Current status of minimally invasive necrosectomy for post-inflammatory pancreatic necrosis

Abstract

Objective:

This paper reviews current knowledge on minimally invasive pancreatic necrosectomy.

Background:

Blunt (non-anatomical) debridement of necrotic tissue at laparotomy is the standard method of treatment of infected post-inflammatory pancreatic necrosis. Recognition that laparotomy may add to morbidity by increasing postoperative organ dysfunction has led to the development of alternative, minimally invasive methods for debridement. This study reports the status of minimally invasive necrosectomy by different approaches.

Methods:

Searches of MEDLINE and EMBASE for the period 1996–2008 were undertaken. Only studies with original data and information on outcome were included. This produced a final population of 28 studies reporting on 344 patients undergoing minimally invasive necrosectomy, with a median (range) number of patients per study of nine (1–53). Procedures were categorized as retroperitoneal, endoscopic or laparoscopic.

Results:

A total of 141 patients underwent retroperitoneal necrosectomy, of whom 58 (41%) had complications and 18 (13%) required laparotomy. There were 22 (16%) deaths. Overall, 157 patients underwent endoscopic necrosectomy; major complications were reported in 31 (20%) and death in seven (5%). Laparoscopic necrosectomy was carried out in 46 patients, of whom five (11%) required laparotomy and three (7%) died.

Conclusions:

Minimally invasive necrosectomy is technically feasible and a body of evidence now suggests that acceptable outcomes can be achieved. There are no comparisons of results, either with open surgery or among different minimally invasive techniques.

Introduction

Patients with acute pancreatitis present to a broad range of general acute care clinicians.¹ Although the majority of episodes of acute pancreatitis are mild and self-limiting, up to one-third of patients will have a severe attack with underlying pancreatic necrosis and some of these patients will go on to develop infected necrosis.² Blunt (non-anatomical) debridement of necrotic pancreatic and peripancreatic tissue at laparotomy is the standard method of treatment for patients with infected post-inflammatory pancreatic necrosis.^3,4 Recognition that laparotomy may in itself add to morbidity by increasing postoperative organ dysfunction³ has led to the recent development of a host of alternative methods for debridement. These alternative methods mostly involve debridement via retroperitoneal, laparoscopic or endoscopic approaches, or combinations of these.⁵ They share the common goal of avoiding laparotomy and collectively are referred to as ‘minimally invasive necrosectomy’.⁶ These techniques continue to evolve and undergo refinement; however, minimally invasive necrosectomy is now recognized as a treatment option in patients withpost-inflammatory necrosis.^5,6 To date, there is no randomized trial evidence comparing these techniques with traditional ‘open’ necrosectomy or, equally importantly, comparing the different techniques of minimally invasive necrosectomy against one another. This presents a problem for clinicians dealing with patients with infected necrosis, who need to refer to the available evidence to guide the selection of optimal treatment options in clinical practice.

This report provides a comprehensive but concise overview of currently employed techniques in minimally invasive necrosectomy, with particular reference to choice of approach and outcome.

Materials and methods

Literature search

A computerized search was performed of the MEDLINE and EMBASE databases for the period from April 1996 to 15 May 2008 using the OVID search engine (Version 10.5.1, Source ID 1.13281.2.21; Ovid Technologies, Inc., New York, NY, USA). The map term to subject (MeSH) headings ‘pancreas’, ‘surgery’, ‘necrosis’, ‘interventional studies’, ‘laparoscopy’, ‘methods’, ‘surgical procedures’, ‘minimally invasive’, ‘necrosis’, ‘pancreatitis’ and ‘acute necrotizing’ were used. Results were combined with the keywords ‘pancreatitis’, ‘procedures’ and ‘necrosectomy’, with the aid of Boolean operators. There were 4063 hits in MEDLINE and 49 742 in EMBASE. The Cochrane systematic reviews methodology⁷ was utilized to cross-reference combined EMBASE and MEDLINE output with all clinical trials and studies, excluding experimental studies. Letters, reviews without original data, non-English-language papers and animal studies were excluded, leaving a final study population of 28 manuscripts. The search strategy and reasons for excluding manuscripts are provided in Fig. 1.

Figure 1.

Search strategy and exclusion criteria. ERCP, endoscopic retrograde cholangiopancreatography; CP, chronic pancreatitis

Data extraction and categorization

According to Windsor,⁶ minimally invasive necrosectomy can be classified by the type of scope used (flexible endoscope,laparoscope or nephroscope) and by the route of access (transperitoneal, transgastric or retroperitoneal). The three most widely reported techniques are the retroperitoneal nephroscopic approach, the laparoscopic route and the endoscopic transgastric route. As Windsor and others^6,8acknowledge that these are relatively ‘mature’ and established approaches, this study focuses on the evidence pertaining to these three techniques. Data on numbers of patients, methods of assessment of necrosis, presence of infection in necrosis and procedure-related outcomes, including morbidity and mortality, were extracted from individual reports.

Results

Characteristics of the overall study population

The 28 manuscripts yield detailed data on 344 patients who underwent minimally invasive necrosectomy. The median (range) number of patients per report is nine (1–53). None of the reports provide information on why a given technique of minimally invasive necrosectomy was chosen in preference to alternatives.

Retroperitoneal necrosectomy

Nine reports on retroperitoneal necrosectomy provide data on a combined total of 141 patients, with a median of 11 (1–47) patients per series (Table 1).^8–16 It is clear that there is no single technique of retroperitoneal necrosectomy. The various approaches may use a small retroperitoneal incision or rely on dilatation of a drain-track. The majority utilize contrast-computed tomography (CT) to guide the placement of small-calibre percutaneous drains into retroperitoneal collections. The drain-track is then dilated using the Seldinger technique and some form of scope (the majority utilize a nephroscope) is placed into the cavity and semi-solid necrotic tissue removed piecemeal. This is achieved using various accessories, including biopsy forceps and baskets.

Table 1.

Retroperitoneal necrosectomy

Study	n	Delay to necrosectomy, days, median (range)	Pre-op infected necrosis	Technique	Post-op irrigation	Procedures per patient, n	Major complications, %	Laparotomy required, n	Intensive care unit stay, days	Inpatient stay, days	Mortality
Shelat & Diddapur 20078	1	5	1	Seldinger + nephroscope	1	8	0	0	0	55	0
Gambiez et al. 20079	20	18 (mean)	13	Lumbotomy + mediastinoscope	20	5 ± 4 (mean)	2	2	N/A	62 ± 21 (mean)	2
Chang et al. 200610	19	35 (average) (14–56)	13	Left flank incision + Blunt dissection	0	N/A	3	1	39 (average) (7–95)	23.2 (average) (4–120)	3
Besselink et al. 200611	18	48 (0–181)	7	Left-sided lumbotomy + VARD	0	2 (1–11)	6	N/A	2 (0–83)	100 (43–240)	2 (11)
Mui et al. 200512	9	N/A	9	Seldinger + choledochoscope	0	3 (2–8)	2	2	0 (0–64)	84 (29–163)	1
Connor et al. 200513	47	28 (3–161)	38	N/A	N/A	3 (1–9)	43	12	0 (0–66)	64 (15–272)	9
Castellanos et al. 200514	11	13 (average) (1–28)	11	Left translumbar + flex endoscope	1	5 (mean) 3–10	N/A	0	33 (3–85)	98 (34–210)	3
Risse et al. 200415	6	48 (mean) (28–92)	6	Seldinger + nephroscope	0	2 (1–4)	1	0	0	26	0
Carter et al. 200016	10	24 (13–187)	10	Nephrostomy + nephroscope + STE	10	3 (1–6)	1	1	4	42 (23–213)	2

N/A, data not available; VARD, videoscope-assisted retroperitoneal debridement; STE, sinus tract endoscopy

In 108 (77%) patients, infection of pancreatic necrosis was confirmed preoperatively. Four of 11 reports refer to the use of postoperative irrigation. We found a consistent trend showing that retroperitoneal debridement needed to be repeated on several occasions in order to achieve clearance and that the technique is used mainly in patients with left-flank collections of necrotic material. All reports which provide data on number of operations refer to more than one procedure per patient. There were 58 (41%) complications in 141 patients. However, there is a lack of standardization in the terminology used to report complications. Eighteen (13%) required formal surgical laparotomy. Inpatient stay in all reports was in excess of 3 weeks. There were 22 (16%) deaths.

Endoscopic necrosectomy

Nine reports provide data on 157 patients undergoing endoscopic necrosectomy, with a median of eight (2–53) patients per report (Table 2).^17–25 Only one report gives data on preoperative confirmation of infection of necrosis. Although detail varies considerably, we found a consistent trend in technique: endoscopic ultrasonography is utilized to localize the collection and to exclude the presence of major vessels and typically a transgastric or transduodenal puncture is made to enter the cavity. The track is dilated and the scope used to achieve irrigation and drainage of debris. Endoscopically placed pigtail stents are left in the cavity. The majority of reports utilize postoperative irrigation and, as with retroperitoneal necrosectomy, multiple procedures are required. Major complications are reported in 31 (20%) patients and death in seven (5%).

Table 2.

Endoscopic necrosectomy

Study	n	Delay to necrosectomy, days, median (range)	Pre-op infected necrosis	Technique	Post-op irrigation	Procedures per patient, n	Major complications, %	Laparotomy required	Intensive care unit stay, days	Inpatient stay, days	Mortality
Voermans et al. 200717	25	84 (21–385)	N/A	EUS + endoscopic cyst entry	25		2	0	N/A	5 (1–45)	0
Papachristou et al. 200718	53	49 (20–300)	N/A	Duodenoscope-19 + posterior transgastric-32 (60%)	53	3 (1–12)	11	12	N/A	13 (0–90)	3
Will et al. 200619	5	N/A	N/A	Guidewire + needleknife	N/A	2 (average) 2–6 (range)	0	0	0	N/A	N/A
Hookey et al. 200620	8	23 (mean) (10–45)	N/A	Transpapillary + EUS	6 Nasocystic	1 (median) (IQR 1–1)	2	2	N/A	18 (10–35)	1
Charnley et al. 200621	13	24 (3–180)	11	EUS + Electrocautery gastrotomy	13 Naso-cavity	4 (1–10)	N/A	1	N/A	N/A	2
Raczynski et al. 200622	2	N/A	N/A	EUS puncture + balloon dilatation + pigtail stent	2	4	0	0	N/A	N/A	0
Seewald et al. 200523	5	14	N/A	EUS + duodenoscope + guide + dilatora	5	7	N/A	(2)	N/A	N/A	0
Baron et al. 200224	43	N/A	N/A	b	N/A	2 (1–6)	16	N/A	N/A	20 (0–75)	1
Seifert et al. 200025	3	N/A	0	EUS + needleknife	N/A	N/A	0	0	N/A	N/A	0

^aSeewald et al. (2005) make a distinction between necrosis (5) and abscess (8). Concomitant pancreatic duct stents were placed if there was duct disruption and fistula tracks were sealed with N-butyl-2-cyanoacrylate

^bBaron et al. (2002): during 1993–1996 the technique was needleknife sphincterotome with electrosurgery, catheter placement and balloon dilatation. From 1996, the technique was Seldinger followed by biliary guidewire, balloon dilatation and drainage

N/A, data not available; EUS, endoscopic ultrasonography; IQR, interquartile range

Laparoscopic necrosectomy

Six reports provide data on 46 patients undergoing laparoscopic necrosectomy, with a median of seven (1–19) patients per report (Table 3).^26–31 Four reports confirm the preoperative presence of infected necrosis. There is considerable variation in technique, but the two main methods involve either a ‘full’ laparoscopic procedure undertaken with a carbon dioxide pneumoperitoneum or a modified laparoscopic procedure aided by the placement of a hand-port. The majority utilize postoperative irrigation. One report describes a need for more than one procedure. Five (11%) patients required laparotomy and there were three (7%) deaths.

Table 3.

Laparoscopic necrosectomy

Study	n	Delay to necrosectomy, days, median (range)	Pre-op infected necrosis	Technique	Post-op irrigation	Procedures per patient, n	Major complications, %	Laparotomy required, n	Intensive care unit stay, days	Inpatient stay, days	Mortality
Bucher et al. 200826	8	31 (13–59)	8	Single-port 5-mm scope	8	1 (2)	0	0	0 (0–4)	N/A	0
Parekh 200627a	19	65 (22–154)	N/A	3 ports and hand-assist	N/A	1 median (1–3)	1	3	6 ± 2	16 ± 4	2
Ammori 200228	1	36	1	5 ports all lap	1	1	0	0	0	14	0
Horvath et al. 200129	6	30 (27–77)	6	2 ports, retroperitoneal No pneumo	3	N/A	N/A	2	N/A	N/A	0
Zhu et al. 200030	10	1–3	N/A	3 ports pneumo	3	N/A	N/A	N/A	N/A	N/A	1
Alverdy et al. 200031	2	20 (18–21)	1	2-ports Seldinger 5-mm scope	0	2 (1–3)	1	0	0	N/A	0

^aParekh (2006) describes a series of 19 from a total of 23 patients undergoing necrosectomy during 2001–2006. One conversion was for intraoperative enteric injury

N/A, data not available; lap, laparoscopic; pneumo, pneumoperitoneum

Discussion

To the best of our knowledge, the data presented here constitute the most comprehensive overview to date of the techniques of minimally invasive necrosectomy. Interpretation of the results must take into consideration the likely positive publication bias inherent in reporting ‘novel’ techniques and the limited value that can be placed on studies with small cohorts (the median number of patients per series was nine [range 1–53]).

Nonetheless, there is a pressing need for guidance in the contemporary management of pancreatic necrosis; patients continue to present with these problems and logical, rational decisions are required to be made about their care on a day-to-day basis.

Firstly, minimally invasive necrosectomy must be set in the context of pre-existing knowledge of the management of pancreatic necrosis and particular reference must be made to the optimal timing of the procedure. A randomized trial comparing early with late necrosectomy showed better outcomes after deferred intervention (after 21 days),³² a finding supported by other researchers.³³ Further, these findings are consistent with the pathobiology of pancreatic necrosis: at early surgery, there is no demarcation of necrotic tissue because this process takes 2–3 weeks to occur. There is no logical reason to suppose that minimally invasive necrosectomy should be exempt from this and intervention should be deferred until beyond the third week.

Next, although there is no randomized trial evidence to suggest that preoperative fine-needle aspiration (FNA) of necrosis reduces mortality, it does aid logical decision making.³⁴ In this regard, it may be of relevance that few patients undergoing endoscopic necrosectomy had reported preoperative FNA (Table 2). In a similar context, as these patients will come to surgery after the initial systemic inflammatory response period, more detailed imaging of necrosis by magnetic resonance scanning may be invaluable in assessing the relative proportions of the fluid and solid content of collections.³⁵ Besselink and colleagues report that minimally invasive ‘drain-guided’ retroperitoneal necrosectomy was feasible in up to 56% of patients with infected necrosis.⁵

As minimally invasive necrosectomy techniques have been available for a decade, it is perhaps not surprising that there is a reasonable degree of concordance regarding each type of procedure. It is accepted that technology and techniques will continue to evolve, However, in order to maintain a fair perspective, it should be noted that the ‘conventional’ operation of open necrosectomy has also evolved and been simplified and, in specialist hepatopancreatobiliary centres, is carried out with very low mortality.^36,37

How then should a generalist treat a patient with suspected pancreatic necrosis? In keeping with international guidelines, patients with pancreatic necrosis represent individuals at high risk for adverse outcome and should be managed by a multidisciplinary team in a specialist unit. At present, there is no direct comparative evidence to suggest that any of the minimally invasive necrosectomy techniques are superior to open surgery. In order to maintain a balanced perspective, however, it should be acknowledged that the available data for the minimally invasive approaches do suggest that good outcomes can be achieved. The traditional limitations of open surgery (worsening of organ dysfunction and postoperative need for critical care) should be balanced against the limited information on the outcomes of the minimally invasive approaches.

Of the minimally invasive techniques, the retroperitoneal approach may be selected in patients with left-sided, predominantly retroperitoneal necrosis with a predominantly semi-solid collection. Endoscopic necrosectomy appears most effective in addressing a uni-locular collection containing fluid of low Hounsfield unit density and located in a principally retrogastric position. Issues of terminology arise here as a retrogastric fluid collection present at more than 4 weeks after an attack of acute pancreatitis would correctly be termed a pseudocyst. The impending revision of the Atlanta terminology will be invaluable in this regard.³⁸ Similarly, laparoscopic necrosectomy seems to be most effective in addressing uni-locular, retrogastric fluid collections, but has the additional advantage of allowing for a wide cystogastrostomy and perhaps better debridement than with endoscopy. A direct comparison of laparoscopic with endoscopic debridement seems feasible.

Percutaneous drainage should also be borne in mind. Although it could be argued that this is simply drainage rather than debridement, in practical terms, percutaneous drainage of infected peripancreatic collections remains a valid option, especially in terms of gaining time to stabilize a critically ill patient.

Fusion of the different approaches has also been reported, such as the ‘laparo-endoscopic’ drainage route.^39,40

An interesting approach taken by the Dutch Acute Pancreatitis group concerns the adoption of a step-up policy commencing with a minimally invasive approach and progressing to open surgery. This study (the PANTER trial) may provide useful evidence in relation to minimally invasive necrosectomy and is a good model for study of these complex techniques.³⁷

In conclusion, it is clear that the era of minimally invasive necrosectomy has arrived, but with a limited body of evidentiary support. The selection of treatment must be guided by the need to ensure the availability of true multidisciplinary expertise in a specialist unit. Techniques should not be selected simply because of the expertise of an individual clinician. As the techniques of minimally invasive necrosectomy become standardized, it is the duty of the specialist hepatopancreatobiliary clinical community to ensure that multicentre comparative studies are undertaken.

Conflicts of interest

None declared.