Introduction

The spectrum of acute pancreatitis ranges from a mild transitory form to a severe disease with high mortality rates. About 20% of cases are complicated by the development of pancreatic necrosis, often resulting in septic complications, multi-organ failure and high mortality rates up to 20–40% [1]. The development of sterile necrosis usually occurs during the first 4 days after onset of clinical symptoms and patients present with the typical picture of a systemic inflammatory response syndrome within the first two weeks [1, 2]. The later course is complicated by bacterial infection of pancreatic or peripancreatic necrosis, which occurs in 40—70%. There seems to be a correlation between the risk of infection and the extent of necrosis [3, 4],

In the past, surgical debridement has been the preferred treatment to control the septic focus [5, 6], which lowered the mortality rates down to 20–40% [7 8 9]. Several approaches were used, including closed and open techniques [10, 11]. However, in the last decade the management of necrotizing pancreatitis has changed. As the mortality rates of critically ill, often septic patients remained high, the preferred primary treatment strategy has shifted to non-surgical techniques, including early transfer of patients to intensive care units at specialized centres [12–14].

In this context, the use of minimally invasive methods has revolutionized patient management. In 1997, Gmeinwieser et al. [15] were the first reporting the percutaneous treatment of infected necrotizing pancreatitis by aggressive drainage and active percutaneous necrosectomy. Since that time a number of studies or case reports with different success rates have been published [16–20].

Whereas passive lavage through large-bore catheters is often not sufficient to remove solid debris, minimally invasive active methods, such as percutaneous necrosectomy, seem to be more effective—either to delay surgery until patients are stabilized or to avoid surgery at all. This hypothesis, however, has not yet been evaluated systematically.

Our hospital is a tertiary referral centre for acute pancreatitis. Since 1992, we treated patients with acute necrotizing pancreatitis minimally invasive with drainage therapy and percutaneous active necrosectomy. This technique, developed further by Gmeinwieser et al. [18] by fragmenting debris, partly under endoscopic control, and removing it actively, is now used routinely in our hospital. This paper analyses our experience regarding minimally invasive techniques as a primary therapy for infected pancreatic necrosis.

Materials and methods

Patient recruitment and characteristics

All patients with the diagnosis of acute pancreatitis treated with drainage therapy and receiving active percutaneous necrosectomy between 1992 and 2004 were identified by searching the radiological, surgical and internal medicine databases. Demographic and clinical data as well as details about surgical procedures were collected by reviewing patient charts and medical and surgical reports.

Radiological characteristics and interventions

The number and characteristics of computed tomographies (CTs) performed per patient and characteristics of percutaneous interventional therapy were collected by reviewing all radiological reports. In particular, the type and number of drainages per patient, as well as the number and details of interventional active necrosectomy procedures, were determined (Fig. 1). Severity/extent of pancreatitis was determined by retrospectively reviewing the CT examination at the time of first drainage placement (see below).

Fig. 1
figure 1

Therapeutic procedures

Full size image

All drainages were placed under CT guidance using the Seldinger technique. The number and the size of catheters were determined individually for each patient by the ICU team and the interventional radiologist. Clinical considerations and the size and location of necrosis, as well as the suspected amount of fluid and debris (based on CT findings), guided this decision. A median of 2.5 drainages per patient were used (range 1–14). The median size was 14 F (12–20). Active necrosectomy was performed when sepsis and necroses persisted despite drainage therapy. The technique has been described in detail before [17]. Briefly, the necrotic cavity was visualized by application of contrast media through the catheter (see Fig. 2). Under local anaesthesia, the drainage catheter was then replaced by a PEEL-AWAY sheath (up to 30 F). The necroses were fragmented by using soft-tip catheters, snares and dormia baskets. This was done under fluoroscopy guidance. The fragments were extracted through the sheath. Smaller necrotic fragments were removed carefully by aspiration using soft-tip catheters and by repeated forced flushing of the necrotic cavity with saline solution with a 50-ml syringe. Lavage was continued until the debris was removed and the fluid became clear. Alternatively, in some cases the cavity was inspected by endoscopy, using a flexible children´s endoscope (GIF-P-Olympus with an outer diameter of 8.7 mm). The endoscope was inserted through the PEEL-AWAY sheath and allowed a direct visualization of the necrotic cavity. Using this approach, necrotic tissue was also removed using the endoscope.

Fig. 2
figure 2

a Acute pancreatitis with extensive organ necrosis. Puncture of necrotic cavity with a 22-G CHIBA needle. The needle for local anaesthesia remains in place for guiding purposes. b CT after insertion of a 24-F Thalquick drainage into the necrotic cavity. c Fluoroscopy after application of contrast media through the catheter: only liquid components of the necrotic cavity are contrasted, solid necrotic tissue presents as filling defects. d The drainage has been replaced by a Peel away sheath. A guide wire has been inserted into the necrotic cavity to secure the access route. Fluoroscopy guided perctaneous necrosectomy with a dormia basket is performed. e CT at discharge from hospital: necroses and fluid collections have completely resolved. Residual vital pancreatic tissue is left. The dense artefact is caused by Lipiodol after percutaneous closure of a fistula to the pancreatic duct with a mixture of Onyx and Lipiodol

Full size image

Scores

The Balthazar score and CT severity index (CTSI) were determined by retrospectively reviewing the CT examination at the time of first drainage placement. The Ranson score was calculated for every patient at admission in our hospital. Furthermore, for patients receiving intensive care treatment, the Acute Physiology and Chronic Health Evaluation (APACHE) II score and the Simplified Acute Physiology Scale (SAPS) II score were determined by reviewing the laboratory data and the intensive car unit (ICU) flow charts.

Statistical analysis

Statistical analysis was made with the SPSS 12.0 statistical software (SPSS, Chicago, Ill.). Endpoints for patient outcome during hospital stay were defined as patient death and length of hospital and/or intensive care unit (ICU) stay. Values are given as total numbers, median with range or as percentages where necessary. A P value ≤0.05 was considered statistically significant.

Results

Patient population

Eighteen patients were included in the study. The characteristics of all 18 patients, including demographics, scores, etiology of pancreatitis, diagnostic and therapeutic procedures (including intensive care treatment modalities), are given in Table 1.

Table 1 Characteristics of 18 patients treated with percutaneous necrosectomy (ERCP endoscopic retrograde cholangiopancreatography, MV mechanical ventilation)
Full size table

Clinical and laboratory data

The Ranson score on day of admission ranged from 1 to 4 with a median of 2. Fifteen patients (83%) were critically ill, with the need of intensive care treatment; five patients required a second ICU stay later on. The APACHE II score ranged from 5 to 35 with a median of 22, the SAPS II score ranged from 20 to 94 with a median of 56. Twelve patients (67%) required mechanical ventilation, non-invasive positive-pressure ventilation was not performed in this setting. Renal replacement therapy was necessary in two patients (11%), plasma separation was performed in two patients (11%).

Inflammatory laboratory parameters were elevated in all patients. The leucocyte count on the day of admission was median 17.8/nl with a range from 4.4/nl to 55.0/nl. Leucocytes decreased over the first four days to a median value of 13.2/nl with a range from 6.1/nl to 25.8/nl. When the decision for drainage therapy was made, the median leucocyte count was 16.6/nl with a range from 7.9/nl to 27.5/nl and decreasing to median 14.5/nl (range 6.5–25.7/nl) within the next two days. The C-reactive protein (CRP) demonstrated elevated levels on the day of admission, in median 211 mg/l (range 7–442 mg/l). After 4 days, the median value was 157 mg/l with a range from 11 to 306 mg/l.

All patients received antibiotic treatment, ten patients (56%) antimycotic treatment.

Radiological data

All patients received contrast-enhanced CT, at a median of eight times, with a range from two to 20. Necrosis was detected in all parts of the pancreas in eight patients, only in the head in three patients, only in the body in one patient and only in the tail in two patients. One patient presented necrosis in the head and the body, two patients in the head and the tail and one patient presented necrosis of the peripancreatic tissue as evidenced by extensive fluid collections containing non-enhancing solid tissue. The clinical course of this latter patient did not differ from the other patients of the study group being critically ill with the need of intensive care treatment for 48 days and a prolonged hospital stay of 241 days.

The CTSI of all 18 patients had a median of 7.0, ranging from 4 to 10. The Balthazar score was 4.0 in all patients.

The first necrosectomy was performed in a median 33 days after hospital admission (range 2–days). For four patients a second necrosectomy session was necessary, in a median 41 days after hospital admission, with a range of 4–102 days. The radiation exposure ranged between 2,250 cGy × cm2 and 11,000 cGy × cm2. The length of the interventions was not documented in the radiology reports. However, according to our experience the procedures took from 45 min up to several hours, depending on the complexity of the anatomy and the size of necrosis.

Outcome and mortality

Length of hospital stay was a of median 75 days, with a range from 17 to 241 days. None of the surviving patients needed a second hospital stay later on. Fifteen patients underwent ICU therapy, the length of stay in the ICU ranged from 2 to 104 days, with median 25 days.

The ICU as well as hospital mortality was 22%, with four out of 18 patients dying in the ICU. Time to death was a median of 50 days after hospital admission (range 48–111), with a 30-day mortality of 0% and a 100-day mortality of 17% with three patients dying within the first 100 days.

For eight patients, full recovery was achieved by drainage therapy and active percutaneous necrosectomy alone, with no need for surgical debridement. In the other ten patients percutaneous necrosectomy alone was not sufficient and further surgical intervention was necessary, in a median 44 days after hospital admission (range 2–102 days). Most of these patients required surgical debridement soon after the percutaneous approach. However, for one patient surgery could be delayed by 39 days. For one patient emergency surgery was necessary because of a bleeding complication. Surgical necrosectomy was performed within the same session. The development of fistulas occurred in seven patients, three patients to the duodenum and one patient each to the stomach, the jejunum, the ileum and to the abdominal wall. No further procedure-associated complications were noticed. Four out of the ten surgically treated patients died: three from multiorgan failure, one from a fulminant pulmonary embolism.

Discussion

The removal of infected necrotic tissue in patients with acute, necrotizing pancreatitis remains the main therapeutic goal. Whereas surgical debridement still represents the “gold standard”, the use of percutaneous catheter treatment has been shown to be successful in some cases [16, 18, 21, 22]. But passive drainage often seems not sufficient to remove debris and devitalized, necrotic tissue from the pancreatic bed and should be limited to liquefied portions of infected necrosis [23, 24]. Another therapeutical option is to combine passive drainage therapy with active percutaneous necrosectomy including mobilization, fragmentation, extraction and lavage under fluoroscopy or endoscopy guidance.

This study outlines the outcome of 18 patients with acute, necrotizing pancreatitis treated primarily with CT-guided drainage therapy followed by active percutaneous necrosectomy. These patients often present in a very poor clinical condition and surgery puts them at risk for an escalation of multiorgan failure and temporary worsening of the septic situation [7, 25]. The minimally invasive technique reduces the surgical trauma and there is no need for general anesthesia. Furthermore, damage to vital pancreatic tissue and neighbouring organs is limited and splenectomy can be avoided. Hospitalization time (median 75 days, range 17–241 days in our series) seems to be comparable to the surgical approach [26]. However, the technique has disadvantages as well. The procedure itself can be time-consuming and can be associated with a considerable radiation exposure for the patient and interventionalist. Moreover, not all patients are suitable for this technique depending on the localization and amount of necrosis. There has to be a safe, preferably retroperitoneal access route. Moreover, percutaneous treatment is not advisable if the area of necrosis is discontinuous or involves the mesenteric root (see Table 2).

Table 2 Advantages and disadvantages of interventional treatment modalities compared with surgery
Full size table

The first investigators reporting minimally invasive techniques were Gmeinwieser et al. [15] in 1997, followed by Freeny et al. [21] in 1998. Freeny et al. reported 34 patients receiving percutaneous drainage therapy with aggressive lavage schemes. Unfortunately, no clear differentiation between patients with pancreatic necrosis, abscesses or pseudocysts was made. The success rate was 53% with regard to avoiding surgery and the mortality rate was 15%. In the same year, Echenique at al. [16] presented data from 20 patients with pancreatic necrosis treated more aggressively, including active aspiration, and in four patients removing solid debris with a dormia basket. The success rate was 100%, with no deaths. However, patient recruitment was biased because only hemodynamically stable patients were treated minimally invasively. Subsequently, the APACHE II scores of these patients were rather low (mean 7.2). Gmeinwieser et al. [15] presented a well-characterized patient population (n = 29) treated by percutaneous drainage therapy and/or percutaneous necrosectomy. Sepsis was controlled in 25 out of 29 patients (86%) and surgery was avoided in 20 patients (69%) with a mortality rate of 27%. We developed this method further by improving the technique for fragmentation of necrotic tissue and active mobilization under fluoroscopic or endoscopic guidance [18]. The minimally invasive approach is, meanwhile, used routinely in our hospital [17].

In our series the success rate was 44%, with eight out of 18 patients being cured without further need of surgery. In ten out of 18 patients surgery was necessary later on because the septic focus could not be entirely removed by the percutaneous approach. However, in one out of these ten patients the minimally invasive procedure stabilized the patient’s condition enough to delay surgery by 39 days. Six of these patients recovered, three died from septic multi-organ failure, one patient from a fulminant pulmonary embolism. Thus, the overall mortality rate was 22%, which is low compared with mortality rates of 20–40% reported in the surgical literature [7–9]. Considering that the patient population presented in this manuscript was seriously ill, which is reflected by very high clinical and radiological scores, a mortality rate of 22% can be regarded as success.

In one patient, a bleeding complication caused by the percutaneous approach occurred. No other procedure-related complications, such as bowel perforation or accidental injury of liver, spleen or kidney, were seen in our series. Although percutaneous necrosectomy means manipulation with potentially harmful devices in close proximity to major vessels, the risk of bleeding seems to be low if the procedure is performed by an experienced interventionalist. Freeny et al. [21] reported one patient with bleeding out of 34 patients (3%), Echenique et al. [16] one out of 20 patients (5%), and Gmeinwieser et al. [15] noticed no bleeding complications.

Regarding long-term outcome, little data exist. Endlicher et al. [27] published a long-term-follow-up study (15–52 months) including nine patients following passive drainage therapy and active percutaneous necrosectomy. The authors reported good long-term results with regard to quality of life after discharge.

Our study has limitations. Data were collected retrospectively and in a single centre. However, the patient population is well characterized.

In conclusion, active, percutaneous necrosectomy can be regarded as a complementary treatment modality in selected patients with pancreatic necrosis. Surgery can be avoided or delayed, thereby reducing perioperative morbidity and mortality [12]. Close communication between the interventionalist, gastroenterologists and surgeon is necessary to determine the optimal treatment strategy for every patient individually. More prospective studies—especially comparing patient outcomes following modern surgical and minimally invasive techniques—are warranted.