Introduction
Several parameters of quality have been defined for optimal colonoscopy, such as cecal
intubation, adverse event rates, duration of withdrawal, and bowel preparation [1]
[2]
[3]
[4]
[5]. High-quality bowel preparation is noted to be of primary importance for an effective
colonoscopic examination.
Among bowel preparation agents, osmotic laxatives (i. e., mainly water-soluble polymers
that are neither digested nor absorbed in the small intestine) have been widely used
for more than three decades [6]. The osmotic action of these agents increases the colonic fluid volume by maintaining
water in the colonic lumen, thereby distending the bowel and indirectly increasing
bowel peristalsis.
More recently, osmotic laxatives containing sodium phosphate have been proposed, in
order to decrease the volume of fluid intake and improve patient acceptability [7]
[8]
[9]
[10]. However, recent studies have reported that these agents, in solution or in a tablet
formulation, may induce aphthous or erosive injuries of the stomach and colon in some
patients [11]
[12]
[13]
[14]
[15]. In addition, after the introduction of sodium phosphate tablets (NaPT) in France,
several individual case safety reports of gastritis or stomach ulcerations were registered.
To date, no risk factors, such as preexisting lesions, history, co-morbidities, and
concomitant drug use, have been identified. Although these lesions are frequently
asymptomatic and can be viewed as “false-positive” findings, it is important to explain
their pathophysiology and evolution more clearly. For instance, although the specific
mechanisms inducing such lesions remain unknown, it has been suggested that an increase
in the production of free radicals may be involved [16]. However, defining these osmotic laxative – induced lesions in humans remains difficult.
Therefore, it was thought that an experimental model would be helpful by allowing
time course studies to be done. Such an experimental model should fulfill several
criteria, enabling (1) prolonged and stable exposure to laxative or control substances,
(2) in vivo and possibly repeated endoscopic examinations, and (3) histologic and
functional analysis of the exposed mucosa.
In an attempt to reproduce the gastric lesions observed in humans, we developed a
pig model to determine the effects of the prolonged and direct application of NaPT
on the gastric mucosa by using a combination of endoscopic, probe-based confocal laser
endomicroscopy (pCLE) and histologic and functional methods.
Materials and Methods
Animals and tablets
A total of 14 pigs (weighing between 30 and 40 kg) were used in this study. Experiments
were done in a dedicated surgical facility of the Laboratoire des Grands Animaux,
INSERM U 643, in accordance with French Veterinary Regulations and Ethics Committee
standards (agreement E.44010). During all procedures, the animals were under general
anesthesia, induced with 5 % isoflurane and 60 % nitrous oxide and subsequently maintained
with 2 % isoflurane.
Each NaPT (Colokit®; Mayoly Spindler, Chatou, France) weighed 1500 mg and was composed of 1102 mg of monobasic sodium phosphate monohydrate
and 398 mg of dibasic anhydrous sodium phosphate. Each placebo tablet (PlaT) weighed
1704 mg and was composed of 170.4 mg of polyethylene glycol, 8.52 mg of magnesium
stearate, 1354.68 mg of lactose (Tablettose®; Meggle, Wasserburg, Germany), and 170.4 mg of microcrystalline cellulose (Avicel®; FMC BioPolymer, Philadelphia, Pennsylvania, USA).
Study design
The study design is summarized in [Fig. 1]. An initial study (study 1) was conducted with five pigs to determine the feasibility
of the procedure and to characterize early lesions observed after NaPT had been applied
for 1.5 hours. Two consecutives studies were then performed to determine the time
course of the lesional aspects at 1.5 hours, 24 hours (study 2, n = 5), and 72 hours
(study 3, n = 4) after the application of NaPT and PlaT. In order to consider potential
mechanical effects of the tablets, a normal mucosal area, hereafter referred as the
control site, was also analyzed. Thus, three different sites of analysis were considered:
control, PlaT, and NaPT.
Fig. 1 Study design. In study 1 (n = 5), analysis was performed on sodium phosphate tablet
(NaPT) sites at 1.5 hours after the application of NaPT tablets and on unexposed mucosa,
referred to as control (CTRL) sites. In studies 2 and 3 (n = 9), analysis was performed
on NaPT, PLaT, and CTRL sites at 1.5, 24, and 72 hours after the application of NaPT
and placebo tablets (PLaT). At each considered time point (0, 1.5, 24, and 72 hours),
the endoscopic aspect was reported, and probe-based confocal laser endomicroscopy
(pCLE) was carried out for 1 minute at each site.
At each considered time (0, 1.5, 24, and 72 hours), both the macroscopic and microscopic
(pCLE) aspects of each site were analyzed in vivo during upper gastrointestinal endoscopy.
At the end of the study, an endoscopic mucosal resection (EMR) was performed on each
site, and specimens were sent for further ex vivo morphologic and functional analysis.
Endoscopic procedure
During all procedures, performed by a single experienced endoscopist (E.C.), the animals
were in prone position and under general anesthesia. The endoscopic appearance of
the gastric mucosa was assessed meticulously with a standard resolution endoscope
(EG450; Fujifilm, Tokyo, Japan). First, particular attention was paid to washing the
gastric cavity with syringes of water. The best area on which to apply both tablets
(PlaT and NaPT) was then defined as a stable area in the fundus with no visible contractions.
One NaPT tablet was captured with an endoscopic snare, stabilized within a cap placed
at the distal tip of the endoscope, and placed into the stomach at the level of the
fundus. The same procedure was immediately repeated with a PlaT, which was placed
2 to 3 cm to the right of the previous tablet. Finally, one endoscopic clip was positioned
between the two tablets in order to allow re-identification of these areas at 24 and
72 hours ([Fig. 2]). The macroscopic aspect of each site of analysis was described and graded as follows:
grade 0, no lesion; grade 1, mild lesion defined by the presence of exanthema; grade
2, moderate lesion defined by the presence of aphthous ulcer and/or black clots; grade
3, severe lesion defined by the presence of ulcer and/or active bleeding.
Fig. 2 Endoscopic views. a The drug tablets – that is, the sodium phosphate tablet (thin arrow) and the placebo
tablet (thick arrow) – a few minutes after endoscopic placement on the fundic mucosa
of pigs. b The same sites at 24 hours. The endoscopic clip was initially positioned between
the two tablets to allow re-identification of the sites after complete dissolution
of the tablets.
Probe-based confocal laser endomicroscopy examination
The pCLE procedure was performed by a single experienced endoscopist (E.C.), as follows.
The confocal miniprobe (ColoFlexUHD, Cellvizio; Mauna Kea Technologies, Paris, France)
was gently positioned onto the gastric mucosa in contact with either the drug tablet
or the placebo tablet, or onto the control site. A volume of 5 mL of 10 % fluorescein
sodium (Novartis Pharma, France) was injected intravenously as a contrast agent. The
pCLE video sequences were recorded from the time of injection to 5 minutes after injection
for further “off-line” analysis.
Following the procedure, each video clip was reviewed by one endoscopist (M.P.), who
was blinded to the procedure. The six best images from each pCLE video sequence were
extracted and transferred to an external drive in PNG (portable network graphics)
format to allow further semiquantitative analysis. The final data set comprised 678
images from 14 pigs. The following parameters were assessed: epithelial lining irregularity,
architectural disorganization, and fluorescein intensity, which was measured in both
the crypt lumen and the intercryptic area. ImageJ 1.42q software was used to measure
fluorescein intensity (National Institutes of Health, Bethesda, Maryland, USA). Epithelial
irregularity and architectural disorganization were assessed with a semiquantitative
score ([Table 1]) adapted from Li et al. [17]
. This score was assessed independently by two procedure-blinded investigators (M.D.
and N.M.).
Table 1
Classification of endomicroscopic lesions.
|
0 (absent)
|
1 (moderate)
|
2 (severe)
|
|
Surface epithelium architecture
|
Regular thickness
|
Irregular thickness
|
Destroyed epithelium
|
|
Crypts architecture
|
Regular arrangement and size of crypts
|
Irregular arrangement of crypts, enlarged
spaces between crypts
|
Crypt destruction
|
Ex vivo assessment of paracellular permeability
Fundic specimens obtained by EMR were immediately placed into Krebs solution (0.187 g/L
NaH2PO4∙2H2O, 6.84 g/L NaCl, 0.35 g/L KCl, 2.10 g/L NaHCO3, 1.98 g/L glucose, 0.368 g/L CaCl2∙2H2O, 0.244 g/L MgCl2∙6H2O) at 4 °C. For each site, the EMR specimen was microdissected to separate the mucosa
from the submucosal layer (in the plane of the submucosal blood vessels). Specimens
of isolated fundic mucosa were then mounted in dedicated Ussing chambers (Physiologic
Instruments, San Diego, California, USA) with a chamber surface of 0.0314 cm2. Each chamber contained 2 mL of Ham’s F12 Nutrient Mixture (Life Technologies, Grand
Island, New York, USA) containing 0.1 % fetal calf serum. The media were continuously
oxygenated by a gas flow of 95 % O2 and 5 % CO2 and maintained at 37 °C. After an equilibration period of 30 minutes, 200 µL of apical
medium was replaced with 200 µL of fluorescein – 5.6 sulfonic acid (1 mg/mL). Every
30 minutes, the fluorescence level of basolateral aliquots of 150 µL was measured
over a period of 180 minutes with a fluorometer (Thermo Scientific). Paracellular
permeability was determined by calculating the slope of the changes in fluorescence
intensity over time with a linear regression fit model (GraphPad Software, La Jolla,
California, USA).
Histologic evaluation
For each site, a portion of the EMR specimen was fixed in 4 % paraformaldehyde in
phosphate-buffered saline. After tissue washing in phosphate-buffered saline, the
specimen was dehydrated and embedded in paraffin. Sequential sections of mucosa (5 µm)
were obtained and, after staining in hematoxylin and eosin, were analyzed with an
Olympus IX50 microscope (Olympus America, Center Valley, Pennsylvania, USA). Pictures
were captured with a DP71 digital camera (Olympus) connected to a computer through
a frame grabber card (Cell^B software, Olympus). To assess the mucosal morphology,
we analyzed mucosal thickness, determined by measuring the distance between the surface
epithelium and the underlying muscularis mucosae (only in study 1).
Statistical analysis
Because of the limited sample size, nonparametric tests were used for statistical
analysis. Two groups were compared with either the Mann – Whitney test or the Wilcoxon
test. More than two groups were compared with either the Kruskal – Wallis test or
the Dunn test. Comparisons between groups, and over time, were performed with a two-way
analysis of variance (ANOVA) followed by a Bonferroni test. All statistical analyses
were performed with GraphPad Prism 5.00 for Windows (GraphPad Software).
Results
Induction of specific, early gastric mucosal lesions with sodium phosphate tablets
All tablets were successfully applied onto the fundic mucosa under endoscopic guidance
([Fig. 2]). At 1.5 hours, all tablets remained in position at the original site and were partially
dissolved.
First, we used white light endoscopy to grade the mucosal changes induced by NaPT.
All sites in all animals were macroscopically normal (grade 0) at the time of NaPT
application. At 1.5 hours, only grade 1 lesions were observed in 8 of the 14 animals
(57 %) at the NaPT sites. No grade 2 or 3 lesions were observed. No lesions were observed
at control sites.
Second, we used pCLE to analyze and quantify microscopic mucosal changes in vivo.
With semiquantitative score, we indicated that all sites initially had normal epithelial
lining and glandular architecture (grade 0). At 1.5 hours, no epithelial irregularity
or architectural disorganization was seen at control sites. In contrast, a significant
increase in epithelial irregularity and architectural disorganization scores was noted
at NaPT sites ([Fig. 3 a, b]). Blinded interobserver analysis showed an acceptable match between two independent
investigators (kappa = 0.52). Further quantitative pCLE analysis showed that changes
in pit intensity over 1.5 hours were significantly greater at NaPT sites than at control
sites. In contrast, changes in intercryptic intensity did not differ between control
and NaPT sites during this time interval ([Fig. 3 c, d]).
Fig. 3 Short-term effects of sodium phosphate tablets (NaPT) 1.5 hours after application
onto the fundic mucosa of pigs, assessed with probe-based confocal laser endomicroscopy
(studies 2 and 3) and graded. a Semiquantitative lesion score of surface epithelium architecture. b Semiquantitative lesion score of crypts architecture. c Variations of fluorescein intensity in the crypt lumen [(H1.5 – H0) × 100 /H0]. d Variations of intercryptic intensity. Wilcoxon signed-rank test: * P ≤ 0.05, *** P < 0.0001; Mann – Whitney test: # P < 0.0025. PlaT, placebo tablet; Ct, control; AU, arbitrary units.
Third, in addition to in vivo endoscopic analysis, we performed both histologic and
functional analysis on EMR specimens collected at the end of this part of the study.
We showed that mucosal thickness at NaPT sites was similar to that at control sites
(0.33 ± 0.05 mm vs. 0.27 ± 0.04 mm; n = 5; P > 0.05). Consistently, fundic paracellular permeability was not changed at NaPT sites
in comparison with control sites (0.28 ± 0.04 vs. 0.33 ± 0.05; n = 5; P > 0.05). Furthermore, there was no correlation between paracellular permeability
and pit or crypt intensity.
Finally, in order to consider potential mechanical effects of NaPT on mucosal lesions,
we performed a comparative study with PlaT. Initially, all sites were macroscopically
normal (grade 0) and remained normal 1.5 hours after application of the PlaT (n = 9).
PlaT induced a slight but significant increase in epithelial irregularity and architectural
scores ([Fig. 3 a, b]). However, epithelial irregularity and architectural disorganization scores were
significantly higher at NaPT than at PlaT sites ([Fig. 3 a, b]). PlaT did not induce a change in pit intensity at 1.5 hours after application,
in contrast to NaPT ([Fig. 3 c]). Similarly, PlaT did not induce a change in intercryptic intensity at 1.5 hours
after application, in contrast to NaPT ([Fig. 3 d]).
Evolution of gastric mucosal lesions induced by sodium phosphate tablets
This part of the study attempted to monitor the evolution of the mucosal lesions induced
by NaPT at 24 and 72 hours after tablet application ([Fig. 4]). At 24 and 72 hours, with white light endoscopy, all application sites could be
easily re-identified by the presence of the endoscopic clips. There were no residual
tablets at these sites.
Fig. 4 Observations made in pigs with probe-based confocal laser endomicroscopy at the sodium
phosphate tablet site. a Normal appearance immediately before tablet application. b, c Severe epithelial irregularity and architectural disorganization at 1.5 and 24 hours,
respectively. d Partial healing of these mucosal alterations at 72 hours.
First, macroscopic examination 24 hours after NaPT application revealed that 2 of
9 pigs had grade 1 lesions (22 %), 1 had grade 2 lesions (11 %), and the remaining
6 had no lesion (67 %). In 2 of 9 pigs (22 %), the PlaT site showed grade 1 lesions,
and the remaining pigs had no lesion at the PlaT site. Finally, no lesion was visible
at the control site in any pig. At 72 hours, no lesion was identifiable in any animal
for any conditions (NaPT, PlaT, and control).
Second, pCLE at PlaT and control sites showed that crypt pit intensity and intercryptic
intensity remained unchanged over the different time points studied ([Fig. 5]). In contrast, at NaPT sites, the increased semiquantitative scores and crypt pit
intensity observed at 1.5 hours remained unchanged at 24 hours. However, at 72 hours,
the semiquantitative scores and crypt pit intensity were significantly reduced in
comparison with the values at 24 hours and were similar to those measured before the
application of NaPT ([Figs. 4, 5]).
Fig. 5 Time course of endomicroscopic aspects at 1.5, 24, and 72 hours after the application
of sodium phosphate tablets (NaPT) or placebo tablets (PlaT) in pigs. Control (Ct):
unexposed mucosal area. a Semiquantitative lesion score of surface epithelium architecture. b Semiquantitative lesion score of crypts architecture. c Variations of fluorescein intensity in the crypt lumen [(H1.5 – H0) × 100 /H0]. d Variations of intercryptic intensity. Wilcoxon signed-rank test: * P ≤ 0.05 vs. value at H0. Mann – Whitney test: # P < 0.05 vs. value of Ct at identical time and ‡ P < 0.05 vs. value of PlaT at identical time. AU, arbitrary units.
Finally, we performed functional analysis on EMR specimens collected at 24 and 48
hours after tablet application. Fundic paracellular permeabilities were similar at
each site and at all time points (0.19 ± 0.07 vs. 0.22 ± 0.07 vs. 0.18 ± 0.07 and
0.10 ± 0.02 vs. 0.09 ± 0.04 vs. 0.10 ± 0.03, respectively, for NaPT, PlaT and Ct sites
at 24 and 72 hours, respectively).
Discussion
The present study demonstrated that a 90-minute application of NaPT onto fundic mucosa
induces endomicroscopically detectable changes in the epithelial lining. These changes
persisted for 24 hours after NaPT application but were resolved at 72 hours. In addition,
standard video endoscopy revealed minor macroscopic lesions, such as local exanthema
or superficial lesions, at the site of NaPT application in a subgroup of animals.
These lesions disappeared by 72 hours after application.
Few studies have specifically determined the gastrotoxic effects of orally administered
drugs, and this characterization remains difficult and limited in humans [18]. In the present study, we developed an original approach for characterizing both
the features of injury (macroscopic and microscopic) and the functional consequences
of direct, drug-induced gastric mucosal toxicity. Our rationale for developing an
experimental animal model to document drug-induced injuries in the gastric mucosa
follows: (1) an animal model is ethical because serial endoscopic examinations over
either a short or prolonged interval cannot be performed in humans; (2) the model
allows the development of lesions to be monitored over a long period of time; and
(3) large submucosal dissections can be done. The pig was chosen for its suitability
as a model in terms of gastric anatomy, gastric size, and physiology, which are considered
to be close to those of humans [19]. Furthermore, in this model, tools similar to those used in humans can be used for
the morphologic and functional assessment of mucosal lesions.
We were able to place the tablets at the chosen sites and, more importantly, to keep
them in place without difficulty for 1.5 hours. This was likely facilitated by the
deep anesthesia of the animals, which helped to inhibit gastric wall motility. In
these conditions of prolonged tablet application, we anticipated the induction of
maximal injury to the exposed area.
Based on the macroscopic appearances, our results show the presence of mucosal lesions
after NaPT application. These observations confirm previous findings in humans, especially
concerning fundic injuries and injuries in the colon [11]
[12]
[14]. Although nonphysiologic and rather drastic experimental conditions were used, with
direct and prolonged application of tablets onto a “dry” area of mucosa, we did not
observe any grade 3 lesions. At 24 hours after the initial procedure, grade 1 and
grade 2 lesions were observed in 3 of the 9 animals (33 %), whereas there were no
lesions in the other 6 animals (66 %). In addition, 72 hours after the procedure,
no lesion could be identified on the previously exposed gastric area. Thus, our design
additionally documents the time course of injuries and their spontaneous reversibility.
Observed injuries were reversible in less than 72 hours, which appears to be consistent
with healing of the gastric mucosa following superficial lesions [20].
The pCLE analysis showed that microscopic architectural disorganization occurred early,
reflected by variation in fluorescein intensity in the crypt lumen as early as 1.5
hours after the application. The placebo tablet did not induce any significant changes
in any of the parameters studied. By contrast, 24 hours after the application of PlaT
and NaPT, significant changes were observed for all parameters except intercryptic
intensity. However, the changes were smaller with PlaT than with NaPT. Overall, this
finding suggests that early macroscopic mucosal changes could be due to active compound
(see below), whereas later effects (24 hours) might involve mechanical as well as
excipient components ([Fig. 5]). In fact, despite the observed pCLE changes, the functional consequences were probably
minimal because (1) there was no change in mucosal permeability assessed with the
Ussing chamber and (2) no morphologic change in mucosal thickness was noted.
The mechanisms and components involved in the occurrence of such lesions remain unknown.
They are likely related to sodium monophosphate because similar observations have
been reported with other compounds containing sodium phosphate [11]
[12]
[14].
Despite the fact that this well-characterized animal model included a time course
analysis, our study has some limitations. First, as always when an animal model is
used, the data may not be considered to reflect human conditions strictly. In addition,
the anesthetic conditions may represent a potential bias because the animals were
maintained in one position, so that gastric motility was inhibited and changes in
mucosal defense mechanisms or in gastric mucosal blood flow could not be formally
excluded. However, despite its nonphysiologic character, this condition does allow
the creation of maximal direct toxic effects on the mucosa. Finally, although endoscopic
sessions were repeated in the animals, the relatively long sampling time period (48
hours between the 24 – and 72-hour sessions) does not allow a precise definition of
the time taken for complete mucosal healing.
Although no severe lesions were observed in this study, the methods presented here
could be valuable for gaining pathophysiologic insight into the evolution of severe
digestive mucosal lesions. For instance, the model could be used to test the potential
additional effects of corticosteroid therapies, especially in combination with other
drugs known to induce gastric injury, such as nonsteroidal anti-inflammatory drugs
and aspirin or platelet inhibitors. Likewise, from a mechanistic point of view, it
could be interesting to use such an experimental model to test for the effects of
simultaneously administered gastroprotective agents.
In conclusion, this study documents, in particularly severe experimental conditions,
the macroscopic and microscopic acute gastric injuries induced by NaPT, which are
commonly used for colonoscopic bowel preparation in humans. It illustrates both the
superficial nature and spontaneous reversibility of the lesions, with healing occurring
in less than 72 hours. Although such experimental conditions are never observed in
clinical practice, and no increased risk factor associated with any drug has been
reported in these conditions, the results indicate that further studies are needed
to document the potential consequences of gastric toxicity in patients with increased
risk factors, such as those with a history of gastric ulcers or gastric hemorrhage.
Financial Support
Declaration of funding interests
This study was funded in part by Mayoly Spindler (experimental costs only) and by
INSERM.
