Clinical Science (2001) 100, 379–386 (Printed in Great Britain)
Improvement in intestinal permeability
precedes morphometric recovery of
the small intestine in coeliac disease
Adrian G. CUMMINS*, Fiona M. THOMPSON*, Ross N. BUTLER†, John C. CASSIDY‡,
David GILLIS‡, Mark LORENZETTI*, Emma K. SOUTHCOTT† and Peter C. WILSON*
*Department of Gastroenterology and Hepatology, DX 465384, The Queen Elizabeth Hospital, Woodville South, SA 5011,
Australia, †Department of Gastroenterology, Women’s and Children’s Hospital, Adelaide, SA 5006, Australia, and ‡Department
of Human Immunology, Institute of Medical and Veterinary Science, Adelaide, SA 5001, Australia
A
B
S
T
R
A
C
T
It is often difficult to assess small bowel recovery in adults with coeliac disease on a gluten-free
diet (GFD). This prospective study compares changes in intestinal permeability with changes in
intestinal biopsy at various intervals after commencing a GFD. Intestinal permeability was
measured by lactulose/rhamnose absorption from 1 week to 24 months after commencing a GFD.
Intestinal morphometry was measured by villus area, crypt length and mitotic count per crypt
at diagnosis and after commencing a GFD. Median intestinal permeability values decreased from
0.47 (n l 35) at diagnosis to 0.25 (n l 17) after 1 week and to 0.16 (n l 18) after 2 months of a
GFD. Rhamnose absorption improved significantly at an early stage, from 6.6 % (untreated) to
15.4 % at 3 months of a GFD, whereas the decrease in lactulose permeation took longer : from
3.4 % (untreated) to 0.8 % after 12 months of a GFD. Mean villus area (n l 29) was reduced to 16 %
of control values at diagnosis, and improved to a maximum of 48 % after 6 months on a GFD, but
did not change thereafter. Mean crypt length and mitotic count per crypt were increased by
222 % and 356 % respectively at diagnosis, and these parameters remained elevated at 172 % and
216 % above control values after 6 months of a GFD. We conclude that intestinal permeability
improves within 2 months after starting a GFD, but that measurable intestinal biopsy
improvement requires ingestion of a GFD for at least 3–6 months, and even then remains
incomplete.
INTRODUCTION
Diagnosis of coeliac disease is made using intestinal
biopsy demonstrating surface villous effacement and
atrophy, crypt hyperplasia and increased numbers of
intra-epithelial lymphocytes [1]. These changes of intestinal damage should respond to a gluten-free diet
(GFD). However, recovery of the intestinal biopsy in
adult-onset coeliac disease remains variable [2–5]. Children generally show better improvement of intestinal
histology than adults [6]. Earlier studies in small numbers
of children and adults at various intervals after starting a
GFD indicated that enterocyte height improved within
days or weeks (but was subject to wide fluctuation), that
dissecting microscopic appearance (flat mosaic to convoluted to leaves) took months to several years to
improve, and that sucrase and maltase disaccharidases
improved, but lactase had delayed recovery [2–7]. The
main problem with these studies has been the arbitrary
intervals of tissue sampling and the small number of
subjects that were studied. A variable histological response in adults with coeliac disease has been observed
even in recent studies [8–12]. This means that it is difficult
to assess recovery alone by intestinal biopsy in adults,
and the questions arises : how long and to what extent
does the small intestine recover in subjects on a GFD ?
Key words : coeliac disease, intestinal permeability, small intestine.
Abbreviations : GFD, gluten-free diet ; sIL-2R, soluble interleukin-2 receptor.
Correspondence : Dr A. G. Cummins (e-mail adrian.cummins!nwahs.sa.gov.au).
# 2001 The Biochemical Society and the Medical Research Society
379
380
A. G. Cummins and others
Intestinal permeability, as measured by a double sugar
permeation technique, is increased in coeliac disease.
While it was suggested initially that this represents a
primary defect [13], it is more likely to be due to
intestinal damage, as abnormal permeability and intestinal damage are associated [14]. We have shown
previously that intestinal permeability in eight subjects
improved rapidly within 1–2 months after commencing a
GFD, but we were unable to detect any morphometric
improvement of the intestinal biopsy in six of the eight
subjects after 3–6 months of a GFD [10]. Thus the
association was not as simple as might be expected, but a
limitation of the study was that the follow-up biopsies
had been grouped together due to the small number of
subjects. The previous study has now been extended
prospectively with additional subjects and over a longer
duration of a GFD. In particular, intestinal biopsies were
taken after 1.5, 3, 6, 12 and 24 months of a GFD, to allow
assessment of recovery.
During the course of the present study, food labelling
guidelines in Australia became stricter, with the designation that ‘ gluten-free food ’ should have no detectable
gluten on immunoassay (sensitivity
0.003 g of gluten\100 g of protein) [15]. This was brought into regulation from 15 March 1995. Prior to that date, food
guidelines followed the Codex Alimentarius Commission
of the World Health Organization and of the Food and
Agricultural Organization of the United Nations. The
Codex allows gluten-free foods to contain 0.3 g of
gluten\100 g of protein. In practice, this meant that the
new guidelines excluded malt and wheat starch from a
GFD. We therefore used this opportunity to compare
biopsies from subjects before and after the introduction
of these stricter guidelines.
The aims of the present study were to compare
intestinal permeability and morphometry as measures of
recovery in coeliac disease, and to discern the relationship
between permeability and histological change. Intestinal
morphometry was also compared in subjects after 6
months of the old Codex and the new, stricter GFD. As
coeliac disease is an immunological intolerance to gluten,
the soluble interleukin-2 receptor (sIL-2R) concentration
was used to assess total T cell activity, and anti-gliadin
and anti-endomysial antibodies were used to assess gluten
withdrawal.
disease included iron- or folate-deficient anaemia,
chronic diarrhoea, epilepsy, symptoms suggestive of
irritable bowel syndrome, recurrent miscarriage, and
metabolic bone disease. One patient had rheumatoid
disease, one had Type I diabetes mellitus, and two had
isolated IgA deficiency without any complicating symptoms. Three patients had aphthous ulcers in the duodenum at endoscopy. Two patients had infertility that
responded to a GFD. A control group of 75 patients
underwent small bowel biopsy while being investigated
for other medical conditions. They were subsequently
found to have normal gastrointestinal function (except
for irritable bowel syndrome). Endoscopic duodenal
biopsies have been shown to be a satisfactory means of
assessing the presence or absence of coeliac disease
[16–18]. The study was commenced at The Royal
Adelaide Hospital and later was transferred to The Queen
Elizabeth Hospital. The research protocol was approved
by the Human Ethics Committees of The Royal Adelaide
Hospital and of The Queen Elizabeth Hospital, and was
carried out in accordance with the Declaration of
Helsinki (1989) of the World Health Organization.
Informed consent was obtained from each subject.
Experimental protocol
METHODS
The initial permeability test in subjects on a normal diet
was performed with the patient unaware of the diagnosis.
Nearly all subjects became members of the Coeliac
Society of South Australia, and received regular food lists
and support. Each subject was interviewed about their
dietary compliance and was encouraged to maintain a
strict GFD at each interval of testing. We endeavoured to
study changes in intestinal permeability in the same
patients after commencing a GFD, but this was not
always possible. However, we found that the mean value
for permeability stabilized with 8–12 patients in each
group, but we recruited at least 15 patients in each group
to ensure good stability of the mean data. Duodenal
biopsies were taken at endoscopy from subjects on a
normal diet, and at 1.5, 3, 6, 12 and 24 months after
commencement of a GFD. Except for the 1.5 month
group, which contained eight subjects, nine subjects had
biopsies on a normal diet and at all intervals on the GFD ;
three further patients had no initial biopsy available for
morphometry at diagnosis but completed all biopsy
intervals of the GFD. Histological recovery in coeliac
subjects was assessed by comparing intestinal morphometry with that of control subjects.
Subjects
Intestinal permeability
A total of 36 subjects with coeliac disease (16 males, 20
females ; age 18–80 years) were studied following diagnosis by small bowel biopsy that showed villous
effacement and atrophy, crypt hyperplasia and increased
intra-epithelial lymphocytes. Subjects were recruited
between 1987 and 1997. The presentations of coeliac
Each subject fasted overnight, and emptied their bladder
before drinking 100 ml of a hypertonic solution (1500
mOsmol) containing 1.0 g of α-L-rhamnose (R-3875 ;
Sigma, St. Louis, MO, U.S.A.), 5.0 g of lactulose and
22.6 g of glucose. Lactulose was used as a 67 % (w\v)
syrup (duphalac ; Solvay Pharmaceuticals, Sydney, NSW,
# 2001 The Biochemical Society and the Medical Research Society
Coeliac disease and intestinal permeability
Australia). Glucose was used as an osmotic filler in
preference to lactose or sucrose, because the latter are
likely to be malabsorbed in villous atrophy. All urine was
collected for the next 5 h. Subjects could drink water
throughout the study, and could eat after the first 3 h.
Urinary rhamnose and lactulose were measured using
HPLC, as previously described and validated [19]. The
overall precision of estimation of the probe sugars varies
from 4.2 to 6.5 % using this method. Intestinal permeability was expressed as the mg ratio of urinary
lactulose to rhamnose, with each expressed as the
percentage of the ingested dose excreted in urine.
Permeability was measured at diagnosis and at 0.25, 0.5,
1, 2, 3, 6, 12 and 24 months after commencement of a
GFD. We determined that the mean intestinal permeability of 38 healthy adult subjects was 0.088, with a 95 %
confidence interval of 0.063–0.112.
Intestinal morphometry
All duodenal biopsies for morphometry were taken from
the distal second or third parts of the duodenum at
endoscopy. Duodenal biopsies were stained with Feulgen
reagent and microdissected [20]. Briefly, rows of villi and
crypts were carefully cut using a cataract knife from
stained tissue suspended in water in a Petri dish lid under
stereomicroscopy. These fragments were transferred into
three drops of water on a microscope slide and mounted
in 45 % (v\v) acetic acid before a cover slip was applied.
This helps to flatten the tissue slices in the wet film.
Measurements were made of villus length, apical villus
width and basal villus width using a i10 lens, and of
length of crypts using a i20 microscope lens and a
calibrated eyepiece graticule. Villus area was calculated
using a trapezoid approximation as described previously,
using measures of apical and basal widths and villus
length [21]. Crypt length and mitotic count were also
determined. Mitotic count was expressed as mitotic
counts per crypt.
Disaccharidases
Lactase, sucrase and maltase activities were assayed as
described previously in detail [22]. The 90 % confidence
intervals for lactase, sucrase and maltase are 3–14, 6–26
and 13–44 µmol:min−":g−" wet weight respectively. The
activity was expressed per wet weight, as this is the
simplest denominator. Measurement of activity per mg of
protein may be subject to a surface-to-volume error, and
assumes that disaccharidase protein is independent of
mucosal changes.
adapted to the Boehringer Mannheim ES300 automated
ELISA system (Boehringer Mannheim\Roche Diagnostics, Sydney, Australia). Gliadin (G-3375 ; Sigma-Aldrich,
Sydney, Australia) was dissolved in 100 % (v\v) ethanol
and used to coat plastic tubes. Serum was diluted 1 : 500
(v\v) and added to the plastic tubes. Bound anti-gliadin
antibodies were detected using anti-IgA (AH10104 ;
1 : 4000 dilution) and anti-IgG (AH10304 ; 1 : 25 000 dilution) human antibodies conjugated to peroxidase (Biosource International, Camarillo, CA, U.S.A.). Chromogen (857424 ; Boehinger Mannheim\Roche Diagnostics)
was added, and the reaction was read after 40 min. A
standard curve was constructed using dilutions of a
known positive control serum (BP140 ; The Binding Site,
Birmingham, U.K.). ELISA values were calculated in
arbitrary units from the standard curve fitted by the
Rodbard Evaluation method. The ELISA was originally
validated using sera from 20 paediatric coeliac subjects,
and yielded a sensitivity of 95 % for IgA anti-gliadin
antibody. The reference ranges for IgA and IgG antigliadin antibodies were 0–58 and 0–78 ELISA units
respectively for control subjects.
Anti-endomysial IgA antibody was detected by immunofluorescence using monkey oesophagus substrate
(FS208.2 ; The Binding Site). Serum was added ‘ neat ’ and
incubated for 30 min. Bound antibody was detected
using anti-IgA human antibody conjugated to fluorescein
(F0204 ; 1 : 20 dilution ; Dako Australia, Sydney, Australia).
sIL-2R levels
sIL-2R levels were measured in sera from coeliac subjects
at diagnosis and at the same intervals of the GFD as for
the permeability study. sIL-2R levels were measured
using a CellFree IL-2R kit (T Cell Sciences, Boston, MA,
U.S.A.). The sensitivity of the assay is 50 units\ml.
Statistics
Data were summarized as mean (S.E.M.), except for
intestinal permeability data, for which the median was
used to indicate central tendency. Intestinal permeability
data were transformed log n(xj1) to normalize the data
"!
and stabilize the variance before significance testing by
Peritz’ F test [23]. Anti-gliadin antibody data that were
more than the upper or less than the lower level of
detection were equated to these respective limits for the
purpose of data analysis. Peritz’ F test was also used for
significance testing of intestinal morphometry, epithelial
cell height, sIL-2R and anti-gliadin antibody data.
Anti-gliadin and anti-endomysial antibodies
Sera stored at k20 mC were available from 28 coeliac
subjects at diagnosis and from 18 subjects after 2 years of
a GFD. Serology was assessed at the end of the study
using this stored sera. Measurement of anti-gliadin
antibodies was performed using an ‘ in house ’ method
RESULTS
Intestinal permeability
A total of 36 subjects had intestinal permeability
measured within a few days of diagnosis of coeliac disease
# 2001 The Biochemical Society and the Medical Research Society
381
382
A. G. Cummins and others
Table 1 Urinary excretion of lactulose and rhamnose in
coeliac subjects at various intervals after commencing a GFD
Results, which are expressed as a percentage of the ingested dose, are given as
means (S.E.M.). Data were transformed by log10n(xj1) before significance testing
using Peritz’ F test. Significance of differences compared with subjects on a normal
diet : * P l 0.038 ; ** P l 0.0013. (Note that there were multiple other
significant differences, but only the important ones are indicated.)
Figure 1 Intestinal permeability in coeliac subjects on a
normal diet, and after 1 week to 24 months of a GFD
Intestinal permeability is given as the mg ratio of lactulose/rhamnose urinary
recovery after oral dosing. The shaded area demarcates the 95 % confidence
interval for healthy control subjects.
by intestinal biopsy, but only 26 of these subjects were
available for sequential prospective permeability testing
at intervals after commencement of the GFD. This was
because the remaining subjects were diagnosed in rural
areas and could not attend for logistical reasons. We
investigated early changes in intestinal permeability in
15–19 patients at 1, 2, 4 and 8 weeks after commencing a
GFD. Ten of these patients completed all these intervals
of testing. Longer-term improvements in permeability
were assessed in 15–23 subjects at 3, 6, 12 and 24 months
of the GFD, including eight patients who completed all
these intervals of permeability testing.
Intestinal permeability lactulose\rhamnose ratios are
given in Figure 1, and the individual urinary excretion
data are given in Table 1. The sensitivity of the intestinal
permeability test for detecting the abnormality of coeliac
disease in 36 subjects on a gluten-containing diet was
89 %. Intestinal permeability improved rapidly within
weeks, with a significant improvement evident from 2
months on a GFD (P l 0.0008). The data for lactulose
and rhamnose excretion, expressed as a percentage of
total dose, showed that the abnormality in initial permeability was due to a combination of increased lactulose
permeation and lowered rhamnose absorption (Table 1).
Lactulose permeation remained abnormal despite institution of a GFD, but had improved by 1–2 years of a
GFD. Rhamnose excretion was unusual, in that excretion
over the period 0.5–3 months after starting the GFD
improved to values beyond those normally seen in
healthy subjects. Thus rhamnose absorption improved
quickly, by 3 months after starting the GFD, whereas
# 2001 The Biochemical Society and the Medical Research Society
Group
Lactulose
Rhamnose
Coeliac/normal diet (n l 35)
Coeliac/GFD
1 week (n l 17)
2 weeks (n l 15)
1 month (n l 19)
2 months (n l 17)
3 months (n l 23)
6 months (n l 21)
12 months (n l 17)
24 months (n l 15)
3.4 (1.0)
6.6 (1.4)
3.2 (0.7)
4.4 (0.9)
2.3 (0.5)
1.9 (0.4)
2.8 (0.7)
2.2 (0.7)
0.8 (0.2)**
0.7 (0.1)
8.2 (2.0)
15.5 (5.7)
11.5 (2.9)
14.3 (4.3)
15.4 (3.2)*
9.5 (2.0)
3.5 (0.7)
7.1 (1.1)
Control group (n l 38)
0.7 (0.09)
10.9 (1.9)
lactulose permeation took 12 months of a GFD to
improve (Table 1).
Intestinal morphometry
A total of 29 patients had duodenal biopsies collected for
morphometry, but 11 of these subjects were not available
for subsequent follow-up for logistical reasons, while
another 10 subjects had already had their diagnostic
biopsy but were available for subsequent biopsies.
Changes in intestinal morphometry before and at progressive intervals after starting a GFD for all available
subjects are given in Figure 2. Coeliac disease was evident
morphometrically at diagnosis, as indicated by a decrease
in mean villus area to 16 %, by a 222 % increase in crypt
length, and by a 356 % increase in mitotic count
compared with control values. Both villus area and crypt
hyperplasia had improved significantly by 3 months of a
GFD (P 0.0001). This stabilized after 6 months of a
GFD. At this time, mean villus area, crypt length and
mitotic count values were 59 %, 175 % and 224 %
respectively of control values. There were 14 subjects
who had biopsies at diagnosis and at 6 months of a GFD.
The mean (S.E.M.) villus area in these subjects was 0.057
(0.025) mm# at diagnosis and 0.165 (0.024) mm# after 6
months of a GFD. These values were equivalent to those
for all subjects who were biopsied at diagnosis and at 6
months of a GFD (Figure 2). There was no further
significant improvement in villus area or mitotic count,
and only a small decrease in crypt length (P l 0.013),
between 6 and 24 months of a GFD. Mean values of villus
area for the nine subjects who completed all the biopsy
intervals (excluding 1.5 months, which had eight subjects)
were similar to those for all subjects biopsied at each of
Coeliac disease and intestinal permeability
Figure 3 Epithelial cell height in coeliac subjectson a normal
diet, and after 1.5 to 24 months of a GFD
Data are given as meansjS.E.M.
GFD from 18 months to 15 years [0.146 (0.023) mm#].
There was no evidence for a bimodal response of
intestinal recovery at 2 years of a GFD, with only three of
20 subjects having villus area greater than the lower
boundary of the 95 % confidence interval of control
subjects. Furthermore, the low values of skewness (0.42)
and kurtois (k0.46) for this group approximated to a
normal rather than a bimodal distribution.
Figure 2 Villus area, crypt length and mitotic count in
duodenal biopsies from coeliac subjects before and after
progressive intervals of a GFD
Changes in enterocyte height
Data are given as meansjS.E.M.
Changes in enterocyte height are given in Figure 3.
Enterocyte height was still depressed at 80 % of control
values at 3 months of the GFD, but had improved by 12
months.
Table 2 Comparison of intestinal morphometry at 6 months
after starting the old Codex GFD or the new GFD after
introduction of stricter food labelling regulations in Australia
Disaccharidases
Data are given as meanspS.E.M.
Diet
Villus area
(mm2)
Crypt length Mitotic count
(µm)
per crypt
Codex GFD (n l 18)
New GFD (n l 8)
0.187 (0.020)
0.112 (0.012)
369 (23)
324 (17)
5.5 (0.6)
5.0 (0.8)
the intervals of a GFD (results not shown). Recovery of
intestinal morphometry was not improved on the stricter
GFD as compared with the older Codex GFD (Table 2).
There was no difference in villus area between patients
positive for anti-endomysial antibody [0.127 (0.024)
mm#] and those negative for anti-endomysial antibody
[0.209 (0.028) mm# ; P l 0.10] after 2 years of a GFD. The
mean (S.E.M.) villus area at 2 years of a GFD in this
longitudinal study [0.185 (0.019) mm#] was comparable
with that of a group of five coeliac subjects, who not been
intensely studied or councilled, at various intervals of a
Lactase activity improved 2.4-fold from initial diagnosis
to 2 years of a GFD (P l 0.008), but remained subnormal.
Sucrase and maltase activities showed no significant
improvement (Table 3).
Anti-gliadin and anti-endomysial antibodies
Anti-gliadin IgA and IgG and anti-endomysial IgA
antibodies were analysed from sera collected at diagnosis
(n l 28) and after 2 years of a GFD (n l 18). Antigliadin IgA antibody was positive in 29 % of subjects at
diagnosis, and in 0 % of subjects after 2 years of a GFD.
The mean (S.E.M.) ELISA values were 59 (11) units at
diagnosis and 29 (13) units after 2 years (P l 0.06). Antigliadin antibody IgG was positive in 54 % of subjects at
diagnosis, and in 22 % of subjects after 2 years of a GFD.
The mean (S.E.M.) values were 125 (18) units at diagnosis
and 67 (19) units after 2 years of a GFD (P l 0.009).
Anti-endomysial IgA was positive in 82 % of subjects at
diagnosis and in 39 % of subjects after 2 years of a GFD.
Of the five patients negative for anti-endomysial anti# 2001 The Biochemical Society and the Medical Research Society
383
384
A. G. Cummins and others
Table 3 Disaccharidase activities of duodenal biopsies from
coeliac subjects on a normal diet, and at various intervals
after starting a GFD
Data are meanspS.E.M.
Activity (µmol:min−1:g−1 wet
weight)
Group
Lactase
Sucrase
Maltase
Coeliac/normal diet (n l 18)
Coeliac/GFD
1.5 months (n l 5)
3 months (n l 18)
6 months (n l 18)
12 months (n l 18)
24 months (n l 16)
Controls (n l 16)
1.4 (0.2)
5.6 (0.7)
10.3 (1.6)
2.2 (0.2)
1.6 (0.2)
2.0 (0.4)
1.7 (0.2)
3.0 (0.5)
5.5 (2.9)*
9.0 (1.5)
5.5 (0.7)
6.3 (1.0)
5.9 (0.6)
8.0 (1.1)
9.6 (1.4)
17.4 (0.9)
13.5 (1.7)
16.5 (2.6)
14.1 (1.8)
17.2 (2.5)
21.8(2.2)
* This excludes subjects with lactase deficiency (n l 8) ; the inclusive values
were 3.7p0.7.
Figure 4 Serum sIL-2R levels in coeliac subjects at diagnosis
and at various intervals after commencing a GFD
Data are given as meanspS.E.M.
body at diagnosis, four patients had elevated levels of
anti-gliadin IgG antibodies.
sIL-2R levels
The mean serum sIL-2 level at diagnosis was 1440
units\ml ; this had decreased to 793 units\ml by 6 weeks
of a GFD (P l 0.016), and to 594 units\ml (P 0.0001)
after 2 years (Figure 4). The mean (S.E.M.) sIL-2R levels
on the old and new GFDs after 6 months were 729 (129)
units\ml (n l 15) and 693 (68) units\ml (n l 8) respectively (not significant).
# 2001 The Biochemical Society and the Medical Research Society
DISCUSSION
Subjects with coeliac disease often report clinical improvement in symptoms before any improvement is
evident on intestinal biopsy [1,2,7]. The present study
was undertaken to investigate whether intestinal permeability (a functional test of the small intestine) improved soon after commencing a GFD. Our investigation
was prospective, using a large number of subjects with
coeliac disease who were investigated at progressive
intervals after starting a GFD. Histological improvement
was quantified morphometrically, unlike the subjective
reporting of the majority of previous studies. Intestinal
permeability improved significantly within 2 months of
commencing a GFD, with the median value of intestinal
permeability decreasing from a ratio of 0.47 to 0.16. This
preceded any improvement in villus area, which only
began to improve after 3 months on a GFD and showed
no further improvement after 6 months of a GFD.
Abnormal intestinal permeability was present in 89 %
of untreated subjects with coeliac disease, which was
greater than the 82 % detected by measurement of antiendomysial IgA antibodies, at least in our population.
The major finding in the present study was that intestinal
permeability decreased exponentially, and had improved
after only 2 months of a GFD (although not completely
to normal), compared with the partial improvement of
intestinal morphometry by 6 months. The improvement
at 2 months was due to a 220 % increase in monosaccharide (rhamnose) absorption and a 44 % decrease in
disaccharide (lactulose) permeation (Table 1). A previous
study of intestinal permeability using lactulose\mannitol
showed a similar improvement in permeability after 2
months of a GFD [24], and an earlier study also showed
improvement in permeability within 5 months of starting
a GFD [25]. It has been suggested that coeliac disease
involves an intrinsic permeability defect [13], but the
rapid improvement after commencement of a GFD
indicates that permeability is related to gluten ingestion.
Conversely, a single 30 g dose gluten challenge causes a
rise in intestinal permeability within 3 days [25].
The simplest possible explanation for the rapid improvement in intestinal permeability is that there was a
greater intestinal surface area available within weeks of
commencing a GFD [26]. It was interesting that rhamnose permeation increased up to 2.2-fold in the period
from 2 weeks to 3 months after starting the GFD. The
mucosal surface area of the small intestine in humans is
amplifed over that of a simple cylinder by submucosal
folds, villi and microvilli, with a proximal-to-distal
gradient such that 50 % of the total intestinal surface area
is within the first 25 % of the proximal intestine [27].
It is unlikely that the improvement in intestinal permeability was due to an increased mucosal surface area,
proximally at least, as intestinal morphometry did not
show any detectable change in villous surface area
Coeliac disease and intestinal permeability
and there was no detectable change in disaccharidase
activities, indicating that microvilli had not improved at
this stage. Therefore this cannot explain the improved
rhamnose absorption.
Controversy exists as to the mechanism of monosaccharide permeation. One proposed mechanism is
transcellular permeation across the enterocyte, which, in
the case of rhamnose, is also mediated passively [26]. This
is presumably dependent on differentiated villous rather
than immature crypt cells. Interestingly, rhamnose absorption was only reduced to 60 % of control values in
untreated coeliac disease, even though villous area was
reduced to 16 % of normal values (Table 1). This suggests
the rhamnose permeation is not entirely dependent on
villous architecture. Rhamnose absorption increased by
220 % between 0.5 and 3 months after starting a GFD,
which is difficult to reconcile with continued villous
atrophy, although there may have been distal ileal
improvement. Another proposed mechanism is paracellular permeation through tight junctions between
enterocytes, with a differential effect between the less
permeable junctions of villous enterocytes and the more
permeable junctions between crypt enterocytes (the
Hollander hypothesis) [26]. However, these theories still
do not explain our findings of increased rhamnose
absorption at 3 months after commencing a GFD. We
suspect that other factors, such as improved ileal compensation and changes in motility, may in some measure
explain these unusual results.
Changes in lactulose permeation are more easily
explained, as being due to an improvement in individual
enterocyte function because of a decrease in inflammatory cytokines in the mucosa. Immunological activity
decreased rapidly by 45 % within 6 weeks after commencing a GFD, as shown by the exponential decrease in
sIL-2R levels (Figure 4). Recent studies have indicated
that mucosal cytokines disrupt intestinal tight junctions.
Fasano et al. [28] described high-level expression of a
novel cytokine, zonulin, in the intestinal lamina propria
of untreated coeliac subjects, and this was found to
disrupt the assembly of tight junctions. Zonulin expression was reduced after treatment, although the period
of a GFD was not indicated. Zonulin is a analogue of
zonula occulens toxin from Vibrio cholerae [29]. Morphometric studies have shown that crypt epithelial
junctions are exposed in untreated coeliac disease, and are
relatively structurally abnormal as compared with tight
junctions on villus epithelial cells [30,31]. These morphometric abnormalities are correlated with increased
paracellular permeability. A recent study has shown that
untreated coeliac subjects have a deficiency of ZO-1
protein, which is a constituent of epithelial tight junctions
[32]. These studies certainly help to explain why lactulose
permeation in our study was increased nearly 5-fold in
subjects with untreated coeliac disease, and normalized
by 12–24 months of a GFD (Table 1).
The present prospective study also confirmed what has
been suspected for a long time in adults with coeliac
disease – namely that the proximal small bowel does not
entirely return to normal after commencing a GFD.
Morphometry demonstrated that villous atrophy persisted even after 2 years of a GFD. We also extended our
morphometry to six subjects after 4 years of a GFD, and
showed no further evidence of improvement. Our
findings agree with those from a study in adult coeliac
subjects which showed that villous epithelial volume
only improved in subjects on a GFD to 61 % of control
values [33]. Although deliberate or inadvertent ingestion
of gluten would probably explain this effect, it is
interesting that no detectable differences in villous
atrophy were seen in the first 6 months of a GFD before
and after gluten-free labelling (and therefore gluten
elimination) became stricter in Australia (Table 2). In
addition, no improvement in villous atrophy was seen in
subjects who were negative for anti-endomysial antibody, as compared with those who were positive, after 2
years of a GFD. The problem is that gluten exclusion is
very difficult, and it is almost impossible to assess
accurately minor indiscretion or, more likely, inadvertent
ingestion of a minuscle quantity of gluten. A GFD
probably represents the ideal end of a continuum rather
than a dichomotous reality of dietary exclusion.
In conclusion, this prospective study has shown that
intestinal permeability, particularly the component of
monosaccharide absorption, improved rapidly after commencing a GFD ; nevertheless villous atrophy, as assessed
morphometrically, takes at least 3–6 months to improve,
and recovery was still incomplete after this duration. This
was in spite of the removal of gluten antigen and
decreased immunological activity. Thus a degree of
villous atrophy may persist indefinitely in coeliac subjects, despite them adhering to an apparent GFD.
ACKNOWLEDGMENTS
We are very grateful to referring physicians who allowed
us to study patients under their care. We thank the
nursing staff of the Endoscopy Units of both The Royal
Adelaide Hospital and The Queen Elizabeth Hospital for
their assistance during this study. We also thank Dr K.
Miki and Mr T. A. Robb, who performed some of the
permeability assays, and Mrs J. Langman, who performed
the disaccharidase assays.
REFERENCES
1 Trier, J. S. (1992) Celiac sprue. Gluten, major
histocompatibility complex and the small intestine.
Gastroenterology 102, 1709–1719
2 Stewart, J. S. (1974) Clinical and morphological response
to gluten withdrawal. Clin. Gastroenterol. 3, 109–126
# 2001 The Biochemical Society and the Medical Research Society
385
386
A. G. Cummins and others
3 Rubin, C. E., Brandborg, L. L., Phelps, P. C. et al. (1960)
Studies of celiac disease. II. The apparent irreversibility of
the proximal intestinal pathology in celiac disease.
Gastroenterology 38, 517–535
4 Rubin, C. E. (1960) Celiac disease and idiopathic sprue :
some reflections on reversibility, gluten, and intestine.
Gastroenterology 39, 260–261
5 Ukabam, S. O. and Cooper, B. T. (1985) Small intestinal
permeability as an indicator of jejunal mucosal recovery in
patients with celiac sprue on a gluten-free diet. J. Clin.
Gastroenterol. 7, 232–236
6 Anderson, C. M. (1960) Histological changes in the
duodenal mucosa in celiac disease. Arch. Dis. Child. 35,
419–427
7 Yardley, J. H., Bayless, T. M., Norton, J. H. and Hendrix,
T. R. (1962) A study of the jejunal epithelium before and
after a gluten-free diet. N. Engl. J. Med. 267, 1173–1179
8 Marsh, M. N. (1988) Studies of intestinal lymphoid tissue.
XI. – The immunopathology of cell mediated reactions in
gluten sensitive and other enteropathies. Scanning Microsc.
2, 1663–1684
9 Gefte, J. M. M., Bouman, J. G., Grond, J., Jansen, W. and
Kleibeuker, J. H. (1988) Slow and incomplete histological
and functional recovery in adult gluten sensitive
enteropathy. J. Clin. Pathol. 41, 886–891
10 Cummins, A. G., Penttila, I. A., LaBrooy, J. T., Robb,
T. A. and Davidson, G. P. (1991) Recovery of the small
intestine in coeliac disease on a gluten-free diet : changes in
intestinal permeability, small bowel morphology and
T-cell activity. J. Gastroenterol. Hepatol. 6, 53–57
11 Jennings, W., Rowland, R., Hecker, R., Gibson, G. E.,
Fitch, R. J. and Reid, D. P. (1976) The significance of
lowered jejunal disaccharidase levels. Aust. N. Z. J. Med. 6,
556–560
12 D’Agostino, L., Daniele, B., Pignata, S. et al. (1987) Post
heparin plasma diamino oxidase increases in patients with
celiac disease during gluten-free diet. Gut 28, 131–134
13 Bjarnason, I. and Peters, T. J. (1982) In vitro
demonstration of small intestinal permeability :
demonstration of a persistent defect in patients with celiac
disease. Gut 15, 145–150
14 Juby, L. D., Dixon, M. F. and Axon, A. T. R. (1987)
Abnormal intestinal permeability assessed by sugar
permeability and jejunal morphometry. J. Clin. Pathol. 40,
714–718
15 Skerritt, J. H. and Hill, A. S. (1991) Enzyme immunoassay
for determination of gluten in foods : collaborative study.
J. Assoc. Anal. Chem. 74, 257–263
16 Dandalides, S. M., Carey, W. D., Petras, R. and Achkar, E.
(1989) Endoscopic small bowel biopsy : a controlled trial
evaluating forceps size and biopsy location in the diagnosis
of normal and abnormal mucosal architecture.
Gastrointest. Endosc. 35, 197–200
17 Gottrand, F., Turck, D., Mitchell, V. and Farriaux, J. P.
(1992) Comparison of fiberendoscopy and Watson capsule
for small intestinal biopsy in infants and children. Acta
Paediatr. 81, 399–401
18 Scott, B. B. and Jenkins, D. (1981) Endoscopic small
intestinal biopsy. Gastrointest. Endosc. 27, 162–167
19 Miki, K., Butler, R., Moore, D. et al. (1996) Rapid and
simultaneous quantification of rhamnose, mannitol, and
lactulose in urine by HPLC for estimating intestinal
permeability in paediatric patients. Clin. Chem. 42, 71–75
20 Goodlad, R. A., Levi, S., Lee, C. Y., Mandir, N., Hodgson,
H. and Wright, N. A. (1991) Morphometry and cell
proliferation in endoscopic biopsies : evaluation of a
technique. Gastroenterology 101, 1235–1241
21 Cummins, A. G., LaBrooy, J. T., Stanley, D. P., Rowland,
D. P. and Shearman, D. J. C. (1990) Quantitative
histological study of enteropathy associated with HIV
infection. Gut 31, 317–321
22 Langman, J. M. and Rowland, R. (1990) Activity of
duodenal disaccharidases in relation to normal and
abnormal mucosal morphology. J. Clin. Pathol. 43,
537–540
23 Harper, J. H. (1984) Peritz ’ F test : BASIC program of a
robust multiple comparison test for statistical analysis of
all differences among group means. Comput. Biol. Med.
14, 437–445
24 Smecuol, E., Bai, J. C., Vazquez, H. et al. (1997)
Gastrointestinal permeability in celiac disease.
Gastroenterology 112, 1129–1136
25 Hamilton, I., Cobden, I., Rothwell, J. and Axon, A. T. R.
(1982) Intestinal permeability in celiac disease : the
response to gluten withdrawal and single-dose gluten
challenge. Gut 23, 202–210
26 Bjarnason, I., MacPherson, A. and Hollander, D. (1995)
Intestinal permeability : an overview. Gastroenterology
108, 1566–1581
27 Wilson, J. P. (1967) Surface area of the small intestine in
man. Gut 8, 618–621
28 Fasano, A., Not, T., Wang, W. et al. (2000) Zonulin, a
newly discovered modulator of intestinal permeability, and
its expression in coeliac disease. Lancet 366, 1518–1519
29 Fasano, A., Uzzau, S., Fiore, C. and Margaretten, K.
(1997) The enterotoxic effect of zonula occludens toxin
(Zot) on rabbit small intestine involves the paracellular
pathway. Gastroenterology 112, 839–846
30 Madara, J. L. and Trier, J. S. (1980) Structural
abnormalities of jejunal epithelial cell membranes in celiac
sprue. Lab. Invest. 43, 254–261
31 Schulzke, J.-D., Bentzel, C. J., Riecken, E.-O. and Fromm,
M. (1998) Epithelial tight junction structure with acute and
treated celiac sprue. Pediatr. Res. 43, 435–441
32 Perry, I., Tselepis, C., Sanders, A. S. A., Jankowski, J. A.,
Cooper, B. T. and Iqbai, T. H. (2000) Reduced expression
of ZO-1 and gamma catenin in the small intestinal mucosa
of untreated celiac disease. Gastroenterology Suppl. 118,
A1964
33 Niazi, N. M., Leigh, R., Crowe, P. and Marsh, M. N.
(1984) Morphometric analysis of small intestinal mucosa. I.
Methodology, epithelial volume compartments and
enumeration of inter-epithelial space lymphocytes.
Virchows Arch. A Pathol. Anat. 404, 49–60
Received 12 July 2000/13 October 2000; accepted 4 January 2001
# 2001 The Biochemical Society and the Medical Research Society