Introduction
Endoscopic ultrasound (EUS)-guided drainage has become the standard of care for the
treatment of symptomatic pancreatic fluid collections (PFCs), including pancreatic
pseudocysts (PPCs) and walled-off necrosis (WON). While PPCs are managed mainly by
the placement of double-pigtail plastic stents (DPPSs), the introduction of lumen-apposing
fully covered metal stents (LAMSs) has substantially increased the ability to treat
necrotic collections [1 ]. Moreover LAMSs facilitate, when needed, direct endoscopic necrosectomy (DEN) [2 ]
[3 ].
In 2017 Bang et al., reporting an interim analysis of an ongoing randomized controlled
trial of WON treatment comparing DPPSs with the only LAMS available at that time (electrocautery-enhanced
Hot-Axios; Boston Scientific Corp., Marlborough, Massachusetts, USA), reported a 25 %
rate of delayed bleeding requiring coil embolization, which occurred at 3 and 5 weeks
after stent implantation [4 ]. No further bleeding episodes were observed after a modification of the protocol
involving LAMS removal within 4 weeks [5 ].
Two subsequent retrospective studies confirmed the increased risk of bleeding, observed
in 13.4 % and 19 % after Hot-Axios placement [6 ]
[7 ]. The concern around bleeding-related adverse events (AEs) after Hot-Axios placement
has been reinforced by data from post-marketing surveillance from the Food and Drug
Administration (FDA) Manufacturer and User Facility Device Experience (MAUDE), which
highlighted the high frequency of AEs relating to use of the Hot-Axios for the treatment
of PFCs, with a bleeding rate of 32.4 % [8 ]. Bleeding occurs as the PFC is resolving and the pseudoaneurysms in the contralateral
collection wall, previously compressed by the pressure of the cavity, come into contact
with the rigid spikes of the terminal end of the intracavitary flange, causing wall
erosion and consequent bleeding [9 ].
A second LAMS (Spaxus; Taewoong Medical Co., Gimpo, Republic of Korea) became available
in 2016 and was recently incorporated into a device with electrocautery capabilities
at its tip (Hot-Spaxus; Taewoong Medical Co.). The Spaxus stent has rounded edges
and foldable flanges, allowing accommodative apposition between the stent and the
cavity wall, which should theoretically reduce intracavity bleeding [10 ]
[11 ]. Available data on PFC drainage have reported bleeding rates of 5.1 %, with one
case requiring angiographic embolization; however, the number of treated patients
in the publication by Teoh et al. [12 ] and in other small reports [13 ]
[14 ] is insufficient to draw any definitive conclusions.
To fill this gap, we performed a retrospective propensity study to compare the occurrence
of bleeding between the Hot-Axios stent and the Spaxus stent, in both its cold and
hot versions. In addition, the technical and clinical success, and overall AE rates
of the two stents were also analyzed.
Methods
Study population
This was a retrospective study conducted on patients who presented with symptomatic
PFCs and were treated using cold or hot LAMSs (Hot-Axios or Spaxus/Hot-Spaxus) from
10 July 2019 to 28 February 2022 at one of 18 high volume endoscopy referral centers.
All of the endoscopists had performed more than 30 LAMS placements before the start
of the study period.
Hospitalized patients were clinically evaluated the day after the procedure and daily
until discharge. After hospital discharge, patients were followed up with outpatient
clinic visits or by phone calls. The protocol was approved by the institutional review
board of Humanitas Mater Domini (no.37/22 HMD). The inclusion criteria were: (i) adult
patients undergoing EUS-guided drainage for a PFC with a LAMS; (ii) PFCs requiring
drainage because of symptoms. Exclusion criteria were: (i) drainage with DPPSs; (ii)
use of the multiple gateway technique; (iii) bleeding disorders or concomitant anticoagulant
therapy (not discontinued); (iv) less than 1 month of follow-up. [Fig. 1 ] represents the flowchart of patients included in the study.
Fig. 1 Flowchart of patients included in the study.
Study device
Electrocautery-enhanced Hot-Axios stent
The Hot-Axios stent and delivery system (Boston Scientific) has the Conformité Européenne
(CE) mark for drainage of PFCs with necrotic content of < 30 %. The self-expanding
Hot-Axios stent is made of braided medical-grade nickel titanium, fully covered with
silicone ([Fig. 2b ]). The largest stents, used to drain PFCs, (saddle part measuring 10–20 mm) are released
through a 10.5-Fr delivery system. The design incorporates flanges on both the distal
and proximal ends to anchor the stent to the luminal walls, and a tip with electrocautery
capabilities. The stent delivery system is Luer-locked onto the endoscope instrumentation
channel inlet port. The dimensions and lengths of the body and flanges are listed
in Table 1 s , see online-only Supplementary material.
Fig. 2 A comparison of the features of: a the Hot-Spaxus stent; and b the Hot-Axios stent, which has metal spikes at the distal extremities of the stent.
Spaxus and Hot-Spaxus stents
This LAMS is a through-the-scope LAMS delivery device that is CE-approved for PPC
drainage. The “cold” version lacks an electrocautery tip that was recently incorporated
into the “hot” version ([Fig. 2a ]). The stent is comprised of braided nitinol, fully covered with silicone, with large
flexible flanges on each end, with a saddle part length of 20 mm for PFC drainage.
The flanges offer accommodative apposition regardless of the wall thickness, and the
stent has a channel in which a 0.035-inch guidewire can be preloaded. The two available
stents for PFC drainage, with body diameters and lengths of 10 × 20 mm and 16 × 20 mm
(flange diameters of 25 and 31 mm, respectively), are delivered using a 10-Fr delivery
catheter.
Procedures
Informed consent was obtained from all of the patients before they underwent treatment
of their PFC. A therapeutic linear echoendoscope was used in all cases, and procedures
were performed by an experienced ultrasonographer in a room with fluoroscopic capabilities,
using carbon dioxide insufflation.
All procedures were performed according to local policies, with deep sedation using
propofol with anesthesiology assistance or with the patient under general anesthesia.
The technique used for LAMS placement – either the freehand technique or 19G fine
needle aspiration (FNA) needle puncture of the target PFC, followed by guidewire placement,
cystotome use, and LAMS placement – was left to the discretion of each endosonographer.
Similarly, the choice of stent diameter was left to endosonographer discretion. Patients
were treated according to their respective local protocols in the event of a procedural
or technical failure.
Distal flange release was performed under EUS control in all cases, whereas the proximal
flange was delivered using the “intrachannel release” technique or under endoscopic/fluoroscopic
view.
Study parameters/end points
The severity of bleeding was classified according to the American Society for Gastrointestinal
Endoscopy (ASGE) lexicon [15 ]. Technical success was defined as adequate access and successful placement of the
stent through the walls of the gastrointestinal tract into the PFC, with drainage
of the intracollection fluids/necrotic material into the stomach/duodenum. Clinical
success was defined as radiological resolution of the fluid collection at 3-month
follow-up, without clinical symptoms attributable to the PFC.
Statistical analysis
Categorical variables were reported as the number of cases and percentage, and differences
between groups were compared using Fisher’s exact test. Continuous variables were
expressed as the median and interquartile range (IQR), and differences between groups
were explored by the Mann–Whitney and Wilcoxon rank tests before and after matching,
respectively. All analyses were two-tailed, and the significance level was set at
≤ 0.05.
To overcome biases owing to the different distribution of covariates among patients
assigned to Hot-Axios or Spaxus placement, a 1-to-1 match was created using propensity
score analysis. The propensity score represents the probability of each patient being
assigned to a particular condition or treatment in a study given a set of known covariates
[16 ].
Multivariate logistic regression was performed to predict the probability of each
patient being submitted to the two groups based on several demographic and collection-related
covariates, including age, sex, collection size, type of collection, use of DEN, and
approach for LAMS placement (transgastric or transduodenal). The predictive values
were then used to obtain a 1-to-1 match using nearest neighbor matching within a predetermined
caliper distance. Nearest neighbor matching within a specified caliper distance selects
as a match an untreated subject whose propensity score is closest to that of the treated
subject (“nearest neighbor matching” approach), with the further restriction that
the absolute difference in the propensity scores of matched subjects must be below
some prespecified threshold (the caliper distance) [17 ]. Therefore, patients whose propensity score could not be matched because of a greater
caliper distance were excluded from further analysis. As suggested by Austin, a caliper
of width equal to 0.2 of the SD of the logit of the propensity score was used, as
this value has been found to minimize the mean squared error of the estimated treatment
effect [18 ]. Subgroup analysis based on LAMS size was performed.
A univariate/multivariate logistic regression analysis was also performed to assess
the correlation between baseline parameters and the bleeding rate. Results were reported
as the odds ratio (OR) and 95 %CI. Significant factors in univariate analysis were
then entered into the multivariate model.
In order to account for eventual center-effects in the analysis, the effect of the
kind of stent used on the occurrence of the primary outcome (bleeding requiring transfusion
and/or intervention) was analyzed through a random-effects analysis fitting a logistic
regression model performed according to the formula log it (πij) = α + βtreatXij + uj
where πij is the probability of an event for the ith patient in the jth center, βtreat indicates the log odds ratio for treatment, Xij indicates whether the patient received the treatment or control, and uj is the effect of the jth center [19 ].
The statistical analysis was conducted using the MatchIt package in R Statistical
Software 3.0.2 (Foundation for Statistical Computing, Vienna, Austria).
Results
Patients
Out of 592 patients initially assessed for eligibility, 363 were retrospectively retrieved
from the databases of the 18 participating centers after excluding patients who did
not fulfil the inclusion criteria ([Fig. 1 ]). In the Spaxus group, 111 (64 %) and 63 (36 %) patients underwent cold and Hot-Spaxus
placement, respectively, whereas the Hot-Axios was placed in 185 patients (98 %) ([Table 1 ]). The number of procedures performed in the participating centers are detailed in
Table 2 s .
Table 1
Baseline characteristics of the 363 patients who were treated with a Hot-Axios or
Spaxus stent (both cold and hot) and whose data were retrospectively retrieved at
one of 18 centers.
Variable
Overall
Spaxus
Axios
P value
Age, median (IQR), years
55 (39–65)
52.5 (37–61)
56 (41–68)
0.01
Sex, male, n (%)
251 (69.1 %)
123 (70.7 %)
128 (67.7 %)
0.61
Collection, n (%)
0.12
123 (33.9 %)
68 (39.1 %)
55 (29.1 %)
236 (65.0 %)
104 (59.8 %)
132 (69.8 %)
4 (1.1 %)
2 (1.1 %)
2 (1.1 %)
Collection maximum diameter, median (IQR), cm
10.0 (7.0–13.0)
9.7 (7.0–12.6)
10.0 (7.0–13.8)
0.50
> 30 % necrosis, n (%)
128 (35.2 %)
43 (24.7 %)
85 (44.9 %)
< 0.001
Rate of infection, n (%)
193 (53.1 %)
95 (54.6 %)
98 (51.8 %)
0.67
Approach, n (%)
0.90
348 (95.7 %)
168 (96.6 %)
180 (95.2 %)
12 (3.3 %)
5 (2.9 %)
7 (3.7 %)
3 (0.8 %)
1 (0.6 %)
2 (1.1 %)
Access type
< 0.001
22 (6.1 %)
22 (12.6 %)
0
93 (25.6 %)
89 (51.1 %)
4 (2.1 %)
19 (5.2 %)
14 (8.0 %)
5 (2.6 %)
229 (63.2 %)
49 (27.9 %)
180 (95.2 %)
Plastic stent inside the LAMS, n (%)
168 (46.3 %)
90 (51.7 %)
78 (41.3 %)
0.02
Necrosectomy, n (%)
140 (38.5 %)
47 (27.0 %)
93 (49.2 %)
< 0.001
IQR, interquartile range; WON, walled-off necrosis; LAMS, lumen-apposing metal stent.
The use of coaxial DPPSs inside the LAMSs was significantly more frequent for Spaxus
stents compared with Hot-Axios stents (52 % vs. 41 %; P = 0.02), whereas necrosectomy was performed significantly more frequently in the
Hot-Axios group than in the Spaxus group (49 % vs. 27 %; P < 0.001). All the other variables/parameters were similar between the two groups.
After 1-to-1 propensity score matching, 264 patients were selected for comparison:
132 were treated with the Spaxus stent (71 [54 %] with the cold Spaxus; 61 [46 %]
with the Hot-Spaxus) and 132 with the Hot-Axios stent. Details of the propensity score
matching are shown in Fig. 1 s . The characteristics of the 264 propensity score-matched patients are reported in
[Table 2 ].
Table 2
Baseline characteristics of the 264 patients who were selected after 1-to-1 propensity
score matching.
Variable
Overall
Spaxus
Axios
P value
Age, median (IQR), years
55 (40–63)
54 (39–61)
55 (40–65)
0.42
Sex, male, n (%)
184 (69.6 %)
92 (69.6 %)
92 (69.6 %)
> 0.99
Collection
0.79
96 (36.4 %)
47 (35.7 %)
49 (37.2 %)
168 (63.6 %)
85 (64.3 %)
83 (62.8 %)
Collection maximum diameter, median (IQR), cm
10.0 (7.0–13.0)
9.7 (7.0–12.5)
10.0 (7.0–13.5)
0.88
> 30 % necrosis, n (%)
90 (34.1 %)
42 (31.8 %)
48 (36.3 %)
0.51
Rate of infection, n (%)
144 (54.5 %)
73 (55.3 %)
71 (53.7 %)
0.90
Approach, n (%)
0.73
255 (93.3 %)
128 (97.0 %)
127 (96.3 %)
9 (6.7 %)
4 (3.0 %)
5 (3.7 %)
Access type, n (%)
< 0.001
21 (7.9 %)
21 (9.8 %)
0
52 (19.6 %)
50 (44 %)
2 (1.5 %)
17 (6.4 %)
14 (10.6 %)
3 (2.8 %)
174 (66.1 %)
47 (35.6 %)
127 (95.7 %)
Plastic stent inside the LAMS, n (%)
141 (53.4 %)
75 (56.8 %)
66 (50.0 %)
0.26
Necrosectomy, n (%)
99 (37.5 %)
47 (35.6 %)
52 (39.4 %)
0.52
IQR, interquartile range; WON, walled-off necrosis; LAMS, lumen-apposing metal stent.
The median age of the selected patients was 55 years (IQR 40–63.4), with equal sex
distribution and no difference in any variable between the two groups (P = 0.42). WON was the most common type of treated collection (64.3 % and 62.8 % in
the Spaxus and Hot-Axios groups, respectively; P = 0.79), with no difference in lesion size (9.7 cm [range 7.0–12.5] in the Spaxus
group vs. 10.0 cm [range 7.0–13.5] in the Hot-Axios group; P = 0.88). Drainage was achieved through the stomach in almost all cases (97.0 % vs.
96.3 % in the Spaxus and Hot-Axios groups, respectively; P = 0.73), with coaxial DPPSs placed in 56.8 % of the Spaxus and 50.0 % of the Hot-Axios
groups, respectively (P = 0.26). Unlike in the overall cohort, the use of DEN was not different between the
two groups (35.6 % vs. 39.4 %; P = 0.52). More than 30 % of necrosis in the PFC was detected in 31.8 % of patients
in the Spaxus group and 36.3 % of patients in the Hot-Axios group (P = 0.51), whereas the rate of infected collections was 55.3 % and 53.7 %, respectively
(P = 0.90).
Outcomes
Study outcomes were reported in [Table 3 ]. The technical and clinical success rates were similar between the two groups. Bleeding,
the primary outcome, occurred in two patients (1.5 %) who had a Spaxus stent compared
with nine patients (6.8 %) who had a Hot-Axios stent placed (P = 0.03) (Fig. 2 s ).
Table 3
Comparison of outcomes for endoscopic ultrasound-guided pancreatic fluid collection
drainage between the 264 propensity score-matched patients selected for comparison.
Overall
Spaxus
Axios
P value
Technical success, n (%)
263 (99.6 %)
132 (100 %)
131 (99.2 %)
> 0.99
Clinical success, n (%)
245 (92.8 %)
122 (92.4 %)
123 (93.1 %)
> 0.99
Adverse event rate, n (%)
Overall
17 (6.4 %)
4 (3.0 %)
13 (9.8 %)
0.04
12 (4.5 %)
2 (1.5 %)
10 (7.5 %)
0.03
11 (4.1 %)
2 (1.5 %)
9 (6.8 %)
0.03
1 (0.3 %)
0
1 (0.7 %)
> 0.99
2 (0.7 %)
0
2 (1.5 %)
0.49
2 (0.7 %)
2 (1.5 %)
0
0.49
LAMS occlusion
39 (14.7 %)
19 (14.3 %)
20 (15.1 %)
0.86
LAMS, lumen-apposing metal stent.
In both of the patients who bled in the Spaxus group, a 16 × 20-mm stent was used,
with two severe bleeding episodes occurring 4 and 8 days post-procedure, at the level
of the site of stent placement in one and an intracavity bleed in the other. The first
patient was treated by stent removal, local epinephrine injection, and application
of argon plasma coagulation, followed by placement of a second Spaxus stent when the
bleeding stopped. In the second case, the intracavitary bleeding was stopped using
an epinephrine injection and hemostatic powder. Both patients were hospitalized for
12 days; in neither case was a blood transfusion required.
Nine bleeds requiring transfusion and/or intervention were recorded in the Hot-Axios
group. The stents placed among the patients who experienced bleeding requiring transfusion
and/or intervention in the Hot-Axios group were a 10 × 10-mm stent (n = 1), a 15 × 10-mm
(n = 5), and a 20 × 10-mm (n = 3) (Table 3 s ). Bleeding occurred early (within 24 hours) in two patients, while two others bled
at 48 hours; the remaining five patients bled late at 12, 16, 18, 21, and 23 days.
Five episodes of bleeding occurred at the site of LAMS placement and were treated
endoscopically, while in four patients (3.0 %), bleeding occurred inside the cavity
and required embolization by an interventional radiologist. In all four of these patients,
a two unit blood transfusion was required; two presented with hematemesis and two
with hypovolemic shock and melena, and their mean (SD) hemoglobin drop was of 4.8
(1.2) g/dL. For all four of them, bleeding resulted in admission to the intensive
care unit for 2 nights, and their length of hospitalization ranged from 10 to 22 days.
There was however no statistical difference in the number of patients who required
embolization between the two groups (P = 0.12). Interestingly in two of these patients, coaxial DPPSs had been placed to
prevent bleeding. Finally, one case of moderate bleeding during Hot-Axios stent placement
(0.7 %) was also observed.
There were three additional AEs, so the overall, AEs were also significantly higher
in the Hot-Axios group than the Spaxus group (9.8 % vs. 3.0 %; P = 0.04). One procedural perforation was successfully closed by an over-the-scope
clip (Ovesco, Tübingen, Germany) and two migrated stents were not replaced at 32 and
45 days because of a healing sign. There were two cases of stent misdeployment (1.5 %)
observed in the Spaxus group.
As reported in [Table 4 ], age (OR 1.16, 95 %CI 1.08–1.34; P = 0.02) and type of stent used (center-effect adjusted OR 0.25, 95 %CI 0.08–0.96;
P = 0.04) were significant predictors of bleeding on multivariate analysis, and also
on univariate analysis (P = 0.04) and (P = 0.03), respectively.
Table 4
Logistic regression analysis for the rates of bleeding requiring transfusion and/or
intervention.
Univariate analysis
Multivariate analysis
Odds ratio (95 %CI)
P value
Odds ratio (95 %CI)
P value
Age
1.13 (1.04–1.21)
0.04
1.16 (1.08–1.34)
0.02
Sex (reference male)
0.88 (0.73–1.83)
0.78
Type of collection (reference pseudocyst)
1.81 (0.75–4.60)
0.18
Diameter (reference < 10 cm)
1.33 (0.89–1.58)
0.15
Approach (reference transgastric)
0.86 (0.61–1.48)
0.80
Stent used (reference Axios)[1 ]
0.20 (0.04–0.95)
0.03
0.25 (0.08–0.96)
0.04
Necrosectomy (reference no necrosectomy)
0.64 (0.35–1.48)
0.62
Use of DPPS inside the LAMS (reference no DPPS)
1.11 (0.36–4.03)
0.60
LAMS, lumen-apposing metal stent; DPPS, double-pigtail plastic stent.
1 Results adjusted for eventual center-effects using random-effects logistic regression.
Discussion
Bleeding is a reported significant AE of Hot-Axios stents when used in patients with
peripancreatic fluid collections and PFCs [6 ]. Driven by the low rate of bleeding episodes reported so far in patients with PFCs
treated with the Spaxus stents [10 ], we performed a retrospective propensity score-matched study with the primary aim
of comparing the occurrence of bleeding requiring transfusion and/or intervention
between the Hot-Axios and the Spaxus stents in this patient population. Globally,
we found the Spaxus stent to be associated with significantly lower rates of bleeding
requiring transfusion and/or intervention episodes, and of overall AEs compared with
the Hot-Axios stent, with similar technical and clinical success rates.
The “cold” Axios and the Hot-Axios were introduced into clinical practice in 2011
and 2013, respectively, with PFC drainage being the first and more common indication.
Initial studies reported negligible rates of bleeding requiring transfusion and/or
intervention [20 ], defined in our studies as endoscopic, radiological, or surgical intervention. Conversely,
in more recent years, rates of bleeding requiring transfusion and/or intervention
of 13.4 %–25 % have been observed [21 ]
[22 ], reaching 32.4 % in post-marketing surveillance from the FDA-MAUDE [8 ]. It is plausible to hypothesize that this risk is related to the conformation of
the intracavitary flange of the Hot-Axios stent, which presents at its terminal portion
rigid spikes that can scrape and/or perforate the surface of the contralateral wall
of the cavity when it shrinks, creating erosion of vessels and subsequent bleeding
[23 ].
The advent of another LAMS in 2016, the Spaxus stent, which has a different terminal
end design, with rounded edges and flanges that fold back and conform to the surface
of the intracavity wall of the PFC, could theoretically reduce the risk of intracavity
bleeding [10 ]
[11 ]. To test this hypothesis, we collected data from 18 tertiary referral high volume
endoscopy centers on a large number of patients with PFCs who underwent drainage with
the Hot-Axios and Spaxus stents. Data were collected on an overall total of 363 patients,
99 of whom were excluded after 1-to-1 propensity score matching, with 132 patients
allocated to each of the two study groups.
Our findings proved that bleeding requiring transfusion and/or intervention occurred
significantly more frequently in the Hot-Axios group than in the Spaxus group (6.8 %
vs. 1.5 %; P = 0.03). The relationship between the occurrence of bleeding requiring transfusion
and/or intervention and the type of stent used was further confirmed by univariate
regression analyses, which revealed that use of the Hot-Axios stent was a significant
predictor of bleeding requiring transfusion and/or intervention (P = 0.03). However, although bleeding requiring arterial embolization was only observed
in patients treated with the Hot-Axios stent, this AE did not reach a statistically
significant difference compared with the Spaxus stents (3.0 % vs. 0 %; P = 0.12).
Caution should therefore be applied in interpreting our results, which do not allow
us to conclude that the less aggressive design of the terminal end of the Spaxus stent,
compared with the Hot-Axios stent, is associated with a decrease in the number of
intracavitary bleeds necessitating interventional radiology. Although there was not
a significant difference, we did however observe that this serious AE never occurred
after Spaxus placement and the four cases registered in patients who were treated
with Hot-Axios stent pose a note of caution that requires further studies with a larger
sample size to draw definitive assumptions on this regard.
Moreover, all stents in the Hot-Axios group were placed using an electrocautery delivery
system, whereas this system was used in only 46 % of the patients in the Spaxus group. In
this regard, we observed that 5/9 cases of bleeding in the Hot-Axios group occurred
at the entry site and not in the cavity, and we cannot be sure that the current used
to place the stent did not have a role in the onset of bleeding. However, the cold
Spaxus stents were placed using a cystotome with the same cutting current applied
as is used for Hot-Axios deployment, thereby suggesting a minor role for the entry
current. Only a prospective multicenter randomized study specifically designed to
answer this question will be able to draw a definitive conclusion.
In our study, the rates of overall bleeding requiring transfusion and/or intervention
were substantially lower than those reported in previous studies. Technical and clinical
success were comparable between the two groups. Notably, in addition to bleeding requiring
transfusion and/or intervention, in the Hot-Axios group, there were three additional
AEs (one perforation and two stent migrations), meaning overall AEs occurred significantly
more frequently in the Hot-Axios group (9.8 % vs. 3.0 %; P = 0.04).
We acknowledge several limitations in our study. First, because of the retrospective
design, we cannot exclude selection bias, which we attempted to minimize by using
the propensity score analysis and balancing the two groups for a number of factors
associated with AEs. Our findings do however need to be further validated in a properly
designed multicenter randomized controlled trial before any definitive conclusion
can be drawn.
Second, our data are derived from experienced centers and operators, and might not
be replicable in other settings. Third, some procedural aspects were not standardized,
and heterogeneity of technical aspects could be found. This however is consistent
with the real world, in which decision-making processes, particularly technical ones,
are subject to operator preference and experience, and vary widely between institutions.
Furthermore, these technical aspects have been shown to impact on technical success
rather than on bleeding rate, which was the primary outcome in our study.
As a further limitation, the etiology of the bleeding episodes could not always be
properly identified. In fact, it is very hard to differentiate between bleeding due
to the natural course of disease (pseudoaneurysm) and direct stent-induced bleeding
for at least two reasons: first, not all pseudoaneurysms are clearly detected on a
computed tomography scan before a subsequent interventional radiology procedure; second,
the two etiologies could coexist and both contribute to the onset of bleeding. It
could be argued that the bleeding episodes observed in our study were detected beyond
the usual timeframe described in the guidelines. However, this axiom has been questioned
in recent large nationwide studies [22 ]
[23 ], hence our results are in line with the recent literature in this field.
In conclusion, our study showed that, in patients with PFCs, bleeding requiring transfusion
and/or intervention occurred significantly more frequently when the Hot-Axios stent
was used than when the Spaxus stent was used, although this was not the case for bleeding
requiring embolization. This higher risk of bleeding has been postulated to be related
to the design of the end of the stent flange. A randomized controlled trial is warranted
to obtain a proper comparison and draw a definitive conclusion.
Correction: Lumen-apposing metal stents for the treatment of pancreatic and peripancreatic
fluid collections and bleeding risk: a propensity matched study
