Enhancing our understanding of small bowel function using modern imaging techniques

Small intestinal function is critical to digestive health and patients believe an abnormal reaction to food is responsible for many of their symptoms. Despite this, our ability to assess disturbed function in clinical practice has been limited, particularly after ingestion of the complex nutrients which make up normal food. Recent advances in both wireless capsules and magnetic resonance imaging have provided new insights. This review will briefly describe the limitations of past techniques and focus on how these newer techniques are changing our understanding, particularly of how patients' gastrointestinal tracts respond to food.

The stimulus for this review has been a burgeoning of new data derived from a range of novel, patient acceptable ways of assessing function. The focus of this review will be on new methods, especially MRI, which have opened up the area, allowing the study of responses to complex mixed nutrient meals and in both healthy volunteers and patients.
The key functions of the small intestine are to receive acidic gastric chyme, neutralize the acid and mix it with bile to optimize the conditions for pancreatic and intestinal enzymes to allow digestion and absorption of complex nutrients. While barium contrast radiology gives excellent definition of anatomy, it is a poor way of assessing function as the barium suspension used lacks normal nutrients.

| INTE S TINAL PERFUS I ON TECHNIQUE S
Assessing absorption/secretion has been performed using intestinal perfusion. This requires the passage of multi-lumen tubes into the intestine through which test solutions incorporating an non-ab-

| B RE ATH TE S TS TO E VALUATE MUCOSAL FUN C TI ON AND TR AN S IT
The breath tests based on the fermentation of unabsorbed carbohydrate by colonic bacteria provided a technique which was easy to apply to large numbers of patients and were initially enthusiastically adopted, though recently recognition of the many limitations, particularly of the lactulose and fructose breath hydrogen test including their very poor reproducibility 7 , means this enthusiasm has waned.

| Lactose breath hydrogen test
This was the earliest breath test to be introduced to clinical practice and remains widely used, though in some laboratories it has been replaced by the more convenient genetic test. 8 It assesses the rise in breath hydrogen after ingesting 50 g of lactose which occurs when malabsorbed lactose enters the colon. This proves a good predictor of response in practice since it simulates real life for example, milk drinking and the signal measured assessed changes in the colon which caused relevant symptoms such as flatulence and diarrhea.

| Xylose breath test
d-Xylose is a pentameric monosaccharide which is absorbed unchanged by an active sodium-dependent transporter in the upper small bowel mucosa. Around 50% of absorbed d-xylose is rapidly excreted in the urine. Absorption is limited allowing its urinary excretion in the 5 hours following ingestion of 25 g to be used as a measure of small bowel function. 9 Excretion of >5 g is normal while subjects with small bowel disease such as celiac disease excrete around 1-2 g.
Collecting urine is inconvenient and often incomplete, so a serum xylose at 1 hour is often used and a level <25 mg/dL considered diagnostic of malabsorption with high sensitivity and specificity (91% and 98%, respectively). 9 These tests of intestinal function were developed at a time when obtaining duodenal biopsies required intubation under radiological screening. They are rarely performed now that more specific diagnoses of the causes of malabsorption can so readily be made by endoscopic duodenal biopsies.

| Lactulose breath test to measure transit
Lactulose, being a disaccharide which cannot be digested by any human produces a steep rise in breath hydrogen on entering the colon, the timing of which has been used as a measure of small bowel transit. It depends on the rapid production of hydrogen (6-8 min after direct infusion into cecum via orocecal tube 10 ) which is true in most individuals, though a minority do not produce detectable hydrogen which is consumed via alternative pathways to yield methane, hydrogen sulfide, or acetate. The quantitative response differs between individuals since it depends on their unique microbiota. These in turn depend on diet as seen by the fact that prefeeding with lactulose has been shown to increase the enzymes needed to ferment lactulose (b-galactosidase). 11 One of its limitations is that it alters what is being assessed, since like lactose in a lactose malabsorbing patient, it induces water secretion and distension of the small bowel 12 with secondary acceleration of transit. 13

| SCINTI G R APHI C A SS E SS MENT OF S MALL BOWEL TR AN S IT
Incorporating radioisotope into a test meal allows assessment of when isotope moves from the confines of the terminal ileum into the larger volume of the cecum and ascending colon, though this is inevitably some what subjective. This subjectivity can be reduced by measuring the % of the dose that has entered the colon at intervals and from extrapolating the plotted curves, the time for 50% of isotope to enter can be estimated. The difference from the time to 50% gastric emptying then gives a measure of small bowel transit.
Unlike lactulose, the isotope has no effect on transit and the method does not depend on the microbiota. It has been used to show delayed small bowel transit in chronic intestinal pseudo-obstruction 14 and the acceleration of small bowel transit by bran. 15 While this has been used to assess drug effects, 13,[16][17][18][19] it has not been widely adopted clinically since isolated disorders of small bowel transit are

Key Points
• Understanding of how the small intestine responds to normal mixed nutrient foods has been limited by difficulties in access • This review outlines advances in non-invasive imaging, both endoscopic and magnetic resonance imaging which have enhanced our understanding • There is a wide range of response to different foods which may underlie some of the food intolerances our patients report uncommon 18 and to some extent it has been replaced by wireless capsule, which avoids exposure to ionizing radiation.

| A SS E SS MENT OF AB NORMALITIE S OF MI CROB I OTA IN S MALL INTE S TINE
The earliest clinical studies used jejunal aspirates to look for fecal organisms and concluded that these were usually associated with blind loops including jejunal diverticulae and surgically created "blind loops" which allowed colonic conditions to develop. 20 Subsequent studies showed that similar abnormal growth of organisms capable of deconjugating bile salts as assessed by the 14 CO 2 glycocholate breath test could also be detected in patients with impaired motility and absence of the migrating motor complex. 21 Numerous studies make it clear that normal commensals of the oropharyngeal cavity including members of the Streptococcus, Staphylococcus, Lactobacillus, Neisseria, Veillonella, and Corynebacterium families will be detected in the small bowel if gastric acid is absent due to proton pump therapy, gastric atropy, or other diseases such as cirrhosis. Prospective studies suggest that prolonged PPI therapy can increase SIBO and is associated with vague abdominal symptoms such as bloating and flatulence. 22 The presence of fecal microbiota including coliforms and Bacteroides which require a microaerophilic environment may indicate more severe abnormalities and also be associated with evidence of damage to the mucosa and frank malabsorption. 14 C-labeled xylose was also used as a non-invasive method to show small intestinal bacteria overgrowth (SIBO) and performed well compared to jejunal aspiration. 23 However, 14 C has a half-life of 100 000 years and is retained within the body so the radiation exposure is unacceptable and these tests are no longer used though 13 C-xylose test is available. Intubation is impracticable in normal clinical practice and most centers will now use duodenal aspirate taken at endoscopy. At the Mayo clinic, the criteria for abnormality are >10 5 colony-forming units(cfu)/mL or counts of anaerobic organisms of >10 4 cfu/mL, the lower threshold recognizing that anaerobic organisms represent a more abnormal state and are more likely to be associated with disease. 24

| L AC TULOS E B RE ATH HYDROG EN TE S T TO DE TEC T S MALL INTE S TINAL BAC TERIAL OVERG ROW TH
Quantitative culture of intestinal aspirates requires samples to be transported anerobically and rapidly processed which is very demanding in terms of time and effort so simpler alternatives would be attractive.
Early reports suggested that small intestinal bacterial overgrowth (SIBO) could be detected by the presence of an abnormal lactulose breath hydrogen test (LBHTs). IBS patients were reported as more likely to show an abnormal LBHT with either (a) a rise in breath hydrogen <90 minutes after ingesting 10 g of lactulose in 240 mL of water or (b) a rise >20 ppm within 180 minutes, or (c) the presence of two distinct peaks. However, subsequent larger studies failed to show any difference between healthy controls and IBS patients using such criteria. [25][26][27] A recent improvement in the lactulose breath hydrogen test incorporating a scintigraphic marker of cecal arrival of the lactulose suggested that most case of early rise in breath hydrogen were due to fast orocecal transit. 28 The proportion of IBS who show a breath hydrogen rise before cecal arrival depends on the thresholds used. Using a 20 ppm threshold suggests only 3% of IBS, and no healthy controls have SIBO, but this rises to 17% of patients and no controls if 10 ppm is used. 29 When the more demanding criteria of jejunal aspirates having >10 5 organisms/mL are used only 4% of IBS have small intestinal bacterial overgrowth. 25 If less stringent criteria of > 5 × 10 3 /mL are used then 12% of controls and 43% of IBS patients meet criteria though this may reflect greater proton pump utilization in IBS which is associated with similar increases in small bowel organisms. Validation against jejunal aspirate has only rarely been performed but suggests the LBHT performs poorly, with low sensitivity and specificity. 30 The small benefit seen with rifaximin treatment cannot be used as evidence of the importance of SIBO since it may reflect changes in the colon rather than small bowel so the role of breath testing for SIBO in IBS remains unclear. 31 More direct studies on microbiota currently being done using small bowel mucosal biopsies for culture and DNA-based assessments may answer these potentially important questions.

| G LUCOS E B RE ATH HYDROG EN TE S T
Measuring the breath hydrogen response after ingesting 50 g of glucose has been widely used and is more sensitive in detecting small intestinal bacterial overgrowth than the 14 C-labeled xylose breath test. 32 However, the performance of these tests requires organisms which ferment glucose to hydrogen and depends on the precise clinical groups studied. Direct comparison with aspirates suggests glucose breath hydrogen test has poor sensitivity, as low as 27% in some series. 30 Furthermore, recent studies show poor correlation between bacterial numbers either by culture or direct DNA-based assessment and breath hydrogen. 33 Fast transit to the colon as shown by simultaneous scintigraphy is a common cause of false positives after gastric surgery. 34

| A SS E SS ING G A S IN THE S MALL BOWEL
Few imaging studies have examined this owing to the overlap of small bowel and colon making it difficult to precisely allocate gas bubbles to gut regions. One study using CT scanning showed that in healthy subjects the gut contained a median(interquartile range) of 95 (71-141) mL of which 68 (58-83)% was in the colon leaving about 1/3rd in the small bowel, mostly in small bubbles containing <6 mL.
Around 5% of the total gas lay in the region of the terminal ileum, which is where bacterial fermentation of poorly absorbed dietary residue is most likely given that the terminal ileum contains around 10 7 organisms per mL. How important this is remains unclear since there was no difference in small bowel gas in patients with FGIDs compared to healthy controls. 35 A novel gas sensing radio-pill, 25 mm long and 9.8 mm in diameter which can detect methane, carbon dioxide, hydrogen, and oxygen has recently been developed and used to study fermentation of carbohydrate in the gut. 36 The oxygen signal falls rapidly after leaving the stomach and its arrival in the anaerobic colon is readily detected so in theory it should be able to detect fermentation in the small bowel suggestive of SIBO. Results in large numbers of IBS patients are awaited with interest.

| WIRELE SS C APSULE TELEME TRY
The idea of using pressure and pH-sensitive pills which could be ingested and pass throughout the gut is not new, being first developed in the 1960s. 37 However, improvements in electronics have permitted the production of much-improved devices allowing both endoscopy as well as assessment of pressure, pH, and temperature. 38,39 The SmartPill capsule measures 26 mm × 13 mm making it relatively easy to swallow, however, its size means that it is likely to require a migrating motor complex to push it through the pylorus. This

| ENDOLUMINAL IMAG E ANALYS IS TO A SS E SS S MALL BOWEL MOTILIT Y
It is a major problem that intubation to perform manometry is stressful and that this undoubtedly alters both gastric and small bowel motility. A novel approach, which largely avoids this problem, has been to use capsule endoscopy to non-invasively quantify wall movements. A study of patients with manometrically diagnosed pseudoobstruction compared with health controls showed the patients had less phasic luminal closures and more images showing no movement.
It also identified a group of symptomatic patients who did not show manometric abnormalities yet did have abnormalities identified with the new technique. 41 Scoring the many thousands of images is very time consuming but a subsequent much larger series of 80 FGID patients using sophisticated automated image analysis showed that the method could identify around 29% with abnormalities, around half having hypodynamic and the other half hyperdynamic motility. 42

| MAG NE TI C TR ACK ING SYS TEM
Magnet Tracking System (MTS-1) is a novel technique involving tracking the movement through the gastrointestinal tract of a small magnetic pill (dimensions: 6 × 15 mm, weight: 0.9 g, density: 1.8 g/ cm 3 ). 43 Its movement from stomach to small intestine can be determined from the cessation of regular 3/min movements due to gastric contractions. Ileocecal passage, which was identified as cessation of the 8-10 min −1 contraction frequency of the small intestine is more problematic, as in the ileum contractions are fewer and more irregular. The most unexpected finding using this technique is that more than half the distance covered during the first 2 hours in the small intestine is due to fast movements which only occupy 5% of the time. This was true both fasting and fed. 44 The MMC may well account for rapid fasting movements but the large rapid movements during feeding are unexpected. However, these may be important in spreading chyme along the small bowel to speedup absorption.
The role of this technique in clinical practice or research remains unclear; however, measuring transit alone is of limited value and usually needs to be combined with assessments of other small bowel functions such as absorption/secretion.

| Clinical use
Modern imaging techniques, which allow high-quality images to be obtained within a few 100 milliseconds, together with the lack of ionizing radiation have made MRI a very attractive way of assessing small bowel anatomy in clinical practice. This is particularly true for patients with Crohn's disease, who often require repeated imaging over a lifetime, which in the past has exposed them to potential harmful doses of radiation. However, the use of large doses of osmotic agents such as mannitol, while providing excellent anatomical definition of mucosal pathology does disturb digestive function and a rather different technique using the undisturbed bowel is required to assess function.

| Measuring small bowel water content
The

| OS MOTI C EFFEC T OF P OORLY ABSORB ED SOLUTE S
Lactose intolerance has been well recognized for over 50

| A SS E SS MENT OF D I G E S TI ON OF COMPLE X FOODS
Most food we eat is not composed of simple solutions but complex mixtures of liquid and solid. Sampling these using intestinal tubes is virtually impossible but MRI allows a non-invasive assessment of fluid fluxes, intestinal distension, and entry of meal residue into the colon with subsequent fermentation, all features which can be linked to a range of symptoms described by patients with FGIDs.
Carbohydrate, derived from polymers used by plants to store energy such as wheat, rice, potatoes, and other tubers, is a dominant part of most patient's diets. Cooked rice is a particularly welltolerated food, being rapidly and fairly completely absorbed as shown by breath hydrogen testing. 50

| G A S TROINTE S TINAL RE S P ON S E TO FODMAPS
The recognition that poorly absorbed but rapidly fermentable dietary carbohydrate, so-called FOMAPs (fermentable oligo-di-and monosaccharides and polyhydric alcohols) was responsible for at least some of the symptoms of the ubiquitous irritable bowel syndrome 51 has led to a substantial improvement in the management of IBS.
Understanding exactly how FODMAPs cause symptoms requires imaging after a provocative meal containing FODMAPs. MRI studies showed clearly that, while the poorly absorbed but osmotically F I G U R E 2 Small bowel water content over time following ingestion of two iso-osmolar solutions of glucose 5% and mannitol 5%. There was a striking rise in water content with mannitol compared to the rapid absorption of the glucose solution 48 This passes through the small bowel without causing distension but is rapidly fermented once it enters the ascending colon, producing obvious gaseous distension. 52 Interestingly, IBS patients did not appear to show any greater physiological response to either fructose or inulin compared to healthy volunteers but had markedly greater symptoms, reflecting their known visceral hypersensitivity. 53 The best correlation with patients' composite symptom score (CSS) was with peak increase in colonic gas volume after inulin. A weaker correlation between SBWC and symptoms was seen after fructose in keeping with its greater impact in the small bowel.
Many patients believe that bread causes symptoms particularly bloating and a low FODMAP diet often involves excluding normal bread which contains about 1% FODMAPs, mostly fructo-oligosaccharides including raffinose and stachyose. 54 While this might lead to excess colonic gas and hence bloating/gas 4-6 hours after ingestion, the bloating developing immediately after eating is more likely due to gastric distension. MRI studies comparing equicaloric rice meal with wholemeal bread showed that while the rice pudding meal separated into a solid and liquid phase, the wholemeal bread formed a rather homogeneous mass with no clear separation of fluid and solid. While the rice meal allowed the stomach to sieve the meal and empty the fluid faster than the solid, sieving was not possible with the more homogeneous bread meal and as a consequence the gastric volume fell more slowly. In contrast once the wheat entered the small intestine, it was rapidly absorbed and the SBWC was consistently lower than after the rice meal postprandially. This suggests than any sensation of bloating at this time was more likely to come from the stomach than the small intestine. 55 Gluten is a key determinant of dough elasticity and thought to contribute to symptoms so we have investigated how it affects the upper intestinal processing of bread meals by comparing gluten-free bread with bread with normal gluten content and a bread supplemented with 3% gluten. The gastric images showed the added gluten did produce a less homogeneous intra-gastric appearance but this was not associated with any difference in gastric emptying, and all meals showed the same pronounced drop in small bowel water postprandially. 56 Unexpectedly, feeding gluten-free bread for 2 days significantly increased the fasting transverse colonic volumes by 36% compared to normal bread, possibly reflecting the effect of the tapioca and potato starches used to replace wheat in gluten-free products. crease after bran appears to be more likely due to stimulation of secretion since unlike psyllium it shows little water trapping. 59 The resulting ileal contents are emptied into the ascending colon after the second large meal which stimulates ileal emptying. This leads to an increase in the fluidity of the ascending colon contents. As already stated, T1 is a time constant reflecting the decay of magnetization after excitation by radiofrequency energy pulses and increases as the colonic contents become more watery. During the psyllium study, the T1 was significantly higher after both 3.5 and 7 g psyllium being 631 ± 37 and 779 ± 48 msec respectively compared to the maltodextrin control at 447 ± 37 milliseconds. We validated the use of T1 measurements as an assessment of colonic water content by showing that T1 of descending colon contents were well correlated with the % water of stool passed subsequently, Pearson's r = 0.65. 57

| EFFEC T OF PHY TO CHEMI C AL S
While carbohydrate is the staple source of energy, much of what we eat is flavored with plant products such as herbs and spices, all of which influence the processes of digestion. Rhubarb is an important ingredient of many Chinese herbal preparations and reported by ileostomists to increase ileal outflow. 60 Imaging the bowel after rhubarb ingestion shows a large increase in SBWC, followed by a rise in ascending colon T1. 61 Rhubarb is known to contain rhein, a mild cyotoxic substance, shown in animal studies to exert a laxative effect which can be blocked by indomethacin. Rhein stimulates prostaglandin E-2 secretion and downregulates expression of Aquaporin-3 in cell lines which is a possible mechanism of its laxative effect. 62 Interestingly, the same study also showed that lettuce which contains a milk-like latex, lactucarium, which like rhein is a mild cytotoxin also increases SBWC. 63 It is likely that many green vegetables similarly increase SBWC and hence delivery of fluid to the colon with likely beneficial effects in preventing constipation.

| AB NORMALITIE S OF S MALL BOWEL IN CELIAC D IS E A S E AND SCLERODERMA
Both celiac disease and scleroderma have a well-defined appearance in barium contrast studies but as mentioned barium is a non-physi-

| AB NORMALITIE S OF S MALL BOWEL IN IBS
One of the big surprises of recent MRI studies was the demonstration that IBS patients with diarrhea (IBS-D) had less small bowel water than normal (Figure 7). 48,64 This was associated with accelerated mouth to cecum transit and anxiety, 40  an amount similar to that seen in IBS-D. CRH also increased the volume of the ascending colon in the following 3 hours, suggesting that stress causes an increased inflow into the colon. 67 Using the same study design but using a fructose meal, we were also able to show not only that CRH constricted the small bowel but that this also enhanced the subsequent increase in ascending

| A SS E SS MENT OF S MALL BOWEL AND COLONIC MOTILIT Y US ING MRI
The high-quality images of the colon and the ability to acquire images in milliseconds have made it possible to produce cine images of the colonic wall over several minutes and to thus quantify colonic motility. 69 A key requirement for this is the ability to allow image acquisition during free breathing by compensating for respiratory movement by means of an image registration technique. By deforming all the different images from the cine acquisition so that they best match a reference image (from the same dataset), it is possible to derive this deformation field information. This information then allows tracking of any location in the bowel through the cine data and provides information on the local expansion and contraction of each pixel in the image over the acquisition time, with large changes in expansion and contraction associated with higher motility. Summary metrics can then be used to quantify the amount of movement seen as a numerical comparison between different people or treatments. The technique was first used to demonstrate reduced motility in Crohn's ileitis. 72 The method has been validated by showing that it can detect the stimulatory and inhibitory effects of neostigmine and buscopan, respectively, on small bowel motility. 73 It has recently been used to show impaired small bowel motility in pseudo-obstruction and also to evaluate patients' response to prokinetic neostigmine. 74 The method continues to evolve and more recently motility has been used to measure the stimulatory effect of nutrients without the need to distend the bowel with mannitol with good reproducibility. 75,76

| FUTURE S TUD IE S
MRI studies are highly patient acceptable, and this makes it possible to perform repeated studies using RCT design, ideal for drug evaluation to provide early signal of effectiveness and mode of action. It also lends itself to mechanistic studies of the impact of different foods designed for specific function in both health and disease. We are currently developing MR suitable markers which could be used to more accurately assess small bowel transit times and hence assess flow.