Introduction
In patients with acute cholecystitis, early laparoscopic cholecystectomy is the gold
standard treatment [1 ]
[2 ]. For patients with serious comorbidities deemed unfit for surgery, a nonoperative
approach with percutaneous gallbladder drainage (PT-GBD) is a safe alternative treatment
option [3 ]
[4 ]. Nonetheless, several challenges remain with the long term use of PT-GBD, including
tube dysfunction, pain, bile leaks, readmission, and reintervention, with adverse
events ranging from 4 % to 51 %. These not only affect patients’ quality of life but
could also result in higher costs of care in the long term [5 ]
[6 ]
[7 ].
With the emergence of endoscopic ultrasound-guided gallbladder drainage (EUS-GBD)
since 2007 [8 ]
[9 ], the technique has rapidly evolved with different stents being used for drainage.
The 2018 Tokyo guidelines also accepts EUS-GBD performed by experienced endoscopists
in high volume centers as an alternative approach to draining the gallbladder in patients
who are high risk candidates for cholecystectomy [1 ]. A few comparative studies with relatively small sample sizes then compared the
outcomes of EUS-GBD with those of PT-GBD in patients with acute cholecystitis who
were unfit for surgery. We therefore conducted a systematic review and meta-analysis
to compare the effectiveness and safety of the two procedures.
Methods
Identification and search strategy
This systematic review was performed in accordance with the Cochrane Handbook for
Systematic Reviews of Interventions [10 ]. It is reported according to the Preferred Reporting Items for Systematic Reviews
and Meta-Analyses (PRISMA) [11 ] and Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines [12 ]. A comprehensive electronic literature search was conducted for all articles published
up to October 2017 using PubMed, MEDLINE, EMBASE, and Cochrane Central Register of
Controlled Trials. Search terms included MeSH and non-MeSH terms relating to “endoscopic
ultrasound-guided gallbladder drainage,” “EUS-guided gallbladder drainage”, “percutaneous
cholecystostomy”, “percutaneous gallbladder drainage” in combination. An additional
search was conducted using hand-searching through references and bibliographies of
relevant studies. Our search was limited to articles published in the English language.
Duplicated, abstract-only, and nonrelevant articles were excluded. Articles were selected
for full-text review based on their title and abstract.
Inclusion and exclusion criteria
We only selected comparative cohorts, both randomized control trials (RCTs) and observational
cohort studies, comparing the effectiveness and safety between EUS-GBD and PT-GBD
in nonsurgical patients with acute cholecystitis. In these studies, patients were
considered by a surgeon to be at too high a risk to undergo cholecystectomy. Only
endoscopic ultrasound-guided transmural drainage of the gallbladder was included,
thus excluding papers that used endoscopic transpapillary drainage. Studies were included
only if they reported all three primary outcomes of interest: 1) clinical success,
defined as resolution of clinical symptoms and/or improvement in biochemical parameters;
2) technical success, defined as the ability to access and drain the gallbladder by
placement of a drainage tube or stent with immediate drainage of bile; and 3) post-procedure
adverse events. Two authors independently identified original articles for review
of eligibility and validity. Any disagreement between reviewers was discussed with
a third reviewer for final consensus.
Study selection and data extraction
Extracted data included study design and period, patient demographics, clinical characteristics,
method of drainage including mode of anesthesia, route of drainage, and type of stent
used in EUS-GBD. Technical success, clinical success, adverse events, post-procedural
pain score, length of hospital stay, need for reintervention, number of unplanned
readmissions, recurrent cholecystitis, 30-day disease-specific mortality, and duration
of follow-up were also evaluated. In the event of missing data, the corresponding
authors were contacted for further details.
Quality assessment
We used the Cochrane tool to assess risk of bias for RCTs [10 ] and the Newcastle-Ottawa scale (NOS) for observational cohort studies [13 ]. The Cochrane tool assesses presence of several biases including selection, performance,
detection, attrition, and reporting. NOS measures quality in three parameters including
selection, comparability, and exposure/outcome with a maximum of 4, 2, and 3 points,
respectively. High quality studies score over 7, moderate quality studies score between
5 and 7, and low quality studies score under 5.
Data synthesis
Primary outcome measures were technical success, clinical success, and post-procedure
adverse events. Technical success was defined as the ability to access and drain the
gallbladder by placement of a drainage tube or stent with immediate drainage of bile.
Clinical success was defined as the resolution of clinical symptoms and/or improvement
in biochemical parameters. The severity of adverse events was graded according to
the lexicon of endoscopic adverse events [14 ].
Secondary outcome measures included length of hospital stay, the need for reintervention,
number of unplanned readmissions, recurrent cholecystitis, and 30-day disease-specific
mortality.
Statistical analysis
Odds ratios (ORs) were calculated for categorical variables and means with standard
deviation were calculated for continuous variables. Weights for individual studies
were based on the model of meta-analysis used to estimate the effect size, which is
represented by the size of the squares on the forest plot. Overall results were analyzed
using a random effects model of meta-analysis. For assessment of heterogeneity, the
I
2 statistic was used, with I
2 values > 50 % representing moderate heterogeneity. Sensitivity analysis was performed
in case of significant heterogeneity. Prediction interval was calculated to estimate
the range of dispersion of the calculated pooled rates [15 ]. Publication bias was illustrated by funnel plots. Subgroup analysis was performed
for studies using lumen-apposing metal stents (LAMS) in EUS-GBD group vs. PT-GBD.
All analyses were carried out using Review Manager software (RevMan Version 5.3 for
Mac; The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark).
Results
Search strategy yield and quality assessment
The electronic literature search identified 189 studies, and 27 additional studies
were identified through manual search of references and bibliographies from relevant
studies. A total of 32 studies were duplicated and 82 studies were removed as abstract-only
articles or were irrelevant by title. Of the 102 abstracts, 61 of them were excluded
because of the type or design of the study. Finally, 41 full-text articles were reviewed
for eligibility, of which 30 were review articles, 4 were systematic reviews of endoscopic
methods of gallbladder drainage, one was a meta-analysis of the effectiveness and
safety of endoscopic gallbladder drainage including both transpapillary and transmural
drainage, and one was a comparative study not using EUS-GBD; all of these studies
were excluded. Ultimately, five studies – one RCT [16 ] and four comparative cohort studies [17 ]
[18 ]
[19 ]
[20 ] – were included in the final meta-analysis. Results of the search strategy are shown
in [Fig. 1 ].
Fig. 1 Flow diagram of the literature search according to the Preferred Reporting Items
for Systematic Reviews and Meta-Analyses.
Three studies [18 ]
[19 ]
[20 ] were multicenter studies and two [16 ]
[17 ] were from a single tertiary care center. A total of 495 patients were included in
the meta-analysis, 206 of whom underwent EUS-GBD and 289 underwent PT-GBD. Four studies
[16 ]
[17 ]
[19 ]
[20 ] used 2013 Tokyo guidelines for the diagnosis of acute cholecystitis [21 ], with typical clinical presentation (fever, right upper quadrant pain, positive
Murphy sign, elevated leukocyte count) and radiological findings consistent with acute
cholecystitis. A total of 376 patients (76.0 %; 146 EUS-GBD and 230 PT-GBD) had calculous
cholecystitis, and 82 patients (16.6 %) had acalculous cholecystitis, of whom 39 underwent
EUS-GBD and 43 underwent percutaneous cholecystotomy; the remaining 7.5 % were listed
as other causes, including malignancy.
All patients were deemed unfit for emergency surgery owing to significant comorbidities,
either American Society of Anesthesiologists grade III or above, or Charlson Comorbidity
Index score ≥ 4. EUS-GBD was performed under general anesthesia in more than half
of the patients (51.5 %) whereas most PT-GBD procedures were performed under local
anesthesia with or without sedation.
Four studies used EUS-GBD as the initial drainage method, whereas in one study, EUS-GBD
was used only after a failed attempt at transpapillary drainage [16 ]. Three studies [18 ]
[19 ]
[20 ] mentioned the route of drainage in the EUS-GBD group: 76 patients were drained via
the transgastric route and 70 patients via the transduodenal route. Three studies
[18 ]
[19 ]
[20 ] used LAMS in the EUS-GBD group and two studies used nasobiliary drains [16 ] and fully covered self-expandable metal stents (FCSEMS) [17 ], respectively.
Two studies [19 ]
[20 ] used the American Society for Gastrointestinal Endoscopy lexicon to grade adverse
events, and three studies [16 ]
[17 ]
[18 ] defined procedure-related adverse event as early (< 24 hours) and delayed (> 1 day
after the procedure). For the follow-up protocol, the cholecystostomy tube was removed
at the time of cholecystectomy in the Jang et al. study [16 ], whereas three studies [17 ]
[19 ]
[20 ] had the tube removed at 6 – 8 weeks, once cystic duct patency was demonstrated on
cholecystogram. The time of removal was not clearly mentioned in the Tyberg et al.
study [18 ]. One author (Teoh AY) was noted to have participated in three studies involving
LAMS for EUS-GBD. It was clarified with the author that there was no overlapping of
cases.
According to the Cochrane tool, the RCT by Jang et al. [16 ] had a low risk of selection, reporting, and attrition bias, but had a performance
bias due to the lack of blinding of both patients and personnel performing the procedure.
All four observational cohort studies [17 ]
[18 ]
[19 ]
[20 ] were of high quality based on NOS. The included studies were also reviewed for confounding
bias. The authors found no significant difference in the baseline characteristics
between the “intervention” and “control” arms in the individual studies and we estimate
the risk of confounding bias to be low. Study characteristics, outcome measures, and
quality assessment of the included studies are summarized in [Table 1 ] and [Table 2 ].
Table 1
Study characteristic and patient demographics.
Study
Design
Study period
Center
Etiology, n
Groups
Patients, n
Male, n
Age, mean (SD), years
Comorbidity index
Type of anesthesia
Stent
Jang et al. 2012[16 ]
RCT
Jun 2010 – Dec 2010
Single center
Calculous 50
EUS-GBD
30
17
62 (15)
ASAIII/IV: 25/5
Midazolam 30
NBD
Acalculous 9
PT-GBD
29
21
68 (10)
ASAIII/IV: 28/1
LA 29
Plastic
Kedia et al. 2015[17 ]
Retro.
July 2011 – Nov 2013
Single center
Calculous 45
EUS-GBD
30
14
62.5 (18.3)
CCI 6.1
GA 30
DP/FCSEMS + DP
Acalculous 28
PT-GBD
43
23
66.9(18.0)
CCI 5.6
Sedation 36
Plastic
Tyberg et al. 2016[18 ]
Retro.
Jan 2010 – Dec 2015
Multicenter 7
Calculous 102
EUS-GBD
42
24
74.0 (14.9)
CCI 4.76
GA 36 Sedation 6
LAMS/FCSEMS/plastic
Acalculous 16
PT-GBD
113
63
74.37 (14.33)
CCI 4.83
GA 22 Sedation 26
Plastic
Malignancy 34
LA 76
Other 3
Irani et al. 2017[19 ]
Retro.
July 2013 – Dec 2015
Multicenter 7
Calculous 61
EUS-GBD
45
29
65 (25 – 87)[* ]
ASAIV/V: 26/3
GA 40
LAMS
Acalculous 29
PT-GDB
45
27
75 (34 – 94)[* ]
ASAIV/V: 31/5
GA 5
Plastic
Teoh et al. 2017[20 ]
Retro.
Nov 2011 – Aug 2014
Multicenter 2
Calculous 118
EUS-GBD
59
30
82.7 (7.9)
CCI 5.6
Sedation/MAC 59
LAMS
PT-GBD
59
30
81.2 (10.4)
CCI 5.8
LA 59
Plastic
RCT, randomized controlled trial; EUS-GBD, endoscopic ultrasound-guided gallbladder
drainage; PT-GBD, percutaneous gallbladder drainage; ASA, American Society of Anesthesiologists;
LA, local anesthesia; NBD, nasobiliary drain; Retro., retrospective cohort; CCI, Charlson
Comorbidity Index; GA, general anesthesia; DP, double pigtail; FCSEMS, fully covered
self-expandable metal stent; LAMS, lumen-apposing metal stent; MAC, monitored anesthesia
care.
* Median (range).
Table 2
Outcome measures and quality assessment of included studies.
Study
Groups
Technical success, n
Clinical success, n
AEs, n
Pain, n
Length of stay, mean (SD), days
Re-intervention, n
Re-admission, n
Recurrent cholecystitis, n
Mortality, n
Follow-up, mean (SD), days
Quality
Jang et al. 2012[16 ]
EUS-GBD
29
29
2
Median 1
NA
NA
NA
NA
NA
90 or surgery
High risk of performance bias
PT-GBD
28
27
1
Median 5
NA
NA
NA
NA
NA
Kedia et al. 2015[17 ]
EUS-GBD
30
26
4
2.1
7.6 (2.5)
4
4
NA
NA
267 (219)
NOS
7
PT-GBD
42
42
17
3.8
16.3 (5.5)
23
17
NA
NA
282 (252)
4/1/2
Tyberg et al. 2016[18 ]
EUS-GBD
40
40
9
NA
8.3 (2.4)
4
6
3
0
Median 133
NOS
7
PT-GBD
112
97
24
NA
17.5 (4.3)
28
27
9
4
Median 231
4/1/2
Irani et al. 2017[19 ]
EUS-GBD
44
43
8
2.5
4.5 (0.8)
11
6
3
1
Median 215
NOS
8
PT-GDB
45
41
14
6.5
16.1 (2.7)
112
22
4
3
Median 265
4/2/2
Teoh et al. 2017[20 ]
EUS-GBD
57
53
19
NA
9.5 (6.1)
1
4
0
5
450.7 (343.1)
NOS
8
PT-GBD
59
56
44
NA
11.2 (6.4)
16
42
4
1
834.1 (416.6)
4/2/2
AE, adverse event; SD, standard deviation; EUS-GBD, endoscopic ultrasound-guided gallbladder
drainage; PT-GBD, percutaneous gallbladder drainage; NA, not assessed; NOS, Newcastle – Ottawa
scale.
Meta-analysis
All included studies compared the effectiveness and safety of EUS-GBD and PT-GBD.
There were no statistically significant differences in technical and clinical success
of EUS-GBD and PT-GBD. The pooled ORs for technical and clinical success were 0.43
(95 %CI 0.12 to 1.58; P = 0.21; I
2 = 0 %) ([Fig. 2 ]) and 1.07 (95 %CI 0.36 to 3.16; P = 0.90; I
2 = 44 %) ([Fig. 3 ]), respectively. The calculated prediction intervals for technical and clinical success
were 0.43 (95 %CI 0.05 to 3.56) and 1.07 (95 %CI 0.05 to 24.00), respectively. The
funnel plots are illustrated in Fig. 1s and Fig. 2s in the online-only Supplementary material. On sensitivity analysis by removing each
study sequentially, there was no change in the effect size for clinical success.
Fig. 2 Comparison of technical success between endoscopic ultrasound-guided gallbladder
drainage (EUS-GBD) and percutaneous gallbladder drainage (PT-GBD). M – H, Mantel – Haenszel;
CI, confidence interval; LAMS, lumen-apposing metal stent.
Fig. 3 Comparison of clinical success between endoscopic ultrasound-guided gallbladder drainage
(EUS-GBD) and percutaneous gallbladder drainage (PT-GBD). M – H, Mantel – Haenszel;
CI, confidence interval; LAMS, lumen-apposing metal stent.
The pooled OR for overall adverse events was 0.43 (95 %CI 0.18 to 1.00; P = 0.05; I
2 = 66 %), in favor of EUS-GBD ([Fig. 4 ]). The calculated prediction interval for adverse events was 0.43 (95 %CI 0.03 to
6.83). The funnel plot is illustrated in Fig. 3s . Moderate heterogeneity was found among included studies; thus, sensitivity analysis
was performed to evaluate the presence of skewness related to the different study
protocols. The study conducted by Jang et al. [16 ] evaluated the early postoperative outcomes and included only adverse events occurring
within 1 week of the procedure. The effect size was more significant when this study
was excluded. The remaining four studies included both early complications and disease-related
adverse events during the follow-up period. The most common adverse events associated
with PT-GBD were tube-related adverse events, ranging from 58.8 % to 75 % in the included
cohorts. The difference appeared to be independent of the difference in duration of
follow-up between the groups. There was no statistically significant difference in
the overall mean duration of follow-up between the EUS-GBD group and the PT-GBD group
(231.1 vs. 340.4 days; P = 0.43). Among the studies, significant differences in duration of follow-up were
observed in two studies only: Teoh et al. adjusted the confounding effects of duration
of follow-up by performing a Cox’s proportional hazards regression for predictors
of overall adverse events, whereas no statistically significant difference in adverse
events were observed in the study reported by Tyberg et al.
Fig. 4 Comparison of overall adverse events between endoscopic ultrasound-guided gallbladder
drainage (EUS-GBD) and percutaneous gallbladder drainage (PT-GBD). M – H, Mantel – Haenszel;
CI, confidence interval; LAMS, lumen-apposing metal stent.
Among the four observational studies [17 ]
[18 ]
[19 ]
[20 ] with a total of 436 patients (176 EUS-GBD and 260 PT-GBD), there were also no statistically
significant differences in technical and clinical success of EUS-GBD and PT-GBD; the
pooled ORs for technical and clinical success were 0.34 (95 %CI 0.08 to 1.47; P = 0.15; I
2 = 0 %) and 0.95 (95 %CI 0.26 to 3.39; P = 0.93; I
2 = 56 %), respectively. The pooled OR for overall adverse events was 0.37 (95 %CI
0.16 to 0.89; P = 0.03; I
2 = 70 %); thus, patients who underwent EUS-GBD experienced significantly fewer adverse
events.
The four observational studies also compared length of hospital stay, reintervention
rates and unplanned readmissions as secondary outcomes. Patients who underwent EUS-GBD
had significantly shorter hospital stays, with a pooled standard mean difference of
– 2.53 (95 %CI – 4.28 to – 0.78; P = 0.005; I
2 = 98 %) ([Fig. 5a ]). The calculated prediction interval for length of hospital stay was – 2.53 (95 %CI – 11.01
to 5.95). Fewer reinterventions were required after EUS-GBD, with a pooled OR of 0.16
(95 %CI 0.06 to 0.042; P < 0.001; I
2 = 32 %) ([Fig. 5b ]). Similarly, patients who underwent EUS-GBD had significantly fewer unplanned readmissions,
with a pooled OR of 0.16 (95 %CI 0.05 to 0.53; P = 0.003; I
2 = 79 %) ([Fig. 5c ]). The calculated prediction intervals for reintervention and unplanned readmission
were 0.16 (95 %CI 0.01 to 3.18) and 0.16 (95 %CI 0.00 to 33.54), respectively. The
number of follow-up procedures performed was also significantly lower in the EUS-GBD
group vs. the PT-GBD group (20/220 procedures vs. 179/465; P < 0.001). Problems related to the cholecystostomy tube were responsible for most
of the readmissions in the PT-GBD group, as mentioned in the two studies [19 ]
[20 ]. Sensitivity analysis performed for length of hospital stay and unplanned readmissions
did not show any change in effect size.
Fig. 5 Comparison between endoscopic ultrasound-guided gallbladder drainage (EUS-GBD) and
percutaneous gallbladder drainage (PT-GBD). a Hospital stay. b Reinterventions. c Unplanned readmissions. SD, standard deviation; IV, inverse variance; CI, confidence
interval; M – H, Mantel – Haenszel.
Three studies [16 ]
[17 ]
[19 ] compared the post-procedure highest pain score within 1 day of intervention between
the two procedures. The use of a visual analog scale was mentioned in two studies.
As different parameters were used (one assessed the median post-procedure pain score
and two used mean pain score), no meta-analysis was performed for this outcome measure.
Overall, all three studies consistently reported that patients experienced significantly
less pain after EUS-GBD than after PT-GBD.
Three studies [17 ]
[19 ]
[20 ] recruited genuine nonsurgical candidates for cholecystectomy as an inclusion criterion.
As a result, 81 patients (16.4 %) underwent cholecystectomy (25 EUS-GBD and 56 PT-GBD).
According to the study by Jang et al. [16 ], there was no statistically significant difference in conversion rate to open surgery
between EUS-GBD (9 %) and PT-GBD (12 %; P = 0.99). The ease of surgery after either endoscopic or percutaneous procedures was
not mentioned.
Subgroup analysis for EUS-GBD using LAMS
Three studies [18 ]
[19 ]
[20 ] using LAMS, with a total 363 patients (146 EUS-GBD LAMS and 217 PT-GBD) were included
in the subgroup analysis. Two studies [19 ]
[20 ] used LAMS exclusively in EUS-GBD drainage, and one study [18 ] used both plastic and metal stents (including FCSEMS and LAMS). One study placed
double-pigtail plastic stents in addition to FCSEMS [17 ]. There were no statistically significant differences in technical and clinical success
between EUS-GBD LAMS and PT-GBD. The pooled ORs for the technical success and clinical
success were 0.21 (95 %CI 0.04 to 1.10; P = 0.06; I
2 = 0 %) ([Fig. 2 ]) and 1.43 (95 %CI 0.42 to 4.81; P = 0.57; I
2 = 45 %) ([Fig. 3 ]), respectively. In addition, no statistically significant difference in adverse
events was observed, with a pooled OR of 0.42 (95 %CI 0.14 to 1.28; P = 0.13; I
2 = 79 %) ([Fig. 4 ]).
The outcomes according to the route of EUS drainage (either transgastric or transduodenal)
were compared in two studies [18 ]
[20 ] and there were no statistically significant differences in terms of technical success
and adverse event rate.
All three studies in the subgroup analysis compared the length of hospital stay, reintervention
rates, and unplanned readmissions between the two procedures. Hospital stays were
shorter for EUS-GBD LAMS than for PT-GBD, with a pooled standard mean difference of – 2.76
(95 %CI – 5.20 to – 0.31; P = 0.03; I
2 = 99 %) ([Fig. 6a ]). Patients who underwent EUS-GBD had fewer unplanned readmissions and reinterventions,
with pooled ORs of 0.14 (95 %CI 0.03 to 0.70; P = 0.02; I
2 = 86 %) ([Fig. 6b ]) and 0.15 (95 %CI 0.02 to 0.98; P = 0.05; I
2 = 63 %) ([Fig. 6c ]), respectively.
Fig. 6 Comparison between endoscopic ultrasound-guided gallbladder drainage (EUS-GBD) using
lumen-apposing metal stents and percutaneous gallbladder drainage (PT-GBD). a Hospital stay. b Reinterventions. c Unplanned readmissions. d Recurrent cholecystitis. e Mortality. SD, standard deviation; IV, inverse variance; CI, confidence interval;
M – H, Mantel – Haenszel.
Furthermore, recurrent cholecystitis and disease-related mortality were also compared
in all three studies as secondary outcome measures. The observed percentage of recurrent
cholecystitis was 4.1 % after EUS-GBD LAMS drainage and 7.83 % after PT-GBD. No statistical
difference in recurrent cholecystitis was observed between the two groups, with a
pooled OR of 0.65 (95 %CI 0.25 to 1.72; P = 0.39; I
2 = 0 %) ([Fig. 6 d ]). The disease-related mortality was also not different between the two groups, at
4.1 % (6 patients) for EUS-GBD LAMS and 3.7 % (8 patients) for PT-GBD, with a pooled
OR of 0.89 (95 %CI 0.12 to 6.46; P = 0.91; I
2 = 49 %) ([Fig. 6e ]).
Discussion
The gold standard treatment for acute cholecystitis is laparoscopic cholecystectomy
[1 ]. In patients who are high risk candidates for cholecystectomy, gallbladder drainage
with PT-GBD in the acute setting is highly efficacious at relieving the obstruction
[3 ]
[4 ]
[5 ]. Alternatively, drainage of the gallbladder can be achieved endoscopically using
either transpapillary stenting via endoscopic retrograde cholangiopancreatography
(ERCP) or transmural drainage by EUS-GBD. However, the transpapillary approach may
be difficult due to a small or tortuous cystic duct, or obstruction by gallstones,
tumor or stents. The route also carries a risk of ERCP-related pancreatitis [22 ]. Thus, EUS-GBD has gained popularity over the endoscopic transpapillary approach
[23 ]
[24 ].
This meta-analysis affirms that EUS-GBD had comparable effectiveness to PT-GBD for
high risk surgical patients with acute cholecystitis, in terms of technical and clinical
success rates. Our analysis suggests that patients with PT-GBD experienced more adverse
events than those treated by EUS-GBD. The majority of adverse events after PT-GBD
were catheter-related, including dislodgment, migration, obstruction, and peri-tubal
leakage. With EUS-GBD, the adverse events were procedure- or stent-related events
including bleeding, perforation, and bile leaks. However, we encountered a moderate
heterogeneity in this estimation. A number of patients in the PT-GBD group were from
historical cohorts with longer follow-up (i. e. lead time bias). However, significant
differences in duration of follow-up were observed in only two studies. Teoh et al.
[20 ] adjusted the confounding effects of duration of follow-up by performing a Cox’s
proportional hazards regression for predictors of overall adverse events, whereas
no statistically significant difference in adverse events were observed in the study
reported by Tyberg et al. [18 ]. Thus, the likelihood of lead time bias producing the observed differences in adverse
events is low.
Apart from the tube cholecystogram to check the patency of the cystic duct and common
bile duct, there is no consensus on the management of PT-GBD after placement. Some
advocate tube removal at 6 – 8 weeks after placement, once cystic duct patency is
demonstrated on the tube cholecystogram [6 ]
[7 ]
[25 ]. This could minimize tube-related complications and patient discomfort but does
nothing to reduce the risk of recurrent cholecystitis. Thus, different practices in
the management of PT-GBD could partly explain the discrepancy in the observed unplanned
remission rates among these studies.
A few published systematic reviews and meta-analyses have addressed the comparability
of EUS-GBD vs. PT-GBD. The first systematic review was reported as a subgroup analysis
with three published studies [26 ]. At least half of the drainage procedures performed in the studies were via the
transpapillary route (i. e. non-EUS drainage) and only six patients underwent transmural
drainage using metal stents. In another meta-analysis, Anderloni et al. [23 ] reported that different types of stents, including plastic, SEMS, and LAMS, used
for EUS-GBD in acute cholecystitis have comparable effectiveness to PT-GBD. Our current
meta-analysis went the extra step with regard to reporting the secondary outcomes
of interest (length of hospital stay, readmission rate, and reintervention rate) that
were missing in the original papers, by contacting the corresponding authors. Jang
et al. [16 ] did not assess those outcomes owing to the study protocol, with drainage procedure
as an interim measure before elective cholecystectomy. Our meta-analysis of the other
four studies favored EUS-GBD, with significantly shorter hospital stays, and fewer
unplanned readmissions and reinterventions. Subgroup analysis with LAMS also showed
similar results. However, this conclusion has to be interpreted with caution because
of the substantial heterogeneity in the studies. The variation in hospital stay for
the two procedures may be related to the difference in post-drainage protocols, discharging
criteria, social support, and cultural belief among various countries. Nevertheless,
patients undergoing EUS-GBD were discharged earlier on average (i. e. mean 7.5 days
vs. 15.3 days in the PT-GBD group).
Undoubtedly, there are several limitations to the current meta-analysis. The number
of included studies was small for meta-analysis. Furthermore, the presence of publication
bias is difficult to address with fewer than 10 studies. Besides, most included studies
were retrospective in nature, which poses a certain risk of selection bias and lead
time bias. The heterogeneity in post-drainage protocol among different centers also
weakens the reliability of assessing the secondary outcome measures. In addition,
estimation of confounding bias was limited by the baseline characteristic or confounding
factors that were considered in individual studies. Further evaluation from an ongoing
international RCT (NCT02212717) with a standardized protocol comparing EUS-GBD with
PT-GBD is expected to shed more light on these limitations.
In conclusion, this meta-analysis demonstrated that EUS-GBD is associated with lower
rates of post-procedure adverse events, shorter hospital stays, and fewer reinterventions
and readmissions compared with PT-GBD in patients with acute cholecystitis who are
unfit for surgery.
