Gastroenterology 2014;147:646–654
CLINICAL—PANCREAS
Administration of Secretin (RG1068) Increases the Sensitivity
of Detection of Duct Abnormalities by Magnetic Resonance
Cholangiopancreatography in Patients With Pancreatitis
Stuart Sherman,1 Martin L. Freeman,2 Paul R. Tarnasky,3 C. Mel Wilcox,4 Abhijit Kulkarni,5
Alex M. Aisen,6 David Jacoby,7 and Richard A. Kozarek8
1
Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana; 2Department of Medicine, University of
Minnesota, Minneapolis, Minnesota; 3Department of Medicine, Methodist Medical Center, Dallas, Texas; 4Department of Medicine,
University of Alabama at Birmingham, Birmingham, Alabama; 5Department of Medicine, Allegheny Center for Digestive Health,
Pittsburgh, Pennsylvania; 6Department of Radiology, Indiana University School of Medicine, Indianapolis, Indiana; 7Repligen Corp,
Waltham, Massachusetts; and 8Department of Medicine, Virginia Mason Medical Center, Seattle, Washington
CLINICAL PANCREAS
See editorial on page 559.
BACKGROUND & AIMS: Administration of secretin improves
noninvasive imaging of the pancreatic duct with magnetic
resonance cholangiopancreatography (MRCP). We performed a
large prospective study to investigate whether synthetic human
secretin (RG1068)-stimulated MRCP detects pancreatic duct
abnormalities with higher levels of sensitivity than MRCP.
METHODS: We performed a phase 3, multicenter, baselinecontrolled study of patients with acute or acute recurrent
pancreatitis who were scheduled to undergo endoscopic
retrograde cholangiopancreatography (ERCP) between March
26, 2008, and October 28, 2009. Patients underwent a baseline
MRCP that was immediately followed by administration of
RG1068 and repeat MRCP and then underwent ERCP within 30
days; they were followed up for 30 days. MRCP and ERCP images were read centrally by 3 radiologists and 2 endoscopists,
respectively, who were all independent and blinded; pancreatic
duct abnormalities were evaluated. The accuracy of MRCP was
evaluated using ERCP as the standard. RESULTS: In total, 258
patients were enrolled in the study; 251 MRCP image sets were
assessed, and 236 patients had evaluable ERCPs. Pancreatic
duct abnormalities were observed in 60.2% of ERCP images. All
radiologists identified duct abnormalities in RG1068-ciné
MRCP image sets with significantly higher levels of sensitivity
(P < .0001) than in images from MRCP, with minimal loss of
specificity. Adverse events were reported in 38.0% of patients
after MRCP and 68.1% after ERCP. Of the 55 patients who
experienced a serious adverse event, 3 (1.2%) and 52 (20.5%)
of the events were reported to be temporally associated with
MRCP and ERCP, respectively. The adverse events most
frequently considered related to RG1068 were nausea,
abdominal pain, and flushing; most were mild. CONCLUSIONS:
Compared with images from MRCP, those from RG1068stimulated MRCP are improved in many aspects and could
aid in diagnosis and clinical decision making for patients with
acute, acute recurrent, or chronic pancreatitis. RG1068enhanced MRCP might also better identify patients in need of
therapeutic ERCP (ClinicalTrials.gov, Number: NCT00660335).
Keywords: Pancreas; Peptide Hormone; Direct Comparison;
Increasing Resolution.
M
agnetic resonance cholangiopancreatography
(MRCP) was introduced as a clinical imaging
method more than 20 years ago; it uses T2-weighted magnetic resonance imaging (MRI) acquisitions to visualize
resident fluid in biliary and pancreatic ducts,1 outlining
their anatomy in situ. MRCP is an excellent modality to
assess the main causes of bile duct obstruction, such as bile
duct lithiasis, strictures, cholangiocarcinoma, and pancreatic
adenocarcinoma.1 However, the resolution of baseline MRCP
is poor for pancreatic ducts because of their small diameter;
nonetheless, visualization can be improved substantially by
intravenous administration of secretin.1–4
Secretin is a natural 27–amino acid peptide hormone that
stimulates the release of pancreatic juice from acinar cells in
the exocrine pancreas into the pancreatic ducts. This results
in increased size and signal of the ducts, with improved
structural delineation during MRCP imaging.4 Thus, secretinstimulated MRCP (S-MRCP) is a noninvasive, ionizing,
radiation-free assessment of the pancreaticobiliary system.5
Secretin is not currently labeled for this use.
Historically, a primary modality for diagnosing and
treating many pancreatic and biliary diseases was endoscopic retrograde cholangiopancreatography (ERCP),6 which
remains the gold standard for visualization of the hepatobiliary and pancreatic ducts and diagnosis of pancreas
divisum.7,8 ERCP is currently used primarily as a therapeutic
procedure; it is estimated that for all indications, including
biliary reasons, approximately 500,000 ERCPs are performed
annually in the United States.9
Even among experienced clinicians, ERCP is associated
with a substantial incidence of complications such as
procedure-related pancreatitis9,10 and, less commonly,
cholangitis, hemorrhage, and perforation.11 Cholangitis may
Abbreviations used in this paper: AE, adverse event; CT, computed
tomography; ERCP, endoscopic retrograde cholangiopancreatography;
EUS, endoscopic ultrasonography; MPD, main pancreatic duct; MRCP,
magnetic resonance cholangiopancreatography; MRI, magnetic resonance imaging; S-MRCP, secretin-stimulated magnetic resonance cholangiopancreatography; SAE, serious adverse event.
© 2014 by the AGA Institute
0016-5085/$36.00
http://dx.doi.org/10.1053/j.gastro.2014.05.035
occur after diagnostic ERCP in patients with malignant biliary
obstruction who have had either no attempt at drainage or
unsuccessful drainage.12 As a result, the use of ERCP as a
purely diagnostic tool declined as less invasive procedures
became available. Nonetheless, due to a lack of adequate
noninvasive imaging, current use of ERCP still includes a
diagnostic role for potential pancreatic duct etiology in cases
of unexplained pancreatitis as well as nonspecific upper
abdominal pain suspected of being pancreatic in origin in
which prior studies have not shown a clear cause.10
Compared with current primary imaging modalities,
such as computed tomography (CT) and transabdominal
ultrasonography, MRCP images show greater extent and
detail of the pancreatic duct anatomy.9 Typically, an S-MRCP
is interpreted as a dynamic study that contains both
anatomic and functional information, with structural delineation of the ducts and flow of pancreatic juice into the
duodenum analyzed as a ciné series of images acquired
sequentially after infusion of secretin. Visualization of the
duct is important to establish etiology, correctly identify
which patients require ERCP, and plan the ERCP therapeutic
strategy. Technical enhancements that improve MRCP image
quality, such as secretin, are likely to further reduce the
need for diagnostic ERCP, with its attendant risks.
RG1068 is a synthetic peptide with an amino acid
sequence identical to that of naturally occurring human
secretin. RG1068 differs from the current commercially
available secretin in a number of respects, including
administration and formulation, which results in beneficial
product characteristics (Repligen Corp, data on file). Previous preliminary studies, including a baseline-controlled
phase 2 MRCP study (Study RG1068-15) in patients with
a history of acute or acute recurrent pancreatitis, showed
that RG1068 has a favorable safety and tolerability profile
and is associated with improved duct visualization, image
quality, and reader confidence (Repligen Corp, data on file).
This report describes the efficacy and safety findings of a
phase 3, multicenter, baseline-controlled, single-dose study
of RG1068 in patients with a history of acute or acute
recurrent pancreatitis. The aim of the study was to show
that RG1068-stimulated MRCP improves sensitivity in the
detection of pancreatic duct abnormalities compared with
baseline MRCP, with minimal loss of specificity, using an
image-based clinical truth standard.
Secretin-Enhanced MRCP in Pancreatitis
647
All authors had access to the final study data and have
reviewed and approved the final manuscript.
Pharmaceuticals for Human Use and all applicable regulations. It
was conducted to comply with the ethical principles described in
the Declaration of Helsinki and local legal requirements. At each
center, institutional review board (United States) or research
ethics board (Canada) approval was obtained. All patients provided written informed consent before the start of the study.
Study visits occurred as follows: visit 1, screening (14 days
before RG1068 dosing); visit 2, MRCP and RG1068 dosing (14
days after visit 1); visit 3, post-RG1068 safety follow-up (20–48
hours after RG1068 dosing); visit 4, ERCP (1–30 days after MRCP;
minimum interval of 20 hours and after completion of visit 3
assessments); and visit 5, post-ERCP safety follow-up (24–72
hours after ERCP). Serious adverse events (SAEs) were reported
for up to 30 days after administration of RG1068.
MRCP and ERCP study images were acquired at the clinical
study sites and sent to a central imaging laboratory (BioClinica,
Newtown, PA) to be read. Randomized baseline and RG1068ciné MRCP image sets were read for the current study by 3
independent radiologists who were blinded to treatment and
patient identifiers. ERCP images, in conjunction with reports
from other imaging modalities, were used to arrive at the
comparative truth standard via a central consensus read by 2
independent qualified and blinded endoscopists.
The results of the current study (NCT00660335) are from a
reanalysis of images collected during an earlier study (RG106816). This reevaluation was prompted after a detailed audit of
the read conducted by the original central imaging laboratory
for RG1068-16 uncovered multiple deviations from the study
protocol read charter, which may have affected image interpretation and study analyses but were not expected to have
adversely affected the quality of the imaging studies themselves. After consultation with the Food and Drug Administration and European Medicines Agency, and in consideration of
factors such as high cost, patient discomfort, the risk associated
with the collection of a new set of images, and that the images
had been collected in a standardized and rigorous manner, it
was deemed appropriate to conduct an independent reanalysis
(or reread) of the original images using the same clinical data
and analytical methodology. NCT00660335 was thus conducted
as a standalone study at a different central imaging laboratory
with new readers and a new independent read charter governing reader training and the read process. Procedures were
implemented to ensure that the conduct of the reread had
minimal potential for bias and provided a valid and reliable
data set for analysis. For example, readers had no prior
knowledge of the study protocol or involvement in or exposure
to data from previous RG1068 clinical studies. To avoid
potential recall bias, each MRCP read session contained both
randomized baseline and RG1068-ciné image sets, and at least
12 days elapsed between review of the baseline and RG1068ciné image sets for a given patient.
Study Design
Objectives
This was a multicenter, open-label, baseline-controlled,
single-dose, phase 3 study (ClinicalTrials.gov, Number:
NCT00660335) of RG1068-stimulated MRCP conducted at 23
sites in Canada and the United States between March 26, 2008,
and October 28, 2009. The study was conducted in accordance
with Good Clinical Practice requirements described in the current revision of the International Conference on Harmonisation
Guidelines of Technical Requirements for Registration of
The primary objective was to show that RG1068-stimulated
MRCP improves sensitivity in the detection of pancreatic duct
abnormalities compared with baseline MRCP, without substantive loss of specificity, using an image-based truth standard.
The secondary objectives were to show the safety of
RG1068-stimulated MRCP compared with ERCP, that RG1068stimulated MRCP improves visualization of the pancreatic
ducts compared with baseline MRCP, that RG1068-stimulated
Patients and Methods
CLINICAL PANCREAS
September 2014
648
Sherman et al
MRCP improves the identification of pancreatic duct abnormalities and visualization of the pancreatic ducts compared
with baseline MRCP in patients with chronic pancreatitis, and
that RG1068-stimulated MRCP improves the ability to ascertain
those patients for whom ERCP is not indicated compared with
baseline MRCP. Image sets were evaluated only for the presence or absence of pancreatic duct or related abnormalities;
biliary tract findings were not reported, because the biliary
tract is relatively well visualized on MRCP and secretin would
not be expected to improve its imaging.
A tertiary objective was to show that RG1068-stimulated MRCP
improves the level of reader confidence in the determination of
pancreatic duct abnormalities compared with baseline MRCP.
Patients
CLINICAL PANCREAS
Medically stable male and female patients (18 years of age or
older) with a history of acute or acute recurrent pancreatitis who
had been scheduled by the enrolling center to undergo ERCP with
pancreatography were enrolled in this study. It was anticipated
that ERCP would reveal that approximately one-third of enrolled
patients would have mild or moderate chronic pancreatitis.
The study inclusion criteria required the performance of
certain imaging within 120 days before administration of
RG1068: CT imaging of the pancreas or endoscopic ultrasonography (EUS) of the pancreas noting 0 to 3 ductal and/or
parenchymal abnormalities (>3 abnormalities required CT of
the pancreas within 120 days before administration of RG1068)
or kidney, ureter, and bladder x-ray and MRI of the pancreas.
Exclusion criteria included any imaging evidence of severe
chronic pancreatitis (which could confound interpretation and
defined as significant pancreatic atrophy or evidence of significant calcification), prior pancreatic surgical procedure
or pancreatic stenting that may affect duct appearance (eg,
Puestow procedure, drainage, or resection), or active acute
pancreatitis requiring pancreatic rest.
Treatment
RG1068 was administered at the time of MRCP examination
(see the following text) as a single dose by intravenous
infusion over 30 seconds (followed by a 10-mL normal saline
flush for 30 seconds; total infusion time of 1 minute [plus a
30-second window]). The following doses were used: patients
weighing 50 kg at screening received a fixed dose of 22.5 mg
RG1068, equivalent to 4.5 mL, and patients weighing <50 kg at
screening received 0.2 mg/kg RG1068.
MRCP and ERCP Image Acquisition
MRCP imaging was performed following accepted acquisition parameters11 with a 1.5-T MRI system using a surface
phased array body coil. Axial sections localized the pancreas
and surrounding anatomy, and baseline MRCP images (T2weighted thick slab breath-holding sequence) were acquired
after final positioning at 1-minute intervals (30 seconds) for 3
minutes. The patient remained in the magnet during infusion of
RG1068, and postdose MRCP images (oblique thick slab), using
the same prospectively determined imaging parameters, were
then acquired 1 minute after the end of the normal saline flush
and at 1-minute intervals (30 seconds) for 10 minutes. The
baseline image set comprised the pre-RG1068 images. Baseline
and RG1068 MRCP images formed a ciné image set.
Gastroenterology Vol. 147, No. 3
ERCP imaging within 30 days of MRCP was a protocol
requirement and typically occurred within 48 hours of MRCP.
For ERCP, an unmagnified image of the pancreatic ducts plus
magnified images of the pancreatic duct in the head, body, and
tail were required. Supportive images included 5 additional images documenting all abnormalities not captured on
required images (if applicable) and scout film (if available).
Required images were obtained even if the pancreatic duct
segments were not fully injected. Any perceived abnormalities
were imaged as clearly as possible. Once acquired at the study
sites, the MRCP and ERCP images were sent to the third-party
central imaging laboratory for analysis.
Efficacy Assessments
Three board-certified radiologists and 2 endoscopists
performed blinded, central assessments of pancreatic duct
abnormalities on MRCP and ERCP, respectively. To ensure that
a standardized interpretation of each of the 10 predefined
abnormality definitions would be applied across the study, the
independent radiologist and endoscopist readers were trained
on abnormality definitions by MRCP and ERCP experts,
respectively, and then tested using nonstudy images that
reflected these abnormalities. Abnormality detection rates
of 50% and 80% and absence rates of 70% and 85%
were required for each radiologist and endoscopist, respectively, before he or she could participate in read sessions.
For the MRCP images, 3 blinded readers separately read all
randomized image sets for abnormalities and duct visualization
and rated image quality and diagnostic confidence. ERCP
images were reviewed by 2 endoscopists, also blinded to
patient identifiers, who had no details of the protocol (except
for information pertaining to the central image truth standard
read procedure), knowledge of the MRCP results, or information regarding the MRCP image sets that were considered
nonevaluable. The ERCP pancreatic duct abnormality data
evaluated by consensus of the 2 endoscopists formed the
image-based truth standard.
In constructing the image-based truth standard, the ERCP
image read was the primary component; however, recognizing
that its use alone has limitations as a truth standard against
which to measure MRCP for detection of abnormalities, the
reference standard was augmented with available reports from
other relevant imaging modalities, thus optimizing the reference standard for each patient as much as possible. ERCP was
always the standard, and if uninterpretable, the MRCP did not
contribute to the analysis even if other imaging data were
available. The detection agreement between the MRCP reads
and the image-based truth standard was used as the basis for
determining sensitivity and specificity. Each of 10 pancreatic
duct abnormalities was determined to be present or not present for every image set considered evaluable. Sensitivity was
defined as the number of individual pancreatic duct abnormalities detected on ERCP that were also present on the
correlated MRCP. Conversely, specificity was defined as the
absence of individual abnormalities on ERCP that were also
absent on MRCP. Sensitivity and specificity were calculated
using a clustered by-abnormality, within-patient analysis.
Secondary endpoints included the efficacy of RG1068stimulated MRCP quantitatively determined by comparing the
proportion of patients with completely or not completely
visualized pancreatic ducts (ie, complete head, body, and tail
segments seen) for baseline and RG1068-ciné image sets. An
additional secondary evaluation was the ERCP prevention
index, a novel negative predictive index, which assessed the
ability of MRCP to identify patients for whom therapeutic ERCP
would not have been indicated. For an MRCP to contribute to
the ERCP prevention index, it had to completely visualize at
least the head and body of the pancreatic duct (the anatomy in
which clinical lesions occur and are most easily accessible and
treatable by endoscopic intervention) and correctly show that
the therapeutic cluster of stenosis/stricture, disruption, or
divisum was not present. Hence, the ERCP prevention index
was intended to estimate the proportion of procedures that
were primarily diagnostic and thus may have been avoided if
the MRCP information had been factored into the decision to
perform ERCP. It therefore evaluated the ability of MRCP,
with and without RG1068, to identify those patients for whom
ERCP was not immediately indicated. RG1068-ciné MRCP
was compared with baseline MRCP for the ability to improve
clinical decision making as estimated by the ERCP prevention
index. In a subset of patients with at least mild severity of
chronic pancreatitis (defined by the Cambridge classification),13 abnormality detection and duct visualization were also
evaluated.
Reader confidence in determining the presence or absence of
abnormalities was assessed for baseline and RG1068-ciné images based on a 5-point scale, ranging from very low to very high.
Safety
Adverse events (AEs) were recorded throughout the study
with respect to their temporal relationship to MRCP (start of
RG1068 administration to before ERCP) and ERCP (start
of ERCP to end of study). The relationship to administration of
RG1068 (investigator determined) was also recorded. Unless
AEs were identified as attributable to the patient’s stable/
chronic condition or intercurrent illness(es), they were monitored until resolution; medical care was provided as appropriate. Blood samples for clinical chemistry analyses were
collected at visits 1 and 3.
Statistical Methods
Based on the phase 2 data, it was estimated that 50% of the
enrolled patients would have pancreatic duct abnormalities on
ERCP and that 5% of those enrolled would die before an
evaluable ERCP was obtained. Hence, it was postulated that 270
patients would need to be enrolled to have 128 patients with
ERCP-demonstrated abnormalities. This sample size was
considered necessary to provide 80% power for detecting a
difference in mean sensitivity of 10% with an SD of 40%
between RG1068-ciné MRCP and baseline MRCP. All patients
whose MRCP image sets had 1 T2-weighted predose images
and 1 T2-weighted postdose images that showed 1 segment
of the pancreatic duct within the field of view were reviewed by
each radiologist. If images were considered technically inadequate and thus could not be interpreted, the radiologist could
call the image set not readable (ie, nonevaluable). Similarly,
ERCP image sets were reviewed by the ERCP readers, who
could deem the image set not readable if the images and
reports could not be interpreted. All patients with an evaluable
ERCP, baseline MRCP, and RG1068-ciné MRCP were included in
the primary efficacy analyses.
Secretin-Enhanced MRCP in Pancreatitis
649
The blinded, independent MRCP assessments completed by
the 3 radiologists for the baseline and RG1068-ciné image sets
were scored using the respective image-based truth standard to
evaluate sensitivity and specificity. To compensate for the
contribution that a single abnormality may make to the sensitivity or specificity because of the unequal prevalence of
abnormalities, a clustered analysis was performed. The 4 clusters were as follows: chronic pancreatitis (abnormal side
branches, dilated main pancreatic duct [MPD], irregular MPD,
pancreatic duct strictures, stenosis), obstructive lesions
(filling defects, side branch intraductal papillary mucinous
neoplasm, main duct intraductal papillary mucinous neoplasm),
ductal leakage and its sequelae (pancreatic duct disruptions,
cysts, pseudocysts), and pancreas divisum. The clustered,
by-abnormality, within-patient analysis scored the presence or
absence of specific matching abnormalities relative to the truth
standard and averaged the by-abnormality results within that
cluster. Each cluster was subsequently averaged to determine
the overall within-patient estimate. Sensitivity and specificity
results were calculated using this clustered by-abnormality,
within-patient analysis and were the primary outcome of the
study.
Baseline MRCP results were compared with RG1068-ciné
MRCP results using the paired t test. Confidence intervals for
the baseline MRCP/ciné MRCP differences were calculated
for the specificity parameter. The coprimary end point was
considered to have been met if there was a significant (P .05)
improvement in sensitivity and noninferiority in specificity for
the RG1068-ciné MRCP compared with the baseline MRCP for 2
of the 3 central radiologists. Noninferiority for specificity was
defined as the lower 95% confidence interval that could not
exceed a 7.5% change from baseline. A robustness analysis
was performed in which sensitivity and specificity were estimated from different evaluable populations. All patients with
an evaluable ERCP (total evaluable) were analyzed; if MRCP
data were missing, they were counted as incorrect. A further
analysis was performed in which notable imaging violations
were excluded (per protocol).
Changes from baseline (pre-RG1068) in the number of
patients with completely visualized pancreatic ducts identified in
the RG1068-ciné image sets were analyzed by logistic regression,
and a responder analysis was performed using the McNemar test.
The ERCP prevention index value for baseline MRCP was
compared with that for the RG1068-ciné MRCP using the ERCP
therapeutic cluster as the truth standard and was considered
positive if a significant difference in favor of the ability of RG1068ciné MRCP to identify those patients for whom ERCP was not
immediately indicated was found by 2 of the 3 radiologists.
For reader confidence, the mean change from the baseline
image sets to the RG1068-ciné image sets was calculated. Shift
tables were analyzed by the Stuart–Maxwell c2 test. The proportion of images that provided high (4) and very high (5)
confidence for the baseline and RG1068-ciné image sets were
tested for baseline versus RG1068-ciné MRCP differences using
the McNemar test.
Results
Patients
In total, 258 patients were enrolled; of these, 252
completed the study (Supplementary Figure 1). Table 1
CLINICAL PANCREAS
September 2014
650
Sherman et al
Gastroenterology Vol. 147, No. 3
Table 1.Patient Demographic and Clinical Characteristics
(Safety Population; N ¼ 258)
Age (y), mean SD
Sex, n (%)
Male
Female
Race/ethnicity, n (%)
White
Black
Asian
Other
Height (cm), mean SD
Weight (kg), mean SD
Medical history affecting body systema
(% of patients)
Genitourinary
Neurological/psychological
Musculoskeletal
Allergy
47.7 (14.6)
93 (36)
165 (64)
231
17
4
5
167.7
79.8
(89.6)
(6.6)
(1.6)
(1.9)
(11.0)
(19.5)
79.8
77.5
75.2
72.1
CLINICAL PANCREAS
Data presented for body systems reported in >70% of
patients; all patients had a gastrointestinal medical history.
a
presents patient demographics and clinical characteristics
for the safety population. Image sets for 251 of the 258
enrolled patients were sent to the readers for assessment;
there were 236 evaluable ERCPs.
Figure 1 presents MRCP images with and without
RG1068 for a single patient with a diagnosis of pancreas
divisum and shows the duct more clearly after infusion of
RG1068.
Efficacy
At least one pancreatic duct abnormality was observed
in 142 of the 236 ERCP image sets (60.2%); the most
commonly reported pancreatic duct abnormalities were
abnormal side branches and dilated and irregular MPDs
(Table 2).
Among patients with evaluable ERCPs, the sensitivity of
MRCP increased from 47.1% to 66.4% and specificity
decreased from 89.5% to 84.7% with administration of
RG1068 during MRCP (Table 3). All 3 radiologists assessed
duct abnormalities with increased sensitivity (P < .0001 for
all) for RG1068-ciné MRCP image sets in comparison to the
baseline MRCP (Table 3). Noninferiority in specificity for
RG1068-ciné MRCP compared with baseline MRCP was
demonstrated for all 3 radiologists (Table 3).
The robustness of the coprimary end point was not
dependent on the evaluable population (Supplementary
Table 1). For all readers, there was a significant improvement in sensitivity for the total ERCP evaluable population
and the MRCP per-protocol population, which was highly
comparable to the primary analysis. Furthermore, in the
subset of patients with chronic pancreatitis shown by ERCP,
the improvement in abnormality detection with RG1068
was also significant (P < .0001 for all readers, for all analyses). Similarly, for specificity, noninferiority of treatment
was met for all populations for all readers.
Analysis of the number of patients with completely
visualized pancreatic ducts found that all 3 readers were
able to visualize the complete pancreatic duct in significantly more patients in the RG1068-ciné image sets than in
the baseline MRCP image sets (P < .0001 for each; Tables 4
and 5).
All 3 readers demonstrated a significant improvement in
the ERCP prevention index for RG1068-ciné MRCP
compared with baseline MRCP (Table 6). RG1068 images
identified a high proportion of patients who did not have an
abnormality in the therapeutic cluster for which intervention is usually indicated (pancreas divisum, duct disruption,
and duct stricture or stenosis) and who may thus not have
required an ERCP had the MRCP information been available
before scheduling the ERCP. Across readers, an average of
74.8% of the 142 evaluable ERCPs that did not have an
abnormality in the therapeutic cluster would have been
identified by RG1068-ciné MRCP, which is improved from
47.9% identified by baseline MRCP. This corresponds to 106
ERCPs potentially being avoided by the results of RG1068ciné MRCP, which is 38 more than would have been
informed by the results of baseline MRCP.
Figure 1. MRCP images
for a single patient. (A)
Baseline MRCP image and
(B) MRCP image taken 5
minutes after administration of RG1068.
651
CLINICAL PANCREAS
6.7
5.9
5.6
5.8
13.1
11.8
10.2
13.1
5.3
4.6
4.5
4.8
12.3
14.3
11.4
13.6
82.0
82.4
89.6
84.7
12.0
12.4
10.3
13.1
87.3
87.0
94.1
89.5
228
218
216
<.0001
<.0001
<.0001
<.0001
43.2
36.9
41.4
36.5
20.8
15.6
21.6
19.3
38.6
40.9
41.0
35.3
71.1
66.1
62.0
66.4
42.6
41.6
40.3
36.5
SD (%)
Mean (%)
Paired t test.
The noninferiority margin was set at 7.5%. Sensitivity was determined from the reader-defined evaluable population with at least one abnormality by the image-based
truth standard. Specificity was determined from the reader-defined evaluable population with at least one of 10 possible abnormalities absent on the image-based truth
standard.
b
This multicenter, baseline-controlled, phase 3 study
showed that use of synthetic human secretin (RG1068) with
MRCP significantly improved identification of pancreatic
duct abnormalities compared with MRCP alone in patients
a
Discussion
50.3
50.6
40.4
47.1
Three patients (1.2%) reported an SAE during the time
associated with MRCP, and 52 patients (20.5%) reported an
SAE during the time associated with ERCP. No SAEs were
considered related to RG1068. There were no deaths in this
study. No patients discontinued the study due to an AE, and
no infusions of RG1068 were interrupted due to an AE.
Table 7 shows the most common AEs (5% of patients)
following MRCP or ERCP. After MRCP, nausea, abdominal
pain, flushing, and feeling hot were most frequently considered potentially related to RG1068. These AEs were generally
mild and transient. Pancreatitis was reported as an SAE in
36 patients (14.2%) following ERCP, all attributed by the
investigators to ERCP, and 1 (0.4%) following MRCP, deemed
unlikely related to RG1068.
Comparing pre-RG1068 (screening) and post-RG1068
(visit 3) clinical chemistry, hematology, and electrocardiographic parameters, there were no changes considered to be
clinically meaningful. The mean (SD) differences before and
after administration of RG1068 for amylase (4.6 [50.5] U/L)
and lipase (8.4 [70.9] U/L) were not significant.
139
131
132
Safety
1
2
3
Overall
Reader confidence in the determination of pancreatic
duct abnormalities substantially improved with administration of RG1068 (P < .0001 for all 3 readers). Similarly,
image quality was also assessed as improved for RG1068 for
all readers (P < .0001).
Mean (%) SD (%)
NOTE. All values are n (%).
a
Defined as diameters greater than 5 mm in the head, 4 mm in
the body, and 3 mm in the tail, corrected for radiographic
magnification.
n
5 (2.1)
5 (2.1)
8 (3.4)
Mean (%) SD (%) P valuea
231 (97.9)
231 (97.9)
228 (96.6)
SD (%)
0 (0.0)
Mean (%)
236 (100.0)
Mean (%) SD (%)
58 (24.6)
0 (0.0)
n
178 (75.4)
236 (100.0)
RG1068-ciné MRCP
(60.2)
(39.0)
(18.2)
(21.6)
(14.8)
Baseline MRCP
142
92
43
51
35
Difference in sensitivity
(39.8)
(61.0)
(81.8)
(78.4)
(85.2)
RG1068-ciné MRCP
94
144
193
185
201
Baseline MRCP
Any abnormality
Abnormal side branch
Dilated MPDa
Irregular MPD
Pancreatic duct stricture or
stenosis
Pancreas divisum
Side branch intraductal papillary
mucinous neoplasm
Main duct intraductal papillary
mucinous neoplasm
Filling defect(s)
Pancreatic duct disruptions
Pancreatic cysts and/or
pseudocysts
Abnormality
Table 3.Comparative Sensitivity and Specificity: Patients With Evaluable ERCP, Baseline MRCP, and RG1068-Ciné MRCP Image Sets
No abnormality
Difference in specificity
Table 2.Pancreatic Duct Abnormality Data, as Evaluated by
the ERCP Consensus Read of 236 Evaluable ERCP
Images
Mean (%) SD (%) Lower limit b
Secretin-Enhanced MRCP in Pancreatitis
Reader
September 2014
652
Sherman et al
Gastroenterology Vol. 147, No. 3
Table 4.Patients With Completely Visualizeda Pancreatic
Ducts
RG1068-ciné
Baseline MRCP
MRCP
Difference
Reader
1
2
3
n
n (%)
Mean
Mean
P valueb
249
248
247
93 (37.3)
32 (12.9)
83 (33.6)
155 (62.2)
119 (48.0)
135 (54.7)
24.9
35.1
21.1
<.0001
<.0001
<.0001
a
All 3 duct segments (head, body, and tail) were visualized
entirely.
b
Responder analysis was analyzed using the McNemar test.
CLINICAL PANCREAS
with acute and acute recurrent pancreatitis. The increase in
sensitivity was clinically significant and associated with
minimal loss of specificity. RG1068-ciné MRCP had
improved sensitivity in all evaluable populations and in the
subset of patients with evidence of chronic pancreatitis on
ERCP, showing the robustness of the primary end points.
RG1068 in the context of MRCP was associated with greater
duct segment visualization and improved information for
identifying patients who may not require therapeutic ERCP.
The data presented here show that inclusion of RG1068stimulated images in an MRCP examination improves
several aspects of imaging that may enhance diagnostic
efficacy compared with baseline MRCP; these include
increased sensitivity for detection of specific duct lesions
and greater visualization of duct anatomy. RG1068 was also
associated with improvement in duct segments visualized,
reader confidence in attributing abnormalities in analysis,
and image quality.
The clinical utility of the identification of pancreatic duct
abnormalities by MRCP both before and after administration
of MRCP was further evaluated by using the ERCP prevention index. This negative predictive index suggested that,
compared with baseline MRCP, RG1068-MRCP images may
be better able to evaluate and identify those patients who do
not have pancreas divisum, duct disruption, duct stricture,
or stenosis. These abnormalities are associated with a high
likelihood of intervention; therefore, accurate identification
Table 6.ERCP Prevention Index
Reader
Baseline
MRCP (%)
RG1068-ciné
MRCP (%)
Difference(%)
P value
1
2
3
Overall
55.0
34.2
55.0
47.9
71.1
75.8
77.2
74.8
16.1
41.6
22.2
34.1
<.0023
<.0001
<.0001
<.0001
of patients who do not have these abnormalities will likely
reduce the number of unnecessary diagnostic ERCP procedures. This finding suggests that RG1068 MRCP is a better
guide than baseline MRCP for predictive decision making
regarding the need for therapeutic ERCP.
Pancreatic imaging is an essential part of the diagnosis of
patients with abdominal pain or suspected pancreatitis. The
number of patients requiring investigation is high; for
example, it is estimated that in the United States, approximately 210,000 patients per year are admitted to hospital
with acute pancreatitis.14 Many modalities may be used to
visualize the pancreatic ducts and the surrounding area,
including contrast-enhanced CT, EUS, MRI/MRCP, and
ERCP.15 CT has a role in early diagnosis; for example, it can
help differentiate acute pancreatitis from other conditions
presenting with abdominal pain or highlight local complications.10,16 MRI may have certain advantages over CT, such
as an improved ability to distinguish necrosis from fluid and
better reliability in staging the severity of acute pancreatitis.10 MRCP, an addition to MRI that directly visualizes the
fluid-filled ductal structures, is an important advance in the
diagnostic imaging arsenal and may be considered a viable
alternative to diagnostic ERCP.5 S-MRCP and secretinstimulated EUS improve the view of the biliopancreatic
ductal system compared with baseline MRCP or EUS.17
Furthermore, in patients with acute recurrent pancreatitis
of unknown etiology, S-MRCP, secretin-stimulated EUS, and
ERCP were found to have similar diagnostic yields.17 EUS is
Table 7.AEs Reported in 5% of Patients in the Total Safety
Population by Temporal Association With MRCP or
ERCP (Safety Population)
Table 5.Proportion of Patients With Complete Visualization of
Pancreatic Ducts
Baseline
MRCP
Pooled, % (lower
to upper 95%
confidence
interval)
26.1
(21.9–30.8)
RG1068-ciné
MRCP
Change from
baseline not
complete to
RG1068-ciné
complete
54.9
(50.3–59.4)
31.1
(27.6–34.9)
NOTE. Derived from logistic regression confidence intervals
from contrast statements.
MRCP
(n ¼ 258)
Any event
Nausea
Abdominal pain
Flushing
Feeling hot
Headache
Vomiting
Postprocedural
nausea
Procedural pain
Pancreatitis
98
37
23
22
13
9
3
2
(38.0)
(14.3)
(8.9)
(8.5)
(5.0)
(3.5)
(1.2)
(0.8)
2 (0.8)
1 (0.4)
NOTE. All values are n (%).
ERCP
(n ¼ 254)
173
37
50
1
0
5
11
26
(68.1)
(14.6)
(19.7)
(0.4)
(0)
(2.0)
(4.3)
(10.2)
39 (15.4)
36 (14.2)
All (N ¼ 258)
200
66
69
23
13
14
14
28
(77.5)
(25.6)
(26.7)
(8.9)
(5.0)
(5.4)
(5.4)
(10.9)
40 (15.5)
37 (14.3)
an evolving technology18 that has become an important
imaging modality19; however, its limitations include the fact
that it is not widely available20 and has incomplete standardization and correlation with pathological findings. ERCP
has been used for many years both as a diagnostic and
therapeutic tool; because of the lack of a widely available,
reproducible imaging modality for visualization of the
pancreatic duct, ERCP remains generally overused. Indeed,
in patients enrolled in the study scheduled for ERCP,
39.8% had no observed pancreatic duct pathology. ERCP is
associated with complications, particularly in a high-risk
group21,22; thus, a sensitive MRI-based test to triage
patients to ERCP and guide therapeutic strategy has major
clinical value.
A number of studies showing the increased effectiveness
of S-MRCP in comparison with baseline MRCP have been
reported. For example, S-MRCP was more effective in
detecting pancreatic duct abnormalities than baseline MRCP
in asymptomatic patients with chronic pancreatic hyperenzymemia (n ¼ 25).23 In this study, S-MRCP boosted the
diagnostic yield of MRCP for the diagnosis of chronic
pancreatitis 4-fold and significantly fewer patients with
chronic pancreatic hyperenzymemia had normal findings
(P < .02) compared with baseline MRCP. More recently,
S-MRCP was shown to detect pancreas divisum with satisfactory specificity (96.8%) and sensitivity (73.3%), whereas
MRCP without secretin was found to be nondiagnostic for
this condition.8 However, there is currently a paucity of
controlled multicenter studies evaluating S-MRCP; indeed,
the present study represents the largest blinded, multicenter assessment of efficacy performed to date.
The safety data reported here are consistent with the
known established safety profile of RG1068. Abdominal pain
and nausea are common AEs associated with S-MRCP,24 so it
was not unexpected to find that these were 2 of the most
common AEs associated with RG1068 in the present study.
The AEs associated with RG1068 were generally related to
the known biological activity of secretin and mild and
transient, in line with its half-life of <3 minutes. Amylase
and lipase levels are typically elevated during the course of
acute pancreatitis; however, there were no clinically relevant increases in these enzyme values after infusion of
RG1068. Overall, there were considerably fewer AEs and
SAEs with MRCP than with ERCP.
The strengths of the present study include the large
number of evaluable images that were assessed and that it
was a prospectively defined, blinded, multicenter study with
independent interpretation of imaging studies. The findings,
however, are limited to the population studied, which was
predominantly white and had a mean age of 48 years.
Further studies may be warranted in other populations,
such as the elderly. Another potential limitation is that the
results were obtained from a randomized, blinded reading
of images, which differs from the clinical setting where the
radiologist will have access to additional clinical information. Instead, this study was designed to show a difference
compared with baseline MRCP based on the images alone.
Access to clinical data and any biliary tract findings during
interpretation would likely improve the sensitivity and
Secretin-Enhanced MRCP in Pancreatitis
653
specificity values, because assessments would be made in
their appropriate clinical context. The ERCP prevention
index is merely an estimate of ERCP procedures that might
be avoided. However, we do not know if these procedures
would actually be avoided in clinical practice. In summary,
compared with baseline MRCP, RG1068-stimulated MRCP
improves many aspects of the collected images of the
pancreatic ducts and subsequently may yield improved
diagnostic outcomes and clinical decision making in patients
with acute, acute recurrent, and chronic pancreatitis.
Furthermore, RG1068-stimulated MRCP has the potential to
better identify those patients needing therapeutic ERCP.
Supplementary Material
Note: To access the supplementary material accompanying
this article, visit the online version of Gastroenterology at
www.gastrojournal.org and at http://dx.doi.org/10.1053/
j.gastro.2014.05.035.
References
1. Maccioni F, Martinelli M, Al Ansari N, et al. Magnetic
resonance cholangiography: past, present and future: a
review. Eur Rev Med Pharmacol Sci 2010;14:721–725.
2. Pascual I, Soler J, Pena A, et al. Morphological and
functional evaluation of the pancreatic duct with secretinstimulated magnetic resonance cholangiopancreatography in alcoholic pancreatitis patients. Dig Dis Sci 2008;
53:3234–3241.
3. Matos C, Cappeliez O, Winant C, et al. MR imaging of the
pancreas: a pictorial tour. Radiographics 2002;22:e2.
4. Tirkes T, Akisik F, Tann M, et al. Imaging of the pancreas
with secretin enhancement. Top Magn Reson Imaging
2009;20:19–24.
5. Matos C, Winant C, Delhaye M, et al. Functional MRCP in
pancreatic and periampullary disease. Int J Gastrointest
Cancer 2001;30:5–18.
6. Cohen S, Bacon BR, Berlin JA, et al. National Institutes of
Health State-of-the-Science Conference Statement:
ERCP for diagnosis and therapy, January 14-16, 2002.
Gastrointest Endosc 2002;56:803–809.
7. Rahman R, Ju J, Shamma’s J, et al. Correlation between
MRCP and ERCP findings at a tertiary care hospital. W V
Med J 2010;106:14–19.
8. Mosler P, Akisik F, Sandrasegaran K, et al. Accuracy of
magnetic resonance cholangiopancreatography in the
diagnosis of pancreas divisum. Dig Dis Sci 2012;
57:170–174.
9. Freeman ML, Guda NM. Prevention of post-ERCP
pancreatitis: a comprehensive review. Gastrointest
Endosc 2004;59:845–864.
10. Banks PA, Freeman ML. Practice guidelines in acute
pancreatitis. Am J Gastroenterol 2006;101:2379–2400.
11. Sandrasegaran K, Lin C, Akisik FM, et al. State-of-theart pancreatic MRI. AJR Am J Roentgenol 2010;195:42–53.
12. Kahaleh M, Freeman M. Prevention and management of
post-endoscopic retrograde cholangiopancreatography
complications. Clin Endosc 2012;45:305–312.
CLINICAL PANCREAS
September 2014
654
Sherman et al
CLINICAL PANCREAS
13. Axon AT. Endoscopic retrograde cholangiopancreatography in chronic pancreatitis. Cambridge classification.
Radiol Clin North Am 1989;27:39–50.
14. Russo MW, Wei JT, Thiny MT, et al. Digestive and liver
diseases statistics, 2004. Gastroenterology 2004;
126:1448–1453.
15. Carroll JK, Herrick B, Gipson T, et al. Acute pancreatitis:
diagnosis, prognosis, and treatment. Am Fam Physician
2007;75:1513–1520.
16. Swaroop VS, Chari ST, Clain JE. Severe acute pancreatitis. JAMA 2004;291:2865–2868.
17. Mariani A, Arcidiacono PG, Curioni S, et al.
Diagnostic yield of ERCP and secretin-enhanced
MRCP and EUS in patients with acute recurrent pancreatitis of unknown aetiology. Dig Liver Dis 2009;
41:753–758.
18. Fabbri C, Luigiano C, Cennamo V, et al. Complications
of endoscopic ultrasonography. Minerva Gastroenterol
Dietol 2011;57:159–166.
19. Hawes RH. The evolution of endoscopic ultrasound:
improved imaging, higher accuracy for fine needle
aspiration and the reality of endoscopic ultrasoundguided interventions. Curr Opin Gastroenterol 2010;
26:436–444.
20. Anandasabapathy S. Endoscopic ultrasound: indications
and applications. Mt Sinai J Med 2006;73:702–707.
21. Kinney TP, Freeman ML. The role of endoscopic retrograde cholangiopancreatography and endoscopic ultrasound in diagnosis and treatment of acute pancreatitis.
Minerva Gastroenterol Dietol 2005;51:265–288.
Gastroenterology Vol. 147, No. 3
22. Cotton PB, Garrow DA, Gallagher J, et al. Risk factors for
complications after ERCP: a multivariate analysis of 11,
497 procedures over 12 years. Gastrointest Endosc
2009;70:80–88.
23. Testoni PA, Mariani A, Curioni S, et al. Pancreatic ductal
abnormalities documented by secretin-enhanced MRCP
in asymptomatic subjects with chronic pancreatic hyperenzymemia. Am J Gastroenterol 2009;104:1780–1786.
24. Murkle EM. Secretin-stimulated MRCP. Gastroenterol
Hepatol 2005;1:159–160.
Received February 20, 2014. Accepted May 28, 2014.
Reprint requests
Address requests for reprints to: Stuart Sherman, MD, University Hospital,
Room 1634, 550 University Boulevard, Indianapolis, Indiana 46202-5149.
e-mail: [email protected]; fax: (317) 944-2751.
Acknowledgments
The authors thank their fellow principal investigators (listed in the
Supplementary Appendix) for their participation in this study as well as Fiona
Boswell, PhD, and Jan Markind, PharmD, CMPP (Caudex Medical, New
York, NY), for their assistance in the preparation of this manuscript, which
was supported by Repligen Corp.
Conflicts of interest
The authors disclose the following: S.S. is a consultant for and has
received research support from Repligen Corp. M.L.F. is a consultant
for Boston Scientific Corp and Cook Endoscopy. P.R.T. is a consultant for
Boston Scientific. C.M.W. discloses no conflicts. A.K. is a consultant for
Boston Scientific. A.M.A. is a consultant for and has received research support
from Repligen Corp. D.J. is an employee of Repligen Corp. R.A.K. has
received research support from Repligen Corp.
Funding
Supported by Repligen Corp.
September 2014
Secretin-Enhanced MRCP in Pancreatitis 654.e1
Supplementary Appendix
The principal investigators in the study were as follows.
United States: Stuart Sherman, Indianapolis, IN; Richard
Kozarek, Seattle, WA; Martin Freeman, Minneapolis, MN;
Paul Tarnasky, Dallas, TX; Eduard De Lange, Charlottesville,
VA; Qiang Cai, Atlanta, GA; Jason Wills, Salt Lake City, UT;
Abhijit Kulkarni, Pittsburgh, PA; Nalini Guda, Milwaukee,
WI; Ronald Szyjkowski, Syracuse, NY; Sridhar Shankar,
Worcester, MA; John Lee, Orange, CA; Stuart Gordon,
Lebanon, NH; Firas Al-Kawas, Washington, DC; C. Mel Wilcox, Birmingham, AL; Nicholas Nickl, Lexington, KY; Yang
Chen, Aurora, CO; Virendra Joshi, New Orleans, LA; Irving
Screened
(n=283)
Failed screening (n=25)
Reasons for failed screening:
Inclusion criteria not met (n=6)
Exclusion criteria met (n=3)
Withdrew consent (n=8)
Other (n=8)
Enrolled
(n=258)
Early termination (n=6)
Reasons for early termination:
ERCP cancelled (n=4)
Protocol nonadherence (no
scheduled Visit 5) (n=2)
Completed
study
(n=252)
Supplemental Figure 1. Patient disposition. ERCP, endoscopic retrograde cholangiopancreatography.
Waxman, Chicago, IL; Ian Scott Grimm, Chapel Hill, NC;
Cuong Nguyen, Scottsdale, AZ; Orhan Ozkan, Galveston, TX;
Mojtaba Olyaee, Kansas City, KS; Nakechand Pooran, Hershey, PA; Janak N. Shah, San Francisco, CA; Steven A.
Edmundowicz, St Louis, MO; Chris Lawrence, Charleston, SC;
Kostaki G. Bis, Royal Oak, MI; Michael Lipton, Bronx, NY;
Laurence Bailen, Newton, MA; Faisal Bukeirat, Morgantown,
WV; Mark Murphy, Savannah, GA; Gerard Isenberg, Cleveland, OH; David Grand, Providence, RI; Waqar Qureshi,
Houston, TX; Shiro Urayama, Sacramento, CA. Canada:
Masoom Haider, Toronto, ON; Jonathan Love Calgary, AB;
Donald MacIntosh, Halifax, NS.
654.e2
Sherman et al
Gastroenterology Vol. 147, No. 3
Supplemental Table 1.Summary of Sensitivity and Specificity Analyses of Different Evaluable Populations (by Reader)
Sensitivity
Evaluable population
Reader
n
Baseline MRCP (%)
RG1068-ciné MRCP (%)
Difference (%)
P-value
All patients with evaluable ERCP
1
2
3
1
2
3
1
142
49.2
46.6
37.5
48.7
45.6
38.3
56.6
71.0
64.5
59.0
70.8
63.7
58.4
70.3
22.8
17.8
25.5
22.1
18.0
20.1
13.6
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
50.3
40.8
65.8
59.3
15.5
18.6
<.0001
<.0001
Per-protocol MRCP
Patient subset with chronic
pancreatitis
131
73
2
3
Specificity
All patients with evaluable ERCP
Per-protocol MRCP
Patient subset with chronic
pancreatitis
Reader
n
Baseline MRCP (%)
RG1068-ciné MRCP (%)
Difference (%)
Lower limit
1
2
3
1
2
3
1
236
84.3
80.8
87.0
84.1
81.2
87.3
85.4
81.6
81.3
87.8
81.0
81.6
87.2
81.6
2.7
0.5
0.8
3.1
0.4
0.1
3.9
4.9
2.2
2.1
5.4
2.4
3.1
6.5
79.8
83.0
78.2
87.2
1.6
4.2
6.1
1.8
2
3
222
73
ERCP, endoscopic retrograde cholangiopancreatography; MRCP, magnetic resonance cholangiopancreatography.