In many cases, the esophagus has been determined to be the source of such pain. Esophageal manometry can identify motor dysfunction in as many as 25% of patients with noncardiac chest pain [6, 7], and additional provocative tests show motor and other abnormalities in another 15% to 25% [6-10]. Most such patients have gastroesophageal reflux disease [3]. Some studies of the sensations evoked by balloon distention in the esophagus have suggested that 48% to 60% of such patients have altered sensory perception [11-13], but other studies have not confirmed this finding [9, 10, 14].

Most studies of esophageal motor function have examined the contractile properties of the esophagus, but little attention has been paid to the biomechanical and sensory properties, primarily because of technical limitations [15-17]. A new technique, impedance planimetry, allows investigation of sensory and biomechanical properties in the intact human gut [15, 16, 18, 19]. This technique provides a way to assess the cross-sectional area of the lumen in a selected plane at a range of distending pressures. It also permits quantification of the resistance offered by the biomechanical and contractile properties of the wall and of the sensory responses that are provoked by balloon distention. We used impedance planimetry to determine whether a sub-group of patients with unexplained chest pain has abnormal biomechanical esophageal muscle properties and abnormal esophageal sensory nerve function.

Methods

Top Methods Results Discussion Author & Article Info References

Participants

We studied 24 consecutive patients with unexplained noncardiac chest pain who were seen on an outpatient basis: 10 men and 14 women (mean age, 48 years [range, 30 to 74 years]). All patients were referred by cardiologists after extensive cardiac evaluations showed no cause for the chest pain and, in many instances, after various empiric therapies proved ineffective. Patients had no comorbid illnesses and had either normal coronary arteries or insignificant coronary artery disease as shown by results of angiography, stress thallium testing, or stress technetium 99m methoxy isobutyl isonitrile (MIBI). In addition, all patients had extensive gastrointestinal evaluations. The results of these evaluations included normal findings on upper gastrointestinal endoscopy with esophageal biopsy, normal esophageal manometric results, and either normal results of 24-hour pH study or no response to 6 weeks of omeprazole, 20 mg per day. Patients were included only if they had at least one episode of chest pain per week and if their symptoms had persisted for more than 3 months.

We also studied 12 healthy controls, 5 men and 7 women (mean age, 35 years [range, 24 to 63 years]), who were recruited by a hospital advertisement. The controls were asymptomatic, were not taking any medication, had not had any previous thoracic or gastrointestinal surgery, and had normal physical examinations. All participants gave written informed consent, and the Human Investigation Review Board of the University of Iowa College of Medicine approved the study.

Impedance Probe and Measuring System

The equipment, which consisted of a probe and a signal-processing system, has been described in detail elsewhere [15, 16, 19]. The flexible plastic probe, 6 mm in diameter, contained four ring electrodes (two outer and two inner electrodes) and five side holes. A thin latex balloon, 4.5 cm long, was tied around the probe to enclose the four ring electrodes and three of the side holes (Figure 1). Two of these were infusion side holes that were used to distend the balloon, and the third was a perfusion side hole that was used to measure intraballoon pressure. The remaining side holes, one located 2.5 cm proximal to the balloon and the other located 2.5 cm distal to the balloon, were used to measure intraesophageal pressures. The side holes were perfused with 0.018% NaCl solution at a rate of 0.2 mL/min, and the intraluminal pressures were measured using a low-compliance pneumohydraulic perfused system (Arndorfer, Milwaukee, Wisconsin) connected to external transducers (Gould, Inc., Essex, United Kingdom). The leveling container, a 400-mL plastic jar half-filled with 0.018% NaCl, was connected to the probe with a plastic tube. The signal processing system Figure 1 consisted of a generator that gave a constant alternating current of 100 µ A at 5 KHz, an amplifier, an impedance detector, an analog-digital converter, and a computer [15, 16, 19].

View larger version (21K): [in this window] [in a new window]   Figure 1. Schematic representation of the probe and the signal-processing system used for impedance planimetry of the esophagus. D equals detection electrode; E equals excitation electrode; I equals infusion channel used for balloon distention; P equals perfusion channel used for measuring intraluminal pressure.

Study Protocol

All participants were studied after an overnight fast. Oropharyngeal anesthesia was achieved with a local spray of dyclonine hydrochloride (Dyclone, Astra, Westboro, Massachusetts). The lubricated probe was passed through the mouth until its tip was located 55 cm from the teeth. Participants were asked to lie on a bed, which was tilted at the head end by 30 degrees. After a rest period of 10 minutes, the catheter was gradually withdrawn until the balloon lay across the lower-esophageal sphincter. The lower-esophageal sphincter was identified as a zone of high resting pressure that decreases with swallows. The catheter was removed and taped in position so that the balloon lay 10 cm above the lower-esophageal sphincter, where all measurements were obtained.

After a second rest period of 10 minutes, the balloon pressure was set at 0 according to the resting intraesophageal pressure by adjusting the height of the leveling container, which was adjacent to the participant but behind a screen. Thus, participants were blinded to the degree of inflation. Next, the balloon pressure was increased in steps of 5 cm H₂O to 50 cm H₂O or greater by raising the leveling container and infusing 0.018% NaCl (at 37 °C) into the balloon. After each distention, the balloon was deflated by lowering the leveling container and was reinflated after a rest period of 3 minutes. Each inflation was maintained for 3 to 5 minutes or until the cross-sectional area reached a new stable baseline between reactive contractions. At this steady state, the cross-sectional area was measured and the sensory responses were noted.

Participants were given a symptom chart and were asked to grade their sensation as follows: grade 0, no sensation; grade 1, a sensation of fullness or distention; grade 2, moderate (tolerable) discomfort; and grade 3, severe pain. Because no controls had discomfort or pain at inflation pressures as great as 50 cm H₂O, serial inflations of the balloon were continued until the controls reported pain or until a pressure of 65 cm H₂O was reached. In the patient group, the inflations were continued until typical chest pain was evoked or until a pressure of 50 cm H₂O was reached. During inflations, participants were asked to refrain from swallowing and to signal the onset of reflex swallows. Reflex swallows were noted to distinguish primary (that is, aborad) propagating contractions, which are initiated by swallowing, from secondary peristalsis, which consists of aborad propagating contractions induced by local distention of the esophagus.

Measurements and Data Analysis

The balloon cross-sectional area was measured according to the field gradient principle [20-22]. When a current, I, is induced by the two outer-ring electrodes in a cylinder of conducting medium, the potential difference, V, between the two inner (detection) electrodes is V equals I x R (Ohm's law), where R is the impedance of the conducting substance. R can also be expressed as d x c^–1 x the cross-sectional area^–1, where d is the distance between the detection electrodes and c is the conductivity of the fluid. If I, d, and c are constants, V is inversely proportional to the cross-sectional area. The computer software provided output that was directly proportional to the cross-sectional area. The data for the cross-sectional area and intraluminal pressure were stored on diskettes. The records, visualized off-line on a computer, were analyzed with a software program by an author who was blinded to the participants' clinical details and results of sensory function.

The following are definitions of and formulas used to calculate the variables.

Sensory Responses

The balloon pressure that induced first sensations of fullness, moderate discomfort, and pain were noted for each participant, and the mean values for the threshold pressures that induced each sensation were calculated.

Passive Biomechanical Properties

Cross-sectional area: At each level of balloon inflation, the radius was calculated from the new steady-state baseline cross-sectional area. Because previous studies [18, 19, 23] had shown that the measurements of the cross-sectional area conform to the linear portion of the calibration curve at pressures as great as 40 cm H₂O, and because 35% of patients could not tolerate balloon distention beyond this level, the cutoff pressure for calculations of biomechanical variables was 40 cm H₂O.

Circumferential wall tension (T): The total force applied to stretch a segment of the wall was calculated [15, 16] as T = r x dP, where r is the balloon radius (r = square root of the cross-sectional area x P^–1) and dP is the transmural pressure difference (Law of Laplace). Transmural pressure was defined as the difference between the balloon pressure and the resting pressure in the esophagus, with the assumption that esophageal intraluminal pressure at rest is the same as intrathoracic pressure.

Strain: We calculated strain (a ratio that refers to the relative deformity produced by the application of stress [24]) [15, 24, 25] for each level of inflation as e equals (r_x – r₅) r₅^{– 1}, where r_x is the balloon radius at a given pressure and r₅ is the radius at an inflation pressure of 5 cm H₂O. The association between the change in tension and the change in strain was plotted during stepwise inflations.

Pressure elastic modulus (wall stiffness): The intramural wall pressure in relation to the fractional change in radius was calculated [15, 24, 25] as r x dP x dr^{– 1}, where r is the radius and dP and dr^{– 1} are the changes in balloon pressure and the radius between two consecutive steps of balloon distention.

Active Biomechanical Properties

Peristaltic properties: Balloon distention induced tertiary contractions, that is, localized, nonpropagating pressure waves confined to the level of the balloon and secondary peristaltic contractions consisting of pressure waves that propagated aborad across three pressure ports. To exclude artifacts induced by respiration, only pressure waves with an amplitude greater than 13 cm H₂O were scored as contractions; we calculated the mean frequencies, mean amplitudes, and mean durations of these contractions. A motility index was calculated for each pressure channel by summing the areas under the curves of all pressure waves that were observed during distention.

Reactivity: Balloon distention induced reactive esophageal contractions that produced a transient decrease in the cross-sectional area. Reactivity was measured as the difference in height between the steady-state cross-sectional area and the minimum cross-sectional area that was observed during each balloon inflation.

Statistical Analysis

The mean values (±SD) for the cross-sectional area, wall tension, strain, pressure elastic modulus, and reactivity for each level of balloon distention were compared between patients and controls using multifactorial analysis of variance. In patients who developed pain and could not tolerate higher levels of balloon distention, we excluded the missing data for the higher pressures. Because the tension-strain curves show a variation in two dimensions, we compared them by defining the highest tension that was common for all individual curves. For this tension, the corresponding strain values were identified and compared between the two groups using the Mann-Whitney test. The mean values for the frequency, amplitude, and duration of peristaltic contractions and the motility index for each inflation pressure and for the three pressure channels were compared using the Student t-test. Multiple comparisons can be susceptible to a type 1 error. However, rather than using the conservative Bonferroni correction, we report each significant P value. We compared the thresholds required to induce first sensation and to induce reactive contractions between the two groups using the Student t-test. The Fisher exact test was used to compare the number of participants in each group who developed moderate or severe pain. We used Systat and Microsoft Excel (Redmond, Washington) software to compute statistical differences.

Results

Top Methods Results Discussion Author & Article Info References

Participants

Table 1 lists the patients' age and sex distributions, the characteristics of chest pain, the associated gastrointestinal symptoms, and the results of diagnostic tests that preceded the study. Briefly, 12 of 24 patients (50%) reported more than one episode of chest pain per day, and 12 of 24 patients (50%) reported less than one episode per day but more than one per week. Twelve of 24 patients (50%) reported moderately severe pain, 11 (46%) reported severe pain, and 1 (4.2%) reported mild recurrent pain. Pain lasted 10 to 30 minutes in 12 patients (50%) and more than 30 minutes in the other 12 patients (50%). In addition, 5 patients (21%) reported dysphagia, and 8 (33%) reported heartburn.

Chest Pain and Sensory Responses

Graded balloon inflation at pressures less than 50 cm H₂O induced typical chest pain in 20 of 24 patients (83%). In contrast, no controls had discomfort or pain up to this level of distention. Five controls reported discomfort at 55 cm H₂O, and three reported pain at 65 cm H₂O. The threshold pressure at which distention was first perceived was significantly lower in patients than in controls (15 ± 9 cm H₂O compared with 30 ± 11 cm H₂O; P < 0.001) (Figure 2).

View larger version (18K): [in this window] [in a new window]   Figure 2. Individual values for the inflation pressures that induced a first sensation, moderate discomfort, and pain in controls (white circles) and patients (black circles). Bars represent the mean and 95% CI. The maximum inflation pressure was 50 cm H2O in controls and 65 cm H2O in patients. NCCP equals noncardiac chest pain.

Passive Viscoelastic Properties

Cross-Sectional Area

Intermittent balloon inflation of the esophagus from 10 to 40 cm H₂O produced a linear increase in the cross-sectional area from 166 ± 63 mm² to 526 ± 166 mm² in patients and from 120 ± 30 mm² to 469 ± 63 mm² in controls (Table 2). Although higher in patients, the mean cross-sectional area did not differ significantly between the two groups (P = 0.09); however, at pressures less than 20 cm H₂O, the cross-sectional area was higher in patients with noncardiac chest pain (P = 0.05). The cross-sectional area considerably varied among participants, but no correlation was seen between cross-sectional area and the sex, height, or weight of patients or controls.

Circumferential Wall Tension and Strain

Balloon distention caused an exponential increase in wall tension from 30 ± 4 mm times cm H₂O to 512 ± 76 mm times cm H₂O in patients and from 26 ± 4 mm times cm H₂O to 488 ± 33 mm times cm H₂O in controls (Figure 3). Strain also increased from 0.3 ± 0.2 to 1.1 ± 0.4 in patients and from 0.2 ± 0.1 to 1.3 ± 0.3 in controls. An exponential-like increase occurred in the tension-strain association Figure 3, and, in patients, the curve significantly shifted to the left (P = 0.02). This suggests that the esophagus is less distensible in patients with noncardiac chest pain than in normal persons.

View larger version (20K): [in this window] [in a new window]   Figure 3. Association between wall tension and strain in response to stepwise increases in balloon pressure in controls (white circles) and patients (black circles). The curve in the patient group is shifted to the left. This suggests that for the same level of tension (force), the degree of strain or the deformability of the esophageal wall was less in patients than in controls (that is, the esophagus was stiffer). Bars represent the mean and 95% CI.

Pressure Elastic Modulus (Wall Stiffness)

In the controls, the elastic modulus insignificantly increased from 46 ± 30 cm H₂O to 72 ± 16 cm H₂O (P = 0.05) Figure 4, left. In contrast, at an inflation pressure of 10 cm H₂O, the esophagus appeared more elastic (less stiff) in patients (elastic modulus, 32 ± 22 cm H₂O) than in controls (P > 0.2). However, when the distending pressure was increased to 40 cm H₂O, the elastic modulus in the patients increased significantly (that is, became stiffer) to 137 ± 74 cm H₂O (P < 0.001). The curve also significantly differed between the two groups (P = 0.003).

View larger version (14K): [in this window] [in a new window]   Figure 4. Left. Effect of serial increments in balloon pressure on the pressure elastic modulus (wall stiffness) of the esophagus. Right. Effect of graded balloon distention on the reactivity of the esophagus. Bars represent the mean and 95% CI. White circles = controls; black circles = patients.

Active Biomechanical Properties

Patterns of Esophageal Contractions

Balloon distention induced both tertiary contractions observed only at the level of the balloon and secondary peristaltic contractions that involved more of the esophagus. Contractions were considered peristaltic if they propagated across two adjacent recording sites. Most (80%) of the secondary peristaltic contractions first appeared at the proximal port.

Peristaltic Threshold

The mean threshold pressure required to induce tertiary contractions was lower than that required to induce secondary peristalsis, but the difference was not significant within the same group. However, a significant difference was seen between patients and controls. The threshold pressure for tertiary contractions was significantly lower in patients than in controls (10.0 ± 11 cm H₂O and 15.4 ± 4.5 cm H₂O, respectively; P = 0.04). The threshold for secondary peristalsis was also significantly lower in patients than in controls (9.6 ± 5.6 cm H₂O compared with 18.8 ± 5.3 cm H₂O; P = 0.03).

Frequency of Contractions

Graded balloon inflations from 5 to 40 cm H₂O induced significantly more contractions at the proximal, balloon, and distal recording sites Table 2 in both patients and controls (P = 0.02). In both groups, more contractions were seen at the balloon channel recording site than at the proximal or the distal sites, but this difference was significant (P = 0.02) only at inflation pressures less than 20 cm H₂O (Table 2). Patients generally had more contractions at all three recording sites than did controls.

Amplitude of Contractions

Graded balloon distention also increased the mean amplitude of contractions in both patients and controls but was significantly higher (P = 0.04) at all three recording sites when inflation was as high as 20 cm H₂O (Table 2). In controls, the amplitude of pressure waves was lower at the distal recording site and was significant for inflation pressures less than 20 cm H₂O. This attenuation of the amplitude of contractions at the distal site was not seen in patients. In fact, the amplitude of contractions at the distal site was significantly higher in patients than in controls (P = 0.04).

Duration of Contractions

The duration of contractions also increased stepwise in response to increasing inflation pressures but did not differ significantly among the recording sites (Table 2). Contractions lasted significantly longer in patients than in controls at inflation pressures of both 10 cm H₂O (P = 0.05) and 20 cm H₂O (P = 0.01).

Motility Index

Stepwise balloon inflation increased the motility index at all three recording sites, and the values were significantly greater (P = 0.03) at the balloon site than at the proximal or the distal sites (Table 2). The proximal site also showed greater motor activity than did the distal site (P = 0.03) (Table 2). Overall, more activity was seen in patients than in controls, but this difference was significant (P = 0.04) at pressures less than 20 cm H₂O at all three sites and at pressures of 30 and 40 cm H₂O at the distal recording site.

Reactivity

The reactivity of the esophagus to balloon distention was significantly higher in patients than in controls (P = 0.01) (Figure 4, right). At 10 cm H₂O, the reactivity was 75 ± 30 mm² in patients and 42 ± 12 mm² in controls (P = 0.001). At 40 cm H₂O, the respective values were 254 ± 119 mm² and 204 ± 69 mm² (P = 0.04).

Discussion

Top Methods Results Discussion Author & Article Info References

It has been suggested that noncardiac chest pain may arise either from stimulation of pH-sensitive receptors in the esophagus [26-28] or from excitation of mechanoreceptors activated by abnormal motility [3, 28]. Both mechanisms can produce esophageal pain, but the cause of such pain is obscure in approximately 50% of patients [2, 3, 6, 8, 10]. We rigorously defined our patient group and tried to elucidate the underlying pathophysiologic mechanisms in patients considered to have unexplained and persistent chest pain after conventional clinical testing. Chest pain typical of the pain our 24 patients had been experiencing was reproduced in 20 patients (83%). These patients reported that the nature and the intensity of the pain induced by balloon distention were similar to that experienced spontaneously. We also observed important differences in tensile and contractile properties between controls and patients. Four patients (17%) did not experience pain. The pain these patients typically experience may be caused by a condition that does not originate in the esophagus, such as microvascular angina or syndrome X [29, 30], abnormal sensitivity of cardiovascular structures [31], or psychological abnormalities [32, 33].

However, in the 20 patients who reported pain, sensory responses during distention of the esophagus were strikingly different from those in controls. Seventeen patients (85%) perceived balloon inflation at values below the normal sensory threshold of 20 cm H₂O. Furthermore, all 20 patients reported moderate discomfort at pressures ranging from 10 to 40 cm H₂O, whereas none of the controls reported discomfort at pressures as great as 50 cm H₂O. When the balloon was inflated to a pressure greater than 50 cm H₂O, five controls reported moderate discomfort at a mean pressure of 58 cm H₂O, and three controls reported pain at 65 cm H₂O. In contrast, typical chest pain was reproduced in 20 patients at pressures ranging from 20 to 50 cm H₂O. Thus, the patients not only perceived balloon distention but also experienced discomfort or pain at significantly lower thresholds than did the controls. This indicates that these patients have a hypersensitive esophagus.

Our study not only confirms previous observations that the esophagus is hypersensitive to stretching in a subset of patients with noncardiac chest pain [10-13] but also offers an explanation for the inconsistent results reported previously. In one study, graded balloon distention with volumes as great as 9 mL of air induced chest pain in 60% of patients with noncardiac chest pain, but 20% of controls also had pain [12]. Using similar volumes, other investigators reported that only 3 of 37 patients [9] and 1 of 20 patients [14] developed pain. In another study [10], 47% of patients with noncardiac chest pain had pain when a 10-mL balloon (diameter, 27 mm) was distended in the esophagus; however, at a lower volume (diameter, 24.6 mm), all 10 controls reported atypical chest pain. Our findings may explain these discrepancies: We found that the cross-sectional area of the esophagus varied among participants and that small changes in pressure could induce significant changes in the sensory and biomechanical variables of the esophageal wall. Thus, a patient with chest pain and a larger cross-sectional area may be able to accommodate a certain balloon volume without discomfort, but a normal person with a normal or smaller cross-sectional area may develop pain at the same volume. Therefore, an assessment of visceral sensation that is based solely on changes in balloon volume may be unreliable, whereas an assessment done under isobaric conditions is less likely to be influenced by these variables. In an organ with a small lumen, such as the esophagus, balloon distention with stepwise increments in balloon pressure induces reproducible changes [15] and may also improve the sensitivity and specificity of the balloon distention test.

Our study also showed important differences in the biomechanical variables between patients and controls. Stepwise increments of intraluminal pressure produced a linear increase in the cross-sectional area of the balloon in both patients and controls. Patients tended to have a larger cross-sectional area in response to stretching, but, paradoxically, the reactivity of the esophageal wall was also greater. In many patients, the reactive contractions were so strong that the distended balloon was almost completely deflated, often at lower inflating pressures. This increased reactivity of the esophagus, particularly at lower levels of inflation, was further corroborated by the measurements of the frequency and amplitude of contractions and the motility index, all of which were higher in patients than in controls. Furthermore, balloon distention normally induces secondary peristalsis with proximal contraction and distal inhibition of the amplitude of pressure wave. This distal inhibitory response was either attenuated or absent in patients, suggesting neuromuscular dysfunction and confirming previous observations [13]. The greater reactivity of the esophageal wall in our patients could be explained in part by the association between tension and strain. Although the amount of tension generated within the esophageal wall was similar in the two groups, the degree of strain was proportionately lower in patients. This suggests that the esophageal wall is stiffer in patients with noncardiac chest pain than in normal persons and may therefore react excessively to luminal distention. This reduced compliance was further confirmed by calculation of the pressure elastic modulus. Stepwise balloon inflations produced little change in the elastic modulus of controls but significantly increased the modulus in patients.

Within the intact human gut, biomechanical variables are probably influenced by peristaltic and other mechanical properties, visceral sensory input, and extramural factors. Therefore, the net change in any one variable is both dependent on and influenced by changes that occur within the wall as well as extramural forces. In vivo, it is therefore difficult to determine the influence of each variable. Nonetheless, the significant alterations in biomechanical function, together with a lower threshold for inducing peristaltic contractions and a lower threshold for sensation in our patient group, supports the existence of an underlying neuromuscular disturbance. Thus, unlike a previous study suggesting that noncardiac chest pain was not related to esophageal contractile events [12], our study shows not only that the esophagus is more reactive to luminal stretching but also that the hyperreactivity is associated with pain and reduced compliance.

A recent study [31] reported abnormal sensitivity to either right ventricular pacing or intracoronary infusion of adenosine in patients with unexplained chest pain. This observation, together with our findings, supports the existence of visceral hyperalgesia of the mediastinal structures as a mechanism for pain in these patients. The afferent viscerosensory fibers overlap extensively with somatic nerve fibers [34, 35], which explains the poor localization and referral of visceral pain and explains why esophageal pain mimics cardiac pain. The afferent sensory neurons receive input from both the low-threshold mechanoreceptors and the high-threshold nociceptors [34, 36]. If the nociceptor tract has been sensitized through previous noxious injury, physiologic stimulation of the mechanoreceptors may additionally activate the nociceptive pathway to trigger acute pain [34-36]. If no overt inflammation is present, this hypothesis provides an explanation for the genesis of visceral hyperalgesia.

Impedance planimetry offers a new way to assess the sensory and biomechanical properties of the esophagus and provides new insights into the pathophysiology of esophageal dysfunction. The technique is simple, safe, inexpensive, comprehensive, and reproducible [15] and could be useful for evaluating patients with persistent chest pain in whom coronary artery disease has been excluded. Although not widely available, impedance planimetry can be developed at a small cost and may prove to be more cost-effective than empiric therapy or cardiac reassessment. The most useful aspect of the technique is that it provides a way to examine esophageal dysfunction. This could facilitate the evaluation of new drugs that may modify neuromuscular disturbances, such as the hypersensitivity and the hyperreactivity of the esophagus to stretching and the poor esophageal compliance that we saw among many patients with unexplained chest pain.

Presented in part at the annual meeting of the American Gastroenterological Association, New Orleans, Louisiana, May 1994.

Dr. Gregersen: Institute of Mechanics and Materials, Department 0404, University of California, San Diego, 9500 Gilman Drive, San Diego, CA 92093-0404.

Mr. Hayek: Department of Internal Medicine Division of GI/Hepatology, University of Iowa Hospitals and Clinics, 4612 JCP, Iowa City, IA 52242.

Dr. Summers: Department of Internal Medicine, Division of GI/Hepatology, University of Iowa Hospitals and Clinics, 4545 JCP, Iowa City, IA 52242

Dr. Christensen: Department of Internal Medicine, Division of GI/Hepatology, University of Iowa Hospitals and Clinics, 4548 JCP, Iowa City, IA 52242