A comparison of direct endoscopic necrosectomy with transmural endoscopic drainage for the treatment of walled-off pancreatic necrosis

ORIGINAL ARTICLE: Clinical Endoscopy

A comparison of direct endoscopic necrosectomy with transmural endoscopic drainage for the treatment of walled-off pancreatic necrosis Timothy B. Gardner, MD, Prabhleen Chahal, MBBS, Georgios I. Papachristou, MD, Santhi Swaroop Vege, MD, Bret T. Petersen, MD, Christopher J. Gostout, MD, Mark D. Topazian, MD, Naoki Takahashi, MD, Michael G. Sarr, MD, Todd H. Baron, MD Rochester, Minnesota, USA

Background: Endoscopic therapy of walled-off pancreatic necrosis (WOPN) via direct intracavitary debridement is described. Objective: To compare direct endoscopic necrosectomy with conventional transmural endoscopic drainage for the treatment of WOPN. Design: Retrospective, comparative study. Setting: Academic tertiary-care center. Patients: Patients referred to Mayo Clinic, Rochester, Minnesota, since April 1998 for endoscopic drainage of WOPN. Interventions: Each patient underwent standard endoscopic drainage that consisted of transmural cavity puncture, dilation of the fistula tract, and placement of a large-bore stent(s). Patients were classified into the direct endoscopic necrosectomy group if, during any of their procedures, adjunctive direct endoscopic necrosectomy was performed; all others were in the standard drainage group. Main Outcome Measurements: Success was defined as resolution of the necrotic cavity without the need for operative or percutaneous intervention. Results: Forty-five patients were identified who met study criteria: 25 underwent direct endoscopic necrosectomy, and 20 underwent standard endoscopic drainage. There were no differences in baseline patient or cavity characteristics. Successful resolution was accomplished in 88% who underwent direct endoscopic necrosectomy versus 45% who received standard drainage (P! .01), without a change in the total number of procedures. The maximum size of tract dilation was larger in the direct endoscopic necrosectomy group (17 mm vs 14 mm, P ! .02). Complications were limited to mild periprocedural bleeding with equivalent rates between groups. Limitations: Retrospective, referral bias, single center. Conclusions: Direct endoscopic necrosectomy achieves higher rates of resolution, without a concomitant change in the number of endoscopic procedures, complication rate, or time to resolution compared with standard endoscopic drainage for WOPN. The need for fewer postprocedural inpatient hospital days and a decrease in the rate of cavity recurrence are also likely benefits of this technique. (Gastrointest Endosc 2009;69:1085-94.)

Abbreviations: BMI, body mass index; NOTES, natural orifice transluminal endoscopic surgery; OR, odds ratio; PFC, pancreatic-fluid collection; TTS, through-the-scope; WOPN, walled-off pancreatic necrosis. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication. See CME section; p. 1148. Copyright ª 2009 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2008.06.061

www.giejournal.org

Historically, acute necrotizing pancreatitis has been characterized by extensive morbidity and mortality.1 However, over the last several decades, improvements in early recognition of disease, appropriate management of infection, and aggressive supportive care have lowered the mortality rate to 10% with sterile necrosis and 25% with infected necrosis.2,3 In addition, much of the improvement in morbidity from necrotizing pancreatitis has resulted from changes in management of pancreatic-fluid collections (PFCs). Volume 69, No. 6 : 2009 GASTROINTESTINAL ENDOSCOPY 1085

Direct endoscopic necrosectomy

Although many types of PFCs exist, the treatment of late or ‘‘walled-off ’’ pancreatic necrosis (WOPN) (formerly called organized pancreatic necrosis), defined as the evolution of acute necrosis into a partially encapsulated, welldefined collection of pancreatic juice and debris, has historically been difficult to manage.4,5 Before a decade ago, WOPN was treated almost exclusively with open, and, more recently, operative laparoscopic debridement in patients with infected or symptomatic collections.6-10 In contrast to WOPN, pancreatic pseudocysts may also arise as a consequence of acute necrotizing pancreatitis, but, because pseudocysts are devoid of any solid debris, they can be drained successfully by endoscopic approaches. However, WOPN was historically believed to be less amenable to endoscopic or percutaneous treatments because of the nonviable solid components.11-14 More recently, however, there has been a paradigm shift in the management of WOPN toward less-invasive approaches. The goal of these newer techniques is to provide a more minimal access functional necrosectomy equivalent to those performed by open operative necrosectomy. Since the late 1980s, percutaneous, large-bore debridement catheters were used with some success by interventional radiologists to manage infected PFCs.15-18 When seeking an even less-invasive approach, the first description of successful, direct endoscopic irrigation of WOPN via cystduodenostomy or cystgastrotomy was reported by Baron et al19 in 1996. Since that time, this endoscopic approach has evolved from an irrigation-based ‘‘debridement’’ to direct entry into the cavity with direct endoscopic necrosectomy.20,21 Several groups reported experience with the latter approach, and one European series recently reported clinical resolution of the collections without the need for operative intervention in 93% of treated patients (n Z 25).20-24 Despite the reported successes of direct endoscopic necrosectomy for WOPN, there has been no comparison between this technique and other endoscopic methods. The purpose of our study was to compare the success of direct endoscopic necrosectomy with standard techniques of transmural endoscopic drainage for treatment of WOPN. We hypothesized that direct endoscopic necrosectomy would successfully resolve WOPN more frequently than standard techniques of endoscopic transmural drainage.

Gardner et al

Capsule Summary What is already known on this topic d

The endoscopic approach to walled-off pancreatic necrosis has evolved from an irrigation-based debridement to entry into the cavity followed by direct endoscopic necrosectomy.

What this study adds to our knowledge d

d

In a retrospective study that compared 2 treatments for walled-off pancreatic necrosis, successful resolution was accomplished in 88% who underwent direct endoscopic necrosectomy versus 45% who received standard drainage. Operative or percutaneous intervention was necessary in 50% of the patients treated with standard endoscopic drainage but in only one patient in the endoscopic necrosectomy group.

The study was approved by the Mayo Clinic Institutional Review Board. All adult patients referred for minimal access endoscopic drainage procedures to the Mayo Medical Center in Rochester, Minnesota, for WOPN were identified retrospectively from our computerized endoscopic database. Patients were referred from a variety of specialists, including gastroenterologists, primary care

physicians, and surgeons from within and outside of our institution. There was no formal interdisciplinary review process in place to determine which patients with WOPN were most appropriate for endoscopic intervention; the decision to proceed with endoscopic management was at the discretion of the consulting endoscopists and the referring providers. However, the direct necrosectomy approach became used after the publication by Seewald et al.21 The first patient was referred in April 1998, and the last patient was referred in October 2007. Data collection was performed retrospectively via comprehensive review of all available medical records. All the patients had been diagnosed with necrotizing acute pancreatitis based on the Atlanta classification before undergoing their first endoscopic drainage procedure.25 Many of the patients were managed initially from the onset of pancreatitis at outside facilities and then were transferred for subsequent treatment. All the patients developed well-circumscribed pancreatic or peripancreatic-fluid collections that had not resolved or responded to observational treatment. All of the patients had some degree of morbidity from their WOPN. Before referral for endoscopy at our medical center, none of the patients had undergone operative or endoscopic treatment for their fluid collections, and the duration of time from initial episode of pancreatitis to referral varied. Before their first endoscopic drainage procedure, all the patients had dynamic CT and/or magnetic resonance imaging performed at our institution. All the CTs were subsequently reviewed by an expert GI radiologist (N.T.), who was blinded to the clinical course. CTs were evaluated for the presence of WOPN, and only patients with solid and/or fat components within the collection were included in the study, because these patients likely would be appropriate candidates for direct necrosectomy. Patients with only fluid and/or air within the cavity would have been unlikely

1086 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 6 : 2009

www.giejournal.org

PATIENTS AND METHODS Patients

Gardner et al

Direct endoscopic necrosectomy

on the presence of fever, leukocytosis, positive Gram stain on fluid-collection aspirates, and/or positive blood cultures. All patients were also candidates for operative or percutaneous necrosectomy at the time of referral.20 The patients were classified into 2 groups for the purposes of this study. Each patient underwent standard endoscopic drainage that consisted of transgastric or transduodenal cavity puncture, dilation of the fistula tract, and stent placement. Patients were classified into the direct endoscopic necrosectomy group if, during any of their endoscopic procedures, the necrotic cavity was entered deliberately with an endoscope and necrosectomy was performed via basket, balloon, forceps, or net; all the other patients were in the standard drainage group.

Procedures Figure 1. Cavity puncture though the duodenal wall by using a needleknife without EUS guidance. Note the extravasation of purulent contents from the fistula tract.

to need direct necrosectomy and thus were not included in the analysis. The radiologic features identified for each collection included the maximum cross-dimension, location, content (presence of gas, debris, septations), extension to the colonic gutters, involvement of the pancreatic parenchyma, discontinuity and deformity of the pancreatic parenchyma and main duct, and the thickness and definition of the wall. A CTseverity index score was also calculated for each CT.26 For comparative purposes, CTs had to have been performed within 2 weeks of the first endoscopic intervention to be included in the statistical analysis that described the cavity characteristics. In contrast, the endoscopic results of patients with CTs performed more than 2 weeks before their initial endoscopic intervention (none more than 8 weeks) were included in the analysis that described success or failure of cavity resolution. Baseline patient characteristics (age, sex, body mass index [BMI], and Charlson comorbidity score) were calculated for each patient.27 The etiology of the initial episode of pancreatitis, as well as the time from the onset of pancreatitis until the time of the first endoscopic intervention, was also recorded. Because not all of the patients had their initial course of acute pancreatitis documented in our medical records, information on the initial duration of stay and organ failure, for example, was not abstracted. Each patient had been nonoperatively managed from the time of onset of pancreatitis and had multiple cross-sectional imaging studies that demonstrated unresolved WOPN. The indications for endoscopic intervention were determined for each patient and included (1) persistent pancreatitis, (2) suspicion of an infected collection, (3) biliary and/or gastric outlet obstruction, (4) persistent pain that required narcotics, (5) inability to eat, and/or (6) failure to thrive. An infected collection was suspected based www.giejournal.org

All patients provided informed consent before each endoscopic intervention. Endoscopies were performed by using a therapeutic side-viewing video duodenoscope (TJF160; Olympus America Corp, Melville, NY) with the patients under conscious sedation. Localization of the most-appropriate access site from within the gastric or duodenal lumen was performed by using 2 techniques. In the majority of the patients, external compression of the gastric or duodenal wall was determined by endoscopy while referencing the most recent CT. In some patients, EUS was used to identify the appropriate site of transmural puncture. If possible, a pancreatogram was performed in all patients before transmural intervention on the collection during the index endoscopy. Once the appropriate site of puncture was identified, the posterior gastric or medial duodenal wall was targeted and punctured by using needle-knife electrocautery or needle aspiration, as previously described (Fig. 1).20,28 In patients undergoing EUS, cyst puncture was performed under direct EUS guidance, with use of color-flow Doppler to help avoid disruption of mural blood vessels at the time of wall puncture. Aspiration of cyst contents and/or demonstration of contrast injection into the cavity under fluoroscopic guidance confirmed cavity access. Aspirated material was sent for culture and Gram stain. Once the collection was accessed, a standard 0.035-inch guidewire was advanced into the collection and coiled within the cavity under fluoroscopic guidance. The fistula tract was then dilated to at least 8 mm in size by using hydrostatic balloons until the waist was obliterated (Fig. 2). In patients who did not undergo endoscopic necrosectomy, the cavity was not entered with an endoscope, and the entire procedure was completed with a side-viewing duodenoscope. Instead, one or usually two 10F doublepigtail stents (Cook Endoscopy, Winston-Salem, NC) were placed into the collection along the fistula tract. In most instances, a 7F pigtail nasobiliary tube or PEG tube with jejunal extension was placed into the collection to allow for aggressive irrigation. Irrigation was performed with 50 to 200 mL 0.9% sodium chloride lavage every 2 Volume 69, No. 6 : 2009 GASTROINTESTINAL ENDOSCOPY 1087

Direct endoscopic necrosectomy

Gardner et al

Figure 3. Initial view from within the necrotic cavity after advancement of the forward-viewing endoscope through the fistula tract, demonstrating extensive necrotic debris. Figure 2. Dilation of the cystduodenostomy fistula tract by using a TTS balloon.

hours for the first 2 days, then every 4 to 6 hours for the ensuing 4 to 6 weeks if necessary. The decision to place an irrigation tube was at the discretion of the endoscopist performing the procedure. In patients in whom direct endoscopic necrosectomy was performed, a forward-viewing therapeutic gastroscope (TJF 160; Olympus) was inserted after dilation of the cyst-access tract either before or after placement of internal plastic stents via the duodenoscope (Fig. 3). By using a through-the-scope (TTS) balloon, the fistulous tract was then further dilated, under fluoroscopic guidance, up to 20 mm. The fluid contents within the cyst were then aspirated through the endoscope until dry, and devitalized necrotic tissue was then removed. To be included in the direct endoscopic necrosectomy group, the endoscopic necrosectomy did not need to be performed at the time of the initial endoscopic intervention, nor with each subsequent endoscopy. Patients needed to have direct necrosectomy performed at least one time to be included in the direct necrosectomy group. Patients were selected to have transmural drainage versus endoscopic necrosectomy at the discretion of the treating endoscopist in a random manner. There were no standard patient factors that led to one procedure being selected over another. However, later in the series, as the endoscopists became more experienced with the direct necrosectomy technique, there was a tendency toward proceeding to direct necrosectomy. Four expert interventional endoscopists experienced with the techniques of direct necrosectomy and transmural drainage performed the endoscopic procedures. The devitalized pancreatic tissue identified within the cavity was removed via a combination of several accessories, including 15-mm biliary-stone retrieval balloons, Roth retrieval net baskets (US Endoscopy, Mentor, Ohio); stone-

retrieval baskets (Olympus); tripod retrieval, rat-toothed, and pelican forceps (Olympus); and 10F irrigation probes (Gold Probe; Microvasive Endoscopy, Boston Scientific Corp, Natick, Mass) (Fig. 4). All devitalized tissue that could be visualized and easily dissociated from the wall of the cavity, without causing significant bleeding, was removed into the stomach or the duodenum. The degree of necrosectomy was performed at the discretion of the therapeutic endoscopist, with the goal of necrosectomy being to uncover pink granulation tissue lining the wall of the collection. Removed tissue was collected within the gastric or duodenal lumens and on occasion removed en bloc via the mouth by using a Roth retrieval net. During each intervention, all patients received intravenous antibiotics, which were continued for at least 3 days after the endoscopy. Serial CTs were performed every 1 to 2 weeks to evaluate the status of the WOPN until resolution. Routine endoscopic removal of stents was typically performed at 4 weeks, and repeated drainage or necrosectomies initially were usually performed at 3-day to 7-day

1088 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 6 : 2009

www.giejournal.org

Figure 4. Active debridement of necrotic debris by using rat-toothed forceps.

Gardner et al

Direct endoscopic necrosectomy

TABLE 1. Baseline patient characteristics

Characteristic No. patients

Endoscopic Standard P necrosectomy drainage value 25

20

61  15

63  10

.758

29  4

29  4

.706

3.3  2.7

3.9  2.3

.437

Gallstone

15 (60)

16 (80)

Alcoholic

1 (4)

3 (15)

Other

9 (36)

1 (5)

17 (68)

11 (55)

.537

6 (24)

9 (45)

.205

74

42

.276

Age, y (mean [SD]) 2

BMI, kg/m (mean [SD]) Charlson score (mean [SD]) Etiology (no. [%])

No. men (%) Positive cavity fluid cultures (%) Mean days from acute pancreatitis to first endoscopic intervention

intervals. When resolution of the cavity appeared assured based on cross-sectional imaging, the frequency of endoscopic reinterventions was decreased to 14-day to 21-day intervals. Endoscopic procedures were discontinued when resolution or near resolution of the collection was documented, or if endoscopic treatment failed and percutaneous and/or operative management was indicated. In some of our patients in whom the collection communicated with the pancreatic duct, transpapillary drainage was also performed at the time of the initial drainage procedure via placement of a pancreatic-duct stent. The following procedural characteristics were evaluated for the purposes of this study: drainage site, use of EUS to guide initial transmural puncture, maximal dilation size of the fistula tract, the number of procedures, the number of direct necrosectomies, whether direct necrosectomy was performed at the time of initial drainage, use of intracystic irrigation catheters, use of a pancreatic drainage tube, and the type of instruments used for necrosectomy.

sectional imaging either in the same location or in a different location within 6 months after documented resolution of the necrotic collection. Patients in whom follow-up had been less than 6 months were not included in the analysis of resolution. Recurrent pancreatitis and the formation of pancreatic fistula were also noted, as was the presence of pancreatic-ductal disruption on the final pancreatogram. Ductal disruptions were categorized as complete or partial. Also, mortality and the need for open, operative intervention for any cause were reviewed. Complications of endoscopic therapy included all periprocedural events, such as bleeding, perforation, or infection. Postprocedural complications, such as the development or exacerbation of pancreatitis, fistulas, or stent migration, were also reviewed.

Statistical analysis The study used a retrospective design in which patients who underwent drainage of WOPN by using direct endoscopic necrosectomy were compared with patients who underwent drainage of WOPN without endoscopic necrosectomy with respect to successful resolution of the cavity. The groups were compared by using the Wilcoxon-Mann-Whitney test for continuous variables and the 2-tailed Fisher exact test for categorical variables. Multivariate logistic regression analysis was performed to estimate the odds of successful resolution of the cavity for the direct endoscopic necrosectomy while controlling for confounders. Potential confounders were examined by means of stratified analyses, and confounding was deemed present when a O10% difference between the crude and adjusted odds ratios (OR) was observed. A P value !.05 was considered statistically significant. With regard to the issue of multiple testing of outcome data arising from individual patients, our approach was to take successful resolution as the main outcome measurement, with all other statistical tests meant to highlight potential differences (such as in the number of procedures or complications) or for the purpose of identifying confounders. Thus, all P values are presented uncorrected for multiple testing. The statistical analyses were performed by using JMP software (Cary, NC) and EXCEL (Microsoft Corp, Redmond, Wash).

RESULTS Outcomes The primary outcome was resolution of the necrotic collection without operative or percutaneous drainage. Resolution was defined as complete or virtually complete (O90%) resolution of the cavity by using only endoscopic techniques. Secondary outcomes included time to resolution of the cavity based on follow-up CTs and the total number of inpatient hospital days from the time of initial therapeutic endoscopic intervention until resolution of the cavity. Recurrent collections were also evaluated, with recurrence defined as return of collection on crosswww.giejournal.org

A total of 45 patients underwent attempted endoscopic management of WOPN during the study period: 25 patients underwent direct endoscopic necrosectomy (cases), and 20 underwent standard endoscopic drainage (controls). Baseline characteristics of each group are displayed in Table 1. The groups were similar in all respects; however, the mean time from initial episode of acute pancreatitis to initial endoscopic intervention tended to be longer in the endoscopic necrosectomy group (74 days vs 42 days, P Z .276). The range was 21 to 345 days (median 39 days) in Volume 69, No. 6 : 2009 GASTROINTESTINAL ENDOSCOPY 1089

Direct endoscopic necrosectomy

Gardner et al

TABLE 2. Baseline CT characteristics

Characteristic No. of patients* CT severity index (mean [SD]) Maximum size in cross section (mean [SD]) (cm)

TABLE 3. Clinical characteristics of drainage procedures

Endoscopic necrosectomy

Standard drainage

17

20

6.5  1.6 14.8  5

P value

Characteristic

Endoscopic Standard P necrosectomy drainage value

Drainage site (%)

6.4  1.6

.781

16.7  5.6

.150

Transgastric

52

60

.764

Transduodenal

48

40

Maximal dilation size (mean [SD]) (mm)

17  2

14  4

3.1  2.4 .089

.016y

Extension to colonic gutter (%)

71

55

.498

No. procedures (mean [SD])

3.6  1.8

Well-defined wall border (%)

29

20

.703

No. necrosectomies (mean [SD])

2.2  0.9

NA

NA

Solid component within cavity (%)

29

20

.703

Necrosectomy at first drainage procedure (%)

44

NA

NA

Fat component within cavity (%)

88

95

.584

Nasocystic/PEG irrigation catheter (%)

84

80

1.000

Gas component within cavity (%)

35

5

.034y

Nasocystic/PEG irrigation catheters (mean [SD]) (d)

13  14

10  11

.669

Cavity septations (%)

82

100

.088

EUS guidance (%)

24

5

.112

Disconnected parenchyma (%)

77

40

.045y

Complications (%)*

32

20

.502

*CTs had to be performed with 14 days of the first necrosectomy; as a result, 8 patients were eliminated from the necrosectomy group because CTs were O14 days. yStatistically significant.

NA, Not applicable. *Complications were all secondary to local bleeding at the time of drainage; all bleeding was controlled with local measures (coagulation, epinephrine injection, or hemoclipping). yStatistically significant.

the necrosectomy group and 22 to 67 days (median 47 days) in the standard drainage group. We acknowledge that the patients managed by necrosectomy were in the latter half of this time period, because our treatment protocol developed over these 9 years. The severity of illness, however, did not appear different, and the etiologies of acute pancreatitis were similar between groups, with biliary disease being the most common. CTs performed within 2 weeks before the initial endoscopic procedure were available in 37 patients (82%). The remaining 18% of patients had CTs performed sooner (range 18-60 days) than 2 weeks before the first endoscopic intervention. The baseline characteristics of WOPN on CT are displayed in Table 2. The collections were similar in terms of size, extension to the gutters, the degree of solid and fat components visualized, and the presence of septations. In contrast, CT findings of gas within the cavity (35% vs 5%, P! .04) and the presence of a ‘‘disconnected parenchyma’’ (77% vs 40%, P ! .05) were more common in the necrosectomy group. About half of the collections in the standard drainage group proved to be infected based on culture results from the cyst fluid, whereas 76% were infected in the necrosectomy group (P! .02). All of the other collections were clinically and biologically sterile.

The clinical characteristics of the endoscopic procedures for each group are highlighted in Table 3. Transgastric access to the WOPN was the most common approach, which represented 56% of the combined total procedures. Patients who underwent endoscopic necrosectomy had larger maximal diameter dilations of the fistula tract compared with the standard drainage group (17 mm vs 14 mm, respectively, P ! .02). The range of maximal dilation was 8 to 20 mm in the standard group and 12 to 20 mm in the necrosectomy group. The number of endoscopic reinterventions and the use of intracystic irrigation tubes were similar between groups. EUS guidance before cyst puncture tended to be used more often in the direct necrosectomy group compared with the standard treatment group (24% vs 5%, respectively). The complication rates related to the procedures are listed in Table 3. There was no statistically significant difference between the groups in terms of the rate of complication. Complications were limited to local bleeding at the time of the procedure. The use of EUS did not prevent bleeding complications, either at the time of initial cavity puncture or during subsequent necrosectomy. All patients with local bleeding were managed with injection of 1:10,000 epinephrine to the bleeding site. Three patients in the direct necrosectomy group had hemoclips placed

1090 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 6 : 2009

www.giejournal.org

Gardner et al

Direct endoscopic necrosectomy

TABLE 4. Primary clinical outcomes

Characteristic

Endoscopic necrosectomy

TABLE 5. Multivariate analysis of possible confounders Standard P drainage value

Successful resolution (%)*

88

45

.003y

Mean months to cavity resolution after initial intervention

4.2

4.0

.989

Need for operative cavity drainage (%)

4%

30%

.034y

Need for percutaneous cavity drainage (%)

0

20

.039y

Total in-hospital days after initial drainage (mean [SD])

15.4  15.2

38.4  60.4 .448

Recurrent collection (%)z

8

40

.014y

Recurrent pancreatitis (%)

0

5

.465

*Success was defined as resolution of cavity without any other interventions besides endoscopic drainage, including surgery or percutaneous drainage. yStatistically significant. zRecurrence of collection O6 mo after initial resolution; 4 patients in the direct necrosectomy group were eliminated from this analysis because their initial debridement occurred !6 mo before data analysis.

locally to control bleeding, whereas none were required in those undergoing standard drainage. No patients required transfusion of packed red blood cells. Bipolar cautery was used in 1 patient in each group. There were no unrecognized, postprocedural bleeding episodes that required repeated endoscopy. Similarly, there were no uncontrolled luminal perforations in either treatment group. The primary clinical outcomes of these 2 treatment groups are displayed in Table 4. Successful resolution of WOPN, without the need for further percutaneous or operative intervention, occurred in 88% of patients treated with necrosectomy versus 45% of patients treated with standard drainage (P ! .01). The time to resolution of the cavity was similar between the groups. In the standard drainage group, 30% of patients eventually needed operative drainage, compared with only 4% of the direct necrosectomy group (P ! .04). Operative interventions were performed a mean of 6.6 months after the initial endoscopic drainage in the standard group and 6.2 months in the direct necrosectomy group (P Z .9). Operations in the standard group (n Z 6) included cystgastrotomy or cystduodenostomy for unresolved WOPN, and, in 1 patient, repair of perforated diverticulitis and simultaneous cystgastrotomy (the PFC was repaired incidentally during the repair of the perforation). In the direct necrosectomy group, only 1 patient required cystduodenostomy for www.giejournal.org

Successful cavity resolution Outcome

OR (95% CI)

1-Crude/ad- P justed OR value

Direct necrosectomy 8.96 (2.01-39.92)

NA

!.01

Adjusted for maximum fistula tract dilation size*

0.10

!.01

Adjusted for gas 8.40 (1.81-38.92) component in cavity on initial CT

0.063

!.01

9.91 (2.04-48.23) Adjusted for disconnected parenchyma on initial CT

0.096

!.01

10.24 (1.98-52.9)

NA, Not applicable. *Analysis was performed for O16 cm and %16 cm, because 16 cm was the median dilation size for both groups.

unresolving WOPN. None of the patients in the direct endoscopic necrosectomy group required percutaneous drainage, whereas 4 patients required this procedure in the standard treatment group (P ! .04). The mean total number of inpatient hospital days after the initial endoscopic intervention did not differ significantly between the 2 groups (15.4 [endoscopic necrosectomy] vs 38.4 [standard drainage] days, P Z .45), although there was a trend toward a shorter duration in the necrosectomy group. The mean duration of followup was 13.6 months for both groups. Recurrent pancreatitis was uncommon (only 1 patient in the entire group), but recurrent fluid collections were more common in the standard drainage group (40% vs 8%, P ! .02). Pancreatograms were performed in 11 patients (55%) in the standard drainage group and 10 (40%) in the necrosectomy group. In the standard drainage group, complete ductal disruptions were present in 5 patients and partial disruptions were noted in 2. In the necrosectomy group, complete disruptions were present in 7 patients and partial disruptions in 2. ORs to evaluate for the potential confounding effect of between-group differences in baseline characteristics are shown in Table 5. Only baseline characteristics that were significantly different between groups were compared in the model. The OR for successful cavity resolution was adjusted to control for the maximum dilation size of the fistulous tract, the presence of gas within the cavity on initial CT, and the presence of a disconnected parenchyma on CT. By adjusting for maximum tract dilation, size slightly confounded the relationship between direct necrosectomy and successful cavity resolution; however, the degree of confounding was minimal at 12.5%. The other 2 variables did not introduce significant confounding. Volume 69, No. 6 : 2009 GASTROINTESTINAL ENDOSCOPY 1091

Direct endoscopic necrosectomy

DISCUSSION

Gardner et al

We found that direct endoscopic necrosectomy achieves higher rates of successful cavity resolution without a concomitant change in the number of endoscopic procedures, complication rate, or time to resolution compared with standard endoscopic drainage for WOPN. The need for operative or percutaneous intervention was necessary in 50% of the patients treated with standard endoscopic drainage but in only 1 patient in the endoscopic necrosectomy group. In addition, although not statistically significant, there were fewer inpatient hospital days from the time of initial therapeutic endoscopic intervention until resolution of the cavity in the direct debridement group. The number of recurrent fluid collections was also less for those undergoing direct debridement. Since the first description in the mid 1990s of minimal access direct endoscopic necrosectomy for the treatment of WOPN, reported case series that describe this technique have been relatively few.20-22 Although successful resolution of postnecrotic peripancreatic collections has been reported in approximately 90% of patients in each series, it was previously unclear as to whether direct necrosectomy with active debridement of intracavitary necrosis conveys any benefit, and whether this more-aggressive approach occurs at the expense of greater complication rates, compared with standard stent-assisted drainage. Our study is the first to comparatively evaluate this technique. Over the past decade, since the initial description of direct endoscopic necrosectomy for WOPN, endoscopic techniques have improved considerably. Endoscopic treatment for WOPN was proposed as a minimally invasive technique to avoid the morbidity and convalescence of open operative necrosectomy.20 Initially, smaller transmural tracts were used to ‘‘drain’’ these collections. Plastic stents were placed through the fistula tract to allow for cavity drainage around and through the stents. Although this technique is effective for pseudocysts, which contain little or no solid material, it became increasingly clear that the necrotic debris inherent in WOPN was unable to be effectively drained or evacuated in this manner. For example, in a Mayo series of 138 patients who underwent endoscopic transmural and/or transpapillary drainage of PFCs, resolution was significantly more frequent in patients with chronic pseudocysts (59/64 [92%]) than acute pseudocysts (23/31 [74%]) or necrosis (31/43 [72%]).29 As a result, direct endoscopic necrosectomy grew from the need for more aggressive attempts at removal of the necrotic contents of WOPN. Initially, necrosectomy was performed through smaller transmural tracts of 12 to 16 mm. However, we have become increasingly more aggressive with our tract dilation. Currently, we regularly dilate the access tract up to 20 mm. In addition, the extent of internal debridement has evolved from a technique of irrigating fluid through

transoral or transgastric intracystic drains in an attempt to ‘‘debride’’ the cavity to our current approach of accessing the cavity with the endoscope and actively removing the devitalized tissue under direct visualization. Our results demonstrate a marked improvement in nonoperative endoscopic success rates by using the direct necrosectomy approach compared with standard drainage techniques. We defined success as complete or virtually complete resolution of the cavity, without the need for percutaneous or operative intervention. We believe that it is important to define success in this manner because it allowed for a direct comparison of endoscopic techniques between groups. For example, percutaneous radiologic interventions, percutaneous laparoscopic necrosectomy, or small incision with focused necrosectomy were all advocated as ‘‘minimally invasive’’ techniques of necrosectomy.30,31 If we had allowed the definition of success to include adjunctive percutaneous radiologic interventions, significant confounding would have occurred. In addition, a percutaneous approach leaves the patient with percutaneous tubes that require management and has limited ability for actual necrosectomy. The 2 groups were similar with regard to baseline characteristics, including age and medical comorbidities, although the time from initial onset of pancreatitis to first endoscopic treatment was greater in the necrosectomy group. As a result, one might conclude that the collections in the necrosectomy group were given more of an opportunity to coalesce before drainage, which could have improved outcomes. This concept is especially true given the recent trend of ‘‘late’’ necrosectomy in light of improved medical therapy for pancreatic necrosis. However, in both groups, endoscopic drainage procedures were performed as early as 3 weeks into the clinical course. None of our patients in either group had complications that could be from intervening on an immature collection of necrosis; thus, it is unlikely that this baseline difference affected our results. In addition, when considering the median time to intervention, the groups were similar. Comparing baseline CTs performed within 2 weeks of the first debridement revealed only minor differences between groups. When we adjusted for gas components in the cavity and a disconnected parenchyma on initial CT, significant confounding was not found. For the purposes of our comparison, it is critical that the groups were similar in terms of the 9 criteria (shown in Table 2) used to evaluate the appearance of the cavities on CT. One significant confounder was the maximum dilation size of the fistula tract. Our data indicate that, when adjusting for maximum tract size, irrespective of direct necrosectomy, patients did have higher rates of successful cavity resolution. Although the degree of confounding was minimal, tract dilation size is likely a dependent variable that does help to determine the ultimate success of endoscopic drainage procedures.

1092 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 6 : 2009

www.giejournal.org

Gardner et al

As we previously reported, the size of the collection more than 15 cm and extension of necrosis into the paracolonic gutters are risk factors for endoscopic failure.20 These patients may benefit from an initial percutaneous and/or operative approach to drainage of WOPN based on the location and extent of necrosis. Therefore, a possible limitation of our study could be the presence of referral bias in our population. Indeed, the treatment approach to necrotizing pancreatitis has evolved tremendously over the last decade. Although every attempt was made to account for all patients referred for endoscopic necrosectomy to our institution since 1998, it is certainly possible that patients in whom endoscopic necrosectomy was thought to be unfeasible could have been missed. These patients may have been referred directly for operative or percutaneous drainage and not be included in our data set. Another limitation of our study is that all of these procedures were performed at a tertiary medical center with extensive experience in this technique. Although we report a success rate for WOPN resolution of close to 90%, this may not be generalizable to routine endoscopic practice. Indeed, a multidisciplinary team that consists of interventional radiologists, pancreatic surgeons, and interventional endoscopists seems most appropriate to determine the appropriateness of each patient for endoscopic drainage. In addition, the role of a ‘‘learning curve’’ for direct necrosectomy should be considered, because the study was conducted over a 10-year period. Although there were no significant advances in endoscopic technology (such as accessories to perform debridement), which might have impacted the procedure during this time, it is likely that the endoscopists continuously improved their individual debridement technique. This may have biased the primary outcome in favor of direct necrosectomy, because most of the debridement procedures were performed in the latter part of the series. We previously reported the results of 53 patients at our institution who underwent transoral-transmural drainage of WOPN from 1996 to 2006.20 In that series, 43 of the patients (81%) were reported as having endoscopic resolution of their WOPN. However, 21 of the patients (40%) required concurrent radiologic-guided catheter drainage of associated or subsequent areas of peripancreatic fluid and/or WOPN. In addition, 12 patients required open operative intervention for resolution. Furthermore, endoscopic necrosectomy was performed in only 22 patients, and patients with air and/or fluid components to their WOPN were included in that series. As such, the patients, inclusion criteria, and definition of successful resolution differ between this study and our previous report, which likely accounts for their slightly incongruous results. As the future of interventional endoscopy continues to expand, most recently with the development of natural orifice transluminal endoscopic surgery (NOTES), the role of www.giejournal.org

Direct endoscopic necrosectomy

minimal access debridement of pancreatic necrosis is also likely to expand. NOTES is likely to drive the creation of improved endoscopic devices and platforms, which may allow interventional endoscopists routine access to the peritoneum and retroperitoneum.32 We think this will improve the ability not only to access necrotic collections and also allow for more aggressive intracavitary debridement. There may even be a time when disconnected pancreatic ducts and parenchyma are suitable for routine endoscopic management. In conclusion, our study compared adjunctive direct endoscopic necrosectomy with standard endoscopic drainage for the treatment of WOPN. We found that direct endoscopic necrosectomy was superior in resolving WOPN without increasing patient risk or delaying time to resolution. Although retrospective and based on a relatively small sample size, the study is the first to compare direct necrosectomy to other drainage techniques and hopefully will serve as a catalyst for prospective comparative studies of this technique in the future.

REFERENCES 1. Banks PA, Freeman ML. Practice guidelines in acute pancreatitis. Am J Gastroenterol 2006;101:2379-400. 2. Vege SS, Baron TH. Management of pancreatic necrosis in severe acute pancreatitis. Clin Gastroenterol Hepatol 2005;3:192-6. 3. Whitcomb DC. Clinical practice. Acute pancreatitis. N Engl J Med 2006;354:2142-50. 4. Adkisson KW, Baron TH, Morgan DE. Pancreatic fluid collections: diagnosis and endoscopic management. Semin Gastrointest Dis 1998;9: 61-72. 5. Baron TH. Endoscopic drainage of pancreatic fluid collections and pancreatic necrosis. Gastrointest Endosc Clin N Am 2003;13:743-64. 6. Connor S, Neoptolemos JP. Surgery for pancreatic necrosis: ‘‘whom, when and what. World J Gastroenterol 2004;10:1697-8. 7. Uhl W, Warshaw A, Imrie C, et al. Guidelines for the surgical management of acute pancreatitis. Pancreatology 2002;2:565-73. 8. Bradley EL 3rd. A fifteen year experience with open drainage for infected pancreatic necrosis. Surg Gynecol Obstet 1993;177:215-22. 9. Fernandez-del Castillo C, Rattner DW, Makary MA, et al. Debridement and closed packing for the treatment of necrotizing pancreatitis. Ann Surg 1998;228:676-84. 10. Ammori BJ. Laparoscopic transgastric pancreatic necrosectomy for infected pancreatic necrosis. Surg Endosc 2002;16:1362. 11. Beckingham IJ, Krige JE, Bornman PC, et al. Long term outcome of endoscopic drainage of pancreatic pseudocysts. Am J Gastroenterol 1999;94:71-4. 12. Yusuf TE, Baron TH. Endoscopic transmural drainage of pancreatic pseudocysts: results of a national and an international survey of ASGE members. Gastrointest Endosc 2006;63:223-7. 13. Monkemuller KE, Baron TH, Morgan DE. Transmural drainage of pancreatic fluid collections without electrocautery using the Seldinger technique. Gastrointest Endosc 1998;48:195-200. 14. Hookey LC, Debroux S, Delhaye M, et al. Endoscopic drainage of pancreatic-fluid collections in 116 patients: a comparison of etiologies, drainage techniques, and outcomes. Gastrointest Endosc 2006;63: 635-43. 15. Segal D, Mortele KJ, Banks PA, et al. Acute necrotizing pancreatitis: role of CT-guided percutaneous catheter drainage. Abdom Imaging 2007;32:351-61.

Volume 69, No. 6 : 2009 GASTROINTESTINAL ENDOSCOPY 1093

Direct endoscopic necrosectomy 16. Freeny PC, Hauptmann E, Althaus SJ, et al. Percutaneous CT-guided catheter drainage of infected acute necrotizing pancreatitis: techniques and results. AJR Am J Roentgenol 1998;170:969-75. 17. Freeny PC, Lewis GP, Traverso LW, et al. Infected pancreatic fluid collections: percutaneous catheter drainage. Radiology 1988;167:435-41. 18. Shankar S, vanSonnenberg E, Silverman SG, et al. Imaging and percutaneous management of acute complicated pancreatitis. Cardiovasc Intervent Radiol 2004;27:567-80. 19. Baron TH, Thaggard WG, Morgan DE, et al. Endoscopic therapy for organized pancreatic necrosis. Gastroenterology 1996;111:755-64. 20. Papachristou GI, Takahashi N, Chahal P, et al. Peroral endoscopic drainage/debridement of walled-off pancreatic necrosis. Ann Surg 2007;245:943-51. 21. Seewald S, Groth S, Omar S, et al. Aggressive endoscopic therapy for pancreatic necrosis and pancreatic abscess: a new safe and effective treatment algorithm. Gastrointest Endosc 2005;62:92-100. 22. Voermans RP, Veldkamp MC, Rauws EA, et al. Endoscopic transmural debridement of symptomatic organized pancreatic necrosis. Gastrointest Endosc 2007;66:909-16. 23. Kang SG, Park DH, Kwon TH, et al. Transduodenal endoscopic necrosectomy via pancreaticoduodenal fistula for infected peripancreatic necrosis with left pararenal space extension. Gastrointest Endosc 2008;67:380-3. 24. Schrover IM, Weusten BL, Besselink MG, et al. EUS-guided endoscopic transgastric necrosectomy in patients with infected necrosis in acute pancreatitis. Pancreatology 2008;8:271-6. 25. Bradley EL 3rd. A clinically based classification system for acute pancreatitis. Ann Chir 1993;47:537-41. 26. Balthazar EJ, Robinson DL, Megibow AJ, et al. Acute pancreatitis: value of CT in establishing prognosis. Radiology 1990;174:331-6. 27. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83.

1094 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 6 : 2009

Gardner et al

28. Baron TH, Morgan DE. Endoscopic transgastric irrigation tube placement via PEG for debridement of organized pancreatic necrosis. Gastrointest Endosc 1999;50:574-7. 29. Baron TH, Harewood GC, Morgan DE, et al. Outcome differences after endoscopic drainage of pancreatic necrosis, acute pancreatic pseudocysts, and chronic pancreatic pseudocysts. Gastrointest Endosc 2002;56:7-17. 30. Carter CR, McKay CJ, Imrie CW. Percutaneous necrosectomy and sinus tract endoscopy in the management of infected pancreatic necrosis: an initial experience. Ann Surg 2000;232:175-80. 31. Horvath KD, Kao LS, Wherry KL, et al. A technique for laparoscopicassisted percutaneous drainage of infected pancreatic necrosis and pancreatic abscess. Surg Endosc 2001;15:1221-5. 32. Giday SA, Kantsevoy SV, Kalloo AN. Principle and history of natural orifice translumenal endoscopic surgery (NOTES). Minim Invasive Ther Allied Technol 2006;15:373-7.

Received March 17, 2008. Accepted June 30, 2008. Current affiliations: Miles and Shirley Fiterman Center for Digestive Diseases (P.C., S.S.V., B.T.P., C.J.G., M.D.T., T.H.B.), Department of Radiology (N.T.), Department of Surgery (M.G.S.), Mayo Clinic, Rochester, Minnesota, Section of Gastroenterology and Hepatology (T.B.G.), DartmouthHitchcock Medical Center, Lebanon, New Hampshire, Department of Medicine (G.I.P.), Division of Gastroenterology, Hepatology and Nutrition, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, USA. Reprint requests: Todd H. Baron, MD, Division of Gastroenterology and Hepatology, Mayo Clinic College of Medicine, 200 First St SW, Charlton 8A, Rochester, MN 55905. If you want to chat with an author of this article, you may contact him at [email protected].

www.giejournal.org