An objective in vivo diagnostic method for inflammatory bowel disease

Inflammatory damage to the bowel, as occurs in inflammatory bowel disease (IBD), is debilitating to patients. In both patients and animal experimental models, histological analyses of biopsies and endoscopic examinations are used to evaluate the disease state. However, such measurements often have delays and are invasive, while endoscopy is not quantitatively objective. Therefore, a real-time quantitative method to assess compromised mucosal barrier function is advantageous. We investigated the correlation of in vivo changes in electrical transmural impedance with histological measures of inflammation. Four platinum (Pt) ball electrodes were placed in the lumen of the rat small intestine, with a return electrode under the skin. Electrodes placed within the non-inflamed intestine generated stable impedances during the 3 h testing period. Following an intraluminal injection of 2,4,6-trinitrobenzene sulfonic acid (TNBS), an established animal model of IBD, impedances in the inflamed region significantly decreased relative to a region not exposed to TNBS (p < 0.05). Changes in intestinal transmural impedance were correlated (p < 0.05) with histologically assessed damage to the mucosa and increases in neutrophil, eosinophil and T-cell populations at 3 h compared with tissue from control regions. This quantitative, real-time assay may have application in the diagnosis and clinical management of IBD.


Introduction
Alterations to mucosal permeability are a key characteristic of IBD [12]. Changes in mucosal permeability can be examined clinically using non-invasive in vivo assays that measure the urinary excretion of large sugar molecules (e.g. lactulose) that are ingested and usually unable to cross the intestinal epithelium if the barrier is uncompromised [13]. However, permeability assays are time-consuming, influenced by confounding factors such as changes in gut transit and altered renal function, and vary considerably between individual patients, making the test hard to standardize [12,14]. Thus, this technique is seldom used for the evaluation of the severity of IBD [12]. A number of animal models of IBD have been developed, the most common being rodents inflamed using 2,4,6-trinitrobenzene sulfonic acid (TNBS) [15][16][17]. In experiments using animal models of IBD, changes in ion permeability can be readily measured ex vivo, using explanted segments of intestine in an Ussing chamber [18][19][20]. In the Ussing chamber, voltage transients in response to applied currents are used to measure the transmucosal impedance [18,21]. In the present study, we have applied the principles of this technique in vivo and hypothesized that the inflammation-induced increase in transmucosal permeability due to the loss of epithelial cells and disruption of tight junctions would result in a decrease in gut wall impedance (figure 1). To evaluate the utility of electrical impedance to evaluate mucosal integrity, we have compared the changes in transmucosal impedance measured in vivo with assessment of inflammation by histopathology and immune cell infiltration.

Animals and surgical procedures
All experiments used male Sprague-Dawley rats (n = 6; 10 weeks old; weight: 357 ± 17 g; Animal Resource Centre, Western Australia) and were approved by the Animal Research and Ethics Committee of the Bionics Institute. The experiments complied with the Australian Code for the Care and Use of Animals for Scientific Purposes (National Health and Medical Research Council of Australia) as well as the United States Army Medical Research and Material Command Animal Care and Use Review Office, protocol SSC-7486.02. Animals were allowed ad libitum access to standard chow, water and fresh food and were kept on a 12 L : 12 D cycle. Prior to surgery, animals were fasted overnight to reduce the amount of gut contents. On the day of surgery, rats were anaesthetized (2% isoflurane, 1% oxygen, flow rate of 1-1.5 l min −1 ; pre-operative analgesia carprofen 50 mg kg −1 subcutaneous), the abdominal cavity was opened and an 8 cm segment of jejunum was selected approximately 30 cm proximal to the ileo-caecal junction [22]. Four platinum (Pt) ball electrodes (1.2 mm diameter; exposed surface area 3.2 mm²), placed 2 cm apart, were inserted into the lumen and secured with sutures (7-0 silk, Ethicon; figure 2a). An 18 G needle placed subcutaneously acted as a return electrode. A 6 cm segment of the jejunum containing three Pt electrodes (E2-E4) was isolated using ligatures, while electrode 1 (E1) was placed 2 cm proximal to the ligated area. Following ligation, 30 min of baseline voltage transient measurements (see the next section for further detail) was generated from all electrodes (E1-E4). Inflammation was then induced within the ligated region by injecting TNBS (1 ml of 0.1% dilution in 50% ethanol, Sigma) into the lumen, similar to the procedure previously described [23,24].

Gut wall impedance monitoring
Gut wall impedance was monitored using biphasic current pulses passed between the intra-and extra-luminal electrodes. The peak voltage at the end of the first phase of the current pulse (total voltage V total , figure 1c,d) was measured as described previously [25]. A 25 μs per phase biphasic current pulse (7 μs interphase gap) was used to minimize the charging of the electrode-electrolyte interface. The current used in this study was 931 µA, which was chosen to produce a peak voltage transient of approximately 500 mV (range: 398-543 mV) when the Pt ball electrodes were placed in sterile 0.9% saline. The V total value was then used to calculate total impedance (Z total ) using Ohm's Law (Z = voltage/current). Following each acute experiment, the Pt ball electrodes were explanted, cleaned and retested in saline. There was no significant difference between pre-and post-experiment in vitro measures (paired T-test; p ≥ 0.05). Triplicate measurements were taken from each intraluminal electrode at T = −30 min and T = 0 prior to TNBS injection to establish a baseline.

Histology and staining
Rats were euthanized (300 mg kg −1 Lethabarb, intracardial injection) at 3 h post-TNBS injection and segments of the jejunum adjacent to electrodes E1-E4 were removed, placed into cold PBS (0.15 M NaCl in 0.01 M sodium phosphate buffer, pH 7.2) and cut longitudinally along the mesenteric border, pinned out onto balsa boards and divided in half. Intraluminal content was gently removed by washing with cold PBS. One half was placed in fixative (2% formaldehyde plus 0.2% picric acid in 0.1 M sodium phosphate buffer, pH 7.4) overnight, embedded in paraffin, sectioned (5 μm), stained with haematoxylin and eosin (H&E) and mounted with DPX [23]. The other half of the tissue was prepared for myeloperoxidase (MPO) staining. Tissue was placed in ice-cold 100% ethanol (10 min), washed in cold 0.1 M sodium phosphate buffer (PB; 3 × 5 min) and then cryoprotected overnight (30% sucrose in PBS). Tissue was immersed overnight in 50% optimal cutting temperature medium and 50% sucrose, frozen (−20°C) and sectioned (14 µm

Immunohistochemistry
Paraffin sections were dewaxed with xylene, graded ethanol and tap water. The sections underwent antigen retrieval (10 mM sodium citrate, 0.05% Tween 20, pH 6 and 1% hydrogen peroxide in PBS heated to 60°C for 10 min). Sections were incubated overnight at 4°C with anti-CD3, a cytotoxic T-cell marker (

Histopathology scoring
Histopathologist (J.B.Fu.), blinded to experimental conditions, used H&E-stained sections to evaluate the degree of inflammation at each electrode site (E1-E4). Histological changes were on a scale of 0-3 for the assessment of the extent and the degree of damage to the mucosa, and on a scale of 0-2 for assessment of the numbers of leucocytes within venules and the extent of haemorrhage within villi [26]. Scores were out of a total of 10 (table 1).

Cell counting
Eosinophilic granulocytes were identified morphologically using H&E-stained sections, by their distinctive nucleus and cytoplasmic staining [23] (identified by arrows in figure 4a,b). MPO activity was used to identify neutrophils (identified in figure 5a,b by arrows). Cytotoxic T-lymphocytes [27] were identified using antibodies to CD3 and counted in H&E sections (identified in figure 6a,b by arrows).
With the observer blinded to the tissue identity, positive cells were counted with a ×40 objective, across 10 fields of view at each electrode position (E1-E4), using a Zeiss Axioplan II microscope, within the following layers: longitudinal smooth muscle, circular smooth muscle, submucosa and mucosa. Images of the total field of view were generated (Axiovision Software, Zeiss, Germany), the area of the total field of view for each layer of tissue was measured using ImageJ and the density of cells per square millimetre was calculated and analysed.

Statistical analysis
Gut wall impedance values (Z total ) were normalized to the baseline measurements (average of data from T = −30 and T = 0 min prior to TNBS injection), set at 100%. A two-way repeated-measures analysis of variance (ANOVA) (Electrode × Time) was used to test for differences and interactions, and Tukey's post hoc test used where appropriate. Infiltration of leucocytes was assessed using a one-way ANOVA and a Tukey's post hoc test. Normalized gut wall impedances were correlated using with leucocyte infiltration and a Pearson's correlation coefficient (R) and two-tailed p-values generated. A p-value of <0.05 was accepted as statistically significant, and data are expressed as mean ± standard error of mean (s.e.m.

Decrease in gut wall impedance following intraluminal 2,4,6-trinitrobenzene sulfonic acid
Soon after implantation into the control jejunum, the in vivo impedance (Z total ) for all electrodes was significantly elevated (1477 ± 158 Ω, n = 24 electrodes, n = 6 rats) compared with measurements made in saline (495 ± 37 Ω). Z total in vivo is presumed to be dominated by the impedance of the gut wall, notably the impedance of epithelial cell barrier of the lining of the intestine, including the impedance provided by the tight junctions between epithelial cells ( figure 1). An increase in transmural permeability is predicted to result in a drop in Z total towards values measured in saline before implantation. Throughout the testing period, gut wall impedance in the non-inflamed region of the jejunum (at electrode 1) remained stable (p > 0.05; n = 6 rats, figure 2a,b). Following TNBS injection, gut wall impedances in inflamed regions (at electrodes E2-E4) rapidly decreased and were significantly less than the impedance measured at E1 at 60 min and remained so for the duration of the experiment (180 min; figure 2b; p < 0.05; n = 6). There were no significant differences between measurements recorded at the three electrodes in the inflamed region (E2, E3 and E4) following TNBS injection (p ≥ 0.05; figure 2b).

Histological scoring of inflammatory damage
In control tissue, villi were long and covered with intact surface epithelium (figure 3a, indicated by arrows), while venules had few to no inflammatory cells within them (figure 3b, indicated by 'ven'). Three hours after infusion of TNBS, we examined the macroscopic appearance of the jejunum and observed patchy redness along the length of lumen exposed to TNBS, but not in other regions. Villi were damaged to various extents. Minimally damaged villi were swollen, and epithelial cells near the tip were flattened and fewer goblet cells were observed. More extensive damage of villi included loss of the surface epithelium from the tips and sides, but retention of the villus core (figure 3c, indicated by arrows). At the highest level of damage, the villus core had detached and the surface epithelium was lost from villus tips. Sloughed off epithelial cells were found in the lumen. There were many microhaemorrhagic foci, consisting of clumps of extravasated red blood cells (figure 3c, indicated by 'm'). These haemorrhages extended to the bases of the villi, but were rare in the region of the crypts. The crypt morphology was similar in samples from control and TNBS-treated jejunum, except in rare cases in which damage extended into the crypts. Venules in the submucosa and at the base of the mucosa after TNBS treatment contained numerous neutrophils and eosinophils (figure 3d, indicated by arrows). Quantitative scoring, based on parameters described in table 1, showed a significant difference between TNBS-exposed and control regions of the jejunum (p < 0.05; figure 3e). Inflammatory score of histological damage in the mucosa was significantly correlated to normalized transmural impedance at 3 h post-TNBS injection (p = 0.0006; R = −0.76; n = 16; figure 3f ). There was no significant correlation between inflamed normalized gut wall impedance values generated from electrodes E2-E4 and inflammatory scores from these electrode areas (p = 0.394; R = −0.27; n = 12). No changes were observed in the external muscle layers. The histopathology reported in this study is similar to that described previously in experimental animals treated with TNBS [23]. The mucosa of the jejunum not exposed to TNBS. The surface epithelium was intact at the tips of the villi (arrows in (a)). Lymphatic vessels (asterisks) were apparent, but there were no red blood cells in the connective tissue spaces. The crypts were of normal appearance and immune cells were not accumulated in venules (ven). Following TNBS (c,d), the villi were shortened and the base of the mucosa can be seen in the same field as villus tips. The surface epithelium was lost from the tips of the villi (arrows). Numerous regions of microhaemorrhage were seen in the villi (m). After TNBS injection, there was marked accumulation of neutrophils in the venules (ven; indicated by arrows in (d)), which was not seen in control tissue (b). The crypts were intact in both control and TNBS-treated tissue. Eosinophils (circled) were more numerous in inflamed tissue (d), than in control (b). (e) Histologically, scored tissue damage was significantly more prevalent within inflamed tissue, than in control.

Discussion
Current clinical diagnostic and assessment methods for IBD include subjective, time-consuming and/or non-specific symptomology, serological, endoscopic and histological scoring, suggesting that there is value in an additional method to characterize disease activity that is real time and objective. We have developed transmural impedance as an in vivo technique that provided real-time quantitative data during experimentally induced inflammation of the jejunum. The decrease in transmural impedance correlated with the increase in T cells, neutrophil and eosinophil numbers, the quantification of which is a reliable and accepted measurement of inflammation. Therefore, the change in transmural impedance reported here can provide an in vivo indicator of mucosal integrity that can be measured in real time. Stable control impedance values were generated from an electrode in an unchallenged region of rat small intestine over the entire 210 min experiment. Following TNBS injection, a rapid decrease in impedance occurred which was correlated with inflammatory damage that was assessed by histology or immune cell infiltration. The TNBS-induced model of jejunitis is an acute model of inflammation and does not fully represent the changes that occur in chronic IBD. That aside, our data support the measurement of transmucosal impedance as a marker for TNBS-induced inflammation.
The technology required to measure the induced voltage transients is standard for many neural prostheses and has a proven long-term safety record [28]. Moreover, the simplest implementation of the system requires only two electrodes (one intraluminal and one extraluminal). Such simplicity lends itself to integration into existing devices such as endoscopes. Thus, quantitative gut wall impedance data could be generated during endoscopy examinations, an idea that has recently been patented [29], and related to qualitative physician scoring of inflammation by visual inspection. In future studies, we aim to evaluate this technology in a clinical setting.
Assessment of mucosal integrity might have diagnostic application in other conditions, such as non-alcoholic fatty liver disease [30], which is characteristically associated with increased permeability of the duodenum, in non-erosive reflux disease (NERD) or in coeliac disease, in which a proportion of patients experience mucosal integrity disruptions and epithelial cell damage without visible signs of macroscopic alteration [31][32][33]. Because endoscopic abnormalities are not accurate indicators of NERD or coeliac disease, an objective measure that is undertaken in real time during the endoscopic procedure, without introducing additional risk, could augment current clinical assessment. The method may also have applicability to the assessment of IBD patients who are in endoscopic remission.
A second application for this technology is the longitudinal recording of real-time measurements of gut inflammation for closed-loop neuromodulation. A number of animal studies demonstrate that vagal nerve stimulation is a feasible, effective therapy for IBD that could mitigate risks associated with pharmacological intervention [17,[34][35][36]. Neuromodulation of the vagus nerve may be preferable in patients who experience adverse side effects to common pharmaceutical therapies such as nonsteroidal anti-inflammatory agents, corticosteroids and/or immunosuppressant drugs [37,38]. Owing to the episodic nature of IBD, in which patients can experience inflammatory episodes several times a year to once every 4 years [1], therapy is often administered to patients longer than is clinically required, thus risking side effects. Precise delivery of stimulation, while minimizing adverse side effects, is desirable for optimal therapy. The measurement described here is a biomarker of gut inflammation that can provide real-time information required for closed-loop control of vagal nerve stimulation in the treatment of IBD.
Other methods of determining tissue impedance exist, including electrochemical impedance spectroscopy and four-electrode impedance techniques [39,40]. While these techniques can provide rich data about the conductance of the tissue, they are typically more challenging to implement clinically and would require additional hardware, either in the form of additional electrodes (for four-electrode impedance measurements) or dedicated hardware for the generation and recording of sinusoidal waveforms (for electrochemical impedance spectroscopy). By contrast, the technique described here of monitoring voltage transients uses technology that is available in contemporary neural stimulators.
In conclusion, we have developed an objective, real-time measure of gut permeability that can be used to assess gastrointestinal inflammation in vivo. We consider this technology and methodology has translatable application in the evaluation of IBD.