We studied esophageal dysfunction as a possible source of pain in patients with angina-like chest pain but normal coronary angiograms. The patients were enrolled consecutively and according to strict cardiologic criteria so that only patients in whom the source of chest pain was truly obscure were evaluated. The study patients did not have coronary artery disease, coronary artery spasms, abnormal motion of the cardiac wall, valvular disease, or cardiomyopathy. In addition, coronary sinus lactate release in response to atrial pacing was measured in all patients to determine whether the patients had ischemic myocardial metabolism despite normal macroscopic coronary anatomy. Because the patients with an abnormal response to exercise electrocardiographic testing might have had an independent disease entity (the cardiologic syndrome X) [18, 19], the patients were grouped according to the outcome of such testing. The study design is shown in Figure 1.

View larger version (29K): [in this window] [in a new window]   Figure 1. Diagram of the study design. ECG equals electrocardiogram.

Methods

Top Methods Results Discussion Author & Article Info References

Participants

All patients had chronic angina-like chest pain and normal coronary angiograms and were referred for routine invasive evaluations of myocardial metabolism at Skejby University Hospital, Aarhus, Denmark, between September 1990 and September 1993. The patients were then screened for the esophageal investigation. The study group consisted of 63 patients: 36 women and 27 men (mean age, 51.7 years [range, 18 to 70 years]). Inclusion criteria were the following: 1) chest pain lasting more than 6 months and normal coronary angiogram [less than 50% luminal narrowing] judged independently by the same two experienced cardiologists; 2) left ventricular ejection fraction greater than 50% as determined by ventriculography; 3) valvular and myocardial diseases excluded by echocardiography and angiography; and 4) coronary artery spasm excluded by hyperventilation testing [20].

Myocardial lactate exchange was evaluated by catheterization of the coronary sinus through the antecubital vein and an arterial catheter in the femoral artery. Lactate levels were measured in blood samples obtained simultaneously from the antecubital vein and the femoral artery at baseline (mean, three samples) and during cardiac pacing to 150 beats/min (mean, two samples) for a maximum of 10 minutes or until angina occurred [21]. We excluded patients with spontaneous or exercise-induced bundle-branch block because such patients may be an independent patient group susceptible to developing substantial deterioration of left ventricular function within a span of years [22]. We did not use the results of any previous testing to select study patients.

We divided the patients into two groups according to the result of electrocardiographic testing during bicycle exercise [exercise began at 25 W, with a 25-W increase every 2 minutes]: 1) patients with a positive exercise electrocardiogram [more than equals 1 mm horizontal or down-sloping ST-segment depression at 80 ms after the J point] and 2) patients with a normal electrocardiogram. All exercise electrocardiograms were read by the same senior cardiologist, who was blinded to patient history.

We selected 22 healthy persons (13 women and 9 men; mean age, 47.6 years [range, 31 to 60 years]) as controls for all esophageal investigations and 24-hour electrocardiographic monitoring. Controls were recruited from hospital staff and blood donors. No control had a history of gastrointestinal, neurologic, endocrinologic, or cardiologic diseases or was taking medication.

All participants gave informed consent for the study, which was approved by the local ethics committee.

Esophageal Symptom Questionnaire

We administered a standard questionnaire to all participants to determine esophageal symptoms. Heartburn was defined as retrosternal burning beginning in the xiphisternal region and ascending to the neck; regurgitation was defined as the passive appearance of gastric or intestinal content, or both, in the mouth; dysphagia was defined as the sticking of a bolus or the subjective sensation of a bolus passing down the esophagus; and odynophagia was defined as the sensation of pain during swallowing.

Study Conditions

All pharmacologic therapy was discontinued at least 72 hours before the investigation began; treatment with proton pump inhibitors (received by one patient) was stopped 5 days before. All studies were done during an unrestricted hospitalization that was not related to the angiographic procedure. Patients and controls were fully ambulatory. Investigations began between 7:00 a.m. and 7:30 a.m. on the morning after an overnight fast and ended at the same time the following morning. Participants could eat and drink between meals but were told to avoid consuming food or beverages that had a low pH (for example, orange juice, cola-flavored drinks, or citrus fruit). However, we did not directly assess compliance with this point. Participants were allowed to smoke and consume alcohol. Participants recorded all intake by operating an event marker on the portable esophagus monitor and by recording the beginning and end of the intake on a diary time sheet. The same procedure was followed for recording pain periods and periods during which the participants were supine.

Esophageal pH and Manometry Monitoring

Portable equipment was used for the 24-hour monitoring. The pH probe contained two antimony electrodes (Monocrystant, Synectics Medical, Stockholm, Sweden) that were 5 cm apart. The manometry catheter contained three strain-gauge microtransducers that were 5 cm apart (Konigsberg, Synectics Medical). The probe, the catheter, and an Ag/AgCl skin electrode (which served as a reference for the pH electrodes) were connected to a belt-mounted microprocessor that had a 2-MB memory and a 5-Hz sampling frequency (Microdigitrapper, Synectics Medical). We calibrated the microprocessor at pH values of 7.01 and 1.07 (pH electrodes) and at 0 and 50 mm Hg by immersing the manometry catheter into a water column. The pH probe was attached to the side of the manometry catheter with tape so that one of the two pH electrodes was positioned midway between the proximal and middle transducers and the other was positioned midway between the distal and middle transducers. The combined assembly was inserted into the participant's nose and placed in the esophagus. The tip transducers were positioned 2.5, 7.5, and 12.5 cm proximally to the lower-esophageal sphincter; thus, the pH electrodes were positioned 5 and 10 cm above the sphincter. The lower-esophageal sphincter was located by a station pull-through technique in which absolute pressure values are shown in a display on the microprocessor. Positioning was confirmed by ingestion of barium contrast and by video fluoroscopy, a technique previously verified in our center [23].

Analysis of Esophageal Monitoring

After the 24-hour esophageal studies ended, data were downloaded to a personal computer-based program (Multigram version 6.01C9, Gastrosoft, Synectics Medical) that has recently been validated in a comparison with manual analysis [24]. The pH data from the two electrodes were analyzed for the percentage of time during which the pH was less than 4.0 (the reflux index). We set an automatic baseline for each pressure channel according to the most frequent level of the smaller pressure variations. The baseline was updated every 20 seconds. A contraction was defined as an increase in the pressure above a 15-mm Hg threshold relative to the baseline that lasted at least 1 second [7, 24-26]. Contractions were temporally classified according to the peaks of the contractions. The analysis program provided values of the median amplitude and duration of the contractions for each pressure channel. The program also reported the percentage of multiple peaked contractions. We defined multiple peaks as secondary peaks in which the amplitude was at least 10% of the amplitude of the main peak, the interval between the local minimum and peak was greater than 0.5 seconds, and the amplitude of the local minimum was at least 15% of the amplitude of the main peak. The analysis report included the percentage distribution of peristaltic, simultaneous, and isolated contractions, the sum of which is 100%. Simultaneous contractions were defined as contraction velocities greater than 15 cm/s. Peristaltic contractions were defined as ranging from 1.3 to 15 cm/s [26].

On the basis of the esophageal recordings, we compared controls with all patients, with patients who had a positive exercise electrocardiogram, and with patients who had a normal exercise electrocardiogram. In the comparison of baseline values with values during pain periods, any abnormal activity could be secondary to the pain rather than the actual cause of the pain. Because of this, the 2-minute prepain interval was introduced [27]. Accordingly, we defined the prepain period as the period that began 2 minutes before and ended after the onset of pain. We defined the pain period as the entire time during which the patient experienced pain. We compared values recorded during prepain and pain periods with those at baseline (the periods during which the participants were upright minus the time spent eating and the pain and prepain periods). Thus, patients served as their own controls.

Electrocardiographic Monitoring

Ambulatory, 24-hour electrocardiograms were recorded by two-channel Oxford Tracker tape recorders (Oxford Instruments, Oxford, United Kingdom) during the esophageal monitoring and provocation testing. The clocks of the esophageal microprocessor and the Holter electrocardiographic recorder were synchronized. The leads were placed to obtain signals comparable to those from leads I and V5 in a standard 12-lead electrocardiogram. A technician, who was blinded to group assignment, did the analysis using a Reynolds Pathfinder II (Reynolds Medical, Hertford, United Kingdom). We defined episodes of transient myocardial ischemia as horizontal or down-sloping ST-segment shifts that were 0.1 mV or more from baseline, occurred 80 milliseconds from the J-point, and lasted at least 60 seconds.

Esophageal Provocation Testing

Two provocation tests were done at the end of the 24-hour monitoring period. The first was conducted with intravenous edrophonium chloride (Tensilon, Roche, Basel, Switzerland), a short-acting anticholinesterase that increases cholinergic activity at muscarinic receptors [28] at a dose (80 µg/kg of body weight) that may produce chest pain and cause manometric activity [25, 28, 29]. Because this test is associated with a risk for airway constriction, it was not done in four patients with bronchial asthma [29]. After edrophonium chloride was injected, patients swallowed five 10-mL glasses of water (as advocated by Richter and colleagues [29]) at 1-minute intervals. The result was considered positive if the test reproduced the patient's typical pain [29, 30].

The second provocation test was acid perfusion, as described by Bernstein and Baker [31]. After the manometric catheter and the pH probe were removed, videofluoroscopy was used to place a thin catheter between the proximal two thirds and distal third of the esophagus. Saline was initially infused, and, although participants were aware that 0.1 moles of HCl per liter would be administered at some point, the actual administration occurred without their knowledge. The HCl was infused at 6 mL/min for no more than 15 minutes or until chest pain was provoked [32]. If pain was provoked, sterile water was infused until the pain disappeared. At this point, HCl was infused again until the pain was reproduced a second time. The test result was considered positive only if both infusions evoked pain. Provocation of heartburn alone was not considered a positive test result [30].

Statistical Analysis

We used Fisher exact tests for categoric baseline variables and the unpaired t-test for continuous baseline variables. Esophageal manometry data recorded during the prepain and pain periods were compared with baseline data in unweighted and weighted analyses. The differences in the unweighted comparison were analyzed using a paired t-test. In the weighted comparison, the differences were weighted according to the time at risk using the following formula: time_pain x time_baseline/time_pain + time_baseline = 1/time_pain + 1/time_baseline^–1. This gives greater weight to data from a person whose pain and baseline periods were of long duration and gives less weight to data from a person whose pain or baseline periods, or both, were of short duration. For a fixed period (pain plus baseline), the maximum weight occurred when pain and baseline periods were of equal duration. The manometry data from prepain periods were compared with baseline data in an analogous manner. Differences in reflux indices were compared using a paired t-test. The number of reflux episodes occurring during pain periods and periods in which participants remained upright was described by Poisson distributions and compared in a stratified analysis using persons as strata. The reflux data from prepain periods were compared with baseline data in an analogous manner. Bivariate associations were evaluated by least-squares regression. Statistical significance was defined as a P value less than 0.05.

Results

Top Methods Results Discussion Author & Article Info References

Of 66 candidates for the study, 3 did not agree to participate. Our study group therefore consisted of 63 patients. Two patients had non-insulin-dependent diabetes mellitus, 6 patients had hypercholesterolemia (cholesterol level more than 8 mmol/L), and 4 patients had bronchial asthma. No patients had other metabolic disorders (such as insulin-dependent diabetes mellitus, hyperthyroidism, or hypothyroidism) or signs of chronic lung disease at the time of the study. Results of coronary angiography were normal in 53 participants. Eight participants had a single stenosis with 10% to 30% luminal narrowing, and 2 participants had a single stenosis with 30% to 50% luminal narrowing. Twenty-two women and 13 men (mean age, 50.7 years [range, 18 to 70 years]) had positive exercise electrocardiograms, and 14 women and 14 men (mean age, 53.0 years [range, 34 to 70 years]) had normal exercise electrocardiograms. Myocardial lactate metabolism could not be evaluated in 3 participants because of technical problems. The fraction of myocardial lactate extraction did not change in response to pacing stress in either patient group. In patients with a positive exercise electrocardiogram, the fraction (mean ±SD) was 19.7% ± 13.4% at rest and 22.5% ± 11.4% during pacing; in patients with a normal exercise electrocardiogram, the fraction was 14.4% ± 13.8% at rest and 13.9% ± 12.6% during pacing (P > 0.2 for both comparisons). Baseline data and the results from the esophageal symptom questionnaire are listed in Table 1; cardiologic data are listed in Table 2.

Esophageal pH and Manometry Monitoring

Complete sets of 24-hour esophageal monitoring data were obtained in 49 of the 63 patients. We excluded the data from a patient who accidentally turned off the recorder after 15 minutes of recording. In the 13 patients who did not have complete data, the recording time was reduced (mean, 19.7 hours [range, 9.5 to 22.6 hours]) because of battery problems; we included data from these recordings. A total of 62 recordings were obtained. In the control group, all esophageal recordings were complete. Holter electrocardiographic recordings could be used to analyze ST-segment depression in 62 patients and in all controls. This analysis could not be done in one recording because of poor technical quality. Bernstein testing was done for all participants.

Forty-six of the 63 patients had a total of 248 chest pain episodes during the esophageal studies (mean, 3.9 episodes [range, 0 to 22 episodes]); each episode lasted a mean of 7.5 minutes (range, 1 to 129 minutes). Pain did not occur more often in patients with a positive exercise electrocardiogram (mean, 4.3 episodes [range, 0 to 22 episodes]) than in patients with a normal exercise electrocardiogram (mean, 3.8 episodes [range, 0 to 20 episodes]) (P > 0.2). No controls reported chest pain during the investigation.

In the unweighted analysis of 24-hour esophageal manometry, few differences were seen between patients or controls during the entire period or the baseline periods. Patients with a normal exercise electrocardiogram had more contractions per minute in the proximal channel (2.13 ± 1.44 contractions per minute during the prepain period compared with 1.48 ± 0.71 contractions per minute at baseline; P < 0.05) and a smaller fraction of multiple-peak contractions in the distal channel (0.15% ± 0.67% during the prepain period compared with 1.05% ± 1.17% at baseline; P < 0.01). Patients with a positive exercise electrocardiogram had a lower amplitude of contractions in the middle channel during the prepain period (37.4 ± 11.3 mm Hg compared with 43.7 ± 17.0 mm Hg at baseline; P < 0.05). In patients with a normal exercise electrocardiogram, the fraction of multiple-peak contractions in the middle channel was higher during pain periods than at baseline (1.80% ± 2.65% compared with 0.60% ± 0.88%; P < 0.05).

Similarly, only a few differences were seen in the time-weighted analysis. In patients with a normal exercise electrocardiogram, the fraction of multiple-peak contractions in the distal channel was smaller during the prepain period than at baseline. Patients with a positive exercise electrocardiogram had fewer contractions per minute in all three pressure channels during the prepain period than at baseline (P values, 0.007 to 0.04). In all patients, contractions in the distal pressure channel lasted longer during the pain periods than at baseline (P < 0.05).

For all patients, reflux occurred significantly less frequently during pain periods at both the distal and proximal recording positions. At the proximal electrode, reflux indices were 0.6% ± 2.4% during pain periods and 1.5% ± 2.3% at baseline (P < 0.05); at the distal electrode, reflux indices were 0.6% ± 2.8% during pain periods and 2.6% ± 3.2% at baseline (P < 0.01). The difference in reflux indices was also statistically significant at the distal position for both patient groups when the data were grouped according to the results of exercise electrocardiography. In patients with a positive electrocardiogram, the reflux indices were 0.5% ± 2.1% during pain periods and 2.3% ± 3.4% at baseline (P < 0.05); in patients with a normal electrocardiogram, the indices were 0.7% ± 3.7% during pain periods and 2.9% ± 3.1% at baseline (P < 0.05). No significant differences in reflux indices were seen between prepain and baseline periods in patients or between baseline periods in patients and controls. When we compared the number of reflux episodes occurring during pain or prepain periods and while the participants were in the upright position (variables were compared as Poisson data stratified by person), the number of reflux episodes at the proximal electrode occurring during prepain periods was slightly higher than would have been expected by random sampling (Table 3).

In Figure 2, the diurnal variation of episodes of chest pain and reflux is shown for the 49 patients who had complete 24-hour esophageal recordings and for the 22 controls. The number of pain episodes that occurred in a 2-hour period correlated well with the number of reflux episodes in the same period (r = 0.83; P < 0.001); the number of pain episodes occurring in patients and the number of reflux episodes in the controls in a 2-hour period were also well correlated (r = 0.85; P < 0.001).

View larger version (24K): [in this window] [in a new window]   Figure 2. Top. Diurnal variation in the percentage of the total number of episodes of chest pain (white bars) and of reflux (black bars) measured 5 cm above the lower-esophageal sphincter in 49 patients with angina-like chest pain, normal coronary angiograms, and complete 24-hour esophageal recordings. Also shown is the percentage of the total number of reflux episodes measured 5 cm above the lower-esophageal sphincter in 22 healthy controls (diagonally striped bars). Results were pooled for 2-hour intervals throughout the day. Bottom. Scatter plots of pain episodes (number of pain episodes occurring in a 2-hour period expressed as a percentage of the total number of episodes). Pain episodes are shown in relation to the percentage of reflux episodes in patients (black circles) (r = 0.83; P < 0.001) and the number of reflux episodes in healthy controls (white circles) (r = 0.85; P < 0.001). The number of reflux episodes is expressed as the number of episodes occurring during a 2-hour period as a percentage of the total number of episodes in each group.

Esophageal Provocation Testing

Edrophonium chloride testing and acid perfusion caused chest pain in 9 patients (14%) and 10 patients (16%), respectively (Table 4). No differences in the number of positive test results were seen between patients with normal exercise electrocardiograms and those with positive exercise electrocardiograms. No controls reported chest pain after provocation testing. During provocation testing, no ST-segment depression was seen with Holter electrocardiography in any participant. Esophageal monitoring variables did not differ between controls and patients with a positive response to one or both provocation tests (n = 16 [25%]) and did not change between baseline and prepain or pain periods.

Electrocardiographic Monitoring

Among 16 patients, a total of 595 minutes of Holter electrocardiographic ST-segment depressions were recorded; this corresponds to 48 episodes (1 patient had 405 minutes of ST-segment depression during 2 episodes). Forty-six of the episodes were seen among patients with a positive exercise electrocardiogram; 2 were seen among patients with a normal exercise electrocardiogram. Only one prepain period and four pain periods were associated with ST-segment depression (Table 4). Reflux occurred during two 2-minute periods preceding ST-segment depression and during three periods of ST-segment depression (two of the three periods occurred in the patient with 405 minutes of ST-segment depression).

Discussion

Top Methods Results Discussion Author & Article Info References

We investigated esophageal function as the possible source of pain in patients with angina-like chest pain despite normal coronary angiograms. In general, reflux indices and esophageal motor function did not differ between patients and healthy controls. Furthermore, esophageal function in the patients did not change appreciably between baseline and the prepain and pain periods. Esophageal provocation testing with HCl and edrophonium chloride, which provoked pain in 25% of the patients but in no controls, was the only indication of an esophageal cause of symptoms. Additional Holter electrocardiographic recordings showed that only 2% of ST-segment depression episodes were associated with prepain periods and that only 8% were associated with pain periods.

We have previously shown that upper endoscopy and distal esophageal biopsy are of limited value in patients with angina-like chest pain [33]. In our previous study, the findings from these procedures did not differ significantly from the results of a large-population screening of randomly selected persons [34, 35]. In many centers, prolonged esophageal monitoring has recently become almost a gold standard for further investigation of chest pain of unknown origin [8, 27, 36]. This approach is based on previous findings indicating that patients with chest pain have many esophageal abnormalities. That we found no evidence of substantially abnormal function may be attributable to factors associated with patient selection, recording technique, and data analysis.

The risk for selection bias is increased in studies of patients who have chest pain but normal coronary angiograms. Thus, investigators often report a high incidence of abnormalities that are within their specific medical discipline [37]. The heterogeneity of patient populations in many gastroenterologic studies has previously been addressed [10, 16, 17]. On the other hand, many cardiologic studies of patients with chest pain but normal angiograms have not included esophageal investigations or have only used short-term esophageal monitoring. We combined extensive cardiologic and gastrointestinal investigations to respond to recommendations from both fields. Patients were selected from a cardiologic specialist referral center that serves an area with about 1.7 million inhabitants. All patients with chest pain of undetermined origin and normal coronary angiograms referred over a 3-year period for invasive cardiac evaluation of myocardial metabolism were offered additional gastrointestinal investigation. The general willingness of patients to participate in the extensive investigations minimized the likelihood of selection bias at this stage and appears to reflect the patients' interest in finding a solution to their symptoms. It is reasonable to assume that the patients' conditions represented true diagnostic dilemmas for the referring physicians.

Our method of patient selection differs from that used in a recent study by Lam and colleagues [7]. In that study, participants had come to the emergency department suspected of having severe angina pectoris or myocardial infarction, which was later ruled out. It could be argued that with such patients, a less selective approach can be used toward the workup of unknown chest pain. However, cardiac conditions such as abnormal metabolic function of the myocardium [13], impaired left ventricular function [14, 15], and coronary artery spasm have not been ruled out in such patients. Furthermore, investigation of the esophagus after an emergency situation is associated with a risk for erroneously diagnosing a gastrointestinal disorder, because esophageal function as measured by manometry is susceptible to nonspecific stress [38].

Esophageal recording techniques have improved during the last few years: More pressure and pH channels are available, and more information can be stored in the portable data logger. Fewer abnormal esophageal findings have been reported in patients with chest pain since these recording techniques were enhanced [9, 39, 40]. In previous studies, brief, conventional esophageal investigations were used [2, 4, 6, 41]. This decrease in the number of abnormalities detected is probably due to the ambulatory (unrestricted) study conditions allowed by the improved techniques. A brief investigation with 10 standardized swallows may be more stressful to a patient with chest pain than to a healthy control.

In the dynamic investigations, we considered two main criteria for linking abnormal organ function to pain episodes: 1) generally abnormal organ function compared with the function in controls and 2) abnormal organ activity coinciding with prepain or pain periods. If a patient group is believed to have a condition representing a homogeneous disease entity, any abnormality shown must differ statistically from conditions in a control group or conditions at baseline. In previous studies of long-term monitoring of esophageal motility, abnormal function was defined as a deviation from baseline (> 97.5th percentile or more than 2 SDs), and the fraction of such episodes could be calculated. With such definitions, a temporal association between chest pain episodes and abnormal motility has been observed in 8.6% to 30.6% of pain episodes [9, 25, 30, 40]. Investigators in these studies did not report whether any differences were seen between groups in mean activity during pain periods and activity at baseline. We could not confirm previous findings of associations between chest pain episodes and gastroesophageal reflux [8, 9, 42, 43]. The number of pain and reflux episodes in the patients correlated with the number of reflux episodes in both patients and controls (Figure 2); this finding supports the hypothesis that pain and esophageal activity may incidentally peak at the same time. Because both phenomena last several minutes, there is a risk for incorrectly assuming that a direct association exists.

The symptom index refers to the number of pain periods during which esophageal abnormalities are noted, expressed as a percentage of the total number of pain episodes. This index is meant to identify patients in whom an esophageal origin of chest pain should be sought [42]. In different centers, symptom indices range from 25% to 75% [25, 36, 42, 44]. Many patients may have only a few chest pain episodes during a 24-hour investigation, which may cause differences between statistical and clinical significance. If one of two chest pain episodes coincides with abnormal esophageal motility, the symptom index would be 50%. In addition, the same patient might have dozens of esophageal abnormalities during the pain-free periods that are similar to those that occurred with pain [45]. Interpreting these abnormalities as "silent esophageal abnormalities" does not seem reasonable.

Chest pain was elicited by esophageal provocation tests in patients but not in controls. Pain has been reported in 10% to 36% of patients with angina and no suspected cardiac disease who receive esophageal acid perfusion provocation [8, 30, 40, 46-49]. The exact mechanism causing pain during acid perfusion is not known. Local sensitization may occur, because edema and hyperemia of the esophageal mucosa reduces the time until pain develops [50]. Although we did not observe ST-segment depression during provocation testing, other researchers [51, 52] have suggested that esophageal acid installation may alter coronary blood flow. The wide use of intravenous drug provocations has been based on the idea that certain pharmacologic agents may act specifically on the smooth musculature of the esophagus. The commonly used spasmomimetic drug edrophonium chloride can produce both chest pain symptoms and alterations in esophageal motility [29]. In previous studies [8, 25, 29, 47, 49, 53], edrophonium chloride has been shown to provoke chest pain symptoms in 4.5% to 55% of patients with angina but no suspicion of cardiac disease. However, in two studies in which edrophonium chloride and saline were administered in a double-blind manner, no patients experienced pain solely after edrophonium [40, 54]. Nonetheless, this drug has never been reported to provoke chest pain in healthy persons, and little is known about the pain mechanism. Because patients with edrophonium-induced pain are not characterized by any single manometric finding, it has been hypothesized that patients with a positive test result may have a heightened sensitivity to esophageal stimuli [29].

A key issue in the study of patients with angina and normal coronary angiograms is whether they have unrecognized cardiac disease. One cardiologic hypothesis of the underlying mechanism of chest pain in this patient group has been microvascular angina, that is, a reduced ability of the minor cardiac vessels to dilate [55-57]. The attractiveness of this idea has faded with the recent finding that restrictions in cardiac flow that may produce pain are almost never associated with detectable myocardial ischemia [58, 59]. An abnormal cardiac nociception is one of the most recent hypotheses [60, 61]. We have previously found altered central nervous system responses to visceral and somatosensory nociceptive input in patients with angina-like chest pain [62].

One limitation of our study was the lack of deglutitive signals during esophageal pressure monitoring. Because of this, we could not distinguish between deglutitive and nondeglutitive motor activity [63]. Optimally, esophageal manometry catheters and pH probes should be placed in a position relative to the lower esophageal sphincter. This is usually done by conventional manometry and the station pull-through technique [9, 30, 64]. We used a modified technique for locating the lower esophageal sphincter [65]; positioning was verified by videofluoroscopy and swallowing of barium contrast, a method previously validated at our medical center [23].

In conclusion, we found no relation between chest pain and abnormal esophageal function. Skepticism about the interpretation of results of esophageal investigations in such patients is reinforced by the demands placed on the patients by such testing [33] and by a poor cost–benefit ratio [66]. Furthermore, clinical trials in which medical therapy was directed against pain presumed to be generated by esophageal abnormalities have been disappointing [67, 68]. Our results question the rationale of routine esophageal investigations in this patient group. On the other hand, the positive results of the pain-provoking procedures indicate that, through still-unknown mechanisms or perhaps a general abnormal visceral nociception, some form of esophageal involvement in the generation of chest pain cannot be entirely ruled out in these patients.

Presented in part at the 68th Scientific Sessions of the American Heart Association, Anaheim, California, 13-16 November 1995.

Dr. Funch-Jensen: KIR-GAS Afdeling 235, Hvidovre Hospital, Kettegaard Alle 30, DK-2650 Hvidovre, Denmark.

Dr. Bagger: Cardiological Sciences, St. George's Hospital Medical School, Cranmer Terrace, London SW17 0RE, United Kingdom.