A novel distinctive form of identification for differential diagnosis of irritable bowel syndrome, inflammatory bowel disease, and healthy controls

Irritable bowel syndrome (IBS) is a functional disorder affecting around 11% of the world population, which diagnostic is mainly based on clinical parameters. IBS shares many symptoms with other gastrointestinal disorders such as inflammatory bowel disease (IBD), which makes positive diagnosis a difficult task.


| INTRODUC TI ON
Irritable bowel syndrome (IBS) is a functional disorder in which recurrent abdominal pain is associated with defecation or changes in bowel habits. 1 IBS affects around 11.0% of the population 2 and only a minority of IBS patients (30%-50%) seek healthcare, accounting for 25% of the visits to a gastroenterologist and up to 12% of the visits to primary care doctors. 3 Thus, it results in the generation of a substantial workload in both primary and secondary care being a significant socioeconomic burden.
IBS pathophysiology is exceptionally complex since it involves different factors such as abnormal intestinal motility, visceral hyperalgesia, increased intestinal permeability, immune activation, altered intestinal microbiota, and disturbance in brain-gut function. 2 Amongst people who meet clinical criteria for IBS diagnosis, symptoms severity varies over a broad spectrum, ranging from very mild to incapacitating, 4 which makes IBS patients have a worse quality of life when compared to the healthy population. 5 Despite the fact that IBS is a bothersome disorder with high prevalence and its pathophysiology is quite well known, to date, no positive test diagnosis exists. Several clinical diagnostic criteria (ie, Kruis, Manning, Rome) have been traditionally used to distinguish IBS patients from those with organic bowel disease in daily clinical practice, 6 being the most recent and commonly used the Rome IV criteria. 2 Nonetheless, there is still considerable overlap between IBS symptoms and those shown by some organic diseases such as Inflammatory Bowel Disease (IBD). Therefore, the current diagnose of IBS consists of conducting a series of tests (including laboratory tests, imaging tests, and endoscopies) in order to rule out some other diseases that may mimic IBS. Currently, it is estimated that up to 45% of patients wait for more than one year for the diagnosis of these conditions and up to 17% wait for more than 5 years.
Nowadays, one of the biomarkers most extensively used in clinical practice to differentiate IBS from IBD is Faecal Calprotectin (FC). Its abundance in faeces highly correlates with the inflammation activity found in the bowel mucosa, 7 with a high sensitivity to discriminate IBD. This biomarker is especially suited for screening IBD since the gold standard method to diagnose it is the colonoscopy. However, colonoscopy procedures are not considered the best option due to their associated costs, ever-increasing waiting lists, risks, and patients' inconvenience. Unfortunately, a considerable number of IBS are diagnosed as false positive, and all these patients must undergo an unnecessary colonoscopy.
Although the aetiology of this disorder has not been determined to date, research findings have revealed that IBS patients feature metabolomic changes and alterations in colonic fermentation, and that gut microbiota may be relevant for the disease pathogenesis. [8][9][10] Some studies report that the change in intestinal microbiota caused by acute gastroenteritis is associated with an increased risk of subsequent development of IBS. [11][12][13] Besides, antibiotic therapy, even when given systemically, has also been significantly associated with IBS. 14 A recent systematic review has reported an exhaustive analysis of the literature, demonstrating the presence of pro-inflammatory species in the gut microbiota of patients with IBS, including phylum Proteobacteria, family Enterobacteriaceae, and genus Bacteroides (phylum Bacteroidetes). 15 Additionally, potentially beneficial bacterial species such as Faecalibacterium prausnitzii were also found in the microbiota associated with IBS patients. 15 The variety of techniques and samples used in the different studies may hamper reaching a consensus on the IBSdysbiosis signature. 10 Methods: A cohort consisting of 165 subjects (52 IBS and 52 IBD patients, and 61 healthy controls) was recruited from the Gastroenterology Department of six hospitals. Each patient provided a stool sample from which DNA was extracted, and microbial markers composing RAID-Dx were analysed by qPCR. The results obtained were used to define and validate the RAID-Dx algorithm.

Results:
The abundance of the biomarkers included in the algorithm differed according to the diagnosis. RAID-Dx shows a high capacity to diagnose IBS with a sensitivity of 82.4% and a specificity of 85.7%. RAID-Dx also reports higher sensitivity and specificity values than faecal calprotectin for IBS and IBD differentiation.
Conclusions: RAID-Dx is a noninvasive tool aimed to diagnose IBS with high sensitivity and specificity. The use of this new tool for IBS diagnosis could significantly improve disease management, minimise its misdiagnosis and increase patients' quality of life.   Table 1.

| Ethical considerations
The study protocols (clinical investigation code: RAIDCD2016_2,

| Faecal sample collection
Patients collected faecal samples from a single bowel movement at home in a sterile container of faeces. Samples were immediately frozen after deposition in a domestic freezer. Patients collected the samples the week prior to their scheduled colonoscopy and before undergoing bowel cleansing. Those patients without a scheduled colonoscopy had no restrictions on sample collection. Then, subjects brought samples to the hospital, where they were kept frozen at −20°C for short-term storage, and at −80°C upon arrival at GoodGut SL facilities in Girona (Spain).

| Faecal calprotectin determination
The concentration of FC was measured at LABCO (SynLab -Barcelona), using a quantitative enzyme-linked immunosorbent assay (ELISA, Buhlman Test). Sensitivity and specificity values obtained using those predetermined cut-off values established for the IBS and IBD discrimination were examined (50 and 150 µg FC/g of faeces). 19

| DNA extraction from stool samples
Genomic DNA was extracted from faecal samples after homogenisation using NucleoSpin® Soil (Macherey-Nagel GMbH & Co., Duren, Germany) by following the manufacturer's instructions. DNA was finally eluted in a 100 µL of SE Elution Buffer and stored at −20 °C until its use.  Table 2), and between 12 and 20 ng of genomic DNA template. Thermal profiles were different depending on the biomarker analysed (Table 3).

| qPCR assay for IBS biomarkers
Primers used in this study were purchased from Macrogen (Macrogen, Seoul, South Korea). All quantitative PCRs were run on an AriaMx Real-time PCR System (Agilent Technologies, Santa Clara, USA). A melting curve step was added at the end of each qPCR when GoTaq qPCR Bryt Master Mix was used to verify the presence of the expected amplicon size as well as to control primer dimer formation.
Data were collected and analysed with the Aria Software version 1.5 (Agilent Technologies, Santa Clara, USA). All samples were amplified in duplicates, which were considered valid when the difference between threshold cycles (Ct) was less than 0.6 or than 1.0 at Ct lower or higher than 28, respectively. Moreover, a nontemplate control reaction was included in each qPCR run.

| Statistical analysis
Data normality was assessed through the Kolmogorov-Smirnov test. The nonparametric Kruskal-Wallis and Mann-Whitney tests were used to analyse differences amongst groups or pairwise comparisons, respectively. All comparisons using microbial markers were performed between the relative abundances, which were normalised by the total bacterial load abundances. In this proof-of-concept study, analysis to determine which combination of microbial markers was capable of distinguishing IBS patients from those healthy controls and IBD patients was performed.
The methodology used consisted of initial training with the 70% of a random partition of the data set and further validation with the 30% left of the data set. RAID-Dx was eventually designed by the combination of eight bacterial markers and one archaeal marker. The final algorithm is based on a Decision Abundance (DA) calculated using the following equation: where Ct is the threshold cycle; b is the intercept point; m is the slope; ind, is the microbial marker; and EUB are eubacteria (total bacterial load). The values for each biomarker are listed in Table 4.
The sample size effects have been calculated using G*Power

| Faecal microbial markers in irritable bowel syndrome patients
Eubacteria have been used to normalise the abundance of all the microbial markers as the total bacterial load in order to avoid data bias. The relative abundance of each microbial marker has been analysed, and significant differences amongst diagnoses were found ( Figure 1). PHGII showed significant differences amongst all the diagnoses (P < .001) being higher in healthy controls and lower in IBD patients. Patients diagnosed with IBS showed a lower relative abundance of BAC than healthy controls and IBD patients (P = .005 and P = .001, respectively), indicating its potential as an IBS biomarker.
Besides, two different microbial markers have been shown to be less abundant in IBD patients when compared to healthy controls and IBS patients, AKK (P < .001 and P = .009, respectively) and MSM (P < .001 and P = .002, respectively). RUM has been shown to be a good biomarker to differentiate healthy controls, since it presented higher relative abundance than in IBS (P = .023) or IBD (P < .001). Also, PHGI presented significant differences between healthy controls and IBD (P = .018). Finally, the total bacterial load presented no significant differences between any of the compared populations (P = .434).
A more in-depth exploratory data analysis was performed com- ROC curves analysis confirmed that the reduction of a single species load in IBS patients is insufficient for diagnostics purposes.

| Microbial markers vs faecal calprotectin for the differential diagnosis of IBS and IBD
Receiver operating characteristics curves analysis was performed to test the accuracy of those biomarkers, which showed abundances significantly different between the two populations ( Figure 3). The best results of AUC were obtained with BAC with a value of 0.689, followed by MSM with a value of 0.680. However, faecal calprotectin obtained the highest AUC, with a value of 0.874 (Table 5). Despite the highest AUC when this situation appears. Therefore, 150 µg/g of faeces is also considered another pre-determined cut-off. In our cohort, this cut-off obtained a sensitivity and specificity of 76.0% and 83.3%, respectively.

TA B L E 4 Slope and intercept point (m and b from Equation 1) with efficiency and r 2 for each microbial marker
ROC curves analysis showed that a single species relative load in IBS could present high values of sensitivity; however, the specificity is highly reduced when compared to FC, the currently used methodology.

| RAID-Dx algorithm development and validation
An algorithm combining the previously analysed biomarkers was developed in order to differentiate IBS patients from IBD patients and healthy controls, and lately validated. As commented above, in this proof-of-concept, initial training with 70% of a random partition of the data set was used for the definition of the algorithm, and the remaining 30% of the data was used for its validation ( Table 6).
The development of the RAID-Dx algorithm was focused on obtaining higher sensitivity and specificity values to diagnose IBS.
The combination of the relative abundance of the above described eight functional species led to the achievement of an algorithm with When the defined algorithm is only applied to IBS and IBD population, it resulted to a substantial increase of the sensitivity value when compared to FC pre-determined cut-off (training partition of 94.3% and validation partition of 88.2%); whereas specificity, in this case for IBD diagnosis, was maintained.  in conjunction with the nonliving components of their environment.

| D ISCUSS I ON
Changes or effects that may occur either due to environment modification or in the balance of the intestinal microbiota will not only affect a single species but the whole system as it was observed in IBD. 37 Therefore, the combination of different species provides more robust and reliable results in the diagnosis of different diseases.
Currently, FC is the most widely used faecal marker, both in primary and secondary care, for differentiating IBS from IBD due to its high accuracy in ruling out intestinal inflammation. 19 As commented, a value of 50 µg FC/g of faeces has been the most commonly adopted cut-off. In the review by Mumolo et al, 19

AUTH O R S H I P
Guarantor of the article: Dr Xavier Aldeguer.
Author contributions: Sara Ramió-Pujol: study design, study conduction, data analysis, data interpretation, and drafting the manuscript. Joan Amoedo: study design, study conduction, sample analysis, data analysis, and data interpretation. Mariona