reach, identify, or cannulate the pancreaticojejunal anastomosis. Prior reports of EUS-guided MPD intervention have described endoscopic methods and technical and clinical success rates and have demonstrated feasibility and safety.1-16 Our study describes the technical and long-term clinical success rates of EUSguided pancreatic duct intervention in the largest published cohort of patients to date and explores the predictors of technical and clinical success.
Our overall technical success rate of any EUS-guided MPD intervention was 73%, which is within the range previously reported.2,4,8,10-18 This success rate is substantially higher than the success rate reported with ERP after pancreaticoduodenectomy,11,17,19-21 and, as a result, EUSguided MPD intervention is generally our first endoscopic intervention in patients with symptomatic pancreatic duct obstruction after a Whipple procedure. The most common site of EUS-guided puncture was the gastric body (pancreaticogastrostomy), which we believe provides enhanced stabilization, echoendoscope positioning, and access to the MPD. Although the most common cause of a failed procedure was the inability to negotiate the guidewire through the site of obstruction or across the papilla or pancreaticojejunal anastomosis, there was no difference in the ability to place a stent in patients with and without complete obstruction seen on pancreatography. In addition, contrary to previous recommendations to intervene only in patients with dilated MPDs,22,23 we were able to access and intervene on ducts as small as 0.8 mm in diameter. Previous studies also report the ability to access a normal-caliber MPD by using EUS guidance.2,10 Despite a higher percentage of failed stent placements with a normal MPD size, this did not reach statistical significance. This emphasizes that MPD diameter is 1 component to factor into the decision making process as to whether a patient may benefit from pancreatic duct stent placement. Other non-anatomic factors such as the patient’s clinical presentation and course should be considered. Additional studies are needed to clarify the role and potential benefit of draining a normal-caliber duct and to identify clinical and imaging characteristics that may be associated with a good response. Prior ERP performed during the same procedure was the only statistically significant risk factor for failed stent placement. It is unclear whether this was because of anatomic factors, inflammation induced by ERP attempts, or less commitment by the endosonographer in the setting of a long, failed ERP attempt. Some of the case series with the highest technical success (O90%) often favored antegrade transluminal stent placement without traversal of the MPD obstruction and papilla or anastomosis.8,13,16 Similarly to an externally placed biliary catheter, this technique provides only 1 route of MPD drainage to the stomach. We favored antegrade stent placement across the papilla and/or pancreaticojejunal anastomosis, which provides dual drainage to the stomach and small bowel and treats the underlying stricture. When the ductal obstruction and papilla and/or anastomosis could not be traversed during initial antegrade stent placement, we placed a transluminal stent into the MPD (pancreaticogastrostomy) and reattempted traversal of the ductal obstruction and papilla and/or anastomosis during a subsequent procedure. Complete symptom resolution occurred in 24 of 29 patients (83%) while stents were in place, and only 4 patients had recurrence of symptoms after removal of stents, with
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 861
Adverse events Moderate to severe adverse events occurred in 3 patients (5.8%). One developed acute pancreatitis resulting in an 11-day hospitalization, 1 had a peripancreatic abscess adjacent to the transgastric stent site that was managed with EUS-guided transmural drainage, and in 1 patient, a 3-cm length of hydrophilic guidewire coating was shaved off by the bevel of the needle during guidewire withdrawal and retained in the retroperitoneum without apparent sequelae. Post-procedure abdominal pain required hospitalization in 13 additional patients for a median of 2 days (range 2-5 days). There was no difference in the adverse event rates in patients with and without prior pancreatic surgery (25% vs 41%; P Z .3). Stent dysfunction occurred in 8 patients (25%) secondary to migration (n Z 5) or occlusion (n Z 3) (Table 4). Four of 5 stent migrations occurred in patients in whom a straight stent was placed; migration occurred after stent placement in 23% of straight stents versus 9% of doublepigtail stents (P Z .62). Symptomatic stent occlusion of 7F stents occurred in 2 patients at 5 and 12 weeks. If either stent occlusion or migration occurred, we attempted to place additional stents during the subsequent procedure to provide flow not only within but also between stents. Additional stents were placed in 4 patients, whereas in 3 patients stents were not replaced because of symptom resolution, and in 1 patient we only could replace the migrated stent. Two patients with stents in for 40 and 42 months developed symptoms related to the stent (abdominal pain and recurrent pancreatitis, respectively), requiring removal.
DISCUSSION
EUS-guided pancreatic duct intervention
Fujii et al
TABLE 4. Patient characteristics for those with stent migration Specific procedure (total number)
Timing (d) after placement
Prior surgery
Stent type
Stent length, cm
Direction/extent of migration*
1
1st (of 1)
1417
Yes
St, 7F
9
Proximal, partial
2
1st (of 1)
42
No
St, 5F
12
Complete
3
2nd (of 5)
98
Yes
2 DP, 5F
5, 9
Complete
3
4th (of 5)
195
Yes
St, 5F
N/A
Complete (2 stents remained)
3
5th (of 5)
149
Yes
St, 7F
5, 7, 9
2 Complete, 1 proximal, partial
4
1st (of 3)
76
No
St, 7F
7
Distal, partial
5
2nd (of 4)
70
Yes
St, 5F
9
Proximal, partial (1 DP, 7F stent remained)
Patient
St, Straight; DP, double-pigtail; F, French, N/A, not available; RAP, recurrent acute pancreatitis. Distal, partial: Indicates that the proximal end of the stent migrated out of the stomach to instead rest within the MPD (ie, no longer traverse gastric wall). Complete: Indicates that the entire stent had migrated out of the MPD and the entire track. *Proximal, partial: Indicates that the distal end of the stent migrated out of the small bowel to instead rest within the MPD.
a median stent-free follow-up period of almost 3 years. Altogether, complete clinical symptom resolution during long-term follow-up (R12 months) was achieved in 68.9% of patients, which is similar to previous reports.4,8,10,12,13 The only factors statistically associated with complete clinical success were stent placement for indications other than anastomotic stricture and initial use of shorter length stents. It is unclear why shorter stents were associated with better long-term clinical success, and this may be a type I error. However, this might also reflect a shorter transmural track to the MPD or drainage of the MPD closer to the obstruction, where the duct is often larger in caliber, presumably allowing better flow around the stent. Patients with normal MPD diameters and/or absence of definite obstruction seen on initial EUS-guided pancreatography typically experienced complete symptom resolution after stent placement. This highlights that MPD obstruction sufficient to cause clinical symptoms may not be evident during pancreatography. Therefore, selected patients with normal-caliber MPD or good flow of contrast material into the small bowel may benefit from MPD stent placement, particularly if they have symptoms of presumed pancreatic origin in the setting of prior pancreatic surgery. Additional studies are necessary to identify specific imaging and clinical criteria for determining which patients with normal MPD diameter and an absence of definite obstruction would benefit from stent placement. Symptom resolution persisted in most patients who underwent stent removal, and 2 patients developed stent-related symptoms. Our findings suggest that many patients will have a durable clinical benefit after pancreatic duct stents are removed. Endoscopist experience is important, because the technical success rate since 2011 in our group is 90%, and no major adverse events have occurred since 2009. EUS-guided pancreatic intervention is a challenging procedure, best performed by endoscopists with both EUS and
ERCP training. These procedures also may be performed by teaming 2 endoscopists with separate EUS and ERCP skills; however, in our experience, procedure time was shorter if the same physician performed all aspects of the procedure. This may vary with the patient’s health and clinical needs, the practice setting, and the availability of other endoscopic and non-endoscopic expertise within an institution. The incidence of major adverse events was relatively low at 5.8%, similar to other reports.2,4,13 One patient had a transection of the guidewire sheath during manipulation, which may have been prevented by minimizing the angulation between wire and needle and the degree of retractive force. Postprocedure abdominal pain was a common occurrence (29%) but was typically shortlived. Adverse events related to the stent itself occurred in 1 patient who developed recurrent acute pancreatitis because of proximal migration of the stent after 45 months. Despite concern for a pancreatic leak and secondary peripancreatic fluid collection after balloon dilation, only 1 patient in our cohort developed a peripancreatic abscess requiring EUS-guided drainage 4 days later. We are uncertain whether balloon dilation is responsible for the peripancreatic fluid collection or whether the need and use of multiple devices in a patient whose duct is difficult to access may be a risk. However, this adverse event occurred in only 1 patient, limiting our understanding of precipitating factors. Stent migration occurred in 5 patients (15.6%), 4 of whom had straight stents placed. Thus, we favor use of double-pigtail stents. In addition, similar to previous reports, we would recommend upgrading the number of stents placed if either stent migration or occlusion occurs.13 Our study has limitations that occur with a retrospective analysis of a diverse patient population. Patients were highly selected and were generally offered EUSguided MPD intervention only when pancreatic surgery
862 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
TABLE 4. Continued Symptoms at time of stent migration
Intervention
Course
RAP
Removed
Symptom resolution for 19 months after removal
None
Unable to identify papilla to replace
Symptom resolution for 36 months
Abdominal pain
Repeat antegrade stent placement
Continued pain
Abdominal pain
Replaced with 3 7F stents
Continued pain
Abdominal pain
Replaced with 1 10F stent
Continued pain, surgery 22 months later
Abdominal pain
Replaced with 1 7F stent
Symptom resolution for 40 months
None
Replaced with 1 5F and 1 7F stent
Symptom resolution for 12 months
was the leading alternative management strategy available. In addition, patient management was individualized, and the choice of stenting method and duration of stenting varied. The reported procedure durations are likely overestimates and reflect limitations of the procedural database in use at our institution during most of the time period studied. Finally, given the retrospective nature of the report, there was no standardized means to report the patient’s pain before and after the procedure. Information regarding patient response was obtained via the clinical notes and telephone interviews and was not graded by using a pain scale. The most important questions pertain to the role of EUS-guided MPD intervention and the patients most likely to benefit from the procedure. At this point in time, MPD stent removal cannot be considered a clear indication for EUS-guided MPD intervention, and further study regarding the utility of this technique is warranted. Opinions vary regarding the role of pancreatic stent placement. A poor indication for ERP-guided stent placement may be an even worse indication for EUS-guided intervention given the inherent risks and the likelihood of multiple follow-up procedures. EUS-guided MPD interventions require a substantial commitment from both the endoscopists and patient that is best justified if the alternative is pancreatic surgery. The indication is especially compelling in patients with pancreatic symptoms after pancreaticoduodenectomy, who might otherwise undergo completion pancreatectomy. In summary, EUS often allows access to and drainage of the pancreatic duct after failed ERCP and can obviate the need for percutaneous and surgical interventions. The use of EUS-guided versus ERP-guided intervention as first-line therapy in patients with altered anatomy should be guided by the specific anatomy and the procedural success and safety within a given institution. These procedures are technically challenging and time demanding. www.giejournal.org
Caution should be exercised, because adverse events are relatively common and can be severe, and patients receiving stents may undergo multiple follow-up procedures to achieve maximal dilation of an underlying pancreatic duct stricture. Additional data are needed to more accurately define the initial clinical predicators of good long-term outcomes as well as the best method of stent placement and the optimal timing of stent removal. Currently, EUS-guided intervention should be performed in carefully selected patients managed by a skilled multidisciplinary team. REFERENCES 1. Harada N, Kouzu T, Arima M, et al. Endoscopic ultrasound-guided pancreatography: a case report. Endoscopy 1995;27:612-5. 2. Barkay O, Sherman S, McHenry L, et al. Therapeutic EUS-assisted endoscopic retrograde pancreatography after failed pancreatic duct cannulation at ERCP. Gastrointest Endosc 2010;71:1166-73. 3. Bataille L, Deprez P. A new application for therapeutic EUS: main pancreatic duct drainage with a “pancreatic rendezvous technique.” Gastrointest Endosc 2002;55:740-3. 4. Brauer BC, Chen YK, Fukami N, et al. Single-operator EUS-guided cholangiopancreatography for difficult pancreaticobiliary access (with video). Gastrointest Endosc 2009;70:471-9. 5. Cooper ST, Malick J, McGrath K, et al. EUS-guided rendezvous for the treatment of pancreaticopleural fistula in a patient with chronic pancreatitis and pancreas pseudodivisum. Gastrointest Endosc 2010;71:652-4. 6. Das K, Kitano M, Komaki T, et al. Pancreatic ductal drainage by endoscopic ultrasound-assisted rendezvous technique for pain caused by ductal stricture with chronic pancreatitis. Dig Endosc 2010;22:217-9. 7. Dumonceau JM, Cremer M, Baize M, et al. The transduodenal rendezvous: a new approach to deeply cannulate the main pancreatic duct. Gastrointest Endosc 1999;50:274-6. 8. Ergun M, Aouattah T, Gillain C, et al. Endoscopic ultrasound-guided transluminal drainage of pancreatic duct obstruction: long-term outcome. Endoscopy 2011;43:518-25. 9. Itoi T, Kikuyama M, Ishii K, et al. EUS-guided rendezvous with singleballoon enteroscopy for treatment of stenotic pancreaticojejunal anastomosis in post-Whipple patients (with video). Gastrointest Endosc 2011;73:398-401.
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 863
EUS-guided pancreatic duct intervention 10. Kahaleh M, Hernandez AJ, Tokar J, et al. EUS-guided pancreaticogastrostomy: analysis of its efficacy to drain inaccessible pancreatic ducts. Gastrointest Endosc 2007;65:224-30. 11. Kinney TP, Li R, Gupta K, et al. Therapeutic pancreatic endoscopy after Whipple resection requires rendezvous access. Endoscopy 2009;41: 898-901. 12. Mallery S, Matlock J, Freeman ML. EUS-guided rendezvous drainage of obstructed biliary and pancreatic ducts: report of 6 cases. Gastrointest Endosc 2004;59:100-7. 13. Tessier G, Bories E, Arvanitakis M, et al. EUS-guided pancreatogastrostomy and pancreatobulbostomy for the treatment of pain in patients with pancreatic ductal dilatation inaccessible for transpapillary endoscopic therapy. Gastrointest Endosc 2007;65:233-41. 14. Vila JJ, Perez-Miranda M, Vazquez-Sequeiros E, et al. Initial experience with EUS-guided cholangiopancreatography for biliary and pancreatic duct drainage: a Spanish national survey. Gastrointest Endosc 2012;76:1133-41. 15. Will U, Meyer F, Manger T, et al. Endoscopic ultrasound-assisted rendezvous maneuver to achieve pancreatic duct drainage in obstructive chronic pancreatitis. Endoscopy 2005;37:171-3. 16. Francois E, Kahaleh M, Giovannini M, et al. EUS-guided pancreaticogastrostomy. Gastrointest Endosc 2002;56:128-33.
Fujii et al
17. Kikuyama M, Itoi T, Ota Y, et al. Therapeutic endoscopy for stenotic pancreatodigestive tract anastomosis after pancreatoduodenectomy (with videos). Gastrointest Endosc 2011;73:376-82. 18. Will U, Fueldner F, Thieme AK, et al. Transgastric pancreatography and EUS-guided drainage of the pancreatic duct. J Hepatobiliary Pancreat Surg 2007;14:377-82. 19. Chahal P, Baron TH, Topazian MD, et al. Endoscopic retrograde cholangiopancreatography in post-Whipple patients. Endoscopy 2006;38: 1241-5. 20. Farrell J, Carr-Locke D, Garrido T, et al. Endoscopic retrograde cholangiopancreatography after pancreaticoduodenectomy for benign and malignant disease: indications and technical outcomes. Endoscopy 2006;38:1246-9. 21. Matsushita M, Takakuwa H, Uchida K, et al. Techniques to facilitate ERCP with a conventional endoscope in patients with previous pancreatoduodenectomy. Endoscopy 2009;41:902-6. 22. Gines A, Varadarajulu S, Napoleon B. EUS 2008 Working Group document: evaluation of EUS-guided pancreatic-duct drainage (with video). Gastrointest Endosc 2009;69:S43-8. 23. Varadarajulu S, Trevino JM. Review of EUS-guided pancreatic duct drainage (with video). Gastrointest Endosc 2009;69:S200-2.
GIE on Twitter GIE now has a Twitter account. Followers will learn when the new issues are posted and will receive up-to-the-minute news as well as links to author interviews, podcasts, and articles. Search on Twitter for @GIE_Journal and follow all of GIE’s tweets.
864 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
APPENDIX 1. Anatomy Based on Type of EUS Procedure Performed and Success of MPD Drainage Total (n[45)
Drainage success (n[33)
Drainage failed (n [12)
Pancreaticoduodenectomy
25
19
6
Partial head resection w/R-Y pancreaticojejunostomy
1
1
0
Pancreas sparing duodenectomy
1
0
1
Billroth II with pancreaticogastrostomy
1
0
1
Middle pancreatectomy with pancreaticojejunostomy
1
1
0
Pancreas divisum
5
2
3
Duodenal stricture
1
1
0
ERCP-induced duodenal perforation
1
1
0
No altered anatomy
9
8
1
Altered Anatomy
APPENDIX 2. Reasons for Prior Failed ERP Reason
(n[37)
Native Papilla (nZ16) Failed cannulation (pancreas divisum)
5
Unable to advance guidewire to desired location
5
Minor papilla not identified (pancreas divisum)
4
Iatrogenic periampullary perforation during ERP
1
Migrated pancreatic stent
1
Surgical Anastomosis (nZ21) Pancreatic anastomosis not identified
10
Pancreatic anastomosis not reached
3
Failed cannulation
7
Unknown
1
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 864.e1
EUS-guided pancreatic duct intervention: outcomes of a single tertiary-care referral center experience Larissa L. Fujii, MD,1 Mark D. Topazian, MD,1 Barham K. Abu Dayyeh, MD,1 Todd H. Baron, MD,1 Suresh T. Chari, MD,1 Michael B. Farnell, MD,2 Ferga C. Gleeson, MD,1 Christopher J. Gostout, MD,1 Michael L. Kendrick, MD,2 Randall K. Pearson, MD,1 Bret T. Petersen, MD,1 Mark J. Truty, MD,2 Santhi S. Vege, MD,1 Michael J. Levy, MD1 Rochester, Minnesota, USA
Background: EUS can provide access to the main pancreatic duct (MPD) for therapeutic intervention. The longterm clinical success of EUS-guided MPD interventions is unknown. Objective: To determine technical and clinical success rates, predictors of success, and long-term outcomes of EUS-guided MPD intervention. Design: Retrospective, single-center study. Setting: Tertiary-care referral center. Patients: Forty-five patients. Intervention: EUS-guided MPD stent retrieval or placement. Main Outcome Measurements: Technical and clinical success rates, adverse events, and long-term clinical outcomes. Results: Among the 45 patients, 37 had undergone failed ERCP, and 29 had surgically altered anatomy. Median follow-up after initial EUS-guided intervention was 23 months. Two patients underwent EUS for stent removal, and EUS-guided MPD stent placement was attempted in 43 patients. Technical success was achieved in 32 of 43 patients (74%) with antegrade (n Z 18) or retrograde (n Z 14) stent insertion. Serious adverse events occurred in 3 patients (6%). Patients underwent a median of 2 (range 1-6) follow-up procedures for revision or removal of stents, without adverse events. Complete symptom resolution occurred in 24 of 29 patients (83%) while stents were in place, including all 6 with nondilated ducts. Stents were removed in 23 patients, who were then followed for an additional median of 32 months; 4 patients had recurrent symptoms. Among the 11 failed cases, most had persistent symptoms or required surgery. Limitations: Retrospective study design, individualized patient management. Conclusion: EUS-guided MPD intervention is feasible and safe, with long-term clinical success in the majority of patients. EUS provides important treatment options, particularly in patients who would otherwise undergo surgery. (Gastrointest Endosc 2013;78:854-64.)
Endoscopic retrograde pancreatography (ERP) is the least-invasive method to access the main pancreatic duct (MPD) and provide therapeutic intervention. Historically,
failed ERP necessitated repeat endoscopic attempts and/or percutaneous radiologic or surgical intervention. EUSguided pancreatic interventions are a new alternative to
Abbreviations: ERP, endoscopic retrograde pancreatography; MPD, main pancreatic duct.
Received March 18, 2013. Accepted May 9, 2013.
DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
Current affiliations: Division of Gastroenterology and Hepatology (1), Division of Gastroenterologic and General Surgery, Mayo Clinic Rochester, Rochester, Minnesota (2).
Copyright ª 2013 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2013.05.016
Reprint requests: Michael J. Levy, MD, Mayo Clinic, Division of Gastroenterology and Hepatology, 200 1st St SW, Rochester, MN 55905.
854 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
Take-home Message EUS-guided pancreatic duct access and stent placement often can provide drainage safely, with good long-term clinical outcomes in selected patients. Prior endoscopic retrograde pancreatogram during the same procedure was associated with failed EUS-guided stent placement, reflecting the importance of determining early which patients would benefit from EUS guidance. Patients with symptoms suggestive of a pancreatic origin but who have normal pancreatogram results during EUS also may benefit from stent placement.
Figure 1. EUS-guided pancreatogram from the gastric body reveals partial pancreatic duct obstruction due to a pancreaticojejunal anastomotic stricture.
ERP, particularly when ERP has failed or is unlikely to succeed (as in surgically altered anatomy). The proximity of the gut lumen to the MPD and the unparalleled image quality and continuous real-time imaging of EUS allow precise targeting of the MPD. Although ductal access is typically easily achieved by using FNA techniques, subsequent interventions may be challenging. Limited data have been published regarding this therapeutic modality, particularly long-term patient outcomes. We assessed technical and clinical outcomes and predictors of success among patients undergoing EUS-guided MPD intervention at our institution.
30 days. Each EUS was performed by 1 of 2 endosonographers (M.J.L. or M.D.T.), both of whom have performed more than 10,000 upper GI tract EUS examinations. Technical success was defined as complete if the intent of the procedure was accomplished; otherwise it was defined as failed. Patient outcomes were monitored for at least 12 months after initial intervention. Patients who had less than 12 months of follow-up within our facility were contacted via telephone to determine clinical outcome. The 3 patients who underwent stent placement and could not be contacted were excluded from the analysis on long-term analysis. All clinical, laboratory, radiologic, and surgical findings after EUS were reviewed. The last available clinical follow-up was used to assess patients’ responses to interventions. Patients were deemed to have complete responses to therapy only if it was clearly documented that all clinical symptoms fully resolved after the procedures. Patients were considered to have partial responses when there was improvement in severity and/or frequency of presenting symptoms. Patients who experienced no benefit were categorized as having no response to endoscopic therapy.
MATERIALS AND METHODS Statistical analysis A prospectively maintained EUS database was reviewed to identify all patients who underwent attempted EUSguided MPD intervention at Mayo Clinic, Rochester, Minnesota. The Institutional Review Board granted approval for retrospective data retrieval and patient contact. Written informed consent was obtained before each procedure. Data collected pertained to all aspects of the clinical, laboratory, and radiologic findings. All endoscopic, interventional radiologic, and surgical procedures performed before and after the attempted EUS-guided MPD access were reviewed. Pancreatograms obtained after initial MPD access (but before guidewire placement) were categorized during the procedure as showing complete obstruction (no contrast material drainage to the small bowel lumen), partial obstruction (slow or poor drainage of contrast material to the small bowel), or no definite obstruction (good drainage of contrast material to the small bowel). Adverse events possibly related to the procedure were identified within
Continuous variables are reported as either a mean and standard deviation (SD) or median and range. Means were reported unless the SD was larger than the mean itself, in which case medians were reported. The t test was used for continuous variables, and a Pearson chi-square analysis was used for categorical variables. A P value of ! .05 was considered statistically significant.
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 855
Technique A therapeutic channel curvilinear array echoendoscope (CF-UCT140-AL5, GF-UC140P-AL5, or GF -UC160P-AT8A; Olympus America, Inc, Center Valley, Pa) by using carbon dioxide insufflation was used to access the MPD. The route was selected based on the site of pathology, to traverse the shortest distance between the transducer and the MPD, to avoid intervening structures, and to provide a needle trajectory to facilitate subsequent track dilatation and stent placement. A 22-gauge or 25-gauge FNA needle (Cook
EUS-guided pancreatic duct intervention
Fujii et al
Figure 2. EUS-guided antegrade stent placement. A, A guidewire is advanced through the needle into the pancreatic duct. B, The track between the gastric body to the pancreatic duct is balloon dilated. C, Antegrade transluminal stent placement. The pancreaticoenteric anastomosis could not be traversed. D, During a subsequent examination, the anastomotic stricture was traversed, and an antegrade transanastomotic stent was placed.
Medical, Bloomington, Ind) was typically used for the initial pancreatogram. When MPD drainage was attempted, a 19gauge FNA needle (Cook Medical) was used for duct access (Fig. 1) followed by guidewire placement via the needle. Guidewires were either 0.035 or 0.025 inches in diameter and included straight and angled hydrophilic wires (Glidewire, Jagwire; Boston Scientific, Inc, Natick, Mass). Attempts were made to advance the guidewire in an antegrade fashion across the site of MPD obstruction to the gut lumen, via the papilla or pancreaticoenteral anastomosis (in postoperative cases) under EUS and fluoroscopic guidance. Pancreatic stent placement was then performed over the guidewire. Stent placement was categorized as (1) retrograde (sometimes referred to as a rendezvous procedure) from the papilla and/or anastomosis into the MPD via another endoscope or (2) antegrade placement across the proximal gut wall into the MPD via the echoendoscope. Antegrade stent placement was further subdivided based on whether the stent crossed the site of ductal obstruction and papilla and/or anastomosis (transpapillary/transanastomotic) or not (transluminal). For retrograde stent placement, the guidewire was left in place, and the echoendoscope was withdrawn. Either
a side-viewing duodenoscope (TJF-160VF; Olympus America) in patients with native anatomy or a pediatric or adult colonoscope (PCF-Q180AL or CF-H180AL; Olympus America) in patients with postoperative anatomy was advanced to the pancreatic-enteric junction. The transpancreatic guidewire was grasped with a snare or biopsy cable and withdrawn through the endoscope, or a second guidewire was passed retrograde alongside the first wire into the MPD, sometimes after sphincterotomy. A stent was then deployed in retrograde fashion, from the small bowel through the papilla and/or anastomosis and into the MPD. For antegrade stent placement (Fig. 2), transmural track dilatation was performed by using hydrostatic balloons (Titan Biliary Dilation Balloon; Cook Medical or Hurricaine Biliary Dilation Balloon; Boston Scientific), tapered catheters (Sohendra; Cook Medical), and/or cannulas (Proforma; ConMed Endoscopic Technologies, Utica, NY). When these devices failed to traverse the transmural track to the pancreatic duct, a wire-guided needle-knife catheter was used (Boston Scientific), initially without electrosurgical current, and if needed, with short bursts of electrosurgical current at standard sphincterotomy current, to access the MPD. If the guidewire did not initially cross the site of
856 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
Figure 3. Overview of the technical success and adverse events for each procedure type. MPD, main pancreatic duct; TS, technical success.
pancreatic duct obstruction, further attempts were made after accessing the MPD with a catheter via the transmural track. Both straight and double-pigtail stents were used (Cook Medical). Typically, the proximal end of the stent was left in the lumen of the proximal GI tract, facilitating subsequent stent exchange, upsizing, or removal. Additional side holes were often created within the intraductal portion of the stent to facilitate drainage.
RESULTS Forty-five patients (mean age 57 y [SD 16 y], 24 men, mean body mass index 23.5 kg/m2 [SD 3.6]) underwent 49 EUS examinations in an attempt to access the MPD under EUS guidance between January 2006 and December 2012 (initial examinations) as well as 59 subsequent examinations to exchange or remove stents (follow-up examinations). The number of total procedures, successful procedures, and adverse events are outlined in Figure 3. Forty-four of the initial 45 examinations were performed with patients under general anesthesia. Intravenous antibiotics (fluoroquinolones or piperacillin-tazobactam) were administered prophylactically in 30 procedures. The median MPD caliber at the site of attempted entry based on EUS or by CT if not measured during EUS was 4.4 mm (range 0.8-10 mm). www.giejournal.org
The clinical and preprocedure structural diagnoses are listed in Table 1. Eighteen patients had more than 1 clinical indication. There was no difference in age, sex, and mean MPD size in patients with and without recurrent acute pancreatitis (the most common clinical symptom). Previous pancreatic surgery had been performed in 29 patients (64%) at 19.9 months (range 3.3-150.6 months) before EUS. Types of prior pancreatic surgery and altered anatomy are listed in Appendix 1, available online at www.giejournal.org. EUS-guided MPD intervention was performed after failed ERP in 37 patients (82%), among whom 21 patients had surgically altered anatomy, and 16 had normal anatomy. Twenty-four ERPs had been attempted on a previous day, whereas 15 ERPs were attempted immediately before EUS-guided intervention. Three of the 37 patients underwent at least 2 separate failed ERP procedures. Reasons for failed ERP are outlined in Appendix 2, available online at www.giejournal.org. Two patients were diagnosed with recurrent pancreatic cancer after EUS-guided MPD intervention.
Attempted stent retrieval Two patients underwent EUS for attempted retrieval of a retained MPD anastomotic stent placed during pancreaticoduodenectomy performed 6 and 1.5 years previously. Both patients underwent EUS for abdominal pain, at which Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 857
EUS-guided pancreatic duct intervention
Fujii et al
In the remaining 43 patients, the procedural intent was to place an MPD stent. Pancreatogram before the attempted stent placement showed complete obstruction in 13 patients, partial obstruction in 11, and no definite obstruction in 13 patients. The remaining 6 patients had either a disconnected duct (n Z 5) or the MPD was not identified (n Z 1). MPD access was achieved by using the gastric body for the majority of cases (41 patients, 95%). One patient had a failed attempt at accessing the MPD from the duodenal
bulb, whereas the remaining patient had a successful EUSguided pancreaticobulbostomy via the retrograde route. Complete technical success with stent placement was achieved in 32 patients (74%). Stent placement was possible during the index examination in 30 patients (94%), whereas the other 2 patients required 1 and 2 additional attempts, respectively. The stents were deployed by using a retrograde or antegrade technique in 14 and 18 patients, respectively; of the 18 antegrade stent placements, the stent crossed the site of pancreatic duct obstruction and the papilla and/or anastomosis in 13 and did not in 5. Among these 5, a stent could be advanced across the site of ductal obstruction in 2 patients during a follow-up procedure (Fig. 2D). The mean procedure duration was 162 minutes (SD 37), 125 minutes (SD 66), and 130 minutes (SD 19) for the retrograde and antegrade methods, respectively; retrograde stent placement was significantly lengthier than either antegrade approach (P Z .05). This is likely explained by a significantly lower mean procedure time if 1 endoscopist performed the entire examination as compared with when separate physicians performed the EUS and ERP portions (109 37 minutes vs 171 44 minutes, respectively; P ! .001). In addition, the retrograde technique was less commonly used in postoperative patients (84% vs 35% antegrade and retrograde techniques, respectively; P Z .006) but more commonly used after failed ERP attempts (100% of patients undergoing the retrograde technique had a prior ERP compared with 61% of patients who had an antegrade approach; P Z .008). Although it was possible to access, inject contrast material, and insert a guidewire into the MPD in 42 patients (98%), failed stent placement (n Z 11) occurred because of the inability to advance the guidewire into the MPD (n Z 1) or through the anastomosis or papilla (n Z 8), inability to adequately dilate the transmural track (n Z 1), or loss of the guidewire during ERP (n Z 1). A needleknife was used to facilitate transgastric ductal access in 6 patients when other efforts failed, allowing transgastric access and examination completion in all 6 patients. When the procedure was successful, the track was dilated to a mean of 5.5 mm (SD 1.5). Baseline characteristics between the patients with and without successful stent placement are shown in Table 2. There was a statistically greater chance of failure when EUS access for stenting was attempted on the same day as a failed ERP. One (n Z 30) or 2 stents (n Z 2) were placed during the initial examination. Straight plastic and double-pigtail plastic and metal stents were used in 17, 11, and 1 case, respectively (3 unknown). The plastic stent diameters were 10F (n Z 1), 7F (n Z 28), 5F (n Z 3), and 3F (n Z 1) with mean stent length of 9 cm (SD 3.3). The 1 covered metal stent used was 10 mm in diameter and 80 mm in length. In 12 patients, stents (11 placed retrograde and 1 placed antegrade) were removed during the subsequent procedure
858 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
TABLE 1. Indications for attempted EUS-guided intervention in 52 patients Indications
No.
Clinical (n Z 79; 18 had more than 1 indication) Recurrent acute pancreatitis
30
Chronic pancreatitis
8
Pancreatic duct leak
3
Retained surgically placed stent
2
Abdominal pain and dilated MPD
23
Weight loss and dilated MPD
8
Steatorrhea and dilated MPD
5
Structural (n Z 45) Benign anastomotic stricture
25
Benign non-anastomotic stricture
5
Malignant non-anastomotic stricture
1
Retained MPD stent
2
Disconnected pancreatic duct
5
Pancreatic duct leak
3
Symptomatic pancreas divisum
3
Papillary stenosis
1
MPD, main pancreatic duct.
time the retained stents were identified and believed responsible for the patients’ pain. EUS transductal displacement into the jejunum was successful by using a 19-gauge FNA needle and guidewire advancement to dislodge the stent in the patient with chronic pancreatitis with an MPD diameter of 1.5 mm. The total procedure duration was 160 minutes (44 minutes of fluoroscopy). After stent removal, the patient’s pain intensity improved but persisted. EUS-guided stent retrieval was unsuccessful in the other patient with a MPD diameter of 1.4 mm. The total procedure duration was 96 minutes (33 minutes of fluoroscopy). The patient’s abdominal pain persisted until his death from recurrent pancreatic cancer 7 months later.
Attempted stent placement
Fujii et al
EUS-guided pancreatic duct intervention
TABLE 2. Patient characteristics relative to technical success of stent placement Stenting successful (n [ 32)
Stenting failed (n [ 11)
P value
57.8 (16.1)
52.7 (18.6)
.39
18
4
.25
23.5 (3.7)
23.6 (3.7)
.96
Prior pancreatic surgery
20
7
.95
Prior ERP attempt
25
11
.09
ERP immediately before EUS
7
7
.04
Fellow present
18
5
.54
Dilated MPD (R3 mm)
3
6
.35
4.9 (2.4)
3.9 (3.2)
.43
Complete MPD obstruction by pancreatogram
5
8
.14
R5 EUS features of chronic pancreatitis
11
6
.24
Duration, total time, mean (SD), min
142 (51)
154 (54)
.49
Duration, EUS mean (SD), min
89 (45)
97 (20)
.68
Duration, fluoroscopy, mean (SD), min
40 (23)
34 (20)
.43
Age, average (SD), y Male, no. Body mass index, mean (SD), kg/m2
Maximum MPD caliber, mean (SD), mm
SD, Standard deviation; ERP, endoscopic retrograde pancreatography; MPD, main pancreatic duct.
after a median of 3 months (range 1-42). In 1 patient who underwent prior pancreaticoduodenectomy, stent retrieval was attempted at 4 months, but the pancreaticojejunal anastomosis could not be reached by using a single-balloon endoscope. It was later removed by using EUS-guidance 46 months after insertion. Of the remaining 11 patients, 5 failed stent replacement (3 because of the inability to identify the papilla and 2 because of the inability to recannulate the MPD), 4 had normal pancreatogram results, and 2 were removed without attempted re-evaluation of the MPD. Two patients did not undergo repeat endoscopy because of their poor prognosis related to underlying malignant disease. The remaining 18 patients (2 retrograde and 16 antegrade stent placements) underwent a median of 2 (range 1-6) additional endoscopic procedures, with stent replacement or upsizing, with the first follow-up examination occurring at a median of 46 days (range 13-77) after the initial procedure. The maximum number of stents placed was 1, 2, 3, 4, and 5 stents in 2, 7, 6, 2, and 1 patients, respectively. In general, these patients underwent stent placement until maximum resolution of pancreatic duct stenosis was obtained. Ten patients had their stents endoscopically removed at a mean of 9 months (range 2.5-39), 1 had her stents removed during completion pancreatectomy, 2 died from underlying disease with stents in place, and 5 remain with stents in place at a median of 42.6 months (range 13-69.5) after initial placement. Detailed follow-up was available for 9 of the 11 patients in whom stent placement failed, with a median follow-up
of 31 months (range 3-80 months). Five patients (45%) required surgery for ongoing symptoms, including lateral pancreaticojejunostomy (n Z 2), revision of pancreaticojejunal (n Z 2), and partial pancreatectomy (n Z 1). After surgery, all but 1 patient had complete resolution of symptoms. Of the 6 patients who have not undergone surgery after EUS, 2 are asymptomatic at 15 and 30 months without intervention, 2 experience ongoing recurrent acute pancreatitis that has been medically managed, and 2 have insufficient follow-up. Complete follow-up was available until death or at least 12 months in 29 of 32 (91%) successfully stented patients. Six died of underlying malignancy or unrelated causes at 11 months (range 2-54 months); of these, 4 patients had complete pancreatic symptom resolution until death. In the remaining 23 patients, complete clinical success with full symptom resolution occurred in 16 patients (69.6%) at the time of last follow-up (median 37 months [range 12-72 months]). Complete symptom resolution while the stent was in place was seen in 24 patients (82.8%). Stents were removed in 23 patients after a median of 4 months (range 1-47 months); the patients were then followed for a median of 32 months (range 15.5-50.5 months), with symptom recurrence in 4 patients after a median of 14 months (range 5-32 months). Seven living patients (30.4%) with complete follow-up had partial clinical success with symptom recurrence at a median of 7 months (range 2-45 months). Of these, 3 had their stents previously removed, whereas 4 had their stents in place at the
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 859
EUS-guided pancreatic duct intervention
Fujii et al
TABLE 3. Patient characteristics relative to long-term clinical success after stent placement* Partial clinical success (n [ 9)
Complete clinical success (n [ 20)
58.8 (12.2)
58.1 (18.7)
.92
5
11
.98
22.2 (4.2)
24.5 (3.3)
.12
Fellow present during first procedure
5
7
.30
Postoperative anatomy
8
11
.08
Structural indication: benign anastomotic stricture
8
10
.046*
Clinical indication: recurrent acute pancreatitis
3
10
.40
Clinical indication: abdominal pain
3
7
.93
Prior ERP attempt
5
17
.09
ERP immediately before EUS
1
4
.87
Dilated MPD (R3 mm)
9
14
.065
5.4 (1.8)
4.4 (2.6)
.33
Complete MPD obstruction by pancreatogram
2
6
.61
Antegrade stent insertion
6
10
.40
Transpapillary/transanastomotic stent placement during first procedure
8
16
.56
Double-pigtail stent placed during first procedure
4
7
.87
Stent length during first procedure (cm)
10.8 (3.3)
8.1 (2.6)
.03*
R5 EUS features of chronic pancreatitis
3
5
.64
Stent exchanged
6
11
.49
Stent eventually removed
6
15
.82
Age, mean (SD), y Male, no. 2
Body mass index, mean (SD), kg/m
Maximum MPD caliber mean (SD), mm
P value
SD, Standard deviation; ERP, endoscopic retrograde pancreatography; MPD, main pancreatic duct. *Includes 29 patients in whom long-term follow-up (until death or O12 months) was available.
time of symptom recurrence. None required surgical or percutaneous intervention. Nineteen patients had a repeat CT scan after stent placement at a median of 335 days (range 30-1393 days) after the initial procedure. The mean MPD size decreased from 5.9 mm (SD 2.3) before the procedure to 4.5 mm (SD 2.2) after stent placement, for a mean difference of 1.4 mm (SD 1.6). Four patients had a slight increase in MPD size ranging between 0.2 and 0.8 mm, which may be related to differences in measurements rather than an actual increase. There was no difference in mean MPD size before and after stent placement or mean difference in MPD size in patients who had complete versus partial and/or no clinical success. Patient and procedural characteristics relative to longterm clinical success are detailed in Table 3. Shorter stent lengths and any structural indication other than benign anastomotic stricture were associated with complete clinical success in univariate analysis. Notably, all 6
patients with normal MPD diameter (defined as diameter !3 mm) had complete responses to stent placement (P Z .065 compared with patients with dilated ducts). The indications for intervention in these patients were recurrent acute pancreatitis (n Z 2), pancreatic duct leakage (n Z 2), abdominal pain (n Z 1), and weight loss (n Z 1) in the setting of surgically altered anatomy (n Z 2) or pancreas divisum (n Z 1). The initial pancreatogram in patients with successful stent placement revealed complete obstruction in 8 patients, partial obstruction in 10 patients, no definite obstruction in 7 patients, and 4 patients with a disconnected pancreatic duct. There was no correlation between pancreatogram findings and likelihood of clinical success (P Z .26 for the comparison between those with partial or complete obstruction versus no definite obstruction). Six of the 7 patients (85.7%) with absence of obstruction on pancreatogram had complete symptom resolution, which was not statistically significant (P Z .26 compared with those with partial or complete
860 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
obstruction). Patients having prior pancreatic surgery experienced a nonsignificant trend toward a lower likelihood of clinical success.
EUS-guided pancreatic duct intervention
The concept of combining ERP with EUS technology led to the first report in 1995 of EUS-guided pancreatography for a patient requiring MPD stone removal after pancreaticoduodenectomy.1 Since that time, there has been increased use and evolution in EUS-guided MPD techniques, particularly in patients in whom previous ERP attempts were unsuccessful. ERP may be particularly challenging after pancreatic head resection because of the inability to reach, identify, or cannulate the pancreaticojejunal anastomosis. Prior reports of EUS-guided MPD intervention have described endoscopic methods and technical and clinical success rates and have demonstrated feasibility and safety.1-16 Our study describes the technical and long-term clinical success rates of EUSguided pancreatic duct intervention in the largest published cohort of patients to date and explores the predictors of technical and clinical success.
Our overall technical success rate of any EUS-guided MPD intervention was 73%, which is within the range previously reported.2,4,8,10-18 This success rate is substantially higher than the success rate reported with ERP after pancreaticoduodenectomy,11,17,19-21 and, as a result, EUSguided MPD intervention is generally our first endoscopic intervention in patients with symptomatic pancreatic duct obstruction after a Whipple procedure. The most common site of EUS-guided puncture was the gastric body (pancreaticogastrostomy), which we believe provides enhanced stabilization, echoendoscope positioning, and access to the MPD. Although the most common cause of a failed procedure was the inability to negotiate the guidewire through the site of obstruction or across the papilla or pancreaticojejunal anastomosis, there was no difference in the ability to place a stent in patients with and without complete obstruction seen on pancreatography. In addition, contrary to previous recommendations to intervene only in patients with dilated MPDs,22,23 we were able to access and intervene on ducts as small as 0.8 mm in diameter. Previous studies also report the ability to access a normal-caliber MPD by using EUS guidance.2,10 Despite a higher percentage of failed stent placements with a normal MPD size, this did not reach statistical significance. This emphasizes that MPD diameter is 1 component to factor into the decision making process as to whether a patient may benefit from pancreatic duct stent placement. Other non-anatomic factors such as the patient’s clinical presentation and course should be considered. Additional studies are needed to clarify the role and potential benefit of draining a normal-caliber duct and to identify clinical and imaging characteristics that may be associated with a good response. Prior ERP performed during the same procedure was the only statistically significant risk factor for failed stent placement. It is unclear whether this was because of anatomic factors, inflammation induced by ERP attempts, or less commitment by the endosonographer in the setting of a long, failed ERP attempt. Some of the case series with the highest technical success (O90%) often favored antegrade transluminal stent placement without traversal of the MPD obstruction and papilla or anastomosis.8,13,16 Similarly to an externally placed biliary catheter, this technique provides only 1 route of MPD drainage to the stomach. We favored antegrade stent placement across the papilla and/or pancreaticojejunal anastomosis, which provides dual drainage to the stomach and small bowel and treats the underlying stricture. When the ductal obstruction and papilla and/or anastomosis could not be traversed during initial antegrade stent placement, we placed a transluminal stent into the MPD (pancreaticogastrostomy) and reattempted traversal of the ductal obstruction and papilla and/or anastomosis during a subsequent procedure. Complete symptom resolution occurred in 24 of 29 patients (83%) while stents were in place, and only 4 patients had recurrence of symptoms after removal of stents, with
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 861
Adverse events Moderate to severe adverse events occurred in 3 patients (5.8%). One developed acute pancreatitis resulting in an 11-day hospitalization, 1 had a peripancreatic abscess adjacent to the transgastric stent site that was managed with EUS-guided transmural drainage, and in 1 patient, a 3-cm length of hydrophilic guidewire coating was shaved off by the bevel of the needle during guidewire withdrawal and retained in the retroperitoneum without apparent sequelae. Post-procedure abdominal pain required hospitalization in 13 additional patients for a median of 2 days (range 2-5 days). There was no difference in the adverse event rates in patients with and without prior pancreatic surgery (25% vs 41%; P Z .3). Stent dysfunction occurred in 8 patients (25%) secondary to migration (n Z 5) or occlusion (n Z 3) (Table 4). Four of 5 stent migrations occurred in patients in whom a straight stent was placed; migration occurred after stent placement in 23% of straight stents versus 9% of doublepigtail stents (P Z .62). Symptomatic stent occlusion of 7F stents occurred in 2 patients at 5 and 12 weeks. If either stent occlusion or migration occurred, we attempted to place additional stents during the subsequent procedure to provide flow not only within but also between stents. Additional stents were placed in 4 patients, whereas in 3 patients stents were not replaced because of symptom resolution, and in 1 patient we only could replace the migrated stent. Two patients with stents in for 40 and 42 months developed symptoms related to the stent (abdominal pain and recurrent pancreatitis, respectively), requiring removal.
DISCUSSION
EUS-guided pancreatic duct intervention
Fujii et al
TABLE 4. Patient characteristics for those with stent migration Specific procedure (total number)
Timing (d) after placement
Prior surgery
Stent type
Stent length, cm
Direction/extent of migration*
1
1st (of 1)
1417
Yes
St, 7F
9
Proximal, partial
2
1st (of 1)
42
No
St, 5F
12
Complete
3
2nd (of 5)
98
Yes
2 DP, 5F
5, 9
Complete
3
4th (of 5)
195
Yes
St, 5F
N/A
Complete (2 stents remained)
3
5th (of 5)
149
Yes
St, 7F
5, 7, 9
2 Complete, 1 proximal, partial
4
1st (of 3)
76
No
St, 7F
7
Distal, partial
5
2nd (of 4)
70
Yes
St, 5F
9
Proximal, partial (1 DP, 7F stent remained)
Patient
St, Straight; DP, double-pigtail; F, French, N/A, not available; RAP, recurrent acute pancreatitis. Distal, partial: Indicates that the proximal end of the stent migrated out of the stomach to instead rest within the MPD (ie, no longer traverse gastric wall). Complete: Indicates that the entire stent had migrated out of the MPD and the entire track. *Proximal, partial: Indicates that the distal end of the stent migrated out of the small bowel to instead rest within the MPD.
a median stent-free follow-up period of almost 3 years. Altogether, complete clinical symptom resolution during long-term follow-up (R12 months) was achieved in 68.9% of patients, which is similar to previous reports.4,8,10,12,13 The only factors statistically associated with complete clinical success were stent placement for indications other than anastomotic stricture and initial use of shorter length stents. It is unclear why shorter stents were associated with better long-term clinical success, and this may be a type I error. However, this might also reflect a shorter transmural track to the MPD or drainage of the MPD closer to the obstruction, where the duct is often larger in caliber, presumably allowing better flow around the stent. Patients with normal MPD diameters and/or absence of definite obstruction seen on initial EUS-guided pancreatography typically experienced complete symptom resolution after stent placement. This highlights that MPD obstruction sufficient to cause clinical symptoms may not be evident during pancreatography. Therefore, selected patients with normal-caliber MPD or good flow of contrast material into the small bowel may benefit from MPD stent placement, particularly if they have symptoms of presumed pancreatic origin in the setting of prior pancreatic surgery. Additional studies are necessary to identify specific imaging and clinical criteria for determining which patients with normal MPD diameter and an absence of definite obstruction would benefit from stent placement. Symptom resolution persisted in most patients who underwent stent removal, and 2 patients developed stent-related symptoms. Our findings suggest that many patients will have a durable clinical benefit after pancreatic duct stents are removed. Endoscopist experience is important, because the technical success rate since 2011 in our group is 90%, and no major adverse events have occurred since 2009. EUS-guided pancreatic intervention is a challenging procedure, best performed by endoscopists with both EUS and
ERCP training. These procedures also may be performed by teaming 2 endoscopists with separate EUS and ERCP skills; however, in our experience, procedure time was shorter if the same physician performed all aspects of the procedure. This may vary with the patient’s health and clinical needs, the practice setting, and the availability of other endoscopic and non-endoscopic expertise within an institution. The incidence of major adverse events was relatively low at 5.8%, similar to other reports.2,4,13 One patient had a transection of the guidewire sheath during manipulation, which may have been prevented by minimizing the angulation between wire and needle and the degree of retractive force. Postprocedure abdominal pain was a common occurrence (29%) but was typically shortlived. Adverse events related to the stent itself occurred in 1 patient who developed recurrent acute pancreatitis because of proximal migration of the stent after 45 months. Despite concern for a pancreatic leak and secondary peripancreatic fluid collection after balloon dilation, only 1 patient in our cohort developed a peripancreatic abscess requiring EUS-guided drainage 4 days later. We are uncertain whether balloon dilation is responsible for the peripancreatic fluid collection or whether the need and use of multiple devices in a patient whose duct is difficult to access may be a risk. However, this adverse event occurred in only 1 patient, limiting our understanding of precipitating factors. Stent migration occurred in 5 patients (15.6%), 4 of whom had straight stents placed. Thus, we favor use of double-pigtail stents. In addition, similar to previous reports, we would recommend upgrading the number of stents placed if either stent migration or occlusion occurs.13 Our study has limitations that occur with a retrospective analysis of a diverse patient population. Patients were highly selected and were generally offered EUSguided MPD intervention only when pancreatic surgery
862 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
TABLE 4. Continued Symptoms at time of stent migration
Intervention
Course
RAP
Removed
Symptom resolution for 19 months after removal
None
Unable to identify papilla to replace
Symptom resolution for 36 months
Abdominal pain
Repeat antegrade stent placement
Continued pain
Abdominal pain
Replaced with 3 7F stents
Continued pain
Abdominal pain
Replaced with 1 10F stent
Continued pain, surgery 22 months later
Abdominal pain
Replaced with 1 7F stent
Symptom resolution for 40 months
None
Replaced with 1 5F and 1 7F stent
Symptom resolution for 12 months
was the leading alternative management strategy available. In addition, patient management was individualized, and the choice of stenting method and duration of stenting varied. The reported procedure durations are likely overestimates and reflect limitations of the procedural database in use at our institution during most of the time period studied. Finally, given the retrospective nature of the report, there was no standardized means to report the patient’s pain before and after the procedure. Information regarding patient response was obtained via the clinical notes and telephone interviews and was not graded by using a pain scale. The most important questions pertain to the role of EUS-guided MPD intervention and the patients most likely to benefit from the procedure. At this point in time, MPD stent removal cannot be considered a clear indication for EUS-guided MPD intervention, and further study regarding the utility of this technique is warranted. Opinions vary regarding the role of pancreatic stent placement. A poor indication for ERP-guided stent placement may be an even worse indication for EUS-guided intervention given the inherent risks and the likelihood of multiple follow-up procedures. EUS-guided MPD interventions require a substantial commitment from both the endoscopists and patient that is best justified if the alternative is pancreatic surgery. The indication is especially compelling in patients with pancreatic symptoms after pancreaticoduodenectomy, who might otherwise undergo completion pancreatectomy. In summary, EUS often allows access to and drainage of the pancreatic duct after failed ERCP and can obviate the need for percutaneous and surgical interventions. The use of EUS-guided versus ERP-guided intervention as first-line therapy in patients with altered anatomy should be guided by the specific anatomy and the procedural success and safety within a given institution. These procedures are technically challenging and time demanding. www.giejournal.org
Caution should be exercised, because adverse events are relatively common and can be severe, and patients receiving stents may undergo multiple follow-up procedures to achieve maximal dilation of an underlying pancreatic duct stricture. Additional data are needed to more accurately define the initial clinical predicators of good long-term outcomes as well as the best method of stent placement and the optimal timing of stent removal. Currently, EUS-guided intervention should be performed in carefully selected patients managed by a skilled multidisciplinary team. REFERENCES 1. Harada N, Kouzu T, Arima M, et al. Endoscopic ultrasound-guided pancreatography: a case report. Endoscopy 1995;27:612-5. 2. Barkay O, Sherman S, McHenry L, et al. Therapeutic EUS-assisted endoscopic retrograde pancreatography after failed pancreatic duct cannulation at ERCP. Gastrointest Endosc 2010;71:1166-73. 3. Bataille L, Deprez P. A new application for therapeutic EUS: main pancreatic duct drainage with a “pancreatic rendezvous technique.” Gastrointest Endosc 2002;55:740-3. 4. Brauer BC, Chen YK, Fukami N, et al. Single-operator EUS-guided cholangiopancreatography for difficult pancreaticobiliary access (with video). Gastrointest Endosc 2009;70:471-9. 5. Cooper ST, Malick J, McGrath K, et al. EUS-guided rendezvous for the treatment of pancreaticopleural fistula in a patient with chronic pancreatitis and pancreas pseudodivisum. Gastrointest Endosc 2010;71:652-4. 6. Das K, Kitano M, Komaki T, et al. Pancreatic ductal drainage by endoscopic ultrasound-assisted rendezvous technique for pain caused by ductal stricture with chronic pancreatitis. Dig Endosc 2010;22:217-9. 7. Dumonceau JM, Cremer M, Baize M, et al. The transduodenal rendezvous: a new approach to deeply cannulate the main pancreatic duct. Gastrointest Endosc 1999;50:274-6. 8. Ergun M, Aouattah T, Gillain C, et al. Endoscopic ultrasound-guided transluminal drainage of pancreatic duct obstruction: long-term outcome. Endoscopy 2011;43:518-25. 9. Itoi T, Kikuyama M, Ishii K, et al. EUS-guided rendezvous with singleballoon enteroscopy for treatment of stenotic pancreaticojejunal anastomosis in post-Whipple patients (with video). Gastrointest Endosc 2011;73:398-401.
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 863
EUS-guided pancreatic duct intervention 10. Kahaleh M, Hernandez AJ, Tokar J, et al. EUS-guided pancreaticogastrostomy: analysis of its efficacy to drain inaccessible pancreatic ducts. Gastrointest Endosc 2007;65:224-30. 11. Kinney TP, Li R, Gupta K, et al. Therapeutic pancreatic endoscopy after Whipple resection requires rendezvous access. Endoscopy 2009;41: 898-901. 12. Mallery S, Matlock J, Freeman ML. EUS-guided rendezvous drainage of obstructed biliary and pancreatic ducts: report of 6 cases. Gastrointest Endosc 2004;59:100-7. 13. Tessier G, Bories E, Arvanitakis M, et al. EUS-guided pancreatogastrostomy and pancreatobulbostomy for the treatment of pain in patients with pancreatic ductal dilatation inaccessible for transpapillary endoscopic therapy. Gastrointest Endosc 2007;65:233-41. 14. Vila JJ, Perez-Miranda M, Vazquez-Sequeiros E, et al. Initial experience with EUS-guided cholangiopancreatography for biliary and pancreatic duct drainage: a Spanish national survey. Gastrointest Endosc 2012;76:1133-41. 15. Will U, Meyer F, Manger T, et al. Endoscopic ultrasound-assisted rendezvous maneuver to achieve pancreatic duct drainage in obstructive chronic pancreatitis. Endoscopy 2005;37:171-3. 16. Francois E, Kahaleh M, Giovannini M, et al. EUS-guided pancreaticogastrostomy. Gastrointest Endosc 2002;56:128-33.
Fujii et al
17. Kikuyama M, Itoi T, Ota Y, et al. Therapeutic endoscopy for stenotic pancreatodigestive tract anastomosis after pancreatoduodenectomy (with videos). Gastrointest Endosc 2011;73:376-82. 18. Will U, Fueldner F, Thieme AK, et al. Transgastric pancreatography and EUS-guided drainage of the pancreatic duct. J Hepatobiliary Pancreat Surg 2007;14:377-82. 19. Chahal P, Baron TH, Topazian MD, et al. Endoscopic retrograde cholangiopancreatography in post-Whipple patients. Endoscopy 2006;38: 1241-5. 20. Farrell J, Carr-Locke D, Garrido T, et al. Endoscopic retrograde cholangiopancreatography after pancreaticoduodenectomy for benign and malignant disease: indications and technical outcomes. Endoscopy 2006;38:1246-9. 21. Matsushita M, Takakuwa H, Uchida K, et al. Techniques to facilitate ERCP with a conventional endoscope in patients with previous pancreatoduodenectomy. Endoscopy 2009;41:902-6. 22. Gines A, Varadarajulu S, Napoleon B. EUS 2008 Working Group document: evaluation of EUS-guided pancreatic-duct drainage (with video). Gastrointest Endosc 2009;69:S43-8. 23. Varadarajulu S, Trevino JM. Review of EUS-guided pancreatic duct drainage (with video). Gastrointest Endosc 2009;69:S200-2.
GIE on Twitter GIE now has a Twitter account. Followers will learn when the new issues are posted and will receive up-to-the-minute news as well as links to author interviews, podcasts, and articles. Search on Twitter for @GIE_Journal and follow all of GIE’s tweets.
864 GASTROINTESTINAL ENDOSCOPY Volume 78, No. 6 : 2013
www.giejournal.org
Fujii et al
EUS-guided pancreatic duct intervention
APPENDIX 1. Anatomy Based on Type of EUS Procedure Performed and Success of MPD Drainage Total (n[45)
Drainage success (n[33)
Drainage failed (n [12)
Pancreaticoduodenectomy
25
19
6
Partial head resection w/R-Y pancreaticojejunostomy
1
1
0
Pancreas sparing duodenectomy
1
0
1
Billroth II with pancreaticogastrostomy
1
0
1
Middle pancreatectomy with pancreaticojejunostomy
1
1
0
Pancreas divisum
5
2
3
Duodenal stricture
1
1
0
ERCP-induced duodenal perforation
1
1
0
No altered anatomy
9
8
1
Altered Anatomy
APPENDIX 2. Reasons for Prior Failed ERP Reason
(n[37)
Native Papilla (nZ16) Failed cannulation (pancreas divisum)
5
Unable to advance guidewire to desired location
5
Minor papilla not identified (pancreas divisum)
4
Iatrogenic periampullary perforation during ERP
1
Migrated pancreatic stent
1
Surgical Anastomosis (nZ21) Pancreatic anastomosis not identified
10
Pancreatic anastomosis not reached
3
Failed cannulation
7
Unknown
1
www.giejournal.org
Volume 78, No. 6 : 2013 GASTROINTESTINAL ENDOSCOPY 864.e1