Early versus late drainage of pancreatic necrotic fluid collections: a systematic review and meta-analysis

Introduction

Acute pancreatitis represents one of the most common gastrointestinal diseases and reasons for hospitalization worldwide, with an annual incidence and mortality of 2,814,972 and 115,053, respectively [1,2]. In the United States, healthcare costs associated with acute pancreatitis hospitalizations exceed $2.5 billion dollars, with over 270,000 admissions per year [2]. While the majority of cases are mild and self-limited, there is a subset of patients, roughly 15-20%, who develop severe pancreatitis associated with organ failure and/or death [3-6]. Necrotizing pancreatitis is a worrisome complication that can prolong hospitalization and lead to a multitude of complications, including necrotic fluid collections (NFCs). NFCs, when infected, can significantly increase the mortality by up to 20-30% [6,7]. NFCs are categorized by their duration and their contents. Encapsulation and maturation of NFCs usually takes at least 4 weeks. Therefore, international guidelines have typically advised postponement of interventional management (drainage and/or debridement) of NFCs in order for the acute necrotic collection to develop into a well-defined, walled-off necrosis [3,8,9].

However, there are clinical scenarios in which patients develop symptomatic fluid collections from necrotizing pancreatitis, which warrant earlier intervention within the first 4 weeks of onset. Historically, patients underwent an open surgical necrosectomy, but the use of that procedure is now limited, because of its high morbidity and the emergence of minimally invasive, endoscopic and percutaneous modalities [10]. An endoscopic step-up approach is recommended as the first-line therapy for symptomatic and infected NFCs, whereas percutaneous intervention can be considered when endoscopic drainage is not technically feasible or unsuccessful [6]. The advent of endoscopic ultrasound (EUS)-guided intervention has challenged the idea that waiting 4 weeks before intervention is required, especially since earlier studies focused primarily on surgical approaches [11]. Clinically this is important, as patients who decompensate despite goal-directed medical therapy may indeed benefit from early intervention, especially in cases of infection, obstruction, nutritional failure or pain [6]. We conducted a systematic review and meta-analysis of studies comparing early drainage (ED) (<4 weeks) vs. late drainage (LD) (≥4 weeks) of NFCs using minimally invasive therapies.

Materials and methods

We performed this systematic review in accordance with the Cochrane handbook of systematic reviews and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [12,13] (Supplementary Table 1).

A comprehensive search of several databases was conducted, from each database’s inception to January 21, 2022, for papers in the English language. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Scopus, and Web of Science. The search strategy was designed and conducted by an experienced librarian with input from the study’s principal investigator.

Controlled vocabulary supplemented with key words were used to search early vs. late drainage of pancreatic NFCs. The actual strategy listing all the search terms and how they were combined is given in Appendix 1.

The key words used in the search included a combination of “pancreatic necrotic fluid collections,” “walled-off pancreatic necrosis (WON),” “drainage/intervention,” and “early vs delayed.” Two authors (SB, AKM) independently reviewed the titles and abstracts from the search results to identify the studies that were relevant to this systematic review and meta-analysis, based on predefined inclusion and exclusion criteria. The remaining full-text articles were then reviewed to include all the studies that met the inclusion criteria with all relevant information. Any discrepancies in study selection were resolved by consensus and in discussion with the senior author (RK). The bibliography sections of the selected studies and related review articles were searched for additional relevant studies.

Studies that met the following inclusion criteria were included in this meta-analysis: 1) studies consisting of mainly endoscopic drainage or an endoscopic “step-up” drainage approach to pancreatic NFC and WON; 2) studies that compared outcomes related to early (<4 weeks) vs. delayed intervention; 3) studies that reported adverse events, mortality, need for reinterventions with subsequent necrosectomies and/or surgery, and length of hospital stay (LOS); and 4) studies with a sample size of more than 10 patients. Studies were excluded if they 1) did not directly compare outcomes between early vs. delayed drainage of NFC/WON; 2) performed only surgical or percutaneous drainage without endoscopic drainage; 3) were not in the English language; 4) were animal studies; 5) were studies on pediatric populations; or 6) were letters to the editor, case reports, editorials or review articles. If multiple publications from the same cohort or a duplication of cohorts was identified, the most recent and the most comprehensive report analyzing the outcomes of interest was included.

After the relevant studies with pertinent outcomes of interest had been identified, 2 authors independently abstracted the data on study characteristics and the outcomes of interest on a standardized form. The abstracted outcomes that were reported as medians, ranges and confidence intervals, were converted into means and standard deviations, as per previously reported methods [14,15]. The risk of bias and quality assessment of the included studies were independently performed by 2 authors using the modified Newcastle-Ottawa scale (NOS) for retrospective studies and the Cochrane risk-of-bias for randomized trials (RoB2) tool [16,17]. Based on the modified NOS scale, the studies were assigned scores under 3 broad perspectives: 1) selection (4 questions with 1 point each); 2) comparability of study groups (2 points maximum); and (3) ascertainment of the outcome of interest (3 questions with 1 point each). An extra point was awarded for studies published as full manuscripts, while no points were awarded to studies published as abstracts. Studies with a total score of ≥9, 6-8 and ≤5 points were considered high, medium and low-quality studies, respectively. Discrepancies in data extraction, quality assessment and assessment of bias were resolved by the combined assessment of 2 authors.

The primary outcome of the current meta-analysis was the complication rates during the study period in the ED vs. LD groups. Secondary outcomes were the number of subsequent endoscopic or minimally invasive necrosectomies, mortality, number of endoscopic sessions needed, and the number of patients requiring salvage surgical intervention during treatment of WON and LOS.

The complications associated with ED and LD were defined and classified based on the American Society for Gastrointestinal Endoscopy lexicon or the Clavien-Dindo classification [18,19]. The need for subsequent necrosectomies was defined as endoscopic and minimally invasive necrosectomies, such as video-assisted retroperitoneal debridement. Patients requiring salvage surgical interventions were those who needed open surgical intervention for the treatment of NFC, or management of complications related to NFC drainage. In addition, the mortality reported during the study period was compared between the ED and LD groups.

Results

A total of 1039 studies were identified based on our search criteria. Eliminating the duplicates resulted in 978 studies that were screened based on their titles and abstracts. There were 753 irrelevant studies and 165 studies pertaining to editorials, review articles, pediatric studies, case reports and case series with less than 10 patients that were excluded. Forty-nine studies were thoroughly reviewed after the exclusion. Twenty-six studies were excluded as they involved cases with surgical or percutaneous drainage only. Fourteen studies were removed as there was no comparison between the timelines of drainage, and the data were incomplete. We identified 9 studies, with a total of 855 patients (320 ED vs. 35 LD), that met the inclusion criteria and were included in this meta-analysis (Fig. 1).

Figure 1 Flowchart summarizing study selection process

One study was a multicenter randomized controlled trial (RCT) [23] and 1 study was a single-center prospective study [24]. The remaining 7 studies were single-center, retrospective studies [11,25-30]. Four studies were based in the United States [11,26,28,29]. One study was from Japan [25], 1 study was from India [27], 2 studies were from Europe [23,24] and 1 study was from Australia [30]. While all the studies compared the outcomes of early vs. late drainage of NFCs using predominantly endoscopic drainage, 2 studies utilized an endoscopically centered “step-up” approach [11,23]. Six studies compared the outcomes of early vs. delayed endoscopic drainage only [24-27,29,30], and 1 study utilized dual modality drainage, which involves simultaneous endoscopic and percutaneous drainage of NFCs [28] (Table 1).

Table 1 Study characteristics

The overall complication rates of ED vs. LD during the study periods were similar, with a rate ratio of 1.060 (95% CI 0.79-1.42; P=0.69; Q=14.46, I2=51.61). Heterogeneity was moderate, with the Q value above the degrees of freedom and an I2 of 51.61 (Fig. 2).

Figure 2 Forest plot for complication rate in early vs. late drainage

CI, confidence interval

The mortality rate was mildly higher in the ED compared to the LD group (odds ratio [OR] 1.94, 95%CI 1.05-3.59; P=0.033; Q=5.71, I2=0). Heterogeneity was minimal (Fig. 3). The number of patients requiring subsequent necrosectomies for the treatment of NFC was similar in both groups (OR 2.15, 95%CI 0.86-5.35; P=0.099; Q=29.72, I2=79.81) (Fig. 4). Similarly, the number of endoscopic sessions needed for the treatment of NFC was similar among the ED and LD groups (OR 3.39, 95%CI 0.78-14.68; P=0.10; Q=128.97, I2=95.34 (Fig. 5).

Figure 3 Forest plot for mortality in early vs. late drainage

CI, confidence interval

Figure 4 Forest plot for number of patients requiring subsequent necrosectomies

CI, confidence interval

Figure 5 Forest plot for the total number of endoscopic sessions in early vs. late drainage

CI, confidence interval

The number of patients requiring open surgical intervention were similar in both groups (risk difference [RD] 0.023, 95%CI -0.009-0.05; P=0.15; Q=3.64, I2=0) (Fig. 6). The LOS was similar among the ED and LD groups (OR 1.01, 95%CI 0.40-2.54; P=0.97; Q=23.6, I2=83.05) (Fig. 7).

Figure 6 Forest plot for salvage surgery requirement in early vs. late drainage

CI, confidence interval

Figure 7 Forest plot for mean length of stay in early vs. late drainage

CI, confidence interval

The risk of bias and quality assessment of the studies were performed using the RoB2 tool or the NOS for RCTs and the remaining studies, respectively. The only RCT involved in the analysis was deemed to be at low risk for bias (Supplementary Fig. 1). According to the NOS assessment of the remaining 8 studies, 5 studies were deemed to be high quality, while 3 were deemed to be medium quality (Supplementary Table 2). Despite the lower number of studies, the assessment of publication bias based on visualization of the funnel plot, imputation adjustments based on Duval and Tweedie’s trim and fill technique, and Egger’s regression test were performed for the complication rate and the mortality. The analyses using these techniques did not show any significant publication bias (Supplementary Fig. 2,3).

Subgroup analyses were performed to analyze the difference between specific adverse events, such as bleeding and perforation, when they were reported in the study. There was no difference between the ED vs. the LD group in bleeding complications (RD 0.015, 95%CI -0.03-0.06; P=0.587). Similarly, there was no difference in perforation complications (RD -0.02, 95%CI -0.05-0.008; P=0.147) (Supplementary Fig. 4).

Subgroup analysis of only the high-quality studies did not reveal any differences between the ED vs. the LD group in terms of complication rate or mortality (Supplementary Fig. 5,6).

A sensitivity analysis was performed by removing 1 study at a time to evaluate if any single study had a major impact on an outcome. No single study had a significant impact on the complication rate (Supplementary Fig. 7). However, when the studies by Trikudanathan et al, Rana et al and McKay et al were individually removed from the analysis, the difference in mortality between ED vs. LD group was obliterated (Supplementary Fig. 8).

Discussion

The present study shows that ED was not associated with a higher incidence of complications or a greater need for subsequent endoscopic or surgical procedures. However, ED was associated with higher mortality compared to LD, probably because these patients were critically ill with symptomatic fluid collections. Although conservative management is the initial treatment of choice for NFCs, clinical deterioration can occur regardless of optimal medical therapy. Infected pancreatic necrosis has a 30% associated mortality rate and approximately half of these infections typically occur within the first week of admission [4]. The traditional approach to delay management in order for the fluid collection to mature was thought to decrease the risk of complications; however, the introduction of minimally invasive therapies has shifted this paradigm. Three recent multicenter randomized trials comparing endoscopic to surgical necrosectomy found that an endoscopic transluminal approach was associated with lower rates of multiorgan failure, proinflammatory responses, pancreatic fistulas and LOS [31-33]. These studies have shifted the doctrine by providing increasing evidence that an endoscopic step-up approach can reduce complications with no change in mortality.

Given these developments, the studies included in this meta-analysis aimed to determine if the timing of drainage could alter clinical outcomes. In our study, there was slightly greater pooled mortality in the ED group (OR 1.94, 95%CI 1.05-3.59; P=0.033). However, only an association, and not causality due to the ED procedure, can be established from this analysis. There have been no clear subgroups which have been identified to benefit from ED, and it has been reported that delaying intervention by 30 days may actually improve mortality [5]. As a result, it is safe to assume that, in the included retrospective studies, ED was performed as an inevitable option in critically ill patients who had failed conservative management, which may have introduced some bias. Moreover, it is important to take into consideration that these mortality rates are notably lower compared to a 2010 meta-analysis of 14 studies (n=1478 patients) that showed that infected pancreatic necrosis was associated with a 32% mortality [34]. Regardless of ED vs. LD, the endoscopic techniques currently available are superior. Our findings support LD as a potentially preferable approach if clinical deterioration is not present.

In terms of adverse events (AE), no difference was seen between ED and LD. While only 1 of the included studies reported a statistically significantly greater incidence of AE in the ED group, the study only reported bleeding complications [27]. Our subgroup analysis of pooled rates of bleeding complications did not show a difference between the ED and the LD group.

While perforation is a theoretical concern in ED, there were no significant differences observed. These findings may be due to improved EUS-guided drainage techniques and equipment in recent years. Furthermore, the advent of lumen-apposing metal stents (LAMS) has greatly improved this procedure, especially since minimally invasive endoscopic necrosectomies can be performed through the LAMS when indicated. In our meta-analysis, most studies, with the exception of 2, utilized LAMS. Six studies also utilized plastic and other metal stents for NFC drainage. It is possible that adverse events could be further reduced with widespread use of LAMS in future studies [35].

Pancreatic NFCs evolve over time, as the solid necrotic content becomes liquefied and encapsulation occurs. Theoretically, procedures done before encapsulation should require a greater number of necrosectomies or reinterventions, since the necrotic contents are not liquefied. However, our study did not show a greater number of subsequent necrosectomies, endoscopic reinterventions or salvage surgeries when comparing ED to LD. The timeline of 4 weeks to allow the optimal morphological changes to occur before drainage of NFCs is arbitrary, and there are instances when encapsulation can occur before 4 weeks. The use of high-resolution imaging studies, such as contrast-enhanced computed tomography and EUS, can help assess and guide an appropriate NFC drainage timeline. Some studies have noted encapsulation in 40% of the patients even before 4 weeks [11,26]. Moreover, it has also been postulated that encapsulation may not be relevant in the era of minimally invasive techniques [7,36]. In the meta-analysis by Bomman et al, the earliest intervention occurred at 10 days following the onset of pancreatitis, although the majority of the included studies performed intervention at 3 weeks [28]. In light of these findings, a multidisciplinary approach is paramount when determining an appropriate treatment course, and it is likely that the contents of the fluid collection are equally as important as the timing.

There are a few limitations to highlight. Firstly, most of the included studies were retrospective, with an inherent risk of bias. In addition, a few studies were reported as abstracts only. Secondly, all studies were conducted at high-volume tertiary care centers with experienced endoscopists. Therefore, the widespread applicability of these techniques may be limited—especially in resource-limited settings. Lastly, the heterogeneity was moderate to high in many of the assessed outcomes. Different modalities were used in the drainage of NFCs. Furthermore, various types of stents such as plastic, LAMS and other metal stents were used in different studies, which could have led to variability in drainage outcomes. It is possible that the increased use of LAMS may impact treatment outcomes in the future. The strengths of our study include the use of a comprehensive literature search of multiple databases, with careful data extraction and elimination of redundant studies.

In conclusion, this is the first meta-analysis to compare the outcomes of early vs. late endoscopic drainage of NFCs. Our findings show that, when clinically indicated, earlier drainage can be performed in carefully selected patients. A multidisciplinary team approach involving expert therapeutic endoscopists, interventional radiologists and hepatopancreaticobiliary surgeons is necessary for considering ED.