Gastroenterology 2023;164:241–255

Intestinal Barrier Healing Is Superior to Endoscopic and

Histologic Remission for Predicting Major Adverse
Outcomes in Inflammatory Bowel Disease: The Prospective

INFLAMMATORY BOWEL DISEASE

ERIca Trial
Timo Rath,1 Raja Atreya,1 Julia Bodenschatz,1 Wolfgang Uter,2 Carol E. Geppert,3
Francesco Vitali,1 Sarah Fischer,1 Maximilian J. Waldner,1 Jean-Frédéric Colombel,4
Arndt Hartmann,3 and Markus F. Neurath1,5
1
Department of Gastroenterology, Ludwig Demling Endoscopy Center of Excellence, University Hospital Erlangen, Medical
Clinic 1, Friedrich-Alexander University Erlangen-Nuernberg, Erlangen, Germany; 2Department of Medical Informatics,
Biometry and Epidemiology, Friedrich-Alexander University Erlangen-Nuernberg, Erlangen, Germany; 3Institute for Pathology,
University Hospital Erlangen, Friedrich-Alexander University Erlangen-Nuernberg, Erlangen, Germany; 4Division of
Gastroenterology, Icahn School of Medicine at Mount Sinai, New York, New York; and 5Deutsches Zentrum für Immuntherapie
(DZI), Friedrich-Alexander University Erlangen-Nuernberg, Erlangen, Germany

See editorial on page 184.

A s shown by recent meta-analyses, achieving endo-
scopic remission in patients with inflammatory
bowel diseases (IBD) is associated with improved long-
BACKGROUND & AIMS: Endoscopic and histologic remission term outcome in Crohn’s disease (CD)1 and with long-
have emerged as key therapeutic goals in the management of term steroid-free clinical remission and colectomy-free
inflammatory bowel diseases (IBD) that are associated with survival in ulcerative colitis (UC).2 Therefore, mucosal
favorable long-term disease outcomes. Here, we prospectively healing is a key therapeutic goal in IBD that is advocated
compared the predictive value of barrier healing with endo- by several guidelines for clinical practice and trial end
scopic and histologic remission for predicting long-term disease points.3-7
behavior in a large cohort of patients with IBD in clinical The composite of symptom control and mucosal healing
remission. METHODS: At baseline, patients with IBD in clinical
is commonly referred to a “deep remission,”5,8,9 and a
remission underwent ileocolonoscopy with assessment of intes-
consensus report by the Selected Therapeutic Targets in
tinal barrier function by confocal endomicroscopy. Endoscopic
Inflammatory Bowel Disease (STRIDE) working group rec-
and histologic disease activity, as well as barrier healing, was
ommends both clinical and endoscopic remission as targets
prospectively assessed along established scores. During subse-
quent follow-up, patients were closely monitored for clinical for routine clinical practice,5 and most recently, both
disease activity and the occurrence of major adverse outcomes treatment targets have been confirmed in the STRIDE-II
(MAOs): disease flares, IBD-related hospitalization or surgery, intiative.6
and initiation or dose escalation of systemic steroids, immuno- However, assessing endoscopic disease activity with
suppressants, small molecules, or biological therapy. RESULTS: white light endoscopy (WLE) cannot accurately assess
The final analysis included 181 patients, 100 with Crohn’s dis- histologic disease activity or detect persistent histologic
ease [CD] and 81 with ulcerative colitis (UC). During a mean inflammation in patients with mucosal healing.10–12
follow-up of 35 (CD) and 25 (UC) months, 73% of patients with Apart from mucosal healing, histologic healing is
CD and 69% of patients with UC experienced at least 1 MAO. another emerging end point in patients with IBD that is
The probability of MAO-free survival was significantly higher in frequently included as a secondary end point in clinical
patients with IBD with endoscopic remission compared with trials. Although histologic healing in UC is associated
endoscopically active disease. In addition, histologic remission with better disease outcome compared with clinical
predicted MAO-free survival in patients with UC but not CD.
Barrier healing on endomicroscopy was superior to endoscopic
and histologic remission for predicting MAO-free survival in Abbreviations used in this paper: CD, Crohn’s disease; CDAI, Crohn’s
both UC and CD. CONCLUSIONS: Barrier healing is associated disease activity index; CLE, confocal laser endomicroscopy; CRP, C-
reactive protein; IBD, inflammatory bowel diseases; IO-IBD, International
with decreased risk of disease progression in patients with clin- Organization for the Study of Inflammatory Bowel Disease; MAO, major
ically remittent IBD, with superior predictive performance adverse outcome; MCS, Mayo Clinical Score; MES, Mayo Endoscopy
compared with endoscopic and histologic remission. Analysis of Score; mRiley, modified Riley score; NHI, Nancy histological index; pCLE,
probe-based CLE; RHI, Robarts histopathology index; SES-CD, simple
barrier function might be considered as a future treatment target endoscopic score for Crohn’s disease; STRIDE, Selected Therapeutic
in clinical trials. ClinicalTrials.gov number, NCT05157750. Targets in Inflammatory Bowel Disease; UC, ulcerative colitis.
Most current article
Keywords: Inflammatory Bowel Diseases; Endoscopy; Histology; © 2023 by the AGA Institute.
Intestinal Barrier; Confocal Laser Endomicroscopy. 0016-5085/$36.00
https://doi.org/10.1053/j.gastro.2022.10.014
 242 Rath et al Gastroenterology Vol. 164, No. 2

WHAT YOU NEED TO KNOW Material and Methods

BACKGROUND AND CONTEXT Ethics Approval and Informed Consent
The study was approved by the Friedrich-Alexander Uni-
Achieving endoscopic and histologic remission in patients
versity Erlangen-Nuremberg Ethics Committee and by the
with inflammatory bowel diseases are key therapeutic
goals that are associated with favorable long-term Medical Faculty Institutional Review Board. All patients gave
disease outcome. The relevance of functional healing of their written informed consent before study inclusion. The
the intestinal barrier is less well understood. study was conducted in accordance to the ethical guidelines of
the Declaration of Helsinki.
NEW FINDINGS
INFLAMMATORY BOWEL DISEASE

When directly comparing endoscopic remission,

histologic remission, and barrier healing in patients with Study Design and Participants
inflammatory bowel diseases in this long-term
prospective study, we found that barrier healing was The study was designed as a prospective observational
highly accurate for predicting the further course of study conducted at the Ludwig Demling Endoscopy Center of
disease and outcompeted endoscopic and histologic Excellence and the IBD outpatient department at the University
remission for predicting a survival free of major adverse Hospital of Erlangen. The study was registered at ClinicalTrials.
outcomes. gov with the following identifier: NCT05157750. All authors
had access to the study data and reviewed and approved the
LIMITATIONS
final manuscript.
Although large in size with multiannual follow-up of the The study enrolled adult patients with an established IBD
included patients, this was a single-center study diagnosis for at least 12 months’ duration presenting in clinical
conducted at a tertiary referral center.
remission between January 2017 and December 2019. Written
IMPACT informed consent was obtained from all patients before the
Barrier healing is associated with decreased risk of disease procedure. Patients with poor bowel preparation, total colec-
progression of inflammatory bowel diseases, with superior tomy, concomitant b-blocker therapy (based on the statement
predictive performance compared with endoscopic and by the Bundesinstitut für Arzneimittel und Medizinprodukte in
histologic remission. Analysis of barrier function might be Germany that the intravenous administration of fluorescein to
considered as a future treatment target. patients on b-blocker therapy is relatively contraindicated),
known allergy to fluorescein, or a planned change in IBD-
related medication were excluded. Clinical disease activity for
remission or endoscopic remission, or both, as evi- patients with UC and CD was determined using the partial
denced by several meta-analyses in the field,13–15 data Mayo clinical disease activity score (MCS) and the Crohn’s
on the relevance of histologic healing in CD are limited disease activity index (CDAI),21 with clinical remission defined
to date. Furthermore, histologic scoring in UC is com- as a CDAI <1509,21 or a partial MCS <2 and no individual
plex, with 26 different histopathologic scores, of which subscore >1,21,22 respectively. Additionally, sociodemographic
only 2 are validated,16 and no score is completely factors, current and past medication, and routine laboratory
representative or validated in patients with CD, thereby parameters were assessed at the time of colonoscopy.
limiting determination and histology-based decision After baseline ileocolonoscopy with CLE, patients under
biological therapy were closely monitored in our IBD outpatient
making in clinical practice.
department every 4 to 8 weeks and those under conventional
Confocal laser endomicroscopy (CLE) is a high-
therapy were monitored every 8 weeks. At each visit, clinical
resolution imaging technology that enables subsurface
disease activity using partial MCS and CDAI, respectively,
imaging of the mucosa in real time during ongoing routine laboratory parameters, and current and past medica-
endoscopy. Apart from the possibility to accurately grade tions were assessed. Further, at each visit, major adverse out-
inflammatory activity in patients with IBDs, CLE enables comes (MAOs) defined as (1) disease relapse, (2) IBD-related
functional assessment of the integrity of the intestinal hospitalization, (3) IBD-related surgery, and (4) necessity for
barrier, and pilot studies have shown that barrier initiation or dose escalation of systemic steroids, immunosup-
dysfunction in patients with IBD correlates to clinical pressants, small molecules, or biological therapy was recorded.
disease behavior and long-term disease outcome.17–20
However, no trial to date has systematically and
comparatively evaluated the relevance of endomicroscopic Colonoscopy and Confocal Laser
barrier healing on IBD disease outcomes in a large pro- Endomicroscopy
spective trial. To directly compare the value of endoscopic All patients received bowel preparation with low-volume
remission, histologic remission, and barrier healing for polyethylene glycol-based bowel lavage in a split dose
predicting long-term disease behavior, we conducted the regimen. Colonoscopy was performed using commercially
Endoscopic Remission, Histologic Remission and Barrier available high-definition endoscopes and video processors
Healing for Predicting Disease Behaviour in IBD (ERIca) (EC38-i10 and Optivista EPK-i7010, both Pentax Medical,
trial, a cross-sectional diagnostic study in which a large Tokyo, Japan). Bowel preparation was assessed using the
cohort of patients with IBD in clinical remission were Boston Bowel Preparation Score (BBPS), with poor bowel
prospectively included and closely monitored during long- preparation defined as a BBPS of <2 in any segment or a total
term follow-up for >2 years. BBPS of <6.
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 243

According to consensus statements, endoscopic remission or patient data. For histopathologic scoring in UC, the Robarts
healing during WLE were defined as follows5,23: In UC, endo- histopathology index (RHI)32 and the Nancy histological index
scopic remission was defined as a Mayo Endoscopy Score (MES) (NHI)33 were used as validated histology scores. Histologic
of
1 and endoscopic healing was defined as an MES of 0. In CD, disease remission was defined as an RHI of
3 without lamina
endoscopic remission was assessed along 2 parameters: (1) propria or epithelial neutrophils or an NHI of
1.
absence/resolution of erosions and ulcerations, as a consensual In the absence of a validated score for grading of histologic
definition of endoscopic remission by the IO-IBD5,23 and (2) the inflammation in patients with CD, we used a modified Riley
simplified endoscopic index of severity (SES-CD) with an SES-CD (mRiley) score, as previously described.34 Apart from including
of <3 for definition of endoscopic remission.24,25 the 6 histologic features for UC (acute inflammatory cell infil-

INFLAMMATORY BOWEL DISEASE

For confocal imaging, a dedicated probe-based CLE (pCLE) trate, crypt abscesses, mucin depletion, surface epithelial
imaging system, consisting of a portable laser station (Cellvizio) integrity, chronic inflammatory cell infiltrate, crypt architec-
and confocal miniprobes (ColoFlex UHD, Mauna Kea Technol- tural irregularities), this score integrates typical histologic
ogies, Paris, France), was used. pCLE imaging was performed in features observed in CD, namely, lymphocyte aggregates,
a standardized fashion in all patients: after reaching the ter- granulomas, and number of eosinophils, thereby assessing a
minal ileum, 5 mL fluorescein, 10%, was intravenously injected total of 9 histologic features. This mRiley score ranges from 0 to
as a contrast agent, and imaging was initiated immediately after 27, with 0 to 4 corresponding to histologic remission, a score of
the injection. For this, the CLE probe was positioned under 5 to histologically active disease, including 5 to 9 low activity,
endoscopic guidance onto the mucosa of the terminal ileum, 10 to 18 moderate activity, and 19 high activity.34
cecum, and rectosigmoid junction in patients with CD and in the Supplementary Table 1 presents the composition of the mRi-
cecum and at the rectosigmoid junction in patients with UC. In ley score. The highest scores obtained during histopathologic
patients without endoscopic remission, areas with the highest scoring of all biopsy samples were used to define presence or
degree of inflammation on WLE were additionally examined by absence of histologic remission.
pCLE.
On target tissue, low-powered blue laser light with a 488-
nm wavelength was activated for tissue illumination. At each End Points, Sample Size, and Statistical Analysis
site, a CLE video of at least 2 minutes’ duration was recorded, The primary end point of this study was to comparatively
with an image acquisition rate of 8 frames/s. All pCLE images assess the predictive values of barrier healing, endoscopic
for each patient were stored on an external hard drive and remission, and histologic remission for predicting occurrence of
were independently reviewed for presence of barrier MAO in patients with IBD in clinical remission. As a secondary
dysfunction by 3 expert readers (T.R., J.B., and F.V.) masked to end point, the predictive value of endoscopic healing in UC
the clinical results of the patients. (MES ¼ 0) for predicting the further course of disease was
Barrier dysfunction in the terminal ileum was assessed calculated. As further secondary end point, the predictive
using the semiquantitative Watson score into 3 grades17–19,26–28: values of the composite between endoscopic remission and
(I) intact epithelial barrier with no fluorescein leakage; (II) histologic remission for predicting MAO in patients with IBD
functional barrier defect with shedding of single epithelial cells were assessed.
and fluorescein leakage into the intestinal lumen; or (III) struc- Demographic and other baseline characteristics were sum-
tural barrier defect with shedding of multiple epithelial cells, marized by tabulating (relative) frequencies or providing
exposure of the lamina propria to the lumen, and fluorescein descriptive statistics. Properties of predictive values of
leakage into the lumen. Barrier dysfunction in the colon was different dichotomized characteristics in predicting later
assessed using a dichotomous distinction, as previously occurrence of MAOs were derived. In this context, the positive
described18,28–30: intact epithelial barrier in the colon was predictive value quantifies the share of patients actually not
characterized by a crypt opening that appeared as a dark center experiencing MAOs during follow-up among those with a
in the crypt. During colonic barrier dysfunction, fluorescein favorable status of the predictor. Conversely, the negative
leaked into the crypt lumen; therefore, the lumen was brighter predictive value indicates the share of patients with occurrence
than the surrounding epithelium.18,27,30,31 The integrity of the of MAOs among all patients diagnosed with an unfavorable
barrier was assessed in each image, and barrier dysfunction at predictor status. To statistically test the different degrees of
the specific imaging site was defined as being present when 1 MAOs between patients with and without barrier healing, an
barrier defects were clearly visible on at least 3 consecutive extension of Fisher’s exact test for a 2  4 contingency table
images. was used. Statistical analyses were performed using R 4.0.x
software (www.r-project.org). All statistical tests were consid-
ered explorative, and hence, no a adjustment was used. More-
Histologic Analysis over, time-to-event analysis by Kaplan-Meier estimates was
From each patient, 2 samples for histopathology were ob- used to examine the time to MAOs (or censoring at end of
tained at the sites where CLE imaging was performed. In follow-up) in the 2 respective strata of the predictor variables.
addition, in case macroscopic inflammation was present during
WLE, areas with highest degree of inflammation on WLE were
also biopsied, matching those areas that were also examined by Results
CLE. In case of ulcerations, biopsies and CLE imaging were
performed at the border of the ulcerations. Each biopsy sample Study Inclusion and Flow of Participants
had a registration number with the corresponding pCLE video Between 2017 and 2019, 296 patients with IBD were
sequence. All samples were scored by an experienced gastro- screened for eligibility. The study excluded 94 patients who
intestinal pathologist (A.H.) masked to clinical and endoscopic exhibited clinically active disease and 31 patients due to
 244 Rath et al Gastroenterology Vol. 164, No. 2

Table 1.Clinical, Endoscopic, and Histologic Characteristics Table 1. Continued

of the Patients With Ulcerative Colitis and Those
With Crohn’s Disease Ulcerative Crohn’s
colitis disease
Ulcerative Crohn’s Variables (n ¼ 81) (n ¼ 100)
colitis disease
Variables (n ¼ 81) (n ¼ 100) Barrier healing present 25 (25)
Histopathology scoring
Clinical characteristics RHI
3 44 (54.3)
Age, y 39 (18–69) 37 (19–68) RHI > 3 37 (45.7)
INFLAMMATORY BOWEL DISEASE

Sex NHI <1 42 (51.9)

Male 39 (48) 58 (58) NHI 1 39 (48.1)
Female 42 (52) 42 (42) Modified Riley score <5 58 (58)
Body mass index, kg/m2 25.6 (17.2–39.2) 26.8 (16–51.9) Modified Riley score 5 42 (42)
Disease duration, y 10 ± 7.9 12.5± 11.9 Erosions or ulcerations
Extent of disease Absent 66 (66.0)
Proctitis 6 (7.4) Present 34 (34.0)
Left–sided colitis 39 (48.1) SES-CD
Pancolitis 36 (44.4) <3 50 (50.0)
Ileum 29 (29) 3 50 (50.0)
Colon 8 (8) Follow-up, mon 25 ± 11.9 35 ± 6.9
Ileocolitis 42 (42)
a
Upper gastrointestinal þ 7 (7) MAO during follow-up
ileum No MAO 25 (30.9) 37 (37)
Upper gastrointestinal þ 2 (2) Occurrence of MAOs
colon
2 months 29 (35.8) 23 (23)
Upper gastrointestinal þ 12 (12) >2 to
4 months 12 (14.8) 12 (12)
ileocolitis >4 to
6 months 8 (9.9) 6 (6)
Extraintestinal 19 (23.5) 31 (31) >6 to
8 months 5 (6.2) 7 (7)
manifestations >8 to
10 months 2 (2.5) 5 (5)
Primary sclerosing 2 (2.5) > 10 to
12 months 0 (0) 4 (4)
cholangitis >12 months 0 (0) 6 (6)

Medication
Mesalamine derivates NOTE. Continuous data are presented as n (%) and cate-
Mesalazine 12 (14.8) 5 (5) gorical data as mean ± standard deviation or mean (range).
Sulfasalazine 2 (2) a
MAOs: disease flare; necessity for initiation or escalation of
Corticosteroids systemic steroids, immunosuppressants, small molecules, or
Budesonide 1 (1) biological therapy.
Budesonide (with colonic 2 (2.5)
delivery)
Prednisolone 4 (4.9) 3 (3)
Prednisolone dose, mg 10 ± 4 5 ± 4.3
poor bowel preparation (n ¼ 18), concomitant b-blocker
Immunomodulator therapy (n ¼ 7), unwillingness to participate in the study
6-Mercaptopurin 1 (1.2) 1 (1) (n ¼ 6), or a planned change in pharmacotherapy (n ¼ 5).
Azathioprine 4 (4.9) 8 (8) Therefore, 181 patients with IBD (CD, n ¼ 100; UC, n ¼ 81)
Biological therapy were finally eligible and included in the study. A flowchart
Anti-tumor necrosis factor 28 (34.6) 43 (43) of the included patients according to the Standards for
Vedolizumab 11 (13.6) 5 (5)
Reporting of Diagnostic Accuracy Studies guidelines35 is
Tofacitinib 3 (3.7) 0 (0)
Ustekinumab 2 (2.5) 17 (17) presented in Supplementary Figure 1.
Combination therapy 8 (9.9) 5 (5) The study included patients with clinical remission, as
No medication 6 (7.4) 10 (10) determined by established clinical activity scores (CDAI
Laboratory parameters
<150 or partial MCS <2 with no individual subscore
Leukocyte count, 109/L 7.9 ± 3.2 8 ± 2.8 >1).5,21,22 Most patients were treated with biological ther-
Hematocrit, % 41.5 ± 4.1 42.3 ± 3.8 apies, and only few patients had mesalamine therapy or low
doses of corticosteroids (Table 1).
Endoscopic and
histopathologic data
MES

1 43 (53.1) Clinical Characteristics and Rates of Endoscopic
>1 38 (46.9) and Histologic Remission and Barrier Healing in
Barrier function Patients With Ulcerative Colitis
Colon
The 81 patients with UC had nonmissing, valid infor-
Barrier healing present 21 (25.9) 27 (27)
Ileum mation regarding the occurrence of MAOs during follow-up.
Data on endoscopic remission and healing and histologic
remission were available for all patients. In 1 patient with
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 245

UC, electronic backup of CLE images failed; hence, no data definition of endoscopic healing was applied, considering
on barrier function in the colon were available for this pa- only patients with an MES of 0, 17 patients with UC
tient. Clinical, endoscopic, and histologic characteristics of exhibited mucosal healing. Of these 17 patients with
the patient cohort with UC are summarized in Table 1. mucosal healing on WLE, 6 experienced MAOs during the
As presented in Table 1, from the 81 patients with UC, course of follow-up, thereby the rate of MAOs in patients
43 (53.1%) had endoscopic remission on WLE at study in- with MES of 0 was 35.3% (Supplementary Table 4). Corre-
clusion. Histologic healing, as defined by RHI and NHI, was spondingly, the probability of remaining without an MAO
present in 54.3% and 51.9%, respectively, during endo- during follow-up was significantly higher in patients with
scopic evaluation. As assessed by the RHI, 36 patients UC with endoscopic healing compared with those with an

INFLAMMATORY BOWEL DISEASE

(44.4%) with UC exhibited endoscopic and histologic heal- MES of >0 (P ¼ .0018) (Figure 1B).
ing. In contrast, an intact barrier function, as defined by lack From the 44 patients with UC with histologic remission,
of fluorescein leakage into the crypt lumen,18,27,30,31 was as defined by the RHI, 23 developed MAOs during follow-up
observed in the colon in only 21 patients (25.9%) during (RHI MAO rate: 52.3%) (Supplementary Table 4), whereas
baseline evaluation with WLE and CLE. MAOs occurred during the course of follow-up in 21 of 42
Clinical, endoscopic, and histologic characteristics of UC patients with histologic remission, as defined by the NHI
patients with and without an intact barrier are summarized (NHI MAO rate: 50%) (Supplementary Table 4).
in Supplementary Table 2. Of the 21 patients with UC with On Kaplan-Meier analysis, patients with histologic
colonic barrier healing, 8 (38.1%) exhibited an MES of 0, remission, as defined by RHI or NHI, had a significantly
whereas 13 (61.9%) had an MES of 1. All patients (n ¼ 8) higher likelihood of remaining without MAOs during follow-
with barrier healing and an MES of 0 also exhibited histo- up compared with patients with UC with histologically
logic remission. Clinical, endoscopic, and histologic charac- active disease (both P < .0001) (Figure 1C and D). From
teristics of patients with UC with combined colonic barrier those 36 patients with combined histologic (as defined by
healing, endoscopic, and histologic remission compared the RHI) and endoscopic remission, 16 experienced MAOs
with those without are summarized in Supplementary during study follow-up (MAO rate, 44.4%) (Supplementary
Table 3. Table 4), and likewise, those patients with combined
In additional studies, we assessed whether the presence endoscopic and histologic remission had a significantly
of barrier healing was associated with changes of serum better course of disease remaining free of MAO on Kaplan-
parameters. We determined C-reactive protein (CRP), al- Meier estimates (P < .0001) (Supplementary Figure 4A).
bumin, and zonulin levels and noted that the serum levels Of the 17 patients with combined histologic remission (as
did not significantly differ between patients with UC with defined by the RHI) and endoscopic healing (as defined by
intact colonic barrier compared with those with colonic an MES of 0), 5 developed MAOs during follow-up, leading
barrier dysfunction (Supplementary Figure 2). Furthermore, to an MAO rate of 29.4% (Supplementary Table 4), and
serum zonulin levels were decreased in patients with correspondingly, patients with combined endoscopic heal-
endoscopic or histologic remission, whereas albumin and ing and histologic remission had a significantly higher
CRP levels were unchanged between UC patients with likelihood for remaining free of MAOs on Kaplan-Meier es-
endoscopic or histologic remission compared with those timates (P ¼ .00039) (Supplementary Figure 4B).
without (Supplementary Figure 3A). In contrast to the aforementioned results, only 4 of 21
patients with UC with barrier healing in the colon developed
MAOs during follow-up, all of which were MAO IV
Follow-up and Occurrence of Major Adverse (Supplementary Table 5); hence, the MAO rate in patients
Outcomes in Patients With Ulcerative Colitis with colonic barrier healing was 19.1% (Supplementary
Mean follow-up in patients with UC was 25 months Table 4). Consistent with this, those patients with UC with
(Table 1). In 25 patients with UC, no MAOs occurred in the barrier healing in the colon had a significantly more favorable
course of follow-up, whereas in the remaining 56 patients, course of disease as shown by Kaplan-Meier analysis (P <
MAOs were noted, with a mean deviation lag of 3.2 (stan- .0001) (Figure 2). Conversely, in most patients experiencing
dard deviation, 52.5 months; range, 1–10 months) as fol- an MAO, no barrier healing was present, and the distribution
lows: MAO I in 16, MAO II in 3, MAO III in 7, and MAO IV of individual MAOs did not differ significantly between UC
in 30. patients with and without barrier healing (P ¼ .2)
Rates of the occurrence of MAOs in patients with endo-
scopic and histologic remission and in patients with barrier
healing are summarized in Supplementary Table 4. As Clinical Characteristics and Rates of Endoscopic
shown, of the 43 patients with endoscopic remission at and Histologic Remission and Barrier Healing in
study inclusion, 21 experienced MAOs; hence the rate of Patients With Crohn’s Disease
MAOs in patients with UC was 48.8%. Time-to-event anal- The 100 patients with CD had nonmissing, valid infor-
ysis using Kaplan-Meier estimates showed that the proba- mation regarding the occurrence of MAOs during follow-up.
bility of remaining free of MAOs during follow-up was Further, data on endoscopic and histologic remission were
significantly higher in patients with UC with endoscopic available for all patients. Because of technical defects in
remission than in patients with endoscopically active dis- image recording, data on barrier function in the terminal
ease (P < .0001) (Figure 1A). When a more stringent ileum were not available in 2 patients with CD. Clinical,
 246 Rath et al Gastroenterology Vol. 164, No. 2
INFLAMMATORY BOWEL DISEASE

Figure 1. Time-to-event analysis for the occurrence of major adverse outcomes in patients with UC with endoscopic remis-
sion, endoscopic healing, and histologic remission. (A) Patients with UC with endoscopic remission (MES of
1) had a
significantly higher probability of remaining free of MAOs compared with patients with endoscopically active disease. (B) In
patients with UC with endoscopic healing (MES of 0), the probability of remaining without MAOs during follow-up was
significantly higher compared with those with a MES of >0. (C) Patients with UC with histologic remission, defined by an RHI of

3, had a significantly higher likelihood of remaining without MAOs during follow-up compared with patients with histologically
active disease according to the RHI. (D) In patients with UC histologic remission, as determined by an NHI of
1, the
probability of remaining without a MAO during follow up was significantly higher compared with those with an NHI of >1. The
shaded areas indicate the 95% confidence interval. (C and D: hematoxylin and eosin staining, original magnification 20.)
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 247

INFLAMMATORY BOWEL DISEASE

Figure 2. Time-to-event analysis for the
occurrence of MAOs in patients with UC
with barrier healing. Patients with an
intact colonic barrier had a significantly
higher probability of remaining without
MAOs during follow-up compared with
patients with barrier dysfunction in the
colon. The shaded areas indicate the
95% confidence interval.

endoscopic, and histologic characteristics of the patient those with ileal or colonic barrier dysfunction
cohort with CD are summarized in Table 1. (Supplementary Figure 2). Furthermore, apart from a
As presented in Table 1, from 100 included patients with decreased serum expression of zonulin in patients with CD
CD, 66 (66%) had endoscopic remission on WLE, as defined by with endoscopic remission, serum albumin and CRP levels
the absence of erosions or ulcerations, or both, at study inclu- were not significantly different between CD with endoscopic
sion. When the SES-CD with an SES-CD of <3 was used for the or histologic remission compared with those without,
definition of endoscopic remission,24,25 50 patients (50%) with respectively (Supplementary Figure 3B).
CD exhibited endoscopic remission during baseline endoscopy.
Histologic remission, as defined by a mRiley score as
previously reported,34 was present in 58 patients (58%) Follow-up and Occurrence of Major Clinical
with CD during the endoscopic evaluation. The combination Events in Patients With Crohn’s Disease
of endoscopic and histologic healing was observed in 49 Mean follow-up in patients with CD was 35 months. No
patients with CD. In contrast, an intact barrier on endomi- MAOs occurred in 37 patients during the follow up, whereas
croscopy without fluorescein leakage was observed in only MAOS were noted in the remaining 63 patients with a mean
25 of 100 patients (25%) with CD in the terminal ileum and lag of 6 months (standard deviation, 6.9 months; range, 1–
in 27 patients (27%) in the colon during the baseline 38 months) as follows: MAO I in 12, MAO II in 4, MAO III in
evaluation with WLE and CLE. Clinical, endoscopic, and 12, and MAO IV in 35.
histologic characteristics of patients with CD with ileal Rates of the occurrence of MAO in patients with endo-
barrier healing compared with those without barrier healing scopic and histologic remission and in patients with barrier
are summarized in Supplementary Table 6. Supplementary healing are summarized in Supplementary Table 8. Of the 66
Table 7 summarizes characteristics of those patients with patients with endoscopic remission at study inclusion,
CD with combined ileal barrier healing and endoscopic and defined by the absence of erosions or ulcerations, 37
histologic remission compared with those without. experienced MAOs; hence the rate of MAOs in patients with
Moreover, we determined CRP, albumin, and zonulin CD in endoscopic remission was 56.9%. When the SES-CD
levels in patients with CD and noted that their serum levels was used to define endoscopic remission, 25 of 50 pa-
did not significantly differ between patients with intact ileal tients with an SES-CD of <3 experienced MAO; therefore,
and colonic barrier on endomicroscopy compared with the MAO rate for SES-CD was 50% (Supplementary Table 8).
 248 Rath et al Gastroenterology Vol. 164, No. 2
INFLAMMATORY BOWEL DISEASE

Figure 3. Time-to-event analysis for the occurrence of MAOs in patients with CD with endoscopic remission and histologic
remission. (A) Patients with endoscopic remission, as defined by the absence of erosions and/or ulcerations, had a signifi-
cantly higher probability of remaining free of MAOs compared with those patients with endoscopically active disease. (B) In
patients with CD with endoscopic remission, as defined by an SES-CD of <3, the probability of remaining without MAOs
during follow-up was significantly higher compared with those with an SES of 3. (C) Patients with CD with histologic
remission, defined by an mRiley score of
4, had a significantly higher likelihood of remaining without MAOs during follow-up
compared with patients with histologically active disease. The shaded areas indicate the 95% confidence interval. (C: he-
matoxylin and eosin staining, original magnification 20.)
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 249

Time-to-event analysis using Kaplan-Meier estimates probabilities for remaining without MAOs during follow-up,
showed that the probability of remaining free of MAOs we sought to directly compare the diagnostic performances
during follow-up was significantly higher in patients with of endoscopic healing, histologic healing, and barrier healing
CD with endoscopic remission compared with patients with for the prediction of the further course of disease.
endoscopically active disease (P ¼ .0084) (Figure 3A), with In UC, endoscopic remission, as defined by an MES of
1,
slight superiority of assessing endoscopic remission with had an overall accuracy of 70.4% for predicting an MAO-free
the SES-CD (P ¼ .0049) (Figure 3B) compared with the course of disease, with positive and negative predictive
definition based on absence of erosions or ulcerations. Of values of 51.2% and 92.1%, respectively (Table 2). When a
the 58 patients with CD with histologic remission, as defined more stringent definition of mucosal healing was applied,

INFLAMMATORY BOWEL DISEASE

by an mRiley score of <5,34 32 developed MAOs during considering only patients with an MES of 0, overall accuracy
follow-up; hence, the MAO rate in patients with CD with for predicting a MAO-free course of disease was increased,
histologic remission was 55.2% (Supplementary Table 8). with an accuracy of 75.3% and positive and negative pre-
Kaplan-Meier analysis showed no significant differences in dictive values of 64.7% and 78.1%, respectively (Table 2).
the probability of remaining without MAOs during follow-up Histologic remission, as defined by the RHI, had an ac-
in patients with CD with and without histologic remission curacy of 66.7%, with negative and positive prediction for
(P ¼ .089) (Figure 3C). the occurrence of MAOs during follow-up of 47.7% and
Of the 49 patients with CD with combined histologic 89.2%, respectively. With an overall accuracy of 69.1% and
remission and endoscopic healing, 25 experienced MAOs positive and negative predictive values of 47.7% and 89.2%,
during study follow-up, for an MAO rate of 51% respectively, assessment of histologic remission as defined
(Supplementary Table 8). Kaplan-Meier analysis revealed a by the NHI score was comparably accurate in prediction of a
significantly higher probability of remaining free of MAOs in MAO-free course of disease (Table 2).
patients with endoscopic and histologic remission compared When the combination of endoscopic remission, as
with those without (P ¼ .0022, Supplementary Figure 5). defined by an MES of
1, and histologic remission, as
Considering only patients with combined histologic remis- assessed by the RHI was used, overall accuracy for pre-
sion and endoscopic ileal remission, as defined by an SES-CD dicting the occurrence of MAOs was increased to 74.1%,
subscore of 0 in the ileum, 16 of 35 patients developed with positive and negative predictive values of 55.6% and
MAOs during follow-up, leading to an MAO rate of 45.7% 88.9%, respectively (Supplementary Table 10). When a
(Supplementary Table 8). more stringent definition for endoscopic remission was
In contrast to the aforementioned results, no MAOs used, including only patients with an MES of 0 (ie, endo-
occurred during follow-up in the 25 patients with CD with scopic healing) and histologic remission as assessed by the
barrier healing in the terminal ileum; hence, the MAO rate in RHI, overall accuracy for predicting the occurrence of MAOs
patients with barrier healing in the ileum was 0% was further increased to 77.8% (Supplementary Table 10).
(Supplementary Table 8). Of the 27 patients with colonic In contrast, diagnostic performances of barrier integrity
barrier healing, 8 exhibited MAOs during follow-up, most of as a marker for the prediction of long-term disease behavior
which were MAO III and MAO IV (Supplementary Table 9). were markedly increased. As such, barrier healing in the
Hence, the MAO rate for patients with colonic barrier healing colon had an overall accuracy of 85%, with a positive and
was 29.6% (Supplementary Table 8). Consistent with this, negative predictive values of 81% and 86.4%, respectively.
patients with CD with barrier healing in the terminal ileum or Diagnostic performances of endoscopic healing, histologic
in the colon had a significantly higher probability for MAO-free healing, and barrier healing for assessing the occurrence of
course of disease compared with patients with CD without MAO in patients with UC are summarized in Table 2.
barrier healing (P < .0001 and P ¼ .00017, respectively) In CD, endoscopic remission, defined by the absence of
(Figure 4). Conversely, in most patients with CD experiencing erosions or ulcerations, exhibited an overall accuracy of
an MAO, no barrier healing was present, and the occurrence of 54% for predicting MAOs during the course of follow-up,
individual MAOs did not differ significantly between patients with a positive predictive value of 43.9% and negative
with CD with vs without barrier healing (P ¼ .2). predictive value of 73.5% (Table 2). Using the SES-CD
As shown in comparative Kaplan-Meier estimates for slightly improved diagnostic accuracy: for an SES-CD of
endoscopic remission, histologic remission, and barrier <3, overall accuracy for predicting MAOs was increased to
healing, barrier healing in the terminal ileum and the colon 62%, with a positive and negative predictive values of 50%
were superior compared with endoscopic and histologic and 74%, respectively (Table 2).
remission, or the combination of the latter, for predicting an Histologic remission, as defined by a mRiley score of
MAO-free course of disease during long term follow-up in <5,34 had an accuracy of 56%, with positive and negative
both UC and CD patients (Supplementary Figure 6). predictive values of 44.8% and 71.4%, respectively
(Table 2). When the combination of endoscopic and histo-
logic remission was used, diagnostic performance was only
Diagnostic Performances of Endoscopic Healing, slightly increased compared with histology alone
Histologic Healing, and Barrier Healing for the (Supplementary Table 11).
Prediction of the Course of Disease Similar to the results observed in UC, diagnostic per-
Based on the observed low MAO rates in CD and UC formances of barrier healing were superior to endoscopic
patients with intact barrier function and the high and histologic remission for predicting the occurrence of
 250 Rath et al Gastroenterology Vol. 164, No. 2
INFLAMMATORY BOWEL DISEASE

Figure 4. Time-to-event analysis for the occurrence of MAOs in patients with CD with barrier healing. (A) Patients with CD with
an intact colonic barrier had a significantly higher probability of remaining without MAOs during follow-up compared with
patients with barrier dysfunction in the colon. (B) In patients with CD with barrier healing in the terminal ileum, the probability of
remaining without MAOs during follow-up was significantly higher compared with those with a barrier defect in the terminal
ileum, as defined by the semiquantitative Watson grading.18,19 The shaded areas indicate the 95% confidence interval.

MAOs during the course of follow-up. As such, barrier symptoms, finding that impaired barrier function, as eval-
healing in the terminal ileum exhibited an overall accuracy uated by CLE, was significantly correlated with severity of
of 88.7% for predicting MAOs in patients with CD, with diarrhea in UC and CD.17 From these observations, the au-
positive and negative predictive values of 100% and 84.7%, thors speculated that resolution of mucosal permeability
respectively (Table 2). In the colon, barrier healing had an beyond mucosal healing might improve outcomes of pa-
accuracy of 72.7%, with positive and negative predictive tients with IBD.17
values of 70.4% and 73.6%, respectively. Diagnostic per- Based on this evidence, we hypothesized that barrier
formances for endoscopic healing, histologic healing, and healing, as assessed by dynamic monitoring of the intestinal
barrier healing for predicting MAO-free course of disease in barrier with CLE, might serve as an accurate parameter that
CD patients are summarized in Table 2. can predict long-term disease behavior in patients with IBD.
For this purpose, we conducted a large prospective study in
which patients with IBD in clinical remission were included
Discussion with subsequent close-meshed and multiannual follow-up,
Impaired barrier function and increased intestinal during which major clinical events were recorded. This
permeability are increasingly recognized as pivotal patho- study used established clinical scores, such as CDAI and
genic factors in IBD.36 Consistent with impaired barrier MCS, rather than biochemical markers or therapeutic regi-
function in patients with IBD, evidence from basic science mens to define clinical remission.5,22 Patients had normal or
studies has revealed impairments in tight junction function only slightly elevated CRP levels, with absent or subclinical,
and epithelial resistance in UC and CD patients,37–39 and mild systemic inflammation.
importantly, alterations in barrier function were found To gain broad insights into the diagnostic and predictive
evenly distributed and independent of focal lesions, such as capabilities of barrier assessment in these patients, we
erosions or ulcerations, in patients with CD.39 Recent comparatively assessed barrier healing against other estab-
studies also have successfully implemented CLE for dynamic lished or emerging treatment end points such as endoscopic
structural and functional assessment of the intestinal bar- and histologic healing. Our data clearly show that barrier
rier in vivo in patients with UC and in those with CD18,19 and healing, especially when present in the terminal ileum, is a
further substantiated the observation that impaired barrier prognostic parameter that by far outcompetes endoscopic and
function is indicative of relapsing disease behavior. Just histologic remission, or their combination, in forecasting the
recently, a prospective study in patients with IBD with occurrence of major clinical events in both UC and CD patients.
endoscopic mucosal healing was able to associate increased In UC, endoscopic remission is a key therapeutic goal
intestinal permeability with persistence of clinical that, as corroborated by several studies in the field including
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 251

Table 2.Diagnostic Performances of Endoscopic Remission, Histologic Remission, and Barrier Healing for Predicting Major
Adverse Outcomes in Ulcerative Colitis and Crohn’s Disease Patients

Accuracy Sensitivity Specificity PPV NPV

Parameter % (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI)

Ulcerative colitis
Endoscopic remission (MES <1) 70.4 (59.2–80) 88 (68.8–97.5) 62.5 (48.6–75.1) 51.2 (42 -60.2) 92.1 (79.8–97.2)
Endoscopic healing (MES ¼ 0) 75.3 (64.5–84.2) 44 (24.4–65.1) 89.3 (78.1–96) 64.7 (43.3–81.5) 78.1 (71.4–83.7)
RHI remissiona 66.7 (55.3–76.8) 84 (63.9–95.5) 58.9 (45–71.9) 47.7 (39–56.6) 89.2 (76.6–95.4)

INFLAMMATORY BOWEL DISEASE

NHI remissionb 69.1 (57.9–78.9) 84 (63.9–95.5) 62.5 (48.6-75.1) 50 (40.6–59.4) 89.7 (77.7–95.7)
Barrier healing–colon 85 (75.3–92) 68 (46.5–85.1) 92.7 (82.4–98) 81 (61.4–91.9) 86.4 (78.2–91.9)
Crohn’s disease
Endoscopic remissionc 54 (43.7–64.2) 76.3 (59.8–88.6) 40.3 (28.1–53.6) 43.9 (37.4–50.7) 73.5 (59.3–84.1)
SES-CDd 62 (51.8–71.5) 65.8 (48.7–80.4) 59.7 (46.5–72) 50 (40.6–59.4) 74 (63.6–82.2)
Histologic remissione 56 (45.7–65.9) 68.4 (51.4–82.5) 48.4 (35.5–61.4) 44.8 (37–52.9) 71.4 (59.4–81)
Barrier healing–colon 72.7 (62.9–81.2) 504 (33.4–66.6) 86.9 (75.8–94.2) 70.4 (53.7–83) 73.6 (66.7–9.6)
Barrier healing–ileum 88.7 (80.6–94.9) 69.4 (51.9–83.7) 100 (94.1– 100) 100 84.7 (77.2–90.1)

CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value.
a
Histologic remission according to the RHI.
b
Histologic remission according to the NHI.
c
Endoscopic remission was defined as absence of erosions and ulcerations.
d
SES-CD <3 ¼ endoscopic remission.
e
Histologic remission was defined according to a mRiley score.

meta-analysis, is associated with favorable disease behavior patients from 15 studies, a meta-analysis by Park et al14
such as long-term corticosteroid-free clinical remission and quantified a risk reduction for clinical relapse in patients
colectomy-free survival.2,40 However, although an MES of with histologic remission of 52%, with superiority of his-

1 is generally accepted to indicate endoscopic remission, tologic remission over clinical and endoscopic remission in
decisive differences exist between patients with an MES of predicting clinical outcomes. Similarly, a recent meta-
0 and 1, as most recently evidenced in a meta-analysis of 17 analysis of 10 studies with patients with endoscopic heal-
studies including 2608 patients with UC in clinical remis- ing showed that patients with UC who achieved histologic
sion, which showed that patients with an MES of 0 had a remission had a 63% lower risk of clinical relapse
52% lower risk of clinical relapse compared with patients compared with patients with persistent histologic
with an MES of 1.15 To reflect these differences, we chose to activity.15
analyze both endoscopic remission (MES of
1) as well as In our study, histopathologic scoring was performed
endoscopic healing (MES of 0) for the prediction of long- using the RHI and the NHI, 2 of the most commonly used
term disease outcome in our study. As shown in time-to- scores, both of which are also validated. As shown in our
event analysis, endoscopic remission and endoscopic heal- study and consistent with data in the literature, histologic
ing were both associated with a significantly more favorable remission, as quantified by RHI and NHI, was associated
course of disease over a mean follow-up period of 25 with significantly lower risk of remaining with major clinical
months. As expected, patients with endoscopic healing were events in our cohort with UC compared with those patients
less likely to experience major clinical events compared who had histologically active disease. Our analysis on the
with patients with endoscopic remission, as shown in the diagnostic performances for predicting disease outcome
Kaplan-Meier analysis. showed both scores were comparably accurate in fore-
A similar result was obtained in CD. Assessing endo- casting the further course of disease. Furthermore, time-to-
scopic remission along an IO-IBD consensus definition that event analysis showed that the combination of endoscopic
is based on the absence or resolution of erosions and ul- remission and histologic remission, as assessed by the RHI,
cerations5,23 and the SES-CD, we found a significantly more increased the predictive values compared with histology or
favorable course of disease over a mean follow-up period of endoscopy alone, although this increase was only incre-
35 months in patients with endoscopic remission, with su- mental in our cohort.
periority of the SES-CD over the IO-IBD definition. Together, We did not find any differences in predicting the long-
these data are consistent with various studies, including term disease outcome between patients with CD with and
meta-analyses, showing that endoscopic remission or without histologic remission. Certainly, assessing histologic
mucosal healing is associated with long-term clinical remission is more complex in CD than in UC for several
remission and reduced need for surgical intervention in reasons. Firstly, no scoring system for assessing histologic
active CD.1,40,41 disease activity in CD has been validated to date.
Histologic remission represents an emerging end point Secondly, due to the discontinuous character of the
in IBD, particularly in UC. Analyzing data from >1500 disease with frequently patchy and focal distribution of
 252 Rath et al Gastroenterology Vol. 164, No. 2

inflammatory lesions, CD is heterogenous, thereby The molecular reasons for these differences between
rendering CD more prone to sampling artifacts compared ileal and colonic disease are currently unclear but might be
with UC. To circumvent these challenges at least partly, we related to the local micromilieu. For instance, there are
took an approach where we obtained biopsy specimens not striking differences in the composition of the mucus layer
only from the site of CLE imaging, but also, if present, from between the ileum and the colon. The small intestine is
those parts of the ileum or colon exhibiting the most severe covered by a single, removable mucus layer that is pene-
inflammation on WLE. Clearly, in case histopathologic trable, with protection provided by antibacterial mediators,
scoring was different between biopsy samples from a single whereas there is a double mucus layer in the colon, in which
patient, the highest score was used for further analysis. To the inner layer is impenetrable to bacteria.46 Furthermore,
INFLAMMATORY BOWEL DISEASE

address the lack of a validated scoring system for CD, we ileum and colon are characterized by a distinct T-cell profile
made use of a comprehensive score with 6 histologic fea- and cytokine signature, with a predominant T helper 1 cells
tures previously described by Riley et al42 in combination profile in the colon and a mixed T helper 1 cell/17 cell
with typical histologic findings observed in CD (ie, profile in the ileum of patients with CD.47,48 Clearly, further
lymphocyte aggregates, granulomas, and eosinophils), so studies are needed to specifically investigate the molecular
that a score with 9 features was used, as previously re- mechanisms that drive barrier integrity in the ileum and
ported.34 However, in light of the lack of differences in long- colon and their implications in determining the further
term outcome between patients with CD with and without course of disease.
histologic remission, as observed in our study, and also the Nevertheless, the high positive and negative predictive
relatively weak differences between patients with and values for barrier healing in the ileum in both diseases
without endoscopic remission, these results indicate that might directly translate into clinical decision making. In this
additional parameters are needed in CD for more accurate context, high positive prediction might help in risk-
forecasting of future disease behavior. stratifying those patients in whom a complicated disease
Barrier healing on functional CLE imaging was superior behavior will occur with high probability, whereas, based on
in its ability to predict the further course of disease in both the high negative prediction, assessment of barrier healing
UC and CD, as shown by comparative analyses of the diag- in the ileum at the same time might allow the identification
nostic performances of the various parameters assessed in of patients in which the occurrence of major clinical event is
this study. Especially in the terminal ileum, barrier healing unlikely to occur. With this, we postulate that functional
had a considerably improved diagnostic performance assessment of the integrity of the intestinal barrier with CLE
compared with endoscopic and histologic remission: overall is a powerful parameter that when used in addition to
accuracy in UC and CD exceeded or closely reached the 90% established or emerging parameters, such as endoscopic and
threshold. Of note, barrier healing in the terminal ileum in histologic remission, may significantly extend the predict-
CD further exhibited a perfect specificity and a perfect ability of future disease behavior.
positive predictive value (both 100%). However, it has to be Consistent with results from our study, previous clinical
kept in mind that 83% of patients with CD included in this studies observed an increased intestinal permeability as
study had ileal involvement. assessed by CLE imaging despite the lack of macroscopic
Against the background that data from clinical behavior, inflammation in patients with IBD.17–19,49 In this regard, it is
epidemiology, genetics, and the gut microbiota suggest that noteworthy that impaired barrier dysfunction has been
ileal and colonic CD should be regarded as at least 2 noted also in nonulcerated epithelia from patients with IBD
different subtypes of CD,43 further studies need to clarify with even distribution of barrier function alterations in
whether barrier function in the ileum can forecast disease patients with CD.39 Furthermore, noninflamed ileum from
behavior equally well in ileal and colonic CD. The higher patients with CD exhibits increased permeability to large
diagnostic accuracy of ileal barrier healing over colonic proteins,50 and even in histologically unaffected ileal tissue
barrier healing for forecasting the further course of disease from patients with CD, increased epithelial uptake of protein
in patients with CD is especially interesting against the antigens has been found to be mediated by tumor necrosis
background that the development of a penetrating disease factor.51
phenotype is significantly higher in patients with ileal dis- The identification of surrogate markers for intestinal
ease compared isolated colonic disease.44 Furthermore, the barrier healing on endomicroscopy requires further in-
cumulative probability of progression from Montreal clas- vestigations. Here, we observed no correlation between
sification phenotype B1 to B2 and B3 is substantially higher barrier healing and serum levels of albumin, CRP, and
in patients with ileal disease (68%) than in those with zonulin. These findings suggested that the presence of in-
colonic disease (23%).44 testinal barrier healing cannot be predicted by these protein
Isolated ileal involvement is also associated increased serum markers.
risk of developing an intestinal complication compared with Furthermore, the structural alterations of the intestinal
patients with CD with isolated colonic involvement.45 These barrier in patients with leakage and absent barrier healing
data clearly indicate that ileal CD is more prone to occur- need future analyses. Although the association between
rence of disease complications than isolated colonic CD, and impaired barrier function and IBD was noted >30 years
it is tempting to speculate that integrity of the ileal barrier ago,52–55 it is unclear to date whether the leakage mainly
could be important to prevent ileal disease-associated occurs through the tight junction, and if so, what tight
complications in CD. junction alterations are present in patients with barrier
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 253

dysfunction. Most studies on the paracellular route of predictive capabilities of barrier function might well exceed
transport describe at least 2 populations of pores regulated established or emerging parameters such as endoscopic and
by tight junctions: (1) the high-capacity charge-selective histologic remission. Therefore, CLE-based dynamic moni-
“pore” pathway allowing paracellular passage of small ions toring of the intestinal barrier during routine ileocolono-
and (2) the low-capacity “leak” pathway permeable to large scopy might be a helpful tool in clinical practice for risk-
ions and molecules irrespective of charge.56 At the molec- stratifying patients with IBD and predicting complicated
ular level, the first pore is mainly regulated by claudins and disease behavior. Finally, our findings suggest that analysis
the latter by the tight junction proteins occludin and the of barrier function might be considered as a future treat-
zonula occludens family, and given these considerations, it ment target in clinical trials.

INFLAMMATORY BOWEL DISEASE

seems clear that future studies implementing clinical tests
assessing the integrity of the intestinal barrier in patients
with IBD in combination with basic science or molecular Supplementary Material
approaches to evaluate barrier structure are highly Note: To access the supplementary material accompanying
warranted. this article, visit the online version of Gastroenterology at
The current study has some limitations that need to be www.gastrojournal.org, and at https://dx.doi.org/10.1053/
acknowledged. Firstly, although large in size with multi- j.gastro.2022.10.014.
annual follow-up of the included patients with IBD, this
study was conducted at a tertiary referral center with high
levels of expertise in the care of patients with IBD and in References
endoscopic and microscopic imaging in these patients. Given
1. Shah SC, Colombel JF, Sands BE, et al. Systematic re-
this, especially the reading of the CLE images for presence
view with meta-analysis: mucosal healing is associated
or absence of barrier healing might be more difficult outside
with improved long-term outcomes in Crohn’s disease.
of expert centers. However, we used a highly reproducible
Aliment Pharmacol Ther 2016;43. 317–133.
semiquantitative scoring system to assess barrier dysfunc-
2. Shah SC, Colombel JF, Sands BE, et al. Mucosal healing
tion and barrier healing in patients with IBD,17-19,49 and
is associated with improved long-term outcomes of pa-
various studies in the field have demonstrated that suc- tients with ulcerative colitis: a systematic review and
cessful use of CLE and image analysis, and especially the meta-analysis. Clin Gastroenterol Hepatol 2016;
identification of barrier dysfunction, can be rapidly learned 14:1245–1255.e8.
with high interobserver and intraobserver agreement and 3. Le Berre C, Peyrin-Biroulet L; SPIRIT-IOIBD study
without the need of prior pathology training.57–59 group. Selecting end points for disease-modification
In this context, we mention that probe-based CLE im- trials in inflammatory bowel disease: the SPIRIT
aging is currently not widely spread and is mainly used at consensus from the IOIBD. Gastroenterology 2021;
expert centers. Although probe-based CLE is compatible 160:1452–1460 e21.
with any existing endoscopy setup with the possibility for 4. Peyrin-Biroulet L, Ferrante M, Magro F, et al. Results
rapid image acquisition after fluorescein injection, CLE from the 2nd Scientific Workshop of the ECCO. I: impact
clearly needs dedicated time during the procedure (in our of mucosal healing on the course of inflammatory bowel
study, w6 minutes per procedure); however, with the disease. J Crohns Colitis 2011;5:477–483.
development of algorithms for automated CLE image ana- 5. Peyrin-Biroulet L, Sandborn W, Sands BE, et al. Select-
lyses, a reduction in the times needed for CLE images ing Therapeutic Targets in Inflammatory Bowel Disease
analysis can be expected and might facilitate transferability (STRIDE): determining therapeutic goals for treat-to-
of this technology to less experienced centers in the target. Am J Gastroenterol 2015;110:1324–1338.
future.20,60 6. Turner D, Ricciuto A, Lewis A, et al. STRIDE-II: an update
Secondly, no allowance was made for multiple compar- on the Selecting Therapeutic Targets in Inflammatory
isons in this exploratory approach. Bowel Disease (STRIDE) initiative of the International
Finally, we did not include central reading for endo- Organization for the Study of IBD (IOIBD): determining
scopic or histopathologic scoring. However, the inclusion of therapeutic goals for treat-to-target strategies in IBD.
IBD-experienced endoscopists in our expert center and 2 Gastroenterology 2021;160:1570–1583.
different masked pathologists, along with the use of 2 of the 7. Walsh AJ, Bryant RV, Travis SP. Current best practice for
most commonly used scores in UC and the observed high disease activity assessment in IBD. Nat Rev Gastro-
concordance between RHI and NHI, might reflect that no enterol Hepatol 2016;13:567–579.
significant bias was present despite the lack of central pa- 8. Zallot C, Peyrin-Biroulet L. Deep remission in inflamma-
thology reading. tory bowel disease: looking beyond symptoms. Curr
Gastroenterol Rep 2013;15:315.
9. Ungaro RC, Yzet C, Bossuyt P, et al. Deep remission at 1
Conclusion year prevents progression of early Crohn’s disease.
In summary, our results show for the first time that Gastroenterology 2020;159:139–147.
barrier healing is highly predictive of the further course of 10. Bryant RV, Burger DC, Delo J, et al. Beyond endoscopic
disease in patients with clinically remittent IBD and that the mucosal healing in UC: histological remission better
 254 Rath et al Gastroenterology Vol. 164, No. 2

predicts corticosteroid use and hospitalisation over 6 activity score for Crohn’s disease: the SES-CD. Gastro-
years of follow-up. Gut 2016;65:408–414. intest Endosc 2004;60:505–512.
11. Iacucci M, Fort Gasia M, Hassan C, et al. Complete 25. Vuitton L, Marteau P, Sandborn WJ, et al. IOIBD tech-
mucosal healing defined by endoscopic Mayo subscore nical review on endoscopic indices for Crohn’s disease
still demonstrates abnormalities by novel high definition clinical trials. Gut 2016;65:1447–1455.
colonoscopy and refined histological gradings. Endos- 26. Lim LG, Neumann J, Hansen T, et al. Confocal endomi-
copy 2015;47:726–734. croscopy identifies loss of local barrier function in the
12. Klenske E, Atreya R, Hartmann A, et al. Magnification duodenum of patients with Crohn’s disease and ulcera-
endoscopy with optical chromoendoscopy shows strong tive colitis. Inflamm Bowel Dis 2014;20:892–900.
INFLAMMATORY BOWEL DISEASE

correlation with histologic inflammation in patients with 27. Karstensen JG, Saftoiu A, Brynskov J, et al. Confocal
inflammatory bowel disease. Endosc Int Open 2019; laser endomicroscopy in ulcerative colitis: a longitudinal
7:E1018–E1026. study of endomicroscopic changes and response to
13. Gupta A, Yu A, Peyrin-Biroulet L, et al. Treat to target: the medical therapy (with videos). Gastrointest Endosc 2016;
role of histologic healing in inflammatory bowel diseases: 84:279–286.e1.
a systematic review and meta-analysis. Clin Gastro- 28. Rath T, Dieterich W, Katscher-Murad C, et al. Cross-
enterol Hepatol 2021;19:1800–1813.e4. sectional imaging of intestinal barrier dysfunction by
14. Park S, Abdi T, Gentry M, et al. Histological disease confocal laser endomicroscopy can identify patients with
activity as a predictor of clinical relapse among patients food allergy in vivo with high sensitivity. Sci Rep 2021;11:
with ulcerative colitis: systematic review and meta- 12777.
analysis. Am J Gastroenterol 2016;111:1692–1701. 29. Li CQ, Liu J, Ji R, et al. Use of confocal laser endomi-
15. Yoon H, Jangi S, Dulai PS, et al. Incremental benefit of croscopy to predict relapse of ulcerative colitis. BMC
achieving endoscopic and histologic remission in pa- Gastroenterol 2014;14:45.
tients with ulcerative colitis: a systematic review and 30. Li CQ, Xie XJ, Yu T, et al. Classification of inflammation
meta-analysis. Gastroenterology 2020;159:1262– activity in ulcerative colitis by confocal laser endomi-
1275.e7. croscopy. Am J Gastroenterol 2010;105:1391–1396.
16. Rath T, Atreya R, Neurath MF. Is histological healing a 31. Rahmi G, Coron E, Perrod G, et al. Probe-based confocal
feasible endpoint in ulcerative colitis? Expert Rev Gas- laser endomicroscopy for in vivo assessment of histo-
troenterol Hepatol 2021;15:665–674. logical healing in ulcerative colitis: development and
17. Chang J, Leong RW, Wasinger VC, et al. Impaired in- validation of the ENHANCE Index. J Crohns Colitis 2021;
testinal permeability contributes to ongoing bowel 15:994–999.
symptoms in patients with inflammatory bowel disease 32. Mosli MH, Feagan BG, Zou G, et al. Development and
and mucosal healing. Gastroenterology 2017; validation of a histological index for UC. Gut 2017;
153:723–731.e1. 66:50–58.
18. Karstensen JG, Saftoiu A, Brynskov J, et al. Confocal 33. Marchal-Bressenot A, Salleron J, Boulagnon-Rombi C,
laser endomicroscopy: a novel method for prediction of et al. Development and validation of the Nancy histo-
relapse in Crohn’s disease. Endoscopy 2016; logical index for UC. Gut 2017;66:43–49.
48:364–372. 34. Knieling F, Neufert C, Hartmann A, et al. Multispectral
19. Kiesslich R, Duckworth CA, Moussata D, et al. Local optoacoustic tomography for assessment of Crohn’s
barrier dysfunction identified by confocal laser endomi- disease activity. N Engl J Med 2017;376:1292–1294.
croscopy predicts relapse in inflammatory bowel dis- 35. Cohen JF, Korevaar DA, Altman DG, et al. STARD 2015
ease. Gut 2012;61:1146–1153. guidelines for reporting diagnostic accuracy studies:
20. Queneherve L, David G, Bourreille A, et al. Quantitative explanation and elaboration. BMJ Open 2016;6:
assessment of mucosal architecture using computer- e012799.
based analysis of confocal laser endomicroscopy in in- 36. Mehandru S, Colombel JF. The intestinal barrier, an
flammatory bowel diseases. Gastrointest Endosc 2019; arbitrator turned provocateur in IBD. Nat Rev Gastro-
89:626–636. enterol Hepatol 2021;18:83–84.
21. Peyrin-Biroulet L, Panes J, Sandborn WJ, et al. Defining 37. Gitter AH, Wullstein F, Fromm M, et al. Epithelial barrier
disease severity in inflammatory bowel diseases: current defects in ulcerative colitis: characterization and quanti-
and future directions. Clin Gastroenterol Hepatol 2016; fication by electrophysiological imaging. Gastroenter-
14:348–354.e17. ology 2001;121:1320–1328.
22. Battat R, Dulai PS, Ma C, et al. Current endpoints of 38. Schmitz H, Barmeyer C, Fromm M, et al. Altered tight
clinical trials in ulcerative colitis: are they valid? Curr junction structure contributes to the impaired epithelial
Treat Options Gastroenterol 2020;18:15–32. barrier function in ulcerative colitis. Gastroenterology
23. Bryant RV, Winer S, Travis SP, et al. Systematic re- 1999;116:301–309.
view: histological remission in inflammatory bowel 39. Zeissig S, Burgel N, Gunzel D, et al. Changes in
disease. Is ‘complete’ remission the new treatment expression and distribution of claudin 2, 5 and 8 lead to
paradigm? An IOIBD initiative. J Crohns Colitis 2014; discontinuous tight junctions and barrier dysfunction in
8:1582–1597. active Crohn’s disease. Gut 2007;56:61–72.
24. Daperno M, D’Haens G, Van Assche G, et al. Develop- 40. Reinink AR, Lee TC, Higgins PD. Endoscopic mucosal
ment and validation of a new, simplified endoscopic healing predicts favorable clinical outcomes in
February 2023 Intestinal Barrier Healing Predicts IBD Outcome 255

inflammatory bowel disease: a meta-analysis. Inflamm 57. Buchner AM, Gomez V, Heckman MG, et al. The learning
Bowel Dis 2016;22:1859–1869. curve of in vivo probe-based confocal laser endomicro-
41. Baert F, Moortgat L, Van Assche G, et al. Mucosal scopy for prediction of colorectal neoplasia. Gastrointest
healing predicts sustained clinical remission in patients Endosc 2011;73:556–560.
with early-stage Crohn’s disease. Gastroenterology 58. Chang J, Ip M, Yang M, et al. The learning curve, inter-
2010;138:463–468; [quiz: e10–e11]. observer, and intraobserver agreement of endoscopic
42. Riley SA, Mani V, Goodman MJ, et al. Microscopic ac- confocal laser endomicroscopy in the assessment of
tivity in ulcerative colitis: what does it mean? Gut 1991; mucosal barrier defects. Gastrointest Endosc 2016;
32:174–178. 83:785–791.e1.

INFLAMMATORY BOWEL DISEASE

43. Atreya R, Siegmund B. Location is important: differenti- 59. Kuiper T, Kiesslich R, Ponsioen C, et al. The learning
ation between ileal and colonic Crohn’s disease. Nat Rev curve, accuracy, and interobserver agreement of
Gastroenterol Hepatol 2021;18:544–558. endoscope-based confocal laser endomicroscopy for
44. Cleynen I, Boucher G, Jostins L, et al. Inherited de- the differentiation of colorectal lesions. Gastrointest
terminants of Crohn’s disease and ulcerative colitis Endosc 2012;75:1211–1217.
phenotypes: a genetic association study. Lancet 2016; 60. Guleria S, Shah TU, Pulido JV, et al. Deep learning sys-
387:156–167. tems detect dysplasia with human-like accuracy using
45. Thia KT, Sandborn WJ, Harmsen WS, et al. Risk factors histopathology and probe-based confocal laser endo-
associated with progression to intestinal complications microscopy. Sci Rep 2021;11:5086.
of Crohn’s disease in a population-based cohort.
Gastroenterology 2010;139:1147–1155.
Received January 28, 2022. Accepted October 13, 2022.
46. Johansson ME, Hansson GC. Immunological aspects of
intestinal mucus and mucins. Nat Rev Immunol 2016; Correspondence
16:639–649. Address correspondence to: Markus F. Neurath, MD, Medical Clinic 1,
Friedrich-Alexander University Erlangen-Nürnberg, Ulmenweg 18, D-91054
47. Kredel LI, Jodicke LJ, Scheffold A, et al. T-cell compo- Erlangen, Germany. e-mail: [email protected].
sition in ileal and colonic creeping fat—separating ileal
CRediT Authorship Contributions
from colonic Crohn’s disease. J Crohns Colitis 2019; Timo Rath, MD (Conceptualization: Lead; Formal analysis: Lead; Investigation:
13:79–91. Lead; Methodology: Lead; Writing – original draft: Lead).
48. Verdier J, Begue B, Cerf-Bensussan N, et al. Compart- Raja Atreya, MD (Formal analysis: Supporting; Writing – review & editing:
Equal).
mentalized expression of Th1 and Th17 cytokines in Julia Bodenschatz, MD (Formal analysis: Equal; Software: Equal).
pediatric inflammatory bowel diseases. Inflamm Bowel Wolfgang Uter, MD (Formal analysis: Lead; Software: Lead; Validation:
Equal; Writing – review & editing: Supporting).
Dis 2012;18:1260–1266. Carol E. Geppert, MD (Formal analysis: Supporting).
49. Buda A, Hatem G, Neumann H, et al. Confocal laser Francesco Vitali, MD (Formal analysis: Equal).
endomicroscopy for prediction of disease relapse in ul- Sarah Fischer, MD (Investigation: Supporting).
Maximilian J. Waldner, MD (Investigation: Supporting).
cerative colitis: a pilot study. J Crohns Colitis 2014; Jean-Frédéric Colombel, MD (Conceptualization: Supporting; Formal
8:304–311. analysis: Equal; Writing – review & editing: Equal).
Arndt Hartmann, MD (Formal analysis: Equal). Markus F. Neurath, MD
50. Soderholm JD, Peterson KH, Olaison G, et al. Epithelial (Conceptualization: Lead; Formal analysis: Equal; Methodology: Lead; Writing
permeability to proteins in the noninflamed ileum of – original draft: Equal).
Crohn’s disease? Gastroenterology 1999;117:65–72.
Conflicts of interest
51. Soderholm JD, Streutker C, Yang PC, et al. Increased The authors disclose the following: Jean-Frédéric Colombel reports research
epithelial uptake of protein antigens in the ileum of grants from AbbVie, Janssen Pharmaceuticals, and Takeda; payment for
lectures from AbbVie, Amgen, Allergan, Inc, Ferring Pharmaceuticals, Shire,
Crohn’s disease mediated by tumour necrosis factor and Takeda; consulting fees from AbbVie, Amgen, Arena Pharmaceuticals,
alpha. Gut 2004;53:1817–1824. Boehringer Ingelheim, BMS, Celgene Corporation, Eli Lilly, Ferring
Pharmaceuticals, Galmed Research, Genentech, GlaxoSmithKline, Janssen
52. D’Inca R, Di Leo V, Corrao G, et al. Intestinal permeability Pharmaceuticals, Kaleido Biosciences, Imedex, Immunic, Iterative Scopes,
test as a predictor of clinical course in Crohn’s disease. Merck, Microba, Novartis, PBM Capital, Pfizer, Sanofi, Takeda, TiGenix, and
Am J Gastroenterol 1999;94:2956–2960. Vifor; and holds stock options in Intestinal Biotech Development. Raja Atreya
has served as a speaker or consultant or received research grants from
53. Katz KD, Hollander D, Vadheim CM, et al. Intestinal AbbVie, Amgen, Arena Pharmaceuticals, Biogen, Boehringer Ingelheim,
permeability in patients with Crohn’s disease and Bristol-Myers Squibb, Cellgene, Celltrion Healthcare, Dr Falk Pharma,
Galapagos, Gilead, InDex Pharmaceuticals, Janssen-Cilag, Lilly, MSD Sharp
their healthy relatives. Gastroenterology 1989; & Dohme, Novartis, Pandion Therapeutics, Pfizer, Roche Pharma, Samsung
97:927–931. Bioepis, Takeda Pharma, and Viatris. Markus F. Neurath has served as
54. Ukabam SO, Clamp JR, Cooper BT. Abnormal small in- advisor or speaker for Pentax, Roche Pharma, Takeda Pharma, Pfizer, MSD,
PPM, Janssen, Gilead, Dr Falk Pharma, Boehringer Ingelheim, Amgen and
testinal permeability to sugars in patients with Crohn’s AbbVie. Timo Rath has served as a speaker for Pentax, AbbVie, Olympus,
disease of the terminal ileum and colon. Digestion 1983; Medtronic, Takeda Pharma, Lilly, Janssen-Cilag, and Galapagos. The other
authors disclose no conflicts.
27:70–74.
55. Wyatt J, Vogelsang H, Hubl W, et al. Intestinal perme- Funding
ability and the prediction of relapse in Crohn’s disease. Raja Atreya and Markus F. Neurath were supported by the Deutsche
Forschungsgemeinschaft (German Research Foundation)
Lancet 1993;341:1437–1439. Sonderforschungsbereiche (SFB) 1181 and Transregio (TRR) 241.
56. Shen L, Weber CR, Raleigh DR, et al. Tight junction pore
Data Availability
and leak pathways: a dynamic duo. Annu Rev Physiol All requests for data should be submitted to the corresponding authors for
2011;73:283–309. consideration.