Efficacy and safety of endoscopic drainage of peripancreatic fluid collections: a retrospective multicenter European study

Introduction

Pancreatitis represents one of the most common gastrointestinal disorders and the most common pancreatic disease. One of the most severe complications of pancreatitis is the development of a peripancreatic fluid collection (PFC) [1]. If this becomes symptomatic or infected, interventional therapy is deemed necessary [2]. Endoscopic therapy was introduced, initially using plastic stents, and has gradually become the method of choice, as it combines safety and cost-effectiveness [3,4]. Plastic stents, usually double plastic pigtail stents (DPPS) even today, remain the type most commonly used. However, as their use has a high migration risk and they are also prone to occlusion due to their small caliber, the placement of multiple stents, necessitating repeated access to the PFC and reinterventions, is often required [5,6]. The metal stents used for PFC drainage were initially fully or partially covered, but more recently their design has evolved: lumen-apposing metal stents (LAMS) were introduced, whose shape minimizes the risk of migration. LAMS are made with wide flanges on both ends, providing anchoring within the PFC and an even distribution of pressure on the luminal walls [7]. However, after the initial enthusiasm over the advantages of LAMS, an increasing number of reports also highlighted various limitations and adverse events associated with their use [8,9]. Thus, in this study from 10 European tertiary centers, we sought to evaluate the efficacy and safety of endoscopic ultrasound (EUS)-guided drainage of PFCs and to compare the technical and clinical outcomes and adverse events of drainage with LAMS and DPPS.

Materials and methods

PFC drainage with LAMS

Upon establishment of the vessel-free tract (Fig. 1A-C), following confirmation that the distance between the EUS probe and the PFC was ≤1 cm, the LAMS (HOT AXIOS, Boston Scientific Corporation, Marlborough, MA, USA) was applied by the “free hand technique”; the stent was inserted into the PFC with cautery assistance, followed by deployment of the distal and the proximal flange under EUS or endoscopic vision, thus completing stent placement (Fig. 1D-F). All LAMS that were used in this study were HOT AXIOS.

Figure 1 (A) Endosonographic view of the catheter of the lumen-apposing metal stent (LAMS) advancing in a peripancreatic fluid collection; (B and C, white asterisk) deployment of the distal flange of the LAMS; (D) endoscopic view of the deployment of the distal flange of the LAMS; (E and F) gradual emptying of the collection within the gastric lumen

Results

Patient demographics and procedure characteristics

During the study period, 159 patients were initially assessed for eligibility, but only 128 of them (mean age 57.2±11.9 years, 71.9% male] met the inclusion criteria and were included in the analysis; these patients comprised 80 patients treated with a LAMS and 48 treated with a DPPS (Fig. 2). Patients’ baseline and procedural characteristics are summarized in Table 1. Ninety-two patients had a pseudocyst (71.9%) and 36 a WON (28.1%), with the majority of the lesions being located in the body of the pancreas (n=56/128, 43.8%) and transgastric (n=117/128, 91.4%) being the predominant approach. Among the 80 patients treated with LAMS, a stent with a diameter of 15 mm was most frequently used (n=61/80, 76.2%), while among the 48 patients treated with DPPS, stents of 3 cm length (35.2%) and 7 Fr diameter (54.9%) were mainly used.

Figure 2 Study flowchart

PFC, peripancreatic fluid collection; WON, walled-off necrosis; LAMS, lumen-apposing metal stents; DPPS, double pigtail plastic stents

Table 1 Patients’ baseline and procedural characteristics

Secondary endpoints

Technical success was achieved in 96.3% (n=77/80; 95%CI 92.1-100) of patients in the LAMS group, compared to 97.9% (n=47/48; 95%CI 93.9-100) in the DPPS group (P>0.99, Fig. 3). Clinical success did not differ significantly between the LAMS and DPSS groups: 76/80 (95.0%, 95%CI 90.2-99.8) vs. 43/48 (89.6%, 95%CI 80.9-98.2); P=0.29] Drainage time was significantly shorter in the LAMS group (88 vs. 35 days; P<0.001). No significant differences between the 2 groups regarding ease of placement, as rated by endoscopists (6.7 vs. 6.7; P=0.96). The rate of early AEs was similar between the 2 groups: 14/80 (17.5%, 95%CI 9.2-25.8) vs. 6/48 (15.2%, 95%CI 3.1-21.9); P=0.61 (Table 2). Of the severe AEs among LAMS patients, bleeding was the most common (n=6), with 4 of them eventually requiring angiographic embolization to be controlled. Finally, 2 cases of perforation occurred in each group. In each perforation case, with either LAMS or DPPS, the stent was immediately retracted and the perforation site was closed with endoscopic clips. The same procedure was repeated with a second cystoenterostomy created successfully at a different location, using LAMS in 2 cases and DPPS in the other 2. It should be noted that the majority of AEs were effectively managed conservatively, requiring no further surgical intervention. The complication rate at stent removal was significantly higher in the LAMS group: 10/80 (13.5%, 95%CI 5.3-19.7) vs. 0/48 (0%, 95%CI 0-0); P=0.03. Tissue overgrowth was the most prevalent complication (n=5/10, 50%), but none of these cases required re-stenting. The mean and median times until LAMS removal were 61.8 and 35 days, respectively.

Figure 3 Comparison of clinical outcomes between the LAMS and DPPS groups

LAMS, lumen-apposing metal stents; DPPS, double pigtail plastic stents; AEs, adverse events; SD, standard deviation; ns, not significant

Table 2 Early adverse events and their management, overall and per group

Technical success was similar for patients undergoing pseudocyst drainage with either a LAMS or DPPS: 51/52 (98.0%, 95%CI 89.7-99.9) vs. 40/40 (100%, 95%CI 91.1-100); P>0.99 (Table 3). The finding was consistent for clinical success and rate of early AEs: 52/52 (100%, 95%CI 97.1-100) vs. 39/40 (97.5%, 95%CI 86.8-99.9), P=0.435; and 7/52 (13.4%, 95%CI 5.6-25.8) vs. 4/40 (10.0%, 95%CI 27.9-23.7); P=0.750, respectively. Contrariwise, the rate of complications at stent removal was significantly higher in the LAMS group: 8/52 (15.4%, 95%CI 6.8-28.1) vs. 0/40 (0%, 95%CI 0-0); P=0.045. Neither sex nor type of drainage approach (transgastric vs. transduodenal) were found to have any impact on drainage outcomes (Supplementary Table 2).

Table 3 Clinical endpoints per peripancreatic fluid collection and stent type

As far as drainage outcomes among patients with WON is concerned, technical success was similar between LAMS and DPPS: 26/28 (92.8%, 95%CI 76.5-99.1) vs. 7/8 (87.5%, 95%CI 47.3-99.7); P=0.541 (Table 3). Clinical success and rate of early AEs did not differ significantly between the LAMS and DPSS groups: 24/28 (85.7%, 95%CI 67.3-95.9) vs. 4/8 (50.0%, 95%CI 15.7-84.3), P=0.05; and 7/28 (25.0%, 95%CI 10.7-44.9) vs. 2/8 (25.0%, 95%CI 31.9-65.1), P>0.99, respectively]. The rate of complications at stent removal was also similar between the 2 groups: 2/28 (7.1%, 95%CI 8.8-23.5) vs. 0/8 (0%, 95%CI 0-0); P>0.99.

Discussion

During the past 2 decades, EUS-guided transmural drainage has revolutionized the approach to symptomatic PFCs, allowing the achievement of technical success rates that exceed 90% and clinical success rates that range between 75% and 90% by performing a minimally invasive procedure [14,15]. Moreover, the advent of LAMS promised to overcome the fundamental caveats of DPPS [3]. Nevertheless, compelling evidence favoring the use of LAMS instead of DPPS in resolving PFCs is so far lacking in the published literature. Hence, current European Society of Gastrointestinal Endoscopy guidelines advocate the implementation of either plastic stents or LAMS for endoscopic transmural drainage [16]. However, one should bear in mind that this is only a weak recommendation based on moderate quality evidence, since data on the efficacy of LAMS remain scant.

Analysis of our primary endpoint showed high rates of technical and clinical success (96.9% and 93%, respectively), along with favorable results and an excellent patient safety profile. Endoscopic transmural PFC drainage can be a challenging procedure [3]. Although early data demonstrated that biliary DPPS can be effectively applied to treat symptomatic PFCs [17], studies that followed revealed the disadvantages associated with their small lumen diameter, the need for multiple stents to allow adequate drainage, occlusion and subsequent infection, significantly worse performance in the case of WON, stent-related AEs (up to 18%) and technical difficulty in the placement of multiple DPPS [5]. In light of these observations, we saw the advent of LAMS, whose bi-flared flanges enable firm tissue apposition, decreasing migration risk, while their wide-caliber lumen allows the performance of endoscopic necrosectomy as well as improved drainage [3]. Beyond their theoretical advantages, the efficacy of these devices was indeed corroborated, as data that followed showed a technical success of 95% and clinical success rates of 85-90%, with minimal risk of migration (5%) [7,18,19].

Our study also demonstrated that EUS-guided drainage with LAMS for the management of PFCs resulted in technical and clinical success rates similar to those of DPPS, according to most of the available data. Although LAMS have been increasingly used for endoscopic drainage of PFCs, large prospective studies comparing them to DPPS in terms of clinical or technical success remain limited, leaving the choice of stent dependent upon several factors—endoscopist’s discretion, local availability, reimbursement policy, etc.—rather than strong evidence-based findings that actually favor one type of stent over the other [16]. In the only randomized controlled trial available to date, no significant difference in treatment outcomes was found between LAMS and plastic stents [20]. This issue has been also at the focal point of several meta-analyses, with the latest and largest reporting no significant difference between LAMS and DPPS in terms of technical success or AEs. Although LAMS use resulted in higher clinical success, less recurrence and fewer additional interventions, one should also pay attention to the fact that in studies of WON LAMS was associated with more perforations [15]. However, a second read between the lines calls for careful interpretation even of these results. More specifically, the majority of individual studies enrolled had a retrospective design, including different types of metal stents or assessing the efficacy of LAMS or DPPS separately, thus raising concerns about biases and heterogeneity [14]. Forthcoming adequately powered, randomized controlled studies should systematically document factors (lesion type, location, endoscopist’s expertise level), that could allow for identification of heterogeneity in the treatment effect and ultimately settle this dispute.

A point that deserves attention is the fact that LAMS deployment resulted in a higher AE rate compared to DPPS, similar to the incidence reported in the most recent and largest meta-analyses on this topic [14,21]. Moreover, we identified bleeding once again as the most worrisome AE-related to LAMS in the setting of PFC drainage, finding that individual studies as well as meta-analyses confirm that it occurs at a significantly higher rate compared to DPPS (19% vs. 1%; P=0.003) [14,21,22]. A plausible explanation for this notion is that the radial force exerted by the LAMS on the walls of the cyst may contribute to hemostasis, but bleeding risk remains, as collapse of the cavity after drainage may lead to erosion of the posterior wall of the cavity by the stent. Other large registry-based studies did not support this notion [9,22,23], but this may be attributed to the inclusion exclusively of WON, or to a mixture of PFCs and variability in the endoscopist’s expertise level, combined with differences in the definition and interpretation of AEs.

LAMS seem to outperform their counterparts in the management of WON, since it is difficult for the solid encapsulated necrotic material to drain spontaneously, requiring additional procedures, i.e., endoscopic necrosectomy in 60% [3,24]. When, however, the drainage of a pancreatic pseudocyst comes into question, literature data are sparse [25]. To make things even more confusing, published studies usually group WON and pancreatic pseudocysts together during their analysis, despite the different outcomes following endoscopic intervention. In this regard, the current study displayed similar performance between DPPS and LAMS groups regarding procedural outcomes of WON and pseudocyst drainage, with only the rate of complications upon removal being higher in the LAMS group.

As far as implications in everyday clinical practice are considered, the nature of the PFC itself might eventually hold the pivotal role in guiding appropriate stent selection for each occasion. Thus, LAMS apparently represent the optimal modality for WON drainage; on the other hand, multiple indwelling DDPS placement should be preferred for the initial management of patients with pancreatic pseudocysts, since this strategy is not only efficient and safe in the long term, as was shown recently [26], but also cost-effective [4]. Nevertheless, there is much work still in progress [27], as factors such as clinical status, pancreatic duct integrity, patient compliance and endoscopist’s expertise level represent “gray” areas whose potential impact on the efficacy of each method remains unknown.

The strengths of this study include its multicenter design, accurately simulating everyday real-world clinical practice and contributing significantly to the generalizability of our findings. The large number of patients enrolled, the application of strict diagnostic criteria and the use of standardized definitions for AEs are also among the study’s assets.

There are also limitations related to our study that merit attention, with the cardinal one attributable to its retrospective design. In this sense, one might repudiate the novelty and quality of the evidence presented, given its relation to previous publications on this topic; however, the current study, by reporting on the accumulated experience of a considerably large cohort of patients from 10 tertiary hospitals across 4 European countries, enriches the literature by providing not only real-world data throughout a continent, but also clinically rich insights into real-world evidence that plays an increasing role in healthcare decisions. Real-world evidence can be generated by different study designs, not only randomized controlled trials, but also large sample observational studies that can accurately replicate everyday clinical practice. In this regard, the current iteration per se constitutes a kind of novelty that may expand our knowledge to support decision-making, improving safety and effectiveness, and ultimately patient-related outcomes. Second, it should be highlighted that only a single design of commercially available LAMS was included in this analysis; thus, is not possible to generalize to other LAMS. Third, the stent management algorithm and follow-up protocol were neither uniform nor standardized across centers, so variability among physicians and clinical settings might have resulted in heterogeneity. Another limitation is that the decision on which type of stent (LAMS, DPPS) to be used was based on each physician’s assessment. Finally, we did not undertake an official cost-effectiveness analysis and we did not report on late AEs, nor on the number of necrosectomy sessions, their efficacy and the PFC recurrence rate after stent removal.

To conclude, the findings of this study suggest that EUS-guided drainage of PFCs is associated with high percentages of technical and clinical success, regardless of the stent used (LAMS or DPPS). Notably, early AEs were similar between the 2 types of drainage, but tended to occur more frequently when LAMS were used. While further prospective randomized trials are needed to enhance our knowledge about PFC treatment and to identify factors that determine clinical success, while minimizing the risk of AEs, the use of DPPS in cases of pseudocyst drainage and the use of LAMS in cases of WON, necessitating further step-up approaches, seems a reasonable approach.