Manometric demonstration of duodenal/jejunal motor function consistent with the duodenal brake mechanism

High‐resolution manometric studies below the stomach are rare due to technical limitations of traditional manometry catheters. Consequently, specific motor patterns and their impact on gastric and small bowel function are not well understood. High‐resolution manometry was used to record fed‐state motor patterns in the antro‐jejunal segment and relate these to fasting motor function.


| INTRODUC TI ON
The duodenum is vital for digestive function, being where secretions are first mixed with acidic gastric chyme. Secretions are stimulated by duodenal mucosal sensing of components of chyme. 1 The effectiveness of duodenal chemical processing is evident from the rapid neutralization of acidic chyme entering the duodenum. [2][3][4] Duodenal mucosal chemosensors also cause immediate modulations of gastric and pyloric motor function that moderate gastric emptying, [5][6][7] ensuring that the rate of emptying is matched to the speed of nutrient processing in the duodenum. Motor functions of the duodenum itself are also influenced by duodenal chemosensors. [8][9][10] Rao et al 9 demonstrated fluoroscopically that intraduodenal infusions of sodium oleate, bile, or hydrochloric acid caused almost instant flow-obstructive occlusion of the distal duodenum that did not occur with infusions of a normal saline/barium sulfate mixture.
This flow-obstructive mechanism was dubbed the "duodenal brake," but the underlying mechanics remained undefined.
Traditional high-resolution manometry catheters are not able to penetrate very far into the small bowel; hence, we have used the larger span of 72 element high-resolution fiber-optic manometry catheters to make 1-cm-spaced recordings of fed-state and fasting pressures from the antrum into the region of the duodenal brake.

| Healthy subjects
Studies were conducted at the Gasthuisberg campus, UZ Leuven, Belgium, after approval from its Medical Ethics Committee and the Federal Agency for Medicines and Health Products, Belgium. A total of 15 female volunteers (26 ± 3 years, BMI between 18 and 25 kg/ m 2 ) gave written consent and fasted for 12 hours prior to the study: none had any history of gastrointestinal diseases, abdominal surgery (appendicectomy allowed), psychiatric illnesses, and nor use of drugs affecting the gastrointestinal tract or central nervous system.

| Manometry catheters
The catheters 3 mm diameter catheters each contained 72 1-cmspaced fiber-optic pressure sensors. The devices were fabricated for investigational use only. Data from the catheter were acquired by a solid-state fiber-optic spectrometer (FBG-scan 804D; FBGS International). Pressures were recorded by a custom-written LabVIEW © program (National Instruments).

| Study protocol
With subjects seated, the catheter was passed transnasally until ~20 sensors were aborad of the lower esophageal sphincter. With the subject supine, the catheter was passed through the pylorus under fluoroscopic guidance until the tip was at or beyond the duodeno-jejunal flexure. The catheter was then taped to the nose. Abdominal X-rays ( Figure 1) documented sensor locations at study start and completion.
Subjects were studied semi-reclined. Fasting recordings were continued until two phase III periods of the migrating motor complex (MMC) were recorded, or for a maximum of 5 hours. Subjects then ingested a standard 200 mL nutritional drink (Vanilla Multifibre Nutridrink, Nutricia; 480 kcal, 42% carbohydrate, 39% lipid, 3% fiber).
Fed-state activity was then recorded for 60-120 minutes.

| Analysis of manometric recordings
Pressures were analyzed both manually with software (PlotHRM, written in Matlab © , The MathWorks) and JavaTM (Sun Microsystems) and with an in-house developed automated system (see below). 15 Fed-state pressures were analyzed in detail for 60 minutes following the drink intake. These were compared to fasted pressure patterns during 30 minutes prior to the drink. Definite phase III MMC episodes that occurred during the fasted period in seven subjects were excluded from analysis (see Section 4) and an equivalent epoch prior to the phase III substituted.
Pressure sensor locations were determined from characteristic frequencies and/or patterns of pressures of the lower esophageal sphincter, antro-pyloric junction, and duodenum. Antro-pyloric junction position was taken as the most aborad point at which the underlying frequency was 3 cycles per minute (cpm) ( Figure 1B).

| Manual pressure event analysis
Propagation was confirmed if a pressure event peak occurred in three or more adjacent channels (ie, ≥2 cm), each with a trough-topeak amplitude of at least 5 mm Hg, and if the upstroke of each adjacent pressure wave commenced during the pressure event in an adjacent channel. 11 Direction of travel, velocity (cm/s), extent (cm), and peak amplitude (mm Hg) of propagating pressure events were scored, and their origins were recorded as either antro-pyloric, duodenal loop, or duodeno-jejunal (see below).

| Automated measurements of duodenal and jejunal pressure events
The automated analysis surveyed the frequencies and amplitudes of pressure events at each recording site. As previously described 12 baseline drift, respiration, coughs, and straining artefacts were removed from the manometry traces.

Key Points
• A previously unreported transition zone that differentiates two different forms of motility is identified in the distal duodenum.
• The transition zone has been identified using a high resolution fibre optic manometry spanning 72 cm between the stomach and proximal jejunum.
• The insights gained from this work and future studies identifying variations in DJC activity could assist with targeted, less morbid, interventions for patients with troublesome slow gastric emptying.
The frequencies of pressure events at each sensor were detected using a wavelet transform. 13 The root-mean-square amplitude over time was derived for the duodenal loop and duodeno-jejunal regions (see below) during fasting and after the drink. The wavelet transform was set up to detect physiological frequencies between 4 and16 cycles per minute (cpm) as this was the frequency range over which the majority of activity occurred. These steps produced curves which represented wavelet amplitudes over all of the assessed frequencies for each of the subjects ( Figure 2).

| Spatial referencing of pressure events
Fed and fasting patterns of duodenal and proximal jejunal pressures revealed a previously undescribed transition point of motor patterns in the distal duodenum. We used this point to divide the duodenum and proximal jejunum into two functional regions, as described below.

| Statistical comparisons
Propagating pressure events Comparisons of numbers, amplitude, velocity, and extent of propagation for propagating pressure waves before and after the nutrient drink were made with a non-parametric Wilcoxon matched-pairs signed-rank test. Comparisons between different propagating motor patterns used a Mann-Whitney test (GraphPad software, Inc).
Statistical comparison of multiple frequencies of pressure events between the different regions, before and after the meal, used Bayesian analysis. 14,15 This was categorized by region ("duodenal loop" and "duodeno-jejunal," defined in the results) and period ("fasted" and "fed"). 16 The probability distribution of the statistical model's parameters was calculated using the Stan software. 17

| RE SULTS
Duodenal intubation succeeded in all subjects. Within the first 5 hours, phase III MMC activity occurred twice in seven subjects, once in 6, and not at all in 2. The nutrient drink was given 238 ± 85 minutes (range 94-300 minutes) after the start of fasting recordings. The average distance of the transition point from the antro-duodenal junction was 18.8 ± 3.7 cm (range 13-28 cm). In three subjects, the depth of insertion of the catheter did not extend significantly beyond the transition point for DJ region motility analysis, though the transition point could be recognized. Accordingly, the data on motor patterns in the DL and DJ regions were from 15 and 12 subjects, respectively.

| Fed-state duodeno-jejunal region activity
Duodeno-jejunal complex activity was the dominant fed-state DJ region motor pattern. This had a mean frequency of 11.5 ± 0.5 cpm F I G U R E 1 A, X-ray image of the fiber-optic manometry catheter in situ. B, Manometry trace, presented as a spatiotemporal color plot, recorded after the nutrient drink. The differing manometric patterns recorded along the catheter illustrate the functional specialization of the antro-pyloric, duodenal loop, and duodeno-jejunal regions ( Figure 2E). There was no clear propagation between the individual DJC pressure events at adjacent recording points ( Figures 4B and 5B).
In nine subjects, DJC activity became the dominant pattern within 90 seconds of starting the nutrient drink. In the remaining three subjects, duodenal phase III-like activity 18,20 occurred within 45-90 seconds of starting the nutrient drink. This activity, which persisted for 5-15 minutes, was followed by an 8-11 minutes quiescent period after which DJC activity commenced.
Once DJC activity commenced, it persisted throughout the 1-hour fed-state recording. In nine of 12 subjects, at 12.3 ± 8.4 minutes after the nutrient drink, discrete clusters of more intense DJC activity were also observed ( Figure 4A,B). In eight subjects, these clusters occurred over ~25 minutes, followed by continuous DJC activity. In one subject, clustered DJC activity continued for the full fed-state recording ( Figure 4A).
Clusters occurred at 7.4 ± 4.9/h, with an interval between them of 3.8 ± 1.2 minutes and lasted 1.4 ± 0.55 minutes. The frequency of pressure events within fed-state clusters was 10.8 ± 0.8 cpm.
These clusters extended to the most aborad pressure sensor over a minimum of 23.5 ± 2.9 cm aborad from the transition point.

| Fed-state motor function in the duodenal loop region
In the DL region, after the nutrient drink, pressure events in individual channels occurred at frequencies between 6 and 12 cpm ( Figure 2D).

F I G U R E 2
Bayesian analysis of pressure wave frequency in the duodenal loop (DL) region (A and D) and the duodeno-jejunal (GJ) region (B and E), both before (left column) and after (middle column) a meal. The end of each column (C, F) compares mean pressure wave frequencies between the DJ and DL regions. When the thick black line appears above the hatched line, activity is significantly greater in the DJ region, and below the hatched line, the activity is significantly greater in the DL region. The end of the two rows (G and H) compares data between the fed-fasted states in the two small bowel regions. The thick black line above the hatched line indicates a significant increase in response to a meal However, while there were short periods of localized 10-12 cpm events they were significantly less frequent and extensive compared to DJC activity ( Figure 2E,F). DL region activity (Tables 1 and 2) was dominated by events that propagated varying distances, predominantly in an aborad direction from the antro-pyloric and proximal DL regions at a rate of 4-6 per minute (Figures 3 and 4C), with more than half extending at least 10 cm (Table 1). Orad propagating and synchronous events in the DL region accounted for 6% and 17% of events, respectively.

| Relationships between fed-state motor function in the duodenal loop and duodenojejunal regions
Of the 1760 aborad propagated pressure events originating in the antro-pyloric or DL regions, 78% did not cross the transition point; the events that did had a greater peak pressure amplitude than those that terminated before or at the transition point (Table 2, P = .007).

| Fasting compared with fed-state motor function
Following the identification of the fed-state DJC, similar DJC activity was also recognized during fasting, but it was less prevalent and of lower amplitude compared to the fed state ( Figures 2B,H   and 3). While short bursts of fasting DJC activity were seen in single or several adjacent channels, there were no prolonged episodes (>3 minutes). Clusters of DJC activity occurred during fasting in eight subjects at 3.6 ± 3.3/h, less than in the fed state (7.4/h; P = .043).
Fasting clusters had a significantly shorter duration (34.1 ± 7.3 seconds; P = .02) and lower frequency of their individual regular pressure events (7.2 ± 1.5/min; P = .01), than fed-state clusters. Fasting clusters frequently did not extend aborad over all of the DJ region sensors ( Figure 3A). Fasting DJC clusters were all preceded by vigorous antro-pyloric pressure events (582.4 ± 213.3 mm Hg) and long (>10 cm) aborad-propagated events in the DL region ( Figure 3A). The overall frequency, amplitude, velocity, and extent of propagating pressure waves in and beyond the duodenal loop region did not differ significantly between the fasted and fed states (Tables 1 and 2).

| D ISCUSS I ON
We propose that DJC activity is the motor mechanism of the "duodenal brake" in humans. Our data show that this is a fundamental physiological mechanism as it is active during gastric emptying of even the modest nutrient intake used in this study. The patterns observed yield insights into how the proposed brake activity relates to gastric and duodenal motor function orad to the distal duodenum. Fluoroscopy and impedance monitoring have shown that the lumen is occluded continuously in segments of duodenum or proximal jejunum encompassed by the phase III MMC 20 : This is likely to be the also the case for DJC activity, given its close similarity. Purely on manometric grounds, the frequency of DJC activity is so high and the duration of the pressure "valley" between each pressure event is so brief ( Figure 5) that the lumen is unlikely to fill with any content between individual events, especially when DJC activity extends for many centimeters. Also, vigorous propulsive, aborally propagated DJ region pressure events that could overcome the resistance from DJC activity are uncommon during DJC activity (Figures 1, 3, 4 and 5, Table 2).
Given our data and those of Rao et al, 9 our hypothesis is that the duodenal brake mechanism actively impedes emptying of chyme from the DL region until its pH has been largely neutralized and it has been thoroughly mixed with and partially processed by the secretions delivered to the duodenal loop. We propose that, with time, these functions alter the chemistry of the duodenal content to the point that it no longer causes major stimulation of DJC activity by duodenal loop chemosensors, allowing it to be delivered into the upper jejunum.
The timing of periods of DJC activity supports the concept that this is modulated by signaling from duodenal chemosensors as the vigor, luminal extent, and prevalence of fasting DJC activity was less than in the fed state (Figures 2 and 3). Secondly, clusters of fed-state DJC activity were strongly associated with propagated antro-duodenal events (Tables 1 and 2) known to cause delivery of pulses of chyme into and along the duodenum in the fed state. [22][23][24] This association is best explained by these pulses of nutrients causing surges of duodenal mucosal chemosensor signaling. The long periods of continuous fed-state DJC activity seen in three subjects could be explained by a relatively constant and slow flow of chyme into the duodenum with a lower, but continuous, stimulus to chemosensors TA B L E 2 The site of origin, count, amplitude, and velocity of propagating motor patterns that terminate at or prior to the transition point and those that propagated over the transition point (gray-shaded rows) The relatively brief DJC clusters recorded during fasting may appear to refute the proposal that these are stimulated by duodenal chemosensors. However, episodic duodenal acidification occurs during fasting, in association with the highly expulsive interdigestive phase II antro-pyloric pressure events 2,4 and it was these events that we found to be strongly associated with fasting DJC activity clusters (Figure 3). That the duodenal brake is potently activated by duodenal acidification alone was shown by Rao et al. 9 The stimulation and modulation of DJC activity during fasting and gastric emptying of the modest caloric intake provided by the nutrient drink indicates that the duodenal brake is active under normal physiological conditions. This refines the findings of Rao et al 9 whose observations might be interpreted as revealing a mechanism possibly triggered only by duodenal chemical "overload," as it is unclear if the infusate stimuli they delivered directly to the duodenum were within the physiological range for the duodenal luminal environment.
Rao et al 9 reported stimulation of the duodenal brake within seconds of the start of the duodenal delivery of infusates, consistent with our finding that clusters of DJC activity started within seconds of motor events previously shown to deliver pulses of gastric content from the stomach to the duodenum. 21,22 The rapidity of the response of the duodenal brake to duodenal chemical stimuli can only be explained through neural mediation. Pharmacological analysis, antral field stimulation, 23 and proximal duodenal transection 24 indicate that ascending intramural nerves are the major pathway for the pyloric stimulatory and antral inhibitory effects of stimulation of duodenal chemosensors. The present study suggests that duodenal chemosensors also signal along descending duodenal intramural pathways to modulate motility and specifically DJC activity.
The episodes of phase III-like activity seen in some subjects following consumption of the nutrient drink are of uncertain significance. These are clearly not DJC activity, as they started in the orad part of the DL region and propagated down the duodenum into the jejunum in the same orderly manner as true Phase III MMC activity.
Rao et al 9 also noted stimulation of duodenal "phase III-like activity" with some episodes of duodenal infusion. Other intraduodenal infusions, 6,8,9 cold stress, 25 and systemic hyperglycemia produced by an intravenous dextrose infusion 18 have also been found to trigger similar activity.
Given that DJC activity had not been recognized prior to this study, a dose-response study was not possible. By removing the variability of gastric emptying rates, future duodenal infusion-based studies would better define control of DJC activity by components of the duodenal content and give better insight into whether the absence of clustered fed-state DJC activity seen in 3 of our subjects was due to predominantly non-pulsatile gastric emptying.
The absence of any imaging data that examine the mechanics of DJC activity could be seen as the second major limitation of the present study, but our analysis, and that of Rao et al, 9 provides strong indirect support that DJC activity impedes flow. The outcomes of the present study now provide the guidance needed for design of ethical imaging studies of DJC activity in the future.
Confirmation of the existence and physiological importance of the duodenal brake may present opportunities for less morbid types of surgery of the stomach and duodenum. Perturbed DJC activity may also contribute to troublesome slow gastric emptying in some patients that might be addressable by interventions targeting DJC activity. Finally, as the rate of delivery of glucose into the upper jejunum, beyond the region of the DJC, is likely to have a major effect on the rate of glucose absorption, it is an intriguing question whether aberrant DJC activity could contribute to impaired glucose tolerance.

D I SCLOS U R E
JWA is managing director of Arkwright Technologies Pty Ltd. A company that makes fiber optic sensors for a range of medical and nonmedical applications. Arkwright Technologies did not fund this study or provide any devices for use in the study. The catheters described in this manuscript are for investigational use only.

AUTH O R CO NTR I B UTI O N S
JD contributed to detailed interpretation and analysis of the data and