Corresponding author J. G. Brasseur: Professor of Mechanical Engineering, Bioengineering and Mathematics, The Pennsylvania State University, 205 Reber Building, University Park, PA 16802, USA. Email: email@example.com
Quantifications of gastro-oesophageal anatomy in cadavers have led some to identify the lower oesophageal sphincter (LOS) with the anatomical gastric sling-clasp fibres at the oesophago-cardiac junction (OCJ). However, in vivo studies have led others to argue for two overlapping components proximally displaced from the OCJ: an extrinsic crural sphincter of skeletal muscle and an intrinsic physiological sphincter of circular smooth-muscle fibres within the abdominal oesophagus. Our aims were to separate and quantify in vivo the skeletal and smooth muscle sphincteric components pharmacologically and clarify the description of the LOS. In two protocols an endoluminal ultrasound-manometry assembly was drawn through the human gastro-oesophageal segment to correlate sphincteric pressure with the anatomic crus. In protocol I, fifteen normal subjects maintained the costal diaphragm at inferior/superior positions by full inspiration/expiration (FI/FE) during pull-throughs. These were repeated after administering atropine to suppress the cholinergic smooth-muscle sphincter. The cholinergic component was reconstructed by subtracting the atropine-resistant pressures from the full pressures, referenced to the anatomic crus. To evaluate the extent to which the cholinergic contribution approximated the full smooth-muscle sphincter, in protocol II seven patients undergoing general anaesthesia for non-oesophageal pathology were administered cisatracurium to paralyse the crus. The smooth-muscle sphincter pressures were measured after lung inflation to approximate FI. The cholinergic smooth-muscle pressure profile in protocol I (FI) matched closely the post-cisatracurium smooth-muscle pressure profile in protocol II, and the atropine-resistant pressure profiles correlated spatially with the crural sling during diaphragmatic displacement. Thus, the atropine-resistant and cholinergic pressure contributions in protocol I approximated the skeletal and smooth muscle sphincteric components. The smooth-muscle pressures had well-defined upper and lower peaks. The upper peak overlapped and displaced rigidly with the crural sling, while the distal peak separated from the crus/upper-peak by 1.1 cm between FI and FE. These results suggest the existence of separate upper and lower intrinsic smooth-muscle components. The ‘upper LOS’ overlaps and displaces with the crural sling consistent with a physiological LOS. The distal smooth-muscle pressure peak defines a ‘lower LOS’ that likely reflects the gastric sling/clasp muscle fibres at the OCJ. The distinct physiology of these three components may underlie aspects of normal sphincteric function, and complexity of sphincter dysfunction.
Underlying the function of the gastro-oesophageal sphincter is a complex anatomy, physiology and mechanics that is necessary to regulate oesophageal emptying into the stomach and to permit air venting and wanted regurgitation, while protecting against unwanted reflux of gastric content. The relative tonic contributions to the protective function of the gastro-oesophageal sphincter and the relative loss of tone during transient sphincteric relaxations (Dent et al. 1980; Mittal et al. 1995b) underlie sphincteric function.
Code et al. (1956) were early reporters of an intraluminal high-pressure zone (HPZ) in the segment separating the oesophageal body from the gastric cardia (see Fig. 1), and suggested that intrinsic smooth muscles of the distal oesophagus may be responsible for maintaining a high-pressure barrier to reflux. Soon thereafter Ingelfinger (1958) argued for crural diaphragmatic contraction as also important to the reflux barrier. Since these works, many manometry-based in vivo studies have led some to argue for the existence of two overlapping sphincteric components in the abdominal oesophagus.
There is general agreement on an extrinsic sphincteric component arising from the skeletal muscles of the right crus that wrap sling-like around the abdominal oesophagus (Fig. 1), attached to the inferior surface of the costal diaphragm above, and to the vertebral column below (Boyle et al. 1985; Mittal et al. 1988, 1993; Martin et al. 1992; Klein et al. 1993; Peck et al. 1995). Less well defined is the intrinsic sphincteric contribution. The partial in vivo suppression of sphincteric pressure in humans by atropine (Dodds et al. 1981; Holloway et al. 1986; Mittal et al. 1990, 1995a, 1997; Fung et al. 1999), and studies in which the HPZ was quantified after removal of the gastro-oesophageal segment during oesophagogastrectomy in humans (Klein et al. 1993) and after removal of the crural sphincter by myotomy in cats (Mittal et al. 1993) has led many to propose a second sphincteric component overlapping spatially with the crural sling and intrinsic to the smooth-muscle of the oesophageal wall (e.g. Kahrilas, 1997). By correlating radiographically manometric pull-throughs with metal clips placed at the oesophago-cardiac junction, Kahrilas et al. (1999) measured the HPZ peak to be 2 cm above the oesophago-cardiac junction in the normal human gastro-oesophageal segment, consistent with the location of the HPZ peak relative to the pressure inversion point just below the costal diaphragm (Stein et al. 1991). Because no anatomic correlate could be found, it has been necessary to assume a purely physiological circular smooth muscle ‘lower oesophageal sphincter’ (LOS) intrinsic to smooth muscle fibres of the abdominal oesophagus (Christensen, 1975, 1987).
Liebermann-Meffert et al. (1979) measured in cadavers a thickened wall of smooth muscle at the junction that coincides with the gastric sling-clasp muscle groups. ‘Sling’ fibres surround the junction of the oesophagus and the greater curvature side of the gastric cardia. Opposing these are ‘clasp’ circular muscle fibres on the lesser curvature junction (Fig. 1). Stein et al. (1995) argued for an anatomical smooth-muscle LOS defined by the sling-clasp muscle groups at the locally thickened oesophago-cardiac junction. This suggestion has led to a number of in vitro muscle strip studies of the pharmaco-physiology of the human and feline sling and clasp fibres, most recently by Preiksaitis & Diamant (1997), Tian et al. (2004) and Heureux et al. (2006). Heureux et al. (2006) show strong nitrergic inhibition of feline clasp fibres, but not of the sling fibres. Whereas both fibres responded to atropine only after cholinergic stimulation in vitro, the inhibition was stronger in the sling fibres, leading Preiksaitis & Diamant (1997) to suggest the maintenance of sphincteric tone in gastric sling fibres by extrinsic neural excitation through a cholinergic mechanism. Different responses of sphincteric muscle to electric field stimulation have been observed by different authors (McKirdy & Marshall, 1985; Preiksaitis & Diamant, 1997; González et al. 2004); the recent Heureux et al. (2006) study indicates a different response in the sling versus clasp fibres of the cat.
The state of understanding of the sphincteric components of the gastro-oesophageal segment, anatomically, physiologically, and functionally, remains confused. Whereas one line of research identifies the intrinsic gastro-oesophageal sphincter with the sling/clasp combination at the oesophago-cardiac junction, another argues for a spatial overlap between and intrinsic smooth-muscle LOS and an extrinsic crural sling located near the peak in manometric pressure within the abdominal oesophagus. Lack of precise anatomical correlates to in vivo pressure and in vitro pharmacological response, the difficulty in measuring separately the overlapping extrinsic and intrinsic components in vivo, and the dynamic changes associated with respiration, strain, etc., have added to the confusion.
In the current study we evaluated the separate contributions to the normal HPZ from the external skeletal-muscle and internal smooth-muscle components of the sphincter using pharmacologic agents to suppress the cholinergic smooth muscle contributions to sphincteric tone in vivo, then recovering the cholinergic contribution by subtraction. To determine the extent to which the cholinergic contribution to the smooth muscle sphincter approximates the full smooth-muscle sphincter, in a separate protocol the external crural sphincter was neuromuscularly blocked, leaving only the smooth-muscle contribution.
Combined ultrasound and manometry instrumentation
A custom assembly was constructed that combined high-frequency ultrasound with water-perfused manometry in order to measure gastro-oesophageal segment cross-sectional images simultaneously with intraluminal pressure. A 20 MHz ultrasonographic transducer was placed within a 6 French (3 mm o.d.) catheter (Microvasive; Boston Scientific, Watertown, MA, USA). The transducer rotated at 15–30 Hz to provide 360 deg oesophageal cross-sectional imaging with a 0.1 mm axial slice thickness and a typical penetration depth of 2 cm. Images were recorded on super VHS videotape on a Kay Elemetrics swallowing workstation (Kay Elemetrics, NJ, USA) at 30 frames s−1. Glued to the first catheter was a second 3 French (1.5 mm o.d.) angiography catheter for perfused manometry. A small side-hole was made in the second catheter at the same level as the ultrasound transducer (Fig. 2A). The manometry system fed into the Kay Elemetrics workstation, designed to temporally synchronize the two data sets to within one video frame (Fig. 2B). A custom pull-through machine provided calibrated proximal retraction of the ultrasound/manometry catheters at 5 mm s−1 from the stomach into the oesophagus (Fig. 1). The video images were digitized in 256 grey-scale levels and stored in 640 × 480 tiff format; pressure was digitized at 250 Hz.
All subjects gave written informed consent to take part in the studies, carried out in accordance with the policies of the National Institute of Health and Temple University School of Medicine and conforming to the standards set by the Declaration of Helsinki. The study was approved by the Temple University Internal Review Board. Exclusion criteria included subjects on any medication that could affect the gastro-oesophageal segment HPZ, including the use of antacids, H2 blockers, proton pump inhibitors, prokinetic agents, erythromycin-type antibiotics and anticholinergics. The following were also exclusion criteria: gastrointestinal (GI) symptoms, conditions and disorders including a history of oesophagitis, GI symptoms such as abdominal pain, heartburn, reflux, regurgitation, chest pain, difficulty swallowing, pain on swallowing, dysphagia, abdominal surgery involving the stomach or oesophagus, nausea or vomiting, diabetes, scleroderma, oesophageal motility disorders, non-cardiac chest pain, achalasia and current pregnancy. To ensure that none of the normal subjects had hiatal hernia, all subjects underwent upper endoscopy using a Pentax 2900 video endoscope (Pentax, Orangeburg, NY, USA) with the topical oral anaesthesia Cetacaine (Getylite Industries, Pennsauken, NJ, USA) without sedation. The gastro-oesophageal junction was carefully inspected using both antegrade and retrograde views. Subjects found to have a hiatal hernia were excluded.
Gastro-oesophageal reflux disease (GORD) was excluded on the basis of a standard questionnaire used at Temple hospital, together with a review of the subject's medical history. This questionnaire consisted of specific questions to rule out oesophageal pathology and, in particular, GORD. In addition to history of hiatal hernia, heartburn or GORD, questions included chest pain, difficulty swallowing solid foods or liquids, pain on swallowing, nausea, vomiting, regurgitation, dysphagia, and history of oesophageal stricture, blockage of the oesophagus, or achalasia. A positive response to any of these questions excluded the patient from the study. In addition, due to the administration of atropine in the protocol, the subject was excluded if there was any history of hypertension or cardiac disease.
In the ‘atropine study’, 15 middle-aged subjects, roughly the same number of males and females, were evaluated (eight male, seven female, 23–47 years, mean age 34 ± 8.5 years). All 15 subjects had normal oesophageal and gastro-oesophageal segment HPZ function as determined by concurrent manometric and endoluminal ultrasound pull-throughs (below). That is, the pressure tracing during the pull-through showed a typical bell-shaped curve distribution of pressure from a minimum on the gastric side of the pull-through to a maximum within the mid portion of the high pressure zone, and then to another minimum on the oesophageal side of the sphincter (see Fig. 1). Prior to insertion of the transducer assembly, the back of each subject's throat was numbed with Cetacaine and the nasal mucosa numbed with lidocaine (lignocaine). An intravenous line was prepared for the later administration of atropine. The subject lay supine with his/her back at approximately 35 deg; the subject was asked to minimize body motion during each study. Ultrasound imaging verified the initial transducer position in the proximal stomach. After ensuring correct catheter placement, the catheter was marked at the nares for accurate repositioning of the assembly to the same ‘pull-through start’ position. The pressure wave from a nasal cannula was recorded to monitor respiration.
Sphincteric contributions to pressure were measured with the costal diaphragm in extreme inferior and superior positions (see Fig. 1). For maximal inferior positioning, the subject was instructed to inhale as deeply as possible and hold his/her breath during ‘full-inspiration’ (FI) pull-throughs. For maximal superior positioning, the subject exhaled as far as possible and held his/her breath during ‘full-expiration’ (FE) pull-throughs. Each pull-through began in the stomach at the pre-marked start location of the transducer and ended well into the oesophageal body. Three pull-throughs were recorded for each FI and FE respiratory state. Swallowing was monitored, and if the subject swallowed at any time during the pull-through, the entire pull-through was discarded.
Subjects were asked to hold their breath after deep inspiration or exhalation in order to quantify the changes in axial pressure variation associated with changes in alignment of smooth versus skeletal muscle sphincteric tone from inferior versus superior displacement of the costal diaphragm. The tonic state of the sphincter with sustained breath hold differs from peak inspiration and expiration during normal respiration (Christensen, 1975). The phasic component of the crural sphincter is absent with breath hold and the crural contribution is reduced relative to respiratory inspiration (Boyle et al. 1985; Klein et al. 1993).
After collecting data in the normal resting state, the cholinergic smooth muscle contribution to the sphincter was attenuated using a pharmaceutical protocol developed by Mittal et al. (1995a, 1997) and Fang et al. (1999), based on the previous studies by Dodds et al. (1981) and Holloway et al. (1986) that showed dose-dependent partial suppression of resting sphincteric pressure. Following the Mittal et al. (1995a, 1997) protocol, we administered an initial bolus of atropine (15 μg kg−1) intravenously, followed by continuous intravenous atropine infusion at 4 μg kg−1 h−1 during the remainder of the study. After waiting 30 min and determining that there was an appropriate increase in heart rate to assure maximal suppression of cholinergic smooth-muscle tone (approximately 40% or greater over baseline), the data were collected for the same positions and respiratory states as done in the absence of atropine with three additional assembly pull-throughs.
Reductions in end-expiratory LOS pressure between 50 and 70% have been reported by Mittal and colleagues with this protocol (Mittal et al. 1995a, 1997). Klein et al. (1993) measured an average reduction of end-respiratory sphincteric pressure of 63% in patients with their LOS removed as compared with the normal sphincter, suggesting that a large portion of smooth-muscle sphincteric tone is attenuated by atropine using the Mittal protocol.
As will be discussed, the cholinergic smooth-muscle contribution to the HPZ will be reconstructed by subtracting atropine-resistant pressure from the full pressure profile. To evaluate the relationship between smooth-muscle sphincteric pressure and cholinergic smooth muscle contribution to pressure, a second protocol was carried out in which patients undergoing general anaesthesia for non-oesophageal pathology were administered cisatracurium to neuromuscularly block the crural sphincter, and the smooth muscle pressure was measured directly. Seven patients (two male, five female, 27–56 years, mean age 43.9 ± 13 years) undergoing general anaesthesia for either thyroid or sinus surgery took part in the study. (The patients in the Cisatracurium study were there for other reasons, so no control over age or gender was possible.) All patients underwent an extensive screening questionnaire to rule out any oesophageal symptoms prior to being entered in the study (the same questionnaire and exclusion criteria that were used in the atropine portion of the study). The patients had standard monitoring and premedication with versed 1–2 mg and fentanyl 1–5 μg kg−1. To suppress muscle fasciculation, a sub-neuromuscular blocking dose of vecuronium (1 mg i.v.), a non-depolarizing muscle relaxant, was given 3–4 min prior to induction of general anaesthesia. After adequate pre-oxygenation, general anaesthesia was performed with propofol 2–2.5 mg kg−1, or thiopental sodium 1–4 mg kg−1. After confirming loss of eyelid reflex, and while manually ventilating the lungs, a short-acting depolarizing muscle relaxant, succinylcholine (1.5 mg kg−1), was given to facilitate tracheal intubation. Endotracheal tube position was confirmed and was secured. General anaesthesia was maintained with 40–60% oxygen and 1–3% end tidal sevoflurane, and the patients were ventilated with a circle system.
Neuromuscular blockade was monitored by a peripheral nerve stimulator (Micro Stim, Neuro Technology Houston, TX, USA) with train-of-four to assess the degree of facial nerve blockade. Cisatracurium (0.2 mg kg−1) was administered to neuromuscularly block the skeletal crus muscles. Train-of-four was used to confirm no twitch (100% neuromuscular blockade) prior to a series of pull-throughs with crural neuromuscular blockade.
Approximately 20 min after intubation, the simultaneous ultrasound/manometry catheter was passed into the stomach transnasally and placed in the proximal stomach at the beginning of each pull-through. After initial placement, the catheter was marked at the nares for later repositioning. Three pull-throughs were performed with manual inflation of the lungs to airway pressures of 35–40 mmHg with inspiratory pause. Since manual inflation of the lungs forces the costal diaphragm inferiorly, the relative displacement of skeletal versus smooth muscle contributions to the gastro-oesophageal segment sphincter approximates FI in the atropine study. Each pull-through ended with the transducer in the oesophageal body. After the study all patients underwent surgery and all were extubated in the operating room without complication.
We reviewed the pharmaceutical agents used to induce and maintain anaesthesia to confirm that there would be no confounding influences on smooth-muscle sphincteric pressure. Martin et al. (1990) showed that fentanyl and vecuronium have no effect on the smooth muscle LOS. Propofal, succinylcholine and thiopental either have no effect or have a duration of action that is too short to significantly affect the smooth muscle LOS in this study (Laitinen et al. 1978; Thorn et al. 2005). Midazolam has a variable effect on the smooth muscle LOS in the literature. In one study (March et al. 1993) it was found to increase LOS pressure while in another study (Fung et al. 1992) it had no effect on LOS pressure. (An increase in smooth muscle LOS pressure is not a negative attribute, in principle, since the contribution of interest is enhanced.) Sevoflurane may cause a very slight, clinically insignificant, decrease in the smooth muscle LOS pressure, but at doses much higher than those used in this study (Kohjitani et al. 1999). It should be noted that the drug Cisatracurium was chosen specifically for this study because of its lack of effect on the smooth muscle components of the LOS.
Anatomic crus, and spatial pressure references Because high-frequency ultrasound penetrates beyond the oesophageal wall, the crural sling could be clearly identified in the ultrasound images. Figure 3, for example, shows a three-dimensional reconstruction of planar cross-sectional ultrasound images stacked and visualized using ray-casting (voxel imaging) software. The crus muscles impinging on the oesophageal wall appear as hypoechoic muscle bundles, identified with ‘D’ in the figure. The proximal (RCp) and distal (RCd) margins of the crural sling adjacent to the oesophageal wall were identified by L.M., and checked independently by Q.D., as the first and last extrinsic crus muscle bundles imaged during each pull-through. The ‘width’ of the crural sling was defined as the axial separation between RCd and RCp. To quantify relative anatomic shifts in crural sling location, both proximal and distal crus locations were used as references. There were no statistical differences in the results when using either reference. Therefore, all results in this paper use the distal marker (RCd) as the spatial reference for the crural sling.
In the atropine study, the spatial excursions of the anatomic crus and the pressure signatures between FI and FE were determined by referencing to the location of the transducer assembly in the stomach at the initiation of each pull-through. This was done both pre and post atropine. The pull-through start reference was not used to evaluate shifts pre to post atropine due to concern with reference drift during the 30+ min pause after atropine was administered.
For the atropine study, pull-throughs were analysed for FI and FE, pre and post atropine. In the Cisatracurium study, pull-throughs were analysed after full lung inflation with inspiratory pause. From the three (or more) pull-throughs per subject for each case, one pull-through was chosen based on best quality of high frequency ultrasound images for determining the anatomic crural sling spatial references. After data collection, separate gastric baseline pressures were determined by Q.D. for each pull-through by averaging the pressure signal 5–10 s just prior to the start of each pull-through, with cough and other obvious artefacts excluded. Secondary verification of the gastric baseline pressures was performed independently by R.U. All pressures were referenced to gastric baseline pressure.
Quantification of shift in pressure profiles with respiration To quantify the spatial shift in the pressure profiles between FI and FE, we computed the convolution (Bracewell, 2000) between the pressure profiles at FI and the pressure profiles at FE as a function of relative shift between the two profiles, with each profile referenced to the pull-through start location. This was done for each subject pre and post atropine. The procedure worked exceptionally well; in each case the convolution displayed a well-defined peak at the optimal displacement between the FI and FE pressure profiles.
Reconstructing the cholinergic smooth-muscle pressure distribution Administration of atropine, as described above, suppresses the cholinergic component of smooth muscle tone in the gastro-oesophageal segment, leaving a pressure distribution associated with the skeletal crural sling as well as residual smooth muscle non-cholinergic or myogenic tone. The cholinergic smooth-muscle contribution to pressure can be reconstructed by subtracting the post-atropine pressures from the full pre-atropine pressures after appropriate spatial referencing. This subtraction was performed for each subject after referencing to the lower margin of the right crus (RCd). In this way the full pressure distribution, the atropine-resistant and cholinergic smooth muscle pressure distributions were obtained relative to the inferior margin of the anatomical crural sling when the costal diaphragm was in extreme superior (FE) and inferior (FI) positions. Ensemble averaging was subsequently carried out across subjects.
Ensemble averaging and statistical significance Ensemble averaging of the individual pressure curves was performed, referenced to the distal margin of the right crus muscle, for the FI, FE data pre and post atropine, and for the post-cisatracurium data. To this end, individual pressure profiles were linearly interpolated onto a grid with time increment of 1/250 s. The number of individuals contributing to the average at a particular point was quantified along with the average; we required that all averages contained >10 samples for the atropine studies and seven samples for the Cisatracurium studies. Statistical significance was determined using paired Student's t tests with 95% confidence level and assuming equal variances. The alternative hypothesis was that the difference in means was not equal to zero (i.e. two sided) unless noted otherwise. Descriptive statistics were used to evaluate the mean ± standard deviation (s.d.) of these values. ANOVA was used to evaluate the changes in position and pressure with pharmacologic manoeuvres.
Structure and displacement of atropine-resistant HPZ and crural sling
The width of the longitudinal segment of the oesophageal wall in contact with the crural sling averaged ∼2.0–2.3 cm, as measured from ultrasound imaging with axial pull-throughs (Table 1), independent of the presence of atropine at both the FE and FI positions (P > 0.25). Furthermore, as shown in Table 1, the lower margin of the crural sling displaced by ∼1.9 cm as the costal diaphragm shifted from its inferior-most (FI) to its superior-most (FE) respiratory positions. This result was independent of the presence of atropine (P= 0.95).
Table 1. Various statistics
FE, full expiration; FI, full inspiration; Rcd, distal margin of the crural diaphragm.
Peak total pressure
Atropine-resistant peak pressure
Cholinergic smooth muscle contribution to pressure: upper pressure peak
Cholinergic smooth muscle contribution to pressure: lower pressure peak
Crural sling width without atropine (±s.d.)
2.36 ± 0.69 cm
2.26 ± 0.72 cm
Crural sling width with atropine (±s.d.)
2.00 ± 0.73 cm
2.01 ± 0.46 cm
Shift in RCd between FI and FE without atropine (±s.d.)
1.93 ± 1.24 (cm)
Shift in RCd between FI and FE with atropine (±s.d.)
1.90 ± 1.32 (cm)
In Fig. 4 we compare the ensemble averaged full pressure profiles of the HPZ (continuous curves) with the averaged pressure profiles after administering atropine (dashed curves), in the FI and FE diaphragmatic positions. All profiles were referenced to the inferior margin of the crus (RCd); the average width of the crus is shown for comparison. Whereas the baseline HPZ extends beyond the anatomic crus, atropine attenuates a major portion of the HPZ such that the anatomic crus is centrally orientated within the atropine-resistant HPZ and approximates the central high-pressure peak.
The two atropine-resistant pressure curves are plotted together in Fig. 5A to show that the central portions of the atropine-resistant FE and FI pressure profiles overlap nearly perfectly and are aligned at the same relative locations above the inferior margin of the anatomic crus, independently of the superior versus anterior positioning of the costal diaphragm. The association between the anatomic crural sling measured on ultrasound and the atropine-resistant HPZ was further evaluated by quantifying the correlation between the shift in the anatomic crus and the shift in the atropine-resistant pressure profile between FI and FE (see Methods). As shown in the scatter plot of Fig. 5B (open symbols), the correlation between the shift in the atropine-resistant pressure profile and anatomic crus is very high (R2= 0.83), implying a strong spatial association between the atropine-resistant HPZ and the anatomic crural sling during maximal displacement of the costal diaphragm. Figure 5B also shows a high correlation between the shift in the baseline pressure profile and anatomic crus (R2= 0.83), suggesting a shift in both cholinergic smooth-muscle and atropine-resistant sphincteric components with respiration.
Structure and displacement of the cholinergic smooth-muscle contribution to sphincteric pressure
Figures 4 and 5A show that the primary contributions to the atropine-resistant contributions to the HPZ displace with the anatomic crural sling during superior and inferior positioning of the costal diaphragm. However, comparison of Fig. 4A and B shows that the total pressure signature of the HPZ widens as the costal diaphragm moves between its inferior-most (FI) and its superior-most (FE) positions. In particular, the half-widths of the pressure profiles (measured from the baseline) increased from 1.93 cm to 2.93 cm between FI and FE.
The cholinergic smooth-muscle contribution to the HPZ, obtained by subtracting each atropine-resistant pressure distribution from each full pressure distribution after referencing to the inferior margin of the crural sling, was compared with the atropine-resistant contribution in Fig. 6. We found an overall widening of the cholinergic smooth-muscle pressure distribution as the costal diaphragm moved from its inferior-most (FI) to its superior-most (FE) positions. However, we also observed a distinctly double-peaked pressure distribution associated with the atropine-sensitive sphincteric contribution at both FI and FE diaphragmatic positions. The separation between the proximal and distal peaks, as shown by the dashed vertical lines in Fig. 6, increased from ∼1.22 cm in FI to ∼2.30 cm in FE, accounting for the increase in the half-widths of the HPZ pressure profiles.
Comparing the vertical dashed lines in Fig. 6A and B, we observed that the proximal contribution to cholinergic smooth-muscle contribution to pressure remained in the same position relative to both the peak in atropine-resistant pressure and the lower margin of the crural sling. The peak in the distal contribution to cholinergic smooth-muscle pressure, on the other hand, shifted 1.1 cm distally relative to the crural sling – from a position that overlaps the distal margins of the atropine-resistant pressure and crural sling at FI, to a position decidedly distal to the both the atropine-resistant pressure peak and the crural sling at FE. These results indicate that the smooth-muscle sphincteric component that generates the upper peak in the cholinergic smooth-muscle pressure profile moves in locked step with the crural sling during respiratory displacement of the costal diaphragm, while the smooth-muscle sphincteric contribution that generates the lower peak in the cholinergic smooth-muscle pressure shifts distally by >1 cm relative to the crural sphincter as the costal diaphragm moves from its inferior-most position (FI) to its superior-most positions (FE).
Comparison between cholinergic smooth-muscle pressure and Cisatracurium-suppressed sphincteric pressure
By following the protocol developed in a long series of previous studies designed to suppress the cholinergic component of smooth-muscle tone in the gastro-oesophageal segment (Dodds et al. 1981; Mittal et al. 1990, 1991, 1995a, 1997; Fang et al. 1999), our aim was to suppress a sufficient percentage of smooth muscle tone to approximate the smooth-muscle contribution to sphincteric tone with the cholinergic smooth-muscle contribution – and, conversely, to approximate the contribution to the HPZ from the crural sling with the atropine-resistant pressure distribution. The strong spatial correlation between the atropine-resistant pressure and anatomic crural sling (Fig. 5) is support for approximating the crural sphincteric contribution by the central peak in the atropine-resistant pressure distribution. Indirectly, this result also supports approximating the smooth-muscle sphincteric contribution with the cholinergic smooth-muscle pressure obtained by subtraction. To directly study the correspondence between smooth-muscle and cholinergic smooth-muscle pressure distributions, we carried out the ‘cisatracurium protocol’ described in Methods, in which the skeletal crural muscle was abolished without affecting the smooth muscle sphincter, and the smooth muscle contribution to the HPZ was measured directly.
In Fig. 7 we compare the smooth-muscle contribution to the HPZ measured indirectly by subtracting individual pressure distributions from the atropine-resistant sphincter from the full pressure distribution, with the averaged smooth muscle HPZ measured directly by ablating the skeletal crus muscle with Cisatracurium. One must keep in mind that the crural sphincter was displaced inferiorly by deep inspiration and breath-holding in the atropine study, and by manually inflating the lungs in the cisatracurium study. Figure 7 suggests that these two approaches lead to the same crural displacement. The cisatracurium pressure distribution displays the same double-peaked structure with virtually identical spatial separation of the two peaks as the atropine study (1.22 cm). The average peak pressure (relative to gastric) associated with the proximal contribution to the smooth muscle sphincter is the same in the two studies (16.5 mmHg). The distal contribution is somewhat lower in the cisatracurium study (10.5 mmHg versus 14.5 mmHg). The smooth muscle pressure curve in Fig. 7A versus B is shifted 2 mm relative to the lower margin of the right crus, an insignificant difference considering the difference in the protocols.
Pressure contributions to the high-pressure zone
Figure 6 indicates that the widening of the HPZ from FI to FE (dotted curves) is a consequence of the shift of the lower cholinergic smooth-muscle sphincteric pressure (continuous) relative to the atropine-resistant sphincteric pressure (dashed) and the crural sphincter. The data summary in Table 1 indicates that the peak pressures associated with the total and atropine-resistant pressures did not change significantly between FE and FI. However, the upper component of the cholinergic smooth-muscle pressure decreased from 23.5 to 16.6 mmHg from FE to FI, while the peak pressure in the lower component to the cholinergic smooth-muscle pressure did not change significantly (Table 1). Based on area under the pressure curves, we estimate that during inspiration the intrinsic cholinergic components account for approximately 38% of the pressure in inspiration and approximately 63% of the pressure in expiration.
There are currently two schools of thought concerning the smooth-muscle intrinsic contribution to the gastro-oesophageal sphincter. One argument places the intrinsic component within the circular muscle of the abdominal oesophagus at roughly the same axial location as the crural sphincter. In this model the gastro-oesophageal sphincter is maintained by two overlapping additive components, one intrinsic and one extrinsic, above the junction between the oesophageal body and gastric cardia. This model has evolved primarily from studies in which sphincteric pressure was measured using the Dent sleeve before and after administering atropine (Dodds et al. 1981; Holloway et al. 1986; Mittal et al. 1990, 1991, 1995a; Fung et al. 1999). These studies could not spatially localize the source of intrinsic sphincteric tone since the Dent sleeve measures only the highest pressure along its length. Nevertheless, this overlapping two-component model of the gastro-oesophageal sphincter is assumed in many studies and reviews (Boyle et al. 1985; Christensen, 1987; Goyal & Paterson, 1989; Martin et al. 1992; Klein et al. 1993; Heine et al. 1993; Mittal et al. 1995a, 1997; Kahrilas, 1997, 1999).
Two studies of the normal gastro-oesophageal sphincter provide important anatomic information relevant to the overlapping two-component model. Using manometric pull-throughs in 50 volunteers, Stein et al. (1991) found that the peak HPZ pressure was spatially close to the pressure inversion point (PIP), with 38% of the integrated pressure above and 62% below the PIP, on average. This places the peak sphincteric pressure well within the abdominal oesophagus and spatially near the oesophageal attachments of the phreno-oesophageal ligaments. Since the sleeve sensor responds to the highest pressure along its length, the Dent sleeve studies showing clear attenuation of sphincteric pressure with atropine (above) and with removal of the smooth-muscle sphincter by oesophagogastrectomy (Klein et al. 1993) combined with the Stein et al. (1991) result leads one to conclude that the attenuated intrinsic smooth-muscle component must overlap the extrinsic skeletal crural component.
This conclusion is more convincingly shown in a study by Kahrilas et al. (1999) in which the HPZ was measured by manometry pull-through concurrent with fluoroscopic imaging of metal clips placed at the squamo-columnar junction and on the build-up of tissue that localizes the gastric sling muscles at the oesophago-cardiac junction (Friedland, 1978; Stein et al. 1995). As shown in Fig. 8A, Kahrilas et al. (1999) measured the peak end-expiratory HPZ pressure (minimizing crural tone) to be 2 cm above the oesophago-cardiac junction (OCJ), 1 cm above the squamo-columnar junction, and 0.5 cm above the hiatus, placing the overlapping extrinsic and intrinsic HPZ sphincteric pressures well within the abdominal oesophagus. They repeated their measurements in patients with hiatal hernia and symptomatic reflux disease and found a two-peaked pressure structure with the proximal contribution displaced axially along with the squamo-columnar junction (Fig. 8B). These results support the model of overlapping extrinsic and intrinsic sphincters with additive pressure contributions that normally sum to create the normal HPZ. Because, the LOS component is attached to the costal/crural diaphragm by elastic phreno-oesophageal ligaments (Friedland, 1978; Kwok et al. 1999), disrupted ligaments in hiatal hernia separate the intrinsic and extrinsic in sphincteric components.
A difficulty with the overlapping two-component model is the lack of research showing clear neurophysiological differentiation of smooth muscle fibres localized spatially to the abdominal oesophagus (Christensen, 1975, 1987). The search for an anatomical LOS has lead to a second school of thought stimulated by measurements in human cadavers of a thickening of the muscularis at the oesophago-cardiac junction (Liebermann-Meffert et al. 1979). Repeating these measurements, Stein et al. (1995) proposed that the observed thickening reflects a pair of specialized smooth-muscle fibres that mark the transition between the tubular oesophagus and gastric cardia at the OCJ. Surrounding the OCJ there are two opposing sets of fibres that do not encircle the lumen (Fig. 1). Gastric ‘sling’ fibres wrap over the greater curvature side and open toward the lesser curvature side. Roughly opposing the sling fibres, wrapping over the lesser curvature side and opening toward the greater curvature side are smooth-muscle fibres often called ‘clasp’ fibres (Tian et al. 2004). From a mechanical perspective the ‘sling-clasp’ complex might form a functional unit of opposing fibres that, when both are in tension, would act together to close the OCJ.
More is understood about the pharmaco-physiology of the smooth-muscle sling and clasp fibres at the oesophago-cardiac junction from in vitro muscle strip studies (Preiksaitis & Diamant, 1997; Tian et al. 2004; González et al. 2004; Heureux et al. 2006). Preiksaitis & Diamant (1997) and Tian et al. (2004) found that human sling fibres were much more sensitive to carbachol and acetylcholine-induced cholinergic stimulation than were clasp fibres, and that neither fibres responded to atropine after stretch. Heureux et al. (2006) found, from in vitro studies of feline muscle strips, a strong nitrergic response in the gastric clasp fibres to l-NG-nitroarginine (l-NNA; a nitric oxide antagonist inhibitor), but not in the sling fibres. However, the gastric sling fibres were strongly suppressed by atropine after electric field stimulation, and the clasp fibres responded significantly to atropine after l-NNA induction of muscle contraction. Thus, it has been proposed that the sling fibres may maintain tone through continual cholinergic stimulation (Preiksaitis & Diamant, 1997).
The cholinergic contribution to sphincter tone from the current study
To clarify the functional components of the gastro-oesophageal sphincter we pharmacologically suppressed the cholinergic smooth-muscle contribution with atropine, and then subtracted the atropine-resistant contribution from the full pressure to obtain the cholinergic smooth-muscle contribution to sphincteric pressure after spatially referencing to the distal margin of the crural sling, identified from concurrent endoluminal ultrasound. The post-atropine pressures were therefore a consequence of the additive contribution from the crural sphincter, and pressure from myogenic and residual cholinergic or non-cholinergic tone within the smooth muscle of the gastro-oesophageal segment.
By subtracting pressures post atropine from pre-atropine pressures, the myogenic and residual smooth-muscle contributions were removed in addition to crural tone, leaving only the atropine-sensitive, or cholinergic smooth-muscle contributions to pressure. To further discover the extent to which the atropine-resistant and cholinergic smooth-muscle pressure contributions approximated the skeletal and smooth-muscle sphincteric contributions, respectively, we carried out a second protocol in which skeletal crural muscle tone was suppressed with the drug cisatracurium and the complete smooth-muscle component was measured directly, with the lungs inflated to approximate full inspiration in the atropine protocol. The cholinergic smooth-muscle contribution to pressure (from the atropine study) and full smooth-muscle pressures (from the cisatracurium study), each referenced to the anatomic crural sling, compared very well (Fig. 7) in most details, including the same two-peaked pressure structure with the same relative separation and magnitudes of the peaks. We conclude that the smooth muscle contribution to the gastro-oesophageal segment pressure profile is, to a large extent, functionally cholinergic. The atropine protocol removed sufficient smooth-muscle tone to suggest that the pressure measured within the atropine-resistant sphincter is largely from the crural sling.
The extrinsic skeletal-muscle sphincteric component
Figure 5A shows that the central atropine-resistant sphincter pressure is virtually identical in the full inspiration and expiration positions of the costal diaphragm when referenced to the lower margin of the crural sphincter, indicating that the atropine-resistant pressure was dominated by the skeletal crus muscles external to the oesophageal wall. Furthermore, the atropine-resistant pressure profile is centred over the anatomic crus, and the shift in the positioning of the anatomic crural sphincter from respiratory-induced movement of the costal diaphragm is strongly correlated to the shift in the atropine-resistant pressure profile (Fig. 5B). We conclude that the atropine-resistant contribution to the HPZ reflects, to a large degree, the anatomic crus muscles and define the extrinsic component of the sphincter. The external crural sphincter covers approximately 2.1 cm of the abdominal oesophagus (Table 1).
The peak crural sphincter pressure (Fig. 5A) and total pressure (Fig. 4) did not significantly change between FE and FI, consistent with Christensen (1975, p. 249), who reported that ‘on deep inspiration, sphincter closure tension, recorded manometrically in humans, falls below the tension recorded in the normal respiratory cycle…’ and with Boyle et al. (1985) and Klein et al. (1993) who report that the phasic respiratory contribution to pressure from the crural sling disappears with sustained breath-holding.
Structure of the cholinergic and intrinsic smooth muscle sphincter pressures
From mechanical principles, it can be argued that the HPZ reflects additive contributions from external skeletal-muscle and internal smooth-muscle sphincteric components of the gastro-oesophageal segment. As the costal diaphragm moved from its inferior-most to its superior-most positions, the crural muscles surrounding the abdominal oesophagus shifted proximally by about 2 cm (Fig. 3B), and the tonic HPZ widened considerably (Fig. 4). Because the width of the crural component was unaffected by its axial displacement (Fig. 5A), widening of the HPZ requires either a widening of the smooth muscle contribution to the sphincter, or a relative displacement of the internal and external sphincteric components.
We discover, however, that both the cholinergic and the complete intrinsic smooth-muscle sphincter have a more complex pressure signature than does the external skeletal-muscle contribution. Figures 6 and 7 show that the pressure signatures associated with these components have a two-peaked structure. The very high structural correlation between the smooth muscle pressure signature obtained by subtraction in the atropine study, and obtained directly in the cisatracurium study, is strong support for the claim that the cholinergic smooth-muscle component is a good approximation of the full smooth muscle sphincter, and allows us to conclude with a high degree of certainty that the cholinergic and complete smooth-muscle contributions to the HPZ are similar in configuration and magnitude and contain two peaks that separate with superior displacement of the diaphragm.
Interestingly, the proximal peak of the smooth-muscle pressure remained fixed relative to the crural sphincter even as the crus shifted >2 cm with respiration (Fig. 6). We conclude that the component of the smooth-muscle sphincter responsible for the proximal pressure peak is as rigidly attached to the costal diaphragm as is the crural sphincter. Since the normal oesophagus is under global longitudinal tension (Liebermann-Meffert et al. 1979; Stein et al. 1995), the portion of the oesophageal body associated with the proximal sphincteric component must be dragged distally by the inferior displacement of the costal diaphragm. The only anatomic structures available to generate a drag force to counter the basal longitudinal force on the abdominal oesophagus are the phreno-oesophageal ligaments that connect the costal diaphragm and the oesophageal wall. We conclude that in the normal oesophago-gastric segment these ligaments are relatively stiff, at least as stiff as the skeletal crural sling.
In contrast to the upper component of the smooth muscle sphincter, responsible for the proximal pressure peak in Figs 6 and 7 and rigidly coordinated with the crural sphincter, the component of the smooth muscle sphincter responsible for the distal pressure peak moves separately from the crural sphincter and costal diaphragm. Figure 6 indicates that the distal peak shifts distally by about 1.1 cm relative to the upper peak as the crural sphincter moves ∼2 cm from its inferior to superior positions. Thus, since the upper component of the smooth-muscle sphincter moves rigidly with the crural sphincter, the lower smooth-muscle component must shift a little under 1 cm proximally as the crural sling and upper sphincteric component shift superiorly by 2 cm and the two sphincteric components separate by 1.1 cm.
Three sphincteric components of the gastro-oesophageal segment
As discussed above, Kahrilas et al. (1999) used hiatal hernia as a model to separate the smooth and skeletal muscle components of the sphincter within the oesophago-gastric segment (see Fig. 8). They went on to hypothesize that the normal HPZ can be reconstructed from the hernia groups by appropriately realigning the proximal and distal pressure contributions in each individual. To realign the intrinsic and extrinsic sphincteric contributions to the HPZ, they assumed (a) that the two sphincteric components are approximately in the same position relative to the hiatal indentation determined fluoroscopically, and (b) that the squamo-columnar junction is positioned approximately the same in each individual relative to the intrinsic sphincter. With these assumptions, the squamo-columnar junctions with attached proximal high pressure contributions were shifted to a position 0.5 cm distal to the hiatal indentation, the average relative position of the squamo-columnar junction in normals. Their result is shown in Fig. 8C. Interestingly, whereas the remodelled pressure distribution approximated reasonably well the normal pressure distribution proximal to the hiatus, the normal HPZ below the hiatus was 2–3 times more extended and at higher pressure in normals than in the reconstructed hiatal hernia group.
Figure 6 suggests a third, smooth-muscle component of the HPZ that probably explains the incomplete reconstruction obtained by Kahrilas et al. (1999) with hiatal hernia. We summarize the three-sphincter model in Fig. 9. We find that the HPZ of the oesophago-gastric segment has three components: one external component associated with the skeletal muscle right crus muscle surrounding the abdominal oesophagus, and two internal components associated with the smooth muscle intrinsic to the oesophageal wall. Figure 5 clearly identified the atropine-resistant (skeletal) component with the external crural sphincter. The proximal smooth-muscle component we propose to identify with the classical intrinsic LOS. This component, we find, overlaps and moves with the crural sling (see Fig. 9), apparently due to rigid attachment of the abdominal oesophagus to the costal diaphragm from relatively stiff phreno-oesophageal ligaments.
In contrast to the upper smooth-muscle sphincteric component of the oesophago-gastric segment (what we shall refer to as the ‘upper LOS’), the distal smooth-muscle component remains more locally in place, shifting slightly less than 1 cm proximally during the superior excursion of the costal diaphragm, while separating from the proximal component by slightly more than 1 cm. We suggest that the distal smooth-muscle component is a ‘lower LOS’ created by sling and clasp fibres within the muscularis propria at the junction between the gastric cardia and tubular abdominal oesophagus (Fig. 9). We propose that the transitional muscle fibres at the OCJ form a third sphincteric component that, like the abdominal oesophagus overall, moves superiorly and inferiorly, but less so than the LOS due to elastic stretch between the two smooth-muscle sphincteric components as gastric ligaments resists the axial motion induced by the rigidly attached phreno-oesophageal ligaments at the upper LOS. We suggest that in the hiatal hernia group described by Kahrilas et al. (1999), the pressure from the lower LOS component was absent, explaining the shorter and lower pressure distal to the reconstructed pressure profile in the hiatal hernia patients relative to the normal HPZ.
Anatomically, it is likely that the lower LOS is associated with the thickened junction between the tubular oesophageal and stomach identified by Stein et al. (1995) as the LOS from measurements in human cadavers. The differences in cholinergic, nitregic and related neurophysiological responses of the sling versus clasp fibres at the OCJ (Preiksaitis & Diamant, 1997; Tian et al. 2004; Heureux et al. 2006) suggest that the gastric sling and clasp fibres may both respond to atropine if in initially stimulated states. Asymmetry of sphincteric pressure, and the asymmetrical effect of atropine on sphincteric pressure, has been reported by Richardson & Welch (1981). However, the measurements are not specific enough to know if the measured asymmetry reflected the asymmetry in forces applied to the oesophageal wall by the crural sling (Fig. 1B), or the asymmetrical structure and neurophysiological response of the gastric sling/clasp combination (Preiksaitis et al. 1994). (Pressure asymmetry was not evaluated in the current study that used only one pressure transducer.)
As summarized in Fig. 9, we hypothesize that the sphincter within the gastro-oesophageal segment has three distinct components, each of which are controlled by different neurophysiology, and which work functionally together and separately in different ways and at different times to both maintain the reflux barrier as well as to allow antegrade or retrograde transport, as the physio-mechanical state of the gastro-oesophageal environment varies. One of these components, the crural sphincter, is extrinsic and composed of skeletal muscle. The other two components are intrinsic to the oesophageal-gastric wall smooth muscle. The proximal component, we argue, is the physiologically mediated LOS (Christensen, 1987), while the distal component is associated with the gastric sling and clasp muscle unit (Stein et al. 1995), each muscle fibre group having a very different physiology (Heureux et al. 2006).
The functional significance of a three-component gastro-oesophageal sphincter follows from the mechanics of sphincter opening and flow. Sphincter opening from below, for example, is a consequence of a change in the balance of forces at the distal-most point of the closed oesophageal lumen, where gastric fluid meets closed lumen. Opening is driven by an increase in gastric fluid (liquid or gas) pressure relative to the pressure external to the gastric wall at this point. The summation of active and passive tension force within the muscle wall at the junction point resists this pressure opening force. Thus, the existence of tension within the sling-clasp complex of a ‘lower LOS’ would act as a barrier to reflux. If this barrier were overcome, then tension forces within the ‘upper LOS’, together with the extrinsic muscles of the crural sphincter provide the next level of protection. Only when all three barriers are overcome will reflux occur. It follows that a complete understanding of the anatomy, physiology and mechanics underling the complex protective functions of the gastro-oesophageal sphincter, and especially the complex interactions that underlie chronic dysfunction, requires a complete understanding of the interactions of the three sphincteric components over a wide range of physiological and mechanical states of the gastro-oesophageal environment.
The authors are grateful to Nicholas Diamant, MD, for useful discussion and to Anupam Pal, PhD, for help with the figures. This work was financially supported through the National Institutes of Health grant R01-DK-59500 (L.S.M., Q.D.); J.G.B. was supported by grant R01-DK-56033. R.U. was supported by C. R. Bard, Inc.