Assessing intestinal permeability in Crohn’s disease patients using orally administered 52Cr-EDTA

Background Intestinal permeability can be assessed by monitoring renal excretion of orally administered radioactively 51Cr-labeled ethylenediaminetetraacetic acid (51Cr-EDTA). Although considered safe, patient participation in using radio-labeled tracers is low. Here, we used orally administered 52Cr-EDTA as non-radioactive alternative to assess intestinal permeability in CD and analyzed the association with disease activity, disease location and gut microbial dysbiosis. Materials and methods 60 CD patients with low (n = 25) and increased (n = 35) fecal calprotectin levels (cut-off: 100 μg/g feces) ingested 20 mL 52Cr-EDTA (20 mmol/L) solution whereafter 24-h urine was collected. Urinary 52Cr-EDTA concentrations were quantified using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). Fecal Enterobacteriaceae and Faecalibacterium prausnitzii were quantified using FISH. Correlations between urinary 52Cr-EDTA excretion and other parameters were established using nonparametric Spearman’s correlation coefficients (ρ). Results CD patients with increased fecal calprotectin levels (> 100 μg/g) demonstrated an elevated urinary 52Cr-EDTA/creatinine ratio (772 vs. 636 μmol/mol, P = 0.132). Patients with primarily colonic disease showed the highest 52Cr-EDTA excretion. Importantly, a positive correlation was observed for the urinary 52Cr-EDTA/creatinine ratio and fecal calprotectin levels (ρ = 0.325, P < 0.05). Finally, urinary 52Cr-EDTA/creatinine ratio negatively correlated with the relative abundance of Faecalibacterium prausnitzii (ρ = -0.221, P = 0.092), while positively correlating with Enterobacteriaceae (ρ = 0.202, P = 0.126). Conclusions Orally administered and renal excreted 52Cr-EDTA may be used to assess intestinal permeability in CD and correlates with fecal calprotectin levels and bacterial species relevant to CD. This test may improve non-invasive detection of disease exacerbations and help monitor disease activity.


Introduction
In Crohn's disease (CD), a disturbed balance between the gut mucosal immune system and intestinal microbiota is intimately associated with diminished functioning of the intestinal barrier. This barrier is instrumental in the protection against adhesion and infiltration of intraluminal antigens into the underlying lamina propria, thereby maintaining a healthy gut. An important functional and measurable feature of the gastro-intestinal barrier constitutes intestinal permeability, which is influenced by both the gut microbiota and mucosal immunity. [1] Intestinal microbiota play a pivotal role in preserving gut epithelial barrier integrity. A strong link exists between microbial dysbiosis and intestinal permeability. [2] Intestinal dysbiosis in CD is characterized by increased numbers of (potentially) pathogenic bacteria, such as several members of the Enterobacteriaceae, and a decreased abundance of commensal bacteria, such as Faecalibacterium prausnitzii. Decreased abundancy of Faecalibacterium prausnitzii is consistently observed in CD and is associated with increased inflammation and presumably disrupted intestinal barrier integrity. [3,4] Similarly, Enterobacteriaceae, e.g. adherent-invasive Escherichia coli (AIEC), are known to possess multiple pathogenic mechanisms that can lead to loss of intestinal barrier function. [5] These typical shifts of many bacterial species that are characteristic for CD dysbiosis are putatively related to either gain or loss of intestinal epithelial barrier function (such as Faecalibacterium prausnitzii and Enterobacteriaceae (e.g. Escherichia coli), respectively). [6,7] An impaired intestinal permeability has been associated with many gastro-intestinal (GI) and non-GI diseases. [1] In inflammatory bowel disease (IBD) research, there has been much scientific interest in unraveling the complex association between a comprised gut wall integrity and mucosal inflammation. Until now, however, it is not fully understood whether a defective mucosal barrier in IBD promotes intestinal inflammation or that impaired barrier function is secondary to the inflammatory process. [8,9] Nonetheless, the link between an increased intestinal permeability and increased inflammatory activity is well-established. [10] There is also increasing evidence that an increased intestinal permeability to macromolecules precedes the onset of IBD and/or is a sign of subclinical disease activity. It may therefore be used as an early predictor of clinical relapses in IBD patients. [11][12][13][14] This is illustrated by the fact that a subset of clinically healthy first-degree relatives of Crohn's disease patients are genetically predisposed to an increased intestinal permeability as compared to healthy controls. [15,16] Despite this strong association, inflammatory disease activity has hitherto been monitored inconsistently in clinical practice. Since a poor association exists between clinically active disease and observed endoscopic disease activity, non-invasive biomarkers have increasingly been studied to predict active intestinal inflammation in CD. [17] For instance, biochemical disease parameters such as the fecal calprotectin (FC) level and serum C-reactive protein (CRP) are being widely utilized, since these markers correlate significantly with endoscopically active disease. [18][19][20] In this study, we use fecal calprotectin levels as an accurate indirect measure of inflammatory disease activity. [21,22] Currently, orally administered tracers for intestinal permeability play no significant role in monitoring disease exacerbations in CD, but may aid in the diagnostic accuracy of the currently available panel of non-invasive disease biomarkers and help predict CD disease course. [19,23,24] The clinical utility and validity of these non-invasive tests for intestinal permeability are still not fully established. [1] An optimal intestinal permeability test should have a number of properties. The test or marker is preferably sensitive, accurate, reproducible and safe in order to assist in diagnosis and monitoring of disease flares. In addition, an in vivo marker should be stable and inert, avoiding metabolism and degradation. [1,14] Numerous studies have evaluated intestinal permeability using radioactively 51 Cr-labeled ethylenediaminetetraacetic acid ( 51 Cr-EDTA). [11,12,[25][26][27][28][29][30][31][32][33] However, these previously published studies have been unable to establish a consistent correlation between 51 Cr-EDTA-measured gut permeability and inflammatory disease activity, likely related to relatively small and heterogeneous IBD study cohorts. Apart from this, most of these studies used the less preferred radioactively 51 Cr-labeled EDTA, followed by urinary analysis of gamma radiation. Conjointly, its individual performance in clinical practice remains mostly unfeasible due to complex and impractical detection methods. [1,29] A similar substance has recently been developed to evaluate intestinal permeability, in which the non-radioactive 52 Cr isotope has been incorporated into 52 Cr-EDTA. This allows us to measure urinary 52 Cr content in a safe, inexpensive and precisely validated manner using the highly sensitive method of Inductively Coupled Plasma Mass Spectrometry (ICP-MS). In contrast to radioactively labeled 51 Cr-EDTA, only few studies have evaluated the performance of 52 Cr-EDTA as a tracer for intestinal permeability. [34][35][36] Cr-EDTA has proven to be a sensitive marker for human intestinal permeability, both in health and disease. [35][36][37][38][39] Cr and EDTA form a stable and inert complex with one of the highest affinity of known metals, without any physicochemical interactions that might be expected to occur. [40,41] However, putative correlations between 52 Cr-EDTA-measured intestinal permeability and CD-specific disease parameters have not yet been established.
The aim of this study is to evaluate intestinal permeability through measuring 24-h urinary excretion of orally administered 52 Cr-EDTA in Crohn's disease and its association with inflammatory disease activity, disease localization and two key bacterial marker strains of CD dysbiosis (Faecalibacterium prausnitzii and Enterobacteriaceae).

Study population
Patients aged 18-65 years were recruited from March 2016 until April 2017 at the outpatient IBD clinic of the University Medical Center Groningen (UMCG), Groningen, The Netherlands. All patients were diagnosed with Crohn's disease (CD), according to clinical, endoscopic and histopathological features. All patients were Caucasians and lived in the Northern part of the Netherlands. Patients were divided into two groups based on fecal calprotectin levels as a measure of inflammatory disease activity (using 100 μg calprotectin / gram feces as our currently used cut-off value for defining low and increased inflammatory disease activity). Standard demographic characteristics, including age, sex, body-mass index (BMI) and smoking history, were recorded, as well as clinical parameters specific for CD, such as the Harvey-Bradshaw index (HBI), Montreal disease classification (including Age at diagnosis (A), Location of disease (L) and the Behavior (B) of the disease) and current maintenance therapy. Disease location was recorded from the most recently performed endoscopic evaluation, which was completed within 12 months of intestinal permeability testing. CD patients who were treated with antibiotics 3 months prior to screening were not eligible for this study (to rule out an effect on the microbiota composition). Furthermore, CD patients with severe disease activity, as reflected by a HBI score > 12, or patients with an indication for remission-induction therapy, were also excluded from this study. Routine laboratory examinations were performed, including hemoglobin, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), leukocyte count, thrombocyte count and creatinine levels. Fecal calprotectin levels were quantified by enzyme-linked immunosorbent assays (ELISA) (BÜ HLMANN Laboratories AG, Switzerland) as a routine measurement in the UMCG. Samples were obtained after patients gave written informed consent. This study has been approved by the Institutional Medical Ethical Review Board (in Dutch: Medisch Ethische Toetsingcommissie, METc) of the University Medical Center Groningen (UMC Groningen) (IRB no. 2014/291) and is in accordance with the principles of the Declaration of Helsinki (2013).

Cr-EDTA solution
The chromium-EDTA oral test solution consists of the active ingredients chromium(III)chloride (CrCl 3 .6H 2 O) and disodium edetate (C 10 H 14 N 2 Na 2 OH.2H 2 O), 70% sorbitol liquid crystalline and raspberry essence as sweeteners, 0.1 M NaOH for pH adjustment to reach a final pH between 4.5-5.5, methylparahydroxybenzoate as preservative and purified water (according to Ph. Eur. requirements) as solvent. The solution contained in total 400 μmol of 52 Cr-EDTA (20 mmol/L). Patients were instructed to drink the 52 Cr-EDTA solution (20 mL) together with a glass of water after an overnight fast. Subsequently, patients were asked to fast for an additional 2 hours. Urine was collected for 24 hours from the moment of ingestion of the 52 Cr-EDTA test solution. Patients were also instructed to refrain from the intake of alcoholic beverages or non-steroidal anti-inflammatory drugs (NSAID's) two days prior to taking the 52 Cr-EDTA solution and during the 24-h urine collection period. Full details on the preparation of the 52 Cr-EDTA solution were described previously. [42] Analytical procedure Urinary content of 52 Cr was quantified using Inductively Coupled Plasma Mass Spectrometry (ICP-MS), which is a highly sensitive analytical method for measuring various metals, including chromium isotopes (Nexion 300X, Perkin Elmer, Medlon BV Enschede, The Netherlands). [43] First, 24-h urine samples were 30-fold diluted with 0.5% HNO 3 and 0.01% Triton X-100, after which 103 Rh was added as internal standard (IS). Subsequently, the sample analyte was nebulized in a cyclonic spray chamber, atomized at extremely high temperature (ranging from 6,000-8,000˚C) and ionized using argon (Ar) plasma. Using a vacuum of~20 mL/min, the resulting ions were deflected into a quadrupole in which ions are separated based on their electrical charge and mass, under the influence of an alternating electrical current. Finally, chromium ions were detected via an electron multiplier. Urinary chromium ( 52 Cr) was measured while applying the kinetic energy discrimination (KED) mode, ensuring a general reduction of polyatomic isobaric interferences (such as that of 40 Ar 12 C). Final urinary 52 Cr concentrations (nmol/L) were divided by urinary creatinine concentrations (mmol/L) as correction for hydration. Urinary creatinine levels were quantified by an enzymatic detection method following supplier's instructions (Roche Diagnostics, Roche Modular P Analyzer, Mannheim, Germany).

Quantification of fecal bacteria using fluorescent in-situ hybridization (FISH)
The quantification of Faecalibacterium prausnitzii and Enterobacteriaceae was performed as previously described, with some minor modifications. [44] Patients were asked to provide fecal samples at the time of the 24-h urine collection. Samples were mixed with 4.5 mL filtered phosphate-buffered saline (PBS) and centrifuged at 700 g for 2-3 min. The resulting supernatant was 4-fold diluted with freshly-prepared 4% paraformaldehyde solution and stored overnight at 4˚C. Before counting, samples were coded and randomized by an independent investigator. Serial dilutions were prepared to allow visual counting of total bacteria, Enterobacteriaceae and F. prausnitzii. Each dilution was spread over gelatin-coated glass slides and dried at room temperature. After addition of the appropriate bacterial probes (see Table 1), Eub338 (Rhodamine) for the total bacteria, Fprau645 (FITC) for F. prausnitzii and Ec1531 (CY3) for Enterobacteriaceae, slides were hybridized overnight at 50˚C. [45][46][47] In each glass slide well, 25 fields were manually quantified using fluorescent microscopes (Leica 2 or Olympus BH20 at 100X magnification). Bacteria were quantified using two different filters, FITC or CY3, based on the probe colour. Based on the absolute bacterial counts, the relative percentage of each bacterial group was calculated.

Statistics
Characteristics of the study population were shown as proportions (%, n), means ± standard deviations (SD) or medians with interquartile range (IQR), as appropriate. Normality testing was performed using Kolmogorov-Smirnov tests. Distributions of urinary 52 Cr-EDTA/Cr excretion were presented as median ± interquartile ranges (IQR) and presented in boxplots (10 th -90 th percentiles) grouped by inflammatory disease activity, as determined by the fecal calprotectin level (using 100 μg/g feces as cut-off value). Differences between groups were tested using the independent sample t-test or Mann-Whitney U-test, as appropriate. Partial correlations between urinary 52 Cr-EDTA/creatinine ratio and other parameters were performed using the nonparametric Spearman's correlation coefficient (ρ). Associations between correlated parameters were visualized in scatter plots with smoothed curves. Smoothing was empirically applied by nonlinear regression using 2 nd order polynomial functions with 1/y 2 weighting. Statistical analyses were performed using SPSS Statistics 23.0 for Windows. Pvalues � 0.05 were considered as statistically significant.

Results
Study cohort characteristics are shown in Table 2. Crohn's disease (CD) patients with low fecal calprotectin levels (n = 25) had a mean age of 44.1 ± 13.3 years and consisted of 8 males (32.0%) and 17 females (68.0%), while patients with increased fecal calprotectin levels (n = 35) had a mean age of 40.5 ± 12.2 years and consisted of 10 males (28.6%) and 25 females (71.4%). Patients with increased levels of fecal calprotectin showed significantly elevated white blood cell counts (WBC) (P < 0.05) and erythrocyte sedimentation rates (ESR) (P < 0.05). CD patients with an increased level of fecal calprotectin had a significantly lower alcohol consumption as compared to CD patients with low fecal calprotectin levels (P < 0.01). No significant differences were observed for any other cohort characteristic. Also, clinical disease activity, as measured by the Harvey-Bradshaw Index (HBI), was not significantly different between patients with either low or increased fecal calprotectin levels. In the total study cohort, median HBI score was 3 (IQR: [1][2][3][4][5], indicating that the majority of patients were in clinical remission at the time of inclusion.

Urinary 52 Cr-EDTA/creatinine excretion proportionally increases with fecal calprotectin levels
Distributions of the 24-h urinary 52 Cr-EDTA/creatinine ratio among both groups are shown in Fig 1A. CD patients with increased fecal calprotectin levels showed a trend of elevated    urinary 52 Cr-EDTA/creatinine excretion and fecal calprotectin levels (ρ = 0.325, P < 0.05, Fig  1B), while controlling for gender, disease localization, ileocecal resection status and alcohol consumption.

Correlation analysis between urinary 52 Cr-EDTA/creatinine excretion and relative abundances of CD-associated bacterial species
The absolute and relative abundance of two CD-associated bacteria, F. prausnitzii and Enterobacteriaceae, was quantified by FISH analysis in fecal samples of all patients (see S1 Fig for representative fluorescence microscopy images and S1 Table for quantifications). Among CD patients with below-median and above-median 52 Cr-EDTA/creatinine excretion, there were no significant differences detected in relative abundances (%) of F. prausnitzii and Enterobacteriaceae. Interestingly, we observed a borderline non-significant negative correlation between 24-h urinary 52 Cr-EDTA/creatinine excretion and relative numbers of F. prausnitzii (ρ = -0.221, P = 0.092, Fig 3A). In contrast, a moderately positive correlation, though non-significant, was found between 24-h urinary 52 Cr-EDTA/creatinine excretion and relative numbers of Enterobacteriaceae species (ρ = 0.202, P = 0.126, Fig 3B).

Discussion
In this study, we show that increased urinary excretion of orally administered 52 Cr-EDTA significantly correlates with elevated levels of fecal calprotectin in CD patients. Furthermore, CD patients with solely colonic disease activity show elevated 52 Cr-EDTA/creatinine urinary excretion as opposed to CD patients having only an affected ileum. In addition, we observed an interesting trend of a negative association between 52 Cr-EDTA-measured intestinal permeability and relative numbers of F. prausnitzii, whereas a trend of a positive association was observed between Enterobacteriaceae counts and urinary 52 Cr-EDTA excretion. These correlations, though they lacked statistical significance, may suggest the possibility of a relationship between intestinal permeability and intestinal dysbiosis. Overall, our findings suggest that the measurement of urinary 52 Cr-EDTA excretion might serve as novel approach for determining compromised intestinal permeability, as is characteristic for severe CD. Chromium-EDTA has been widely used as marker for intestinal permeability, due to its high sensitivity and the stability of the Cr-EDTA complex. [35] However, most studies used the radioactive form 51 Cr-EDTA that may have adverse effects due to radiation exposure. Here, we assessed the non-radioactive and inert 52 Cr-EDTA to evaluate intestinal permeability in CD, in combination with quantifying urinary 52 Cr by highly-sensitive ICP-MS. This method is particularly useful in detecting metal isotopes at very low concentrations in a safe, inexpensive and rigorously validated manner (Nexion 300X, Perkin Elmer, Medlon BV Enschede, The Netherlands). The currently applied ICP-MS method appeared to be highly sensitive in detecting 52 Cr, with an internally determined lower limit of quantification (LLOQ) of 9.8 nmol/L. The average concentration of endogenous chrome in individuals in human blood and urine lies in the range of 0.1-0.5 μg/L (1.9-9.6 nmol/L), which falls below the LLOQ of ICP-MS and is thus negligibly small. For instance, the lowest individual value measured in our study cohort was 1352 nmol/L. [48] This study assessed the 52 Cr-EDTA intestinal permeability test in a relatively large study cohort of CD patients. Interestingly, the majority of these patients (70%) were in clinical remission according to the HBI, but still show a large variation in 52 Cr-EDTA-measured intestinal permeability. Our study cohort is larger and less heterogeneous than all previous performed studies that have evaluated the potential application of Chromium-EDTA-measured intestinal permeability in IBD. [11,12,[25][26][27][28][29][30][31][32][33] Patient participation within our study was high, due to the non-invasive and non-radioactive nature of this test. However, careful test monitoring was required since several factors could affect urinary excretion levels of 52 Cr-EDTA, such as adherence to the experimental protocol and the duration of urine collection. [49] Also, in the present study cohort, no endoscopic disease activity data were available, which are preferentially used as representative of inflammatory disease activity in CD. Alternatively, we used the fecal calprotectin level as surrogate marker for CD disease activity. [19,20,[50][51][52] Here, fecal calprotectin levels were directly related to 52 Cr-EDTA-measured intestinal permeability, which has previously not been established.
Few studies specifically focused on the association between increased 51 Cr-EDTA-measured intestinal permeability and inflammatory disease activity. [26,32,53] Hitherto, Berstad et al. found a significant correlation between 5-h urinary 51 Cr-EDTA excretion and calprotectin levels in gut lavage fluid of IBD patients. [53] Another study observed an association between increased 51 Cr-EDTA-measured intestinal permeability and serum levels of acutephase reactants and 111 In-labeled leukocyte scans. [32] In UC, a significant correlation between 51 Cr-EDTA-measured intestinal permeability and endoscopic disease activity was demonstrated. [26] In CD, it has been shown that anti-TNF treatment attenuates mucosal inflammation and also improves intestinal integrity measured by orally administered 51 Cr-EDTA. [54] Although the relationship between disease activity and Cr-EDTA-measured intestinal permeability seems to be well-established, studies show weak comparability because of various study designs, different intestinal permeability tracers and varying time intervals. Furthermore, it is difficult to clearly demonstrate this relationship since many factors are known to affect intestinal permeability, such as disease localization, surgical history, alcohol consumption, medication use, commensal bacteria and many dietary components. [55] As a consequence, the extent of this relationship remains inconclusive. In addition, all the previously mentioned studies are remarkably contradictory as to the primary intestinal location and degree of increased 51 Cr-EDTA bowel passage. [25][26][27][28][29] In our study cohort, CD patients with exclusively colonic disease activity showed a higher median urinary 52 Cr-EDTA/creatinine excretion as compared to patients with ileal involvement, though non-significantly. Similarly, we could not demonstrate any statistically significant differences in intestinal permeability between three different subgroups of disease localization (i.e. ileal, colonic or ileocolonic disease).
Until now, it remains unclear whether increased intestinal permeability is secondary to subclinical mucosal inflammation in CD or vice versa. [56,57] Accumulating evidence suggests that increased intestinal permeability may be the consequence of a genetically pre-existing epithelial barrier abnormality in certain individuals that predisposes them to the onset of intestinal inflammation. [58] Various studies have shown that a subset of clinically healthy firstdegree relatives of Crohn's disease patients show increased intestinal permeability as compared to healthy controls. [15,16] Moreover, it has been hypothesized that impaired intestinal permeability may already be apparent long before any mucosal inflammation is present and could be predictive of clinical relapse. In addition, another study demonstrated that in firstdegree relatives of CD patients subclinical intestinal inflammation is present, reflected by an increased fecal calprotectin level. [59] In our study, we found a significant correlation between intestinal inflammatory disease activity (as represented by fecal calprotectin levels) and intestinal permeability (as measured by urinary 52 Cr EDTA excretion). Overall, this indicates that 52 Cr-EDTA-measured intestinal permeability might be particularly useful in defining the first stages of development of Crohn's disease in a subgroup of unaffected, genetically susceptible individuals.
CD patients typically demonstrate a decreased gut microbiota diversity as compared to healthy individuals. [60] An important observation in CD is the reduction in the commensal anaerobic, butyrate-producing bacterium F. prausnitzii, as well as an increased number of Enterobacteriaceae (e.g. Escherichia coli). [3,6,61] In the present study, we observed a negative non-significant relationship between 52 Cr-EDTA-measured intestinal permeability and the relative abundance of F. prausnitzii, while a positive non-significant relationship was observed between the relative abundance of Enterobacteriaceae and urinary 52 Cr-EDTA excretion in CD patients. These interesting associations (though moderate and non-significant) support the existence of a link between the gut microbiota composition and the integrity of the intestinal barrier. Future studies are needed to further elucidate this relationship, for example through in vitro modeling of host-microbe interactions using bacteria-gut epithelial co-culture systems. [6] In this way, the influence of specific bacterial species on epithelial markers for intestinal permeability may give more fundamental insight into this complex association.
In conclusion, we show that a moderate positive correlation exists between 52 Cr-EDTAmeasured intestinal permeability and fecal calprotectin levels. In addition, we demonstrate interesting, though non-significant correlations between intestinal permeability and two key gut bacterial species, suggesting the possibility of a relationship between dysbiosis and permeability. Future studies should primarily focus on the utility of orally administered 52 Cr-EDTA as measure of intestinal permeability in relation to endoscopic disease activity and its potential role as predictor of CD disease exacerbations.  Table. Median numbers of F. prausnitzii and Enterobacteriaceae per gram of feces and their relative abundances (%) to total microbiota (Eub338). Data are shown for the total study cohort (n = 60), and for subgroups of CD patients with below-and above-median urinary 52