Introduction
The incidence of acute pancreatitis continues to rise [1 ]. Although biliary lithiasis and/or sludge, and alcohol abuse remain the most common
causes of acute pancreatitis, in approximately 25 % of cases a causative factor cannot
be determined during standard diagnostic work-up [1 ]. Standard diagnostic work-up consists of: a personal and family history; laboratory
tests including serum alanine transaminase (ALT), calcium, and triglycerides; and
transabdominal ultrasonography during admission and after discharge. Patients in whom
such work-up is negative are referred to as having idiopathic acute pancreatitis (IAP)
[2 ].
In comparison to pancreatitis with a known origin, IAP has a relatively high pancreatitis
recurrence rate of 25 % within 3 years. Furthermore, the risk of recurrence in patients
who have already had one recurrent IAP episode is twice as high as in patients with
a first episode of IAP [3 ].
For many years, undetected microlithiasis and biliary sludge have been considered
to be the major causes of IAP, and even routine cholecystectomy has been suggested
in these patients [4 ]. In recent years, however, a more diverse view has arisen as studies have shown
that other etiologies such as pancreatic cancer may not be as rare in this situation
as previously thought [5 ].
Partly owing to the heterogeneity of occult etiologies in IAP, no consensus existed
among physicians regarding the use of various additional diagnostic modalities. The
International Association of Pancreatology/American Pancreatic Association (IPA/APA)
evidence-based guidelines on management of acute pancreatitis recommend endoscopic
ultrasonography (EUS) as the first step after negative standard diagnostic work-up
(GRADE 2C, weak agreement) [2 ], with discussion as to whether this should be performed after the first episode
or only after recurrent episodes of IAP. Considering the hypothesis that IAP recurrences
may be caused by occult and subsequently untreated underlying causes, it is suggested
that by improving the detection rate of etiology in IAP by implementing EUS in the
diagnostic work-up, recurrence rate in this patient group could be reduced [3 ].
The goal of this study was to determine the diagnostic yield of EUS for etiological
factors in IAP. Secondary objectives were to determine whether the diagnostic yield
is altered by the presence of the gallbladder during EUS and in recurrent pancreatitis,
as opposed to a first episode of pancreatitis; and whether the detection of etiology
by EUS and subsequent treatment of underlying etiologies may be associated with a
reduced pancreatitis recurrence rate and occurrence of biliary events.
Methods
Study design and registration
This study was designed as a systematic review with meta-analysis of the diagnostic
yield of EUS in patients with IAP. The review was written in accordance with the Preferred
Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [6 ] and Meta-analyses Of Observational Studies in Epidemiology (MOOSE) guidelines (see
the PRISMA and MOOSE checklists in Appendix 1 s , available in online-only Supplementary Material) [7 ]. This study was registered in the PROSPERO database under registration number CRD42019120730.
Search strategy
With the aid of an expert librarian, the databases PubMed, EMBASE, and the Cochrane
library were systematically searched for relevant articles between inception and 19
November 2019. The search contained the following key words: pancreatitis, pancreas,
acute, recurrent, relapsing, idiopathic, unexplained, unknown, endoscopic, ultrasonography,
ultrasound, and EUS. The complete searches are listed in Appendix 2 s . After performing the search, duplicates were removed and the search results were
uploaded to Covidence systematic review software (Veritas Health Innovation, Melbourne,
Australia).
Study selection
Two independent reviewers (D.U. and C.R.) screened potentially relevant articles by
title and abstract. Disagreements were resolved in a discussion between the two reviewers.
The full text of the potentially eligible studies was read by both reviewers individually.
Mutual agreement was required for inclusion of potential studies, while disagreement
on the eligibility of an article was resolved after joint re-evaluation of the article
by the two reviewers. Reasons for exclusion were recorded during screening by title
and abstract, and full-text screening.
Peer-reviewed articles in English that included patients with IAP in whom, after presentation
with IAP, a diagnostic EUS was performed, and articles reporting the diagnostic yield
of EUS by etiology were considered. Studies with patients younger than 18 years or
with known chronic pancreatitis, of animals, or where the EUS was performed for indications
other than detection of etiology were excluded. Letters, comments, case reports, reviews,
book sections, conference abstracts, and case – control studies were excluded as well.
Finally, all patients in whom an etiology was found prior to EUS were excluded from
the meta-analysis.
Outcome measures
The primary outcome was the diagnostic yield of EUS for detection of etiology in IAP.
Subgroup analyses were made for diagnostic yield of EUS for biliary etiology, chronic
pancreatitis, and neoplasms. Comparative analyses were made for diagnostic yield in
patients with a first episode of IAP versus patients with recurrent pancreatitis and
in patients with a gallbladder in situ versus post-cholecystectomy patients.
Secondary outcome measures were the treatment of underlying etiologies, pancreatitis
recurrence rate, and the occurrence of biliary events (i. e. cholangitis, cholecystitis,
acute biliary pancreatitis, and biliary colic) during follow-up.
Data extraction
After the studies that met the inclusion criteria had been selected, all relevant
data from these studies were extracted by two reviewers using a standardized form.
Relevant data included: study characteristics (period and country of inclusion, year
and journal of publication, study design, number of patients, and follow-up time);
patient characteristics (sex, age, recurrence and severity of pancreatitis, previous
cholecystectomy); the use of diagnostic tests prior to EUS; type of scope and EUS
technique used; definitions of positive imaging; diagnostic yield (including yield
for biliary etiology, chronic pancreatitis, neoplasms, and other anomalies); treatment
of etiology after EUS; and pancreatitis recurrence and occurrence of biliary events
during follow-up. The definition of positive imaging used in each of the included
studies is provided in Appendix 3s .
No attempt was made to communicate with the corresponding authors concerning missing
data. Missing data were reported as “not reported” and excluded from pooled analyses.
Quality assessment
The quality of the included articles was appraised at study level by two independent
reviewers (D.U. and C.R.) using the QUADAS-2 score for quality assessment of diagnostic
accuracy studies, adjusted for the study designs of the included studies by omitting
the third domain regarding the reference standard [8 ]. Risk of bias in patient selection was scored based on whether the study included
a consecutive or random sample, the type of center in which patients were included,
and whether additional diagnostics were performed before EUS. Risk of bias in the
use of EUS was scored based on whether the type of scope, expertise level of the endoscopist,
use of sedation, and definitions for biliary etiology and chronic pancreatitis were
reported. Risk of bias in the timing of EUS was scored based on whether the EUS was
performed before or after clinical recovery from the acute pancreatitis episode. Applicability
in patient selection was scored based on whether a minimal standard diagnostic work-up
was performed before EUS, and applicability in use of EUS was scored based on whether
the EUS technique described was similar to conventional EUS techniques [9 ].
Disagreement on the appraisal was resolved by joint re-evaluation by the two reviewers.
Data analyses
The study and patient characteristics, as well as the diagnostic work-up prior to
EUS were reported using descriptive statistics.
The pooled estimates of the primary outcome measure, diagnostic yield, were reported
as proportions with a 95 % confidence interval (CI) with 95 % prediction intervals
(PI), estimated using a random-effects model with DerSimonian – Laird estimator [10 ], implemented in R [11 ]. A random-effects model, in which some studies are appointed more weight in pooled
proportions than others, as opposed to a fixed-effects model, was chosen to correct
for the between-study heterogeneity on baseline level.
Sensitivity analyses for diagnostic yield were performed based on study type, year
of publication, type of pancreatitis (first versus recurrent disease), presence of
the gallbladder, timing of EUS, definition of chronic pancreatitis, and diagnostic
work-up prior to EUS. For comparative analyses of diagnostic yield in patients with
a first episode of IAP versus recurrent pancreatitis and post-cholecystectomy patients
versus patients with a gallbladder in situ and pancreatitis recurrence rate, the risk
ratio with 95 %CI was also reported. This was done in a random-effects model with
the inverse variance method and DerSimonian – Laird estimator [10 ], using Review Manager software [12 ]. Other secondary outcome measures were reported descriptively. No correction for
multiple testing was performed.
Between-study heterogeneity was assessed using the I
2 statistic. I
2 values of < 25 %, 25 % – 50 %, 50 % – 75 %, and > 75 % were classified as low, moderate,
high, and very high heterogeneity [13 ].
Publication bias was evaluated using the Egger’s linear regression method [14 ].
Results
Search results and included studies
A systematic literature search yielded 567 unique records. After screening and full-text
eligibility assessment, 22 studies with 1490 patients with IAP who underwent EUS were
included in the qualitative and quantitative synthesis [3 ]
[15 ]
[16 ]
[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
[22 ]
[23 ]
[24 ]
[25 ]
[26 ]
[27 ]
[28 ]
[29 ]
[30 ]
[31 ]
[32 ]
[33 ]
[34 ]
[35 ]. The exclusion criteria for the excluded studies are listed in [Fig. 1 ].
Fig. 1 PRISMA flowchart of study screening and selection, as per Moher et al. [6 ]. More information available from: www.prisma-statement.org.
Study characteristics
The 22 included studies, published between 1999 [23 ] and 2019 [22 ], comprised 16 prospective cohort studies [3 ]
[15 ]
[16 ]
[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
[23 ]
[24 ]
[27 ]
[28 ]
[30 ]
[32 ]
[33 ]
[34 ], two post-hoc analyses of prospective cohorts [22 ]
[31 ], three retrospective cohort studies [25 ]
[29 ]
[35 ], and one study with unclear design [26 ]. Mean or median follow-up time ranged between 0 and 73.7 months. A complete overview
of the study characteristics is provided in [Table 1 ].
Table 1
Characteristics of the 22 included studies of endoscopic ultrasonography (EUS) for
the investigation of idiopathic acute pancreatitis (IAP).
First author
Year of publication
Country
Study design
Period of inclusion
Follow-up time, months
Ammori [15 ]
[1 ]
2003
UK
Prospective cohort study
2000 – 2001
NR
Choudhary [16 ]
2016
India
Prospective cohort study
Period of 2 years
NR
Frossard [17 ]
[2 ]
2000
France
Prospective cohort study
1991 – 1995
NR
Garg [18 ]
2007
India
Prospective cohort study
1995 – 2003
17.63 (mean)
Govil [19 ]
2014
India
Prospective cohort study
2010 – 2012
NR
Kim [20 ]
[3 ]
2011
South Korea
Prospective cohort study
NR
36.4 (median)
Liu [21 ]
[4 ]
2000
China
Prospective cohort study
1996 – 1997
20, 22 (median)
Lopes [22 ]
2019
Brazil
Post-hoc analysis of a prospective EUS database
2012 – 2017
31.7 (mean)
Maes [23 ]
1999
France
Prospective cohort study
1994 – 1995
3.65 (mean or median not reported)
Mariani [24 ]
2009
Italy
Prospective cohort study
NR
NR
Morris-Stiff [25 ]
2009
UK
Retrospective cohort
2000 – 2004
73.7 (median)
Norton [26 ]
2000
UK
NR
NR
3 – 28 (range)
Poves [27 ]
2010
Spain
Prospective cohort study
Period of 18 months
21.5 (mean)
Queneau [28 ]
2002
France
Prospective cohort study
1995 – 1997
36 (median)
Rana [29 ]
[5 ]
2012
India
Retrospective cohort
NR
5 – 36 (range)
Repiso Ortega [30 ]
2011
Spain
Prospective cohort study
2005 – 2009
16 (mean)
Tandon [31 ]
2001
USA
Post-hoc analysis of a prospective EUS database
NR
16 (mean)
Thevenot [32 ]
2013
France
Prospective cohort study
2008 – 2010
22 (mean)
Vila [33 ]
2010
Spain
Prospective cohort study
2004 – 2007
28.95 (mean)
Wilcox [3 ]
2016
USA
Prospective cohort study
2003 – 2013
37 (mean)
Yusoff [34 ]
2004
Canada
Prospective cohort study
2000 – 2003
NR
Zhan [35 ]
2011
China
Retrospective cohort
2006 – 2009
NR
NR, not reported.
1 Ammori et al. included eight patients with IAP who underwent EUS but four of these
patients has significant liver enzyme abnormalities and were for that reason excluded
from this review.
2 Frossard et al. included 168 patients with IAP who underwent EUS but one of these
patients had furosemide-induced pancreatitis and another had Coxsackie virus-related
pancreatitis. These patients were also excluded from the review.
3 Kim et al. included 31 patients with IAP who underwent EUS but seven of them already
had signs of biliary etiology on previous imaging (i. e. a dilated common bile duct).
These patients were excluded. Kim et al. also included two patients with cholecystolithiasis
but owing to missing data these patients could not be excluded.
4 Liu et al. reported a median follow-up time of 20 and 22 months for EUS-positive
patients and EUS-negative patients, respectively.
5 Rana et al. did not report the years of inclusion but did mention they included patients
during “the last 3 years.”
Patient characteristics
In total, 1679 IAP patients were included, of whom 1490 underwent diagnostic EUS. The
average ages ranged between 22.25 and 53.75 years, with 53 % of the patients (869/1647)
being men (reported in 21 studies [3 ]
[15 ]
[16 ]
[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
[22 ]
[23 ]
[24 ]
[25 ]
[28 ]
[29 ]
[30 ]
[31 ]
[32 ]
[33 ]
[34 ]
[35 ]), and 49 % (649/1318) had had more than one episode of acute pancreatitis before
undergoing EUS (reported in 16 studies [3 ]
[15 ]
[16 ]
[18 ]
[19 ]
[20 ]
[21 ]
[23 ]
[24 ]
[29 ]
[30 ]
[31 ]
[32 ]
[33 ]
[34 ]
[35 ]). Cholecystectomy was performed before EUS in 28 % (343 /1217; reported in 13 studies
[3 ]
[16 ]
[19 ]
[20 ]
[22 ]
[24 ]
[26 ]
[28 ]
[30 ]
[31 ]
[32 ]
[33 ]
[34 ]) and 13 % (117 /919) had severe pancreatitis before EUS (reported in 14 studies
[3 ]
[15 ]
[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
[23 ]
[24 ]
[25 ]
[30 ]
[32 ]
[33 ]
[35 ]). All patient characteristics are listed in [Table 2 ].
Table 2
Patient characteristics in the 22 included studies of endoscopic ultrasonography (EUS)
for the investigation of idiopathic acute pancreatitis (IAP).
Study
All included IAP patients, n
IAP patients with EUS, n
Male sex, n (%)[1 ]
Age, years
Recurrent pancreatitis, n (%)[1 ]
Previous cholecystectomy, n (%)[1 ]
Severe pancreatitis, n (%)[1 ]
[2 ]
Ammori [15 ]
68
4
23 (34)
55 (median)
NR
NR
12 (18)
Choudhary [16 ]
192
192
131 (68)
34.6 (mean)
102 (53)
57 (30)
NR
Frossard [17 ]
168
166
102 (61)
50 (mean)
NR
NR
15 (9)
Garg [18 ]
75
10
60 (80)
31.9 (mean)
75 (100)
NR
15 (20)
Govil [19 ]
51
51
35 (69)
36.7 (mean)
0 (0)
2 (4)
6 (12)
Kim [20 ]
31
24
11 (35)
51.3 (mean)
31 (100)
6 (19)
8 (26)
Liu [21 ]
18
18
9 (50)
68 (median)
13 (72)
NR
10 (56)
Lopes [22 ]
35
35
10 (29)
51.9 (mean)
NR
10 (29)
NR
Maes [23 ]
18
6
11 (61)
55.5 (mean)
3 (17)
NR
3 (17)
Mariani [24 ]
44
44
20 (45)
48.9 (mean)
44 (100)
7 (16)
0 (0)
Morris-Stiff [25 ]
42
42
25 (60)
53 (mean)
NR
NR
0 (0)
Norton [26 ]
44
43
20 (45)
53.5 (median)
10 (23)
8 (18)
NR
Poves [27 ]
32
32
NR
NR
NR
NR
NR
Queneau [28 ]
48
17
21 (44)
51 (mean)
NR
0 (0)
NR
Rana [29 ]
40
40
26 (65)
17 – 72 (range)
17 (43)
NR
NR
Repiso Ortega [30 ]
49
49
24 (49)
58 (mean)
16 (33)
9 (18)
5 (10)
Tandon [31 ]
31
31
12 (39)
48.8 (mean)
17 (55)
3 (10)
NR
Thevenot [32 ]
45
38
25 (56)
53.8 (mean)
8 (18)
7 (16)
7 (16)
Vila [33 ]
44
44
31 (70)
61.45 (mean)
19 (43)
11 (25)
9 (20)
Wilcox [3 ]
201
201
95 (47)
53 (mean)
121 (60)
99 (49)
27 (13)
Yusoff [34 ]
370
370
165 (61)
53.4 (mean)
169 (63)
124 (46)
NR
Zhan [35 ]
33
33
13 (39)
46.5 (mean)
4 (12)
NR
0 (0)
Total
1679
1490
869 (53)
22.25 – 53.75 (range)
649 (49)
343 (28)
117 (13)
NR, not reported.
1 Only concerns the studies that reported this parameter.
2 Severity was defined by the Atlanta score in four studies (Ammori et al., Maes et
al., Repiso Ortega et al., Thevenot et al.); by the Ranson criteria in four studies
(Frossard et al., Kim et al., Liu et al., Zhan et al.); by clinical criteria in two
studies (organ failure at the time of admission [Govil et al.]) and hospitalization
of more than 10 days [Wilcox et al.]); and was not reported in four studies (Garg
et al., Mariani et al., Morris-Stiff et al., Vila et al.).
Critical appraisal
A comprehensive quality assessment is provided in Appendix 4s . There were 10 studies that were considered to have a low risk of bias regarding
patient selection [3 ]
[15 ]
[17 ]
[24 ]
[27 ]
[28 ]
[30 ]
[32 ]
[33 ]
[34 ], while one study was appraised as having low applicability concerns [23 ]. Regarding risk of bias in the use of EUS, four studies were assessed as having
high risk of bias [15 ]
[23 ]
[25 ]
[35 ] and 10 studies were considered to have low applicability concerns in the use of
EUS [17 ]
[19 ]
[21 ]
[24 ]
[25 ]
[26 ]
[28 ]
[29 ]
[30 ]
[35 ]. Finally, five studies were appraised as having high risk of bias in flow and timing
of the study [23 ]
[25 ]
[27 ]
[28 ]
[35 ].
An Egger’s linear regression indicated low risk of publication bias (P = 0.74) (Appendix 5 s ).
Diagnostic work-up prior to EUS
Standard diagnostic work-up, as defined by the IAP/APA evidence-based guidelines for
management of acute pancreatitis (i. e. personal and family history; laboratory tests
including serum ALT, calcium, and triglycerides; and transabdominal ultrasonography
during admission and after discharge) [2 ], and additional diagnostic tests are listed in Appendix 6 s . Definitions of positive tests are summarized in Appendix 7 s . The majority of studies excluded patients with known alcohol abuse, drugs associated
with acute pancreatitis, hypercalcemia, and hypertriglyceridemia, although definitions
for all of these exclusion criteria differed among studies.
Regarding the diagnosis of biliary etiology, most studies required at least one transabdominal
ultrasound, while only two studies required a repeat ultrasound after clinical recovery
from the episode of pancreatitis in all included patients [18 ]
[26 ], and six studies explicitly excluded patients with abnormal liver function tests
[3 ]
[15 ]
[16 ]
[18 ]
[19 ]
[20 ].
Importantly, none of the studies performed a complete standard diagnostic work-up,
according to the IAP/APA guidelines, prior to EUS.
Diagnostic yield of EUS
Out of the 1490 IAP patients who underwent EUS, EUS found a possible etiology in 874
patients (59 % in random-effects model, 95 %CI 52 % – 66 %, 95 %PI 32 % – 82 %), as
shown in [Fig. 2 ]. In random-effects models, biliary etiology was detected in 30 % of patients (429 /1490;
95 %CI 21 % – 41 %, 95 %PI 5 % – 77 %), chronic pancreatitis in 12 % (271 /1490; 95 %CI
8 % – 19 %, 95 %PI 2 % – 51 %), and neoplasms in 2 % (45 /1490; 95 %CI 1 % – 4 %,
95 %PI 0 % – 17 %), as shown in [Table 3 ]. The neoplasms included 22 intraductal papillary mucinous neoplasms (IPMNs), 12
pancreatic carcinomas, three ampullary adenomas, two ampullary cancers, one malignant
IPMN, one gastric adenocarcinoma invading the pancreatic parenchyma, two (suspected)
neuroendocrine tumors, one cystic lesion not further specified, and one pancreatic
mass suspicious for malignancy where pathology after resection showed inflammation.
Fig. 2 Pooled overall diagnostic yield of endoscopic ultrasonography (EUS) for the etiology
of idiopathic acute pancreatitis (IAP). CI, confidence interval.
Table 3
Diagnostic yield of endoscopic ultrasonography (EUS) in patients with idiopathic acute
pancreatitis (IAP).
Diagnostic yield, n[1 ]
Proportion, %
95 %CI
95 %PI
Risk ratio (95 %CI)
Heterogeneity, I
2 , %
Overall positive for etiology
874/1490
59
52 – 66
32 – 82
NA
82
Overall negative for etiology
616/1490
41
34 – 48
18 – 68
NA
82
Biliary disease
429/1490
30
21 – 41
5 – 77
NA
92
Chronic pancreatitis
271/1490
12
8 – 19
2 – 51
NA
89
Neoplasms
45/1490
2
1 – 4
0 – 17
NA
68
Other
138/1490
4
2 – 8
0 – 32
NA
87
Prospective studies
739/1266
58
50 – 66
31 – 81
NA
83
Non-prospective studies
135/224
61
47 – 73
31 – 84
NA
73
Studies < 2010
475/751
62
53 – 71
36 – 83
NA
75
Studies > 2009
399/739
56
47 – 65
29 – 81
NA
82
Rosemont criteria for chronic pancreatitis
264/403
65
52 – 77
40 – 85
NA
78
Other criteria for chronic pancreatitis
610/1087
57
50 – 65
30 – 81
NA
80
EUS after standard diagnostics
302/475
67
52 – 80
31 – 91
NA
86
EUS after additional diagnostics
572/1015
56
50 – 62
37 – 73
NA
62
First episode
264/494
56
46 – 65
33 – 76
0.89 (0.71 – 1.11)
69
Recurrent disease
299/594
52
34 – 69
11 – 91
Previous cholecystectomy
105/211
50
43 – 56
43 – 56
0.81 (0.70 – 0.95)
1
Gallbladder in situ
310/496
64
54 – 73
42 – 82
Early EUS[2 ]
24/50
48
35 – 62
NA
NA
0
Late EUS[3 ]
80/1360
61
53 – 68
33 – 83
NA
84
CI, confidence interval; PI, prediction interval; NA, not applicable.
1 The overall diagnostic yield as well as the yield for several etiologies in a random-effects
model is shown. See Appendix 9 s for the forest plots made to facilitate the subgroup analyses.
2 Before clinical recovery from pancreatitis episode.
3 After clinical recovery from pancreatitis episode.
Other reported etiologies included: pancreas divisum (n = 87), ascariasis (n = 3),
autoimmune pancreatitis (n = 3), cystic fibrosis (n = 3), impaired outflow through
the papilla of Vater (n = 2), abnormal pancreaticobiliary junction (n = 1), choledochocele
(n = 1), and diverticulum (n = 1). Impaired outflow through the papilla of Vater was
diagnosed in one study by measuring the diameter of the pancreatic duct after secretin
injection, and was considered suggestive of sphincter of Oddi dysfunction [24 ].
A sensitivity analysis showed no statistically significant difference in the diagnostic
yields reported by prospective versus non-prospective studies, studies published before
2010 versus after 2009, studies using the Rosemont criteria for chronic pancreatitis
[36 ] versus other criteria, or studies performing only standard diagnostic work-up prior
to EUS versus additional diagnostic work-up ([Table 3 ]).
In random-effects models, there was no difference in diagnostic yield between patients
with a first episode of IAP and patients with recurrent pancreatitis before EUS (264/494
[56 %] versus 299/594 [52 %]; risk ratio 0.89, 95 %CI 0.71 – 1.11). Diagnostic yield
of EUS in random-effects models was, however, significantly higher in patients with
their gallbladder in situ compared with those patients who had undergone a previous
cholecystectomy (105/211 [50 %] versus 310 /496 [64 %]; risk ratio 0.81, 95 %CI 0.70 – 0.95).
When EUS was performed after clinical recovery from the pancreatitis episode, the
diagnostic yield in a random-effects model was 61 % (80/1360; 95 %CI 53 % – 68 %),
while the diagnostic yield in EUS before clinical recovery was 48 % (24/50; 95 %CI
35 % – 62 %).
Between-study heterogeneity was very high in the analyses of overall diagnostic yield,
yield for biliary etiology, yield for chronic pancreatitis, and yield for EUS after
clinical recovery, as exemplified by I
2 statistics of 82 %, 92 %, 89 %, and 84 %, respectively. The I
2 statistic showed high heterogeneity in the analyses of yield for neoplasms and in
the comparison of yield between patients with a first episode of IAP versus patients
with recurrent IAP (68 % and 69 %, respectively). The heterogeneity of the analyses
of yield in EUS after clinical recovery and the comparison of yield between post-cholecystectomy
patients versus patients with gallbladder in situ was low (0 and 1 %, respectively).
Interventions, pancreatitis recurrence, and biliary events after EUS
Interventions and adverse events during follow-up were scarcely reported and not systematically
assessed. Thirteen studies reported performing endoscopic sphincterotomy and/or cholecystectomy
after the EUS uncovered a biliary etiology [3 ]
[19 ]
[20 ]
[21 ]
[22 ]
[25 ]
[26 ]
[27 ]
[29 ]
[30 ]
[31 ]
[32 ]
[33 ]; in one study, endoscopic sphincterotomy was performed during endoscopic retrograde
cholangiopancreatography (ERCP) in one patient with choledocholithiasis and in three
patients without evidence of biliary etiology [23 ]. Other treatments included enzyme replacement therapy and endoscopic therapy for
chronic pancreatitis [19 ]
[20 ], surgery and/or chemoradiation for malignancies [20 ]
[23 ]
[30 ]
[33 ], and pancreatic stenting for pancreas divisum [3 ]
[26 ]
[29 ].
Six studies reported no recurrence during follow-up [19 ]
[21 ]
[27 ]
[28 ]
[29 ]
[32 ]. Seven studies reported seven recurrences in 76 patients with confirmed biliary
etiology versus 49 recurrences in 138 patients with unknown etiology (9.2 % versus
35.5 %, respectively; risk ratio 0.71, 95 %CI 0.21 – 2.41) (Appendix 8 s ) [3 ]
[18 ]
[20 ]
[22 ]
[26 ]
[30 ]
[31 ].
Regarding biliary events, one study reported jaundice in a patient in whom biliary
etiology was confirmed during EUS [25 ], and one study reported acalculous cholecystitis in a patient in whom no etiology
was found during EUS [26 ].
Discussion
This meta-analysis including 1490 patients who underwent EUS for IAP found an overall
diagnostic yield of 59 %. EUS mostly detected a biliary etiology (30 %), while chronic
pancreatitis was diagnosed in 12 % of patients. Strikingly, in 2 % of patients, neoplastic
lesions were detected.
While the number of pancreatitis episodes before EUS did not affect the diagnostic
yield, we did establish that the diagnostic yield is higher in patients with a gallbladder
in situ. Additionally, we found that an EUS after clinical recovery has a higher yield
than EUS during acute pancreatitis (61 % versus 48 %). This was not statistically
significant, potentially because only 50 patients had EUS during an acute pancreatitis
episode versus 1360 patients thereafter.
Owing to limited reporting on interventions and biliary events after EUS, no meta-analysis
could be performed for these secondary outcome measures. However, we did establish
that the pancreatitis recurrence rate after EUS tended to be lower when biliary etiology
was detected compared with when no etiology was uncovered. This finding supports the
hypothesis that uncovering the etiology by EUS may prevent recurrence.
None of the studies included in this systematic review performed diagnostic work-up
according to the IAP/APA Guidelines on management of acute pancreatitis [2 ]. This is exemplified by the fact that in the quality assessment, 21 out of 22 studies
were deemed to have selected patients that were not representative of the IAP patient
population. Most importantly, only two studies required a repeat transabdominal ultrasound
after clinical recovery, five studies performed magnetic resonance imaging (MRI) or
magnetic resonance cholangiopancreatography (MRCP) before EUS, and only six studies
excluded patients based on abnormal liver function tests. It has previously been established
that a repeat ultrasound has a diagnostic yield of 20 % for biliary stones and sludge
[37 ], and that in particular an elevated ALT level on admission indicates high probability
of biliary etiology [15 ]. This incomplete diagnostic work-up may have led to an overestimation of the overall
diagnostic yield and of the diagnostic yield for biliary etiology using EUS.
There appeared to be considerable between-study heterogeneity in this systematic review,
particularly with regard to the inclusion of patients with a first IAP or recurrent
episode of IAP, inclusion of post-cholecystectomy patients, timing of the EUS, and
the definition of a positive EUS. This is demonstrated by a high I
2 statistic in the meta-analysis. In this review, we have attempted to make the data
more homogeneous by performing a sensitivity analysis.
In this systematic review, most detected etiologies were treatable (e. g. biliary
etiology), and EUS appeared to lower recurrence rates. Additionally, some neoplastic
lesions were found. Early detection of neoplasm is obviously essential. In a considerable
proportion of the included patients, a pancreas divisum was present. As the prevalence
of divisum is similar in the general population and in patients with a first episode
of IAP, the role of pancreas divisum in acute pancreatitis is debated [38 ]. This may imply that the reported diagnostic yield of EUS in the studies included
in this systematic review is higher than the clinically relevant diagnostic yield
of EUS.
The diagnostic yield of EUS in a first episode of IAP was as high as the yield in
recurrent IAP. Although the diagnostic yield was significantly higher in patients
with a gallbladder in situ, the diagnostic yield in post-cholecystectomy patients
was still 50 %. These findings underline the importance of additional diagnostic work-up
for etiology in IAP, even in a first episode of IAP and in post-cholecystectomy patients.
EUS also appeared to have a higher yield when performed after clinical recovery. Therefore,
physicians should consider delaying EUS until patients have recovered.
The results of this systematic review are similar to those of a recently published
review, in which a diagnostic yield of 62 % (CI 56 % – 68 %) was reported [39 ]. However, this review does not report on the quality of the included studies, the
definitions of a positive EUS, gallbladder status, timing of cholecystectomy, or on
the statistical methods used to perform and interpret the meta-analyses. Therefore,
although they report considerable between-study heterogeneity with an I
2 of up to 87 %, no sensitivity analysis was performed. In our systematic review, we
have attempted to increase the quality of the meta-analyses by including only peer-reviewed
studies, critically appraising these studies, and extracting sufficient data to perform
a sensitivity analysis.
The main limitations of this systematic review are that all studies lacked a complete
standard diagnostic work-up of IAP before EUS, including consideration of abnormal
liver functions tests and repeat imaging after clinical recovery, and that most of
the included studies lacked homogeneous data on patients with either first episode
IAP or recurrent disease. Multiple previous studies have confirmed the association
between elevated liver function tests and biliary etiology, with a positive predictive
value of 85 % for an ALT above 150 U/L within 48 hours after onset of symptoms [15 ]. Future studies should focus on including homogeneous patient groups who truly have
IAP, according to current guidelines. Therefore, the Dutch Pancreatitis Study Group
has decided to conduct the multicenter, prospective cohort PICUS study, including
106 patients after a first episode of IAP with complete standard diagnostic work-up.
In conclusion, this systematic review shows that EUS can detect a potential etiology
in the majority of patients with IAP and that detection and subsequent treatment of
the etiology may be associated with a reduction of pancreatitis recurrence. There
is, however, a paucity of prospective homogeneous data on the diagnostic yield of
EUS in IAP after a complete standard diagnostic work-up according to international
guidelines. The etiology appears to be mostly biliary stones or sludge, and chronic
pancreatitis, but neoplastic causes are also found in a substantial proportion of
these patients.
The protocol for this systematic review was registered on PROSPERO (CRD42019120730)
and is available in full on the NIHR website (https://www.crd.york.ac.uk/prospero/).