Professional Documents
Culture Documents
To cite this article: Nicholas Yuen, Pauline OShaughnessy & Andrew Thomson (2017): New
classification system for indications for endoscopic retrograde cholangiopancreatography
predicts diagnoses and adverse events, Scandinavian Journal of Gastroenterology, DOI:
10.1080/00365521.2017.1384053
Download by: [Australian Catholic University] Date: 29 September 2017, At: 04:11
SCANDINAVIAN JOURNAL OF GASTROENTEROLOGY, 2017
https://doi.org/10.1080/00365521.2017.1384053
ORIGINAL ARTICLE
CONTACT Nicholas Yuen u5513193@anu.edu.au Australian National University Medical School, Florey Building, 54 Mills Road, Acton ACT 2601, Australia
2017 Informa UK Limited, trading as Taylor & Francis Group
2 N. YUEN ET AL.
Publicaons idened
through database
search (n=1396)
Excluded aer
screening
tle/abstract (n=1360)
analysis (n=29)
Publicaons from
database search
included in study
(n= 7)
Publicaons sourced
from bibliographies
and others (n=5)
Total publicaons
included in study
(n=12)
literature published in English, and that did not comprise more than 2% of indications in any
published in the last 10 years (2004 onwards). single study have been excluded for simplicity.
hospital stay following ERCP. The ERCP indication categories such as parenchymal liver disease or Gilberts syndrome. A
used are listed and defined in Table 1. small group of patients did not fall into any of the primary
or secondary indications and were grouped together under
Primary indications the indication of other.
There were six primary indication categories such as cholan- The findings at ERCP were classified into seven categories.
gitis, biliary leak, acute (biliary) pancreatitis, positive intrao- These are summarized in Table 2. Findings classified as N/A
perative cholangiogram (IOC), change/removal of stent for indicate that no cholangiogram was conducted, commonly
malignant disease, and change/removal of stent for benign due to lack of biliary access or when a cholangiogram was not
disease. The indications were prioritized in that if there were necessary. Post-ERCP, patients were followed proactively with
more than one, then the one higher on the list (as specified telephone calls by the endoscopist both on the evening of the
in Table 1) was used. For instance, if a patient presented procedure and the next day. If the patient was an inpatient,
with both cholangitis and pancreatitis, then cholangitis was either the endoscopist reviewed the patients in person or tele-
used as the indication. Similarly, if a patient who had previ- phoned the nursing staff supervising the patients both on the
ously undergone biliary stenting for a malignant stricture, evening of the ERCP and the following day. In our hospital,
presented with cholangitis secondary to blockage of the most patients undergo ERCPs as outpatients. Inpatient ERCPs
stent, then this was classified as cholangitis. If and only if a were almost all performed on patients whose clinical status
patient did not fall into one of these six categories, an appro- mandated inpatient care. ERCP-related adverse events includ-
priate secondary indication was invoked. In addition, if an ing the occurrence of either post-ERCP pancreatitis or other
indication was itself a diagnosis, clearly defined parameters, instances of unplanned hospital admission/prolongation of
which were as objective as possible were used. hospital stay were documented contemporaneously.
Statistical analysis was conducted on the ERCP data using
Secondary indications the IBM SPSS Statistics 22.0 program (Armonk, NY). To inves-
Secondary indications, which were only used if a primary tigate the association between ERCP indications, findings and
indication was not applicable, consisted of a combination of occurrence of adverse events, separate chi-squared tests
two or three of: epigastric or right upper quadrant pain (P), were performed for every combination of these variables.
imaging suggestive of significant abnormalities of the bile The Fishers exact test was used in cases where total sample
duct (I), and significant abnormalities in LFTs (L). No precise size was small and the expected count in contingency table
numeric criteria were used in determining if abnormal LFTs cells was less than 5. Results were assessed as statistically sig-
were considered significant, but in general there had to be nificant if p values were smaller than .05.
either a sharp, recent increase over a day or two in more
than one biochemical marker (alanine transaminase, bilirubin, Results
and gamma-glutamyl transferase) or a persistent level of ele-
Literature review
vation in at least two of these markers, one of which was bili-
rubin. Care was taken to ensure as far as possible that Twelve papers meeting the above criteria were identified,
alternative causes of elevated LFTs/bilirubin were excluded covering a date range from 2004 to 2014, and a total of
4 N. YUEN ET AL.
53,394 ERCP procedures. A summary of the publications New a-priori system for ERCP indications and its
included in the study is shown in Table 3. The indications evaluation
found to be used in the reviewed literature are grouped and
ERCP indications were closely associated with the ERCP find-
shown in Table 4 in terms of how many papers utilized each
ings. Figures 2 and 3 show for each ERCP indication the
indication. For example, only two of 12 papers used Imaging
respective ERCP findings. Statistical analysis of the results is
abnormalities as an indication for ERCP.
shown in Table 5.
There were relatively few indications in categories that
were not based on pre-ERCP diagnoses. The most frequent
indication was stones, followed closely by malignant sten- Primary indications (listed in Table 1)
osis and biliary leak (reported 10 times each), and this pre- Eight hundred and seventy-six patients (49.8%) had primary
ERCP diagnosis of stones was qualified as suspected five indications. The most common primary indication was
times. There was only a single study in which stones/sus- change/removal of stent for benign disease (16.2%). As
pected stones was not considered to be an indication for expected, well over half the patients with the indication of
ERCP [4]. Out of the indications that were not based on a cholangitis had stones. Similarly, patients whose indication
pre-ERCP diagnosis, the most frequent was jaundice was an abnormal IOC during cholecystectomy had a finding
(reported six times). of common bile duct stone at subsequent ERCP in 52.5% of
Importantly, it is not made clear on the basis of which cases (74 of 141 cases). However, in keeping with previous
parameters the suspected or apparent diagnoses were made. studies a substantial proportion of patients with abnormal
Table 4 also shows the frequency with which defining param- IOC were found to have normal cholangiograms at ERCP (51
eters were mentioned, specifically for indications that presup- of 141 cases) [8]. Of the patients who underwent ERCP for
pose a diagnosis. When suspected or apparent diagnoses clinical evidence of a bile leak, cholangiographic evidence of
were provided as indications, defining parameters were pro- a bile leak was found in 25 out of 44 cases. Of the 193
vided in only 8% of the cases. As examples, Christensen et al. patients undergoing ERCP for acute pancreatitis, 91 were nor-
[5] described the basis of suspected stones as Suspected bile mal and 44 had a stone at ERCP, reflecting the fact that small
duct stones at ultrasound, magnetic resonance or chol- stones, which may not be evident at ERCP, often cause pan-
angiography. An American prospective study using the creatitis. Acute biliary pancreatitis was associated with no
American Society for Gastrointestinal Endoscopys (ASGE) cholangiogram being obtained (31 of 193 cases, v2(1) 10.1,
indications [6] described the basis of a pancreatic malignancy p .001). The failure to gain access to the biliary tree in
as signs/symptoms suggesting pancreatic malignancy when these cases may be attributable to swelling and distortion of
direct imaging is equivocal or normal [7]. Notwithstanding the ampulla and the neighboring duodenum related to pan-
these examples, in the majority of cases where suspected or creatic swelling.
apparent diagnoses were used as indications, no definitions
of the diagnoses were stated and there was no mention of Secondary indications (used only if patients did not fall
which signs, symptoms or investigation results led to sus- into one of the six primary indications)
pected diagnoses. In short, the majority of ERCP indications Eight hundred and eighty-two patients (50.2%) had second-
are provided as suspected or apparent diagnoses, and with- ary indications at least two of pain (P), imaging evidence
out definition, it is not clear how objectively these diagnoses (I), and significant abnormalities in LFTs (L).
were made nor how justified were the suspicions of the diag- The subtypes of the secondary indications were predictive
nosis, in which a suspected diagnosis was used as an of ERCP findings. The subtype with both imaging abnormal-
indication. ities and LFT abnormalities (IL) and the subtype with all
6 N. YUEN ET AL.
Biliary leak
60% 91
9 N/A
1 39
50%
Other
48 23
40% Normal
6
101 Dilated duct
30% 23
2 74 Stricture
2
20% 111 CBD stone
6
10% 44
2
Downloaded by [Australian Catholic University] at 04:11 29 September 2017
2
0% 1 2
Cholangis Biliary AP IOC C/ROSM C/ROSB
leak
Primary indicaons
Figure 2. Percentages of the endoscopic retrograde cholangiopancreatography (ERCP) findings in those performed with primary indications. AP: acute (biliary) pan-
creatitis; IOC: intraoperative cholangiogram; C/ROSM: change or removal of stent for malignant disease; C/ROSB: change or removal of stent for benign disease.
three components (PIL) were both associated with the pres- stones (35 of 128 cases, v2(1) 4.7, p .031) and stricture (11
ence of biliary stricture (113 of 179 cases, v2(1) 255, of 128 cases, v2(1) 9.5, p .002).
p < .001, and 104 of 475 cases, v2(1) 3.8, p .049, respect-
ively). Only if all three components (PIL) were present, there
was a statistically significant association with choledocholi- Adverse events
thiasis (225 of 475 cases, v2(1) 35.3, p < .001). Conversely, Figure 4 shows the ERCP indications compared against the
the secondary indication combination of pain and imaging occurrence of post-ERCP adverse events. Only two ERCP indi-
abnormalities (PI) was associated with a finding of dilated cations were significantly associated with the occurrence of
bile duct (14 of 53 cases, v2(1) 18.2, p < .001) and the adverse events: the combination subtype with PI (15 of 53
absence of stricture (3 of 53 cases, v2(1) 6.2, p .012). The cases, v2(1) 17.0, p < .001), and the other group (10 of 47
subtype of pain and LFT abnormalities (PL) was associated cases, v2(1) 5.3, p .021). Each was significantly associated
with normal cholangiography at ERCP (59 of 128 cases, with unplanned/prolonged hospitalization from all adverse
v2(1) 44.8, p < .001), but negatively associated both with events. Interestingly, neither PI nor other were significantly
SCANDINAVIAN JOURNAL OF GASTROENTEROLOGY 7
90%
80%
70%
38
60% 37 No adverse
423 108 166
160 264 142 128 events
50% 43 57
Adverse
40%
events
30%
20%
15
10% 10
52 20 27
19 20 12 13
0% 1 3
Indicaons
Figure 4. Percentages of endoscopic retrograde cholangiopancreatography (ERCP) related adverse events according to ERCP indications.
associated with post-ERCP pancreatitis. The unplanned/pro- such an approach is intrinsically unscientific as presupposing
longed hospitalization was thus due to factors other than a diagnosis can never substitute for an objective evaluation
post-ERCP pancreatitis. of a patient before the diagnosis (and any associated therapy
such as sphincterotomy or biliary stenting performed during
ERCP) is made. Although the a-priori indication system that
Discussion was used in our study uses some indications that presuppose
a diagnosis (e.g., acute biliary pancreatitis and cholangitis) as
For a new indication system for ERCP to be considered use-
primary indications, the diagnoses we used as indications, in
ful, it should be established that: first, it constitutes a signifi-
contradistinction to previous unselected series, were carefully
cant improvement on currently used indication criteria and
defined by clear clinical parameters that could be ascertained
second, it is helpful in predicting, on the one hand, which
prior to ERCP.
patients are most likely to benefit from ERCP and on the The new classification system was shown to be predictive
other hand, in which patients ERCP constitute an unaccept- of findings found at ERCP. This was entirely expected for
able risk. indications such as cholangitis in which a majority of affected
Regarding the first aim, it is clear that in previous series of patients had stones demonstrated within the common bile
unselected ERCPs there has been widespread use of ERCP duct. The other primary indications [clinical evidence of bil-
indications that pre-suppose diagnoses or suspected diagno- iary leak, acute (biliary) pancreatitis, abnormal IOC, change/
ses. This could be reflective of the trend away from using removal of stent for benign disease, and change/removal of
diagnostic ERCP in favor of therapeutic ERCP [2]. This trend stent for malignant disease] are also uncontroversial and the
may have influenced the classification of ERCP indications as findings in patients with these indications were also unsur-
it is easier to define therapeutic goals if there has been a prising. The exact clinical context in which acute biliary pan-
presumptive diagnosis made before the procedure. However, creatitis should lead to an ERCP and the timing of such an
8 N. YUEN ET AL.
ERCP are hotly debated in the literature [9]. It is clear none- been previously reported by our group [16]. In addition, pan-
theless that biliary pancreatitis is an appropriate indication creatic indications such as endotherapy for pancreatic ductal
for ERCP in many instances. There are other causes of pan- strictures and stones, minor ampullary pancreatic sphincterot-
creatitis, for which ERCP has been used but these are either omy for recurrent pancreatitis in the context of pancreas divi-
quite rare (pancreas divisum) or controversial (idiopathic pan- sum are also uncommon and often performed in quaternary
creatitis without any evidence of gallstone disease). In add- referral centres. The development of indication classification
ition, the pre-procedure exclusion of IgG4-related pancreatic systems related to these more uncommon scenarios would
disease may be important in some clinical scenarios. be complementary rather than competitive to the indications
The proposed classification system is not only worthwhile classification system outlined in the current study. Sphincter
for the relatively uncontroversial and straightforward primary of Oddi dysfunction or suspected sphincter of Oddi dys-
indications, but also contributes significantly to establishing a function was avoided as indications as this also presupposes
system of indications in clinical scenarios in which the role of a pre-procedure diagnosis or suspected diagnosis. The exact
ERCP is not so clear. Abdominal pain, imaging abnormalities nature of this entity is controversial in any case [17].
of the bile duct, and deranged LFTs were used as secondary The presented classification system is only a starting point
indications. Different combinations of these parameters, for refining indications in ERCP. It is often the pattern of the
which do not presuppose a diagnosis, have different associa- symptoms and investigation abnormalities which can sway a
Downloaded by [Australian Catholic University] at 04:11 29 September 2017
tions with ERCP findings. Patients presenting with the com- clinician to proceed with ERCP. For instance, a sudden
bination of PI commonly had a subsequent ERCP that change in LFTs and a sudden resolution or partial resolution
demonstrated the presence of a dilated bile duct in the increase the likelihood of choledocholithiasis being found.
absence of a stricture or stone. The combination of PIL was Similarly, severe pain with sudden onset is more suggestive
similarly associated with subsequent ERCP demonstration of of choledocholithiasis rather than a chronic mild pain of less
stones and strictures. The combination of PL was, however, acute onset. Specifying these patterns in indication defini-
associated with a normal cholangiogram a finding that is tions, however, risks over-defining without any improvement
less likely to lead to an ERCP-related benefit. Interestingly, in objectivity.
the combination of IL was associated with lack of access In conclusion, we have presented an ERCP indication clas-
into the bile duct. This analysis of secondary indications sification system, which is simple, easy to understand and
shows that patients with status of PI and PL may require intrinsically more scientific than its predecessors. In our ser-
additional evidence before ERCP is considered. In this ies, it has been shown to be useful in predicting the success
respect, endoscopic ultrasound may have a significant role as rate, findings, and adverse events of ERCP and we are confi-
it becomes more widespread and refined.
dent that it will prove useful in advancing the science of
These findings complement research that supports the
ERCP in general and thereby in the longer-term reducing
idea that a greater focus on objective parameters can result
unnecessary procedures and their related morbidity and
in indications for ERCP that reduce unnecessary procedures
mortality.
[1013]. Parnaby et al. [10] in a small study of 153 consecu-
tive patients developed an algorithm based on clinical,
laboratory, and radiological data and showed that this pre- Acknowledgements
dicted the need for ERCP. With respect to choledocholithiasis,
The authors thank Dr. Linda Lee of The Brigham and Womens Hospital,
Jovanovic et al. [12] have also evaluated an effective model Boston, for her help in reviewing this paper.
for predicting the need for therapeutic ERCP using biochem-
ical and ultrasound results.
It is acknowledged that a similar approach to ours has Disclosure statement
been used previously but only in terms of attempting to No potential conflict of interest was reported by the authors.
determine the likelihood of choledocholithiasis. In an ASGE
endorsed guideline, Maple et al. [13] proposed a series of
parameters to determine the likelihood of choledocholithia- References
sis. Adams et al. [14] and Suarez et al. [15], however, have [1] Chapman R. Complications of gastrointestinal endoscopy: compli-
found these criteria to be lacking in diagnostic accuracy. In cations of ERCP. BSG Guidelines Gastroenterol 2006;2025.
addition, some of the parameters used in the ASGE guideline [2] Adler DG, Lieb JG, Cohen J, et al. Quality indicators for ERCP.
were not precisely defined: clinical gallstone pancreatitis Gastrointest Endosc 2015;81:5466.
and clinical ascending cholangitis. [3] Cohen S, Bacon BR, Berlin JA, et al. NIH state-of-the-science state-
ment on endoscopic retrograde cholangiopancreatography (ERCP)
It is also acknowledged that our classification system can-
for diagnosis and therapy. NIH Consen State Sci Statements
not cover all the scenarios in which ERCP is indicated. 2002;19:126.
Uncommon indications such as an ampullary lesion requiring [4] Balik E, Eren T, Keskin M, et al. Parameters that may be used for
ampullectomy and rendezvous procedures for previously predicting failure during endoscopic retrograde cholangiopan-
percutaneously placed transhepatic catheters have unique creatography. J Oncol 2013;2013:201681.
[5] Christensen M, Matzen P, Schulze S, et al. Complications of ERCP:
endoscopic challenges and adverse event profiles. This is
a prospective study. Gastrointest Endosc 2004;60:721731.
confirmed in our study, in which unusual indications (other) [6] Baron TH, Petersen BT, Mergener K, et al. Quality indicators for
were shown to predict a greater risk of unplanned hospital- endoscopic retrograde cholangiopancreatography. Gastrointest
ization/prolongation of hospital stay. These findings have Endosc 2006;63:S29S34.
SCANDINAVIAN JOURNAL OF GASTROENTEROLOGY 9
[7] Colton JB, Curran CC. Quality indicators, including complications, [16] Chan E, Neeman T, Thomson A. Follow-up to determine
of ERCP in a community setting: a prospective study. Gastrointest unplanned hospitalization and complications after endoscopic
Endosc 2009;70:457467. retrograde cholangiopancreatography. ANZ J Surg 2016.
[8] Metcalfe MS, Ong T, Bruening MH, et al. Is laparoscopic intraoper- [17] Baillie J. Sphincter of Oddi dysfunction: overdue for an overhaul.
ative cholangiogram a matter of routine? Am J Surg Am J Gastroenterol 2005;100:1217.
2004;187:475481. [18] Cote GA, Imler TD, Xu H, et al. Lower provider volume is associ-
[9] Petrov MS, van Santvoort HC, Besselink MGH, et al. Early endo- ated with higher failure rates for endoscopic retrograde cholan-
scopic retrograde cholangiopancreatography versus giopancreatography. Med Care 2013;51:10401047.
conservative management in acute biliary pancreatitis without [19] Cotton PB, Garrow DA, Gallagher J, et al. Risk factors for compli-
cholangitis: a meta-analysis of randomized trials. Ann Surg cations after ERCP: a multivariate analysis of 11,497 procedures
2008;247:250257. over 12 years. Gastrointest Endosc 2009;70:8088.
[10] Parnaby C, Jenkins J, Ferguson J, et al. Prospective validation [20] Ekkelenkamp VE, Koch AD, Haringsma J, et al. Quality evaluation
study of an algorithm for triage to MRCP or ERCP for investiga- through self-assessment: a novel method to gain insight into
tion of suspected pancreatico-biliary disease. Surg Endosc ERCP performance. Frontline Gastroenterol 2014;5:1016.
2008;22:11651172. [21] Katsinelos P, Lazaraki G, Chatzimavroudis G, et al. Risk factors for
[11] Sharma S, Larson K, Adler Z, et al. Role of endoscopic retrograde therapeutic ERCP-related complications: an analysis of 2,715 cases
cholangiopancreatography in the management of suspected chol- performed by a single endoscopist. Ann Gastroenterol
edocholithiasis. Surgical Endoscopy 2003;17:868871. 2014;27:6572.
[12] Jovanovic P, Salkic NN, Zerem E, et al. Biochemical and ultra- [22] Siiki A, Tamminen A, Tomminen T, et al. ERCP procedures in a
sound parameters may help predict the need for therapeutic Finnish community hospital: a retrospective analysis of 1207
Downloaded by [Australian Catholic University] at 04:11 29 September 2017