Correspondence to: Dr H. H. -X. Xia, Department of Medicine, The University of Hong Kong, Queen Mary Hospital, Pokfulam, Hong Kong, China. E-mail: firstname.lastname@example.org
Aim : To determine whether symptomatic response to lansoprazole predicts abnormal acid reflux in endoscopy-negative patients with non-cardiac chest pain.
Methods : Patients who complained of chest pain, but had normal coronary angiography, were asked to undergo upper endoscopy. Those without gastric and oesophageal lesions were recruited for 24-h ambulatory oesophageal pH monitoring, and were randomly given lansoprazole 30 mg or placebo, both daily for 4 weeks. Chest pain symptoms were recorded before and 1 month after treatment on a locally validated questionnaire. The symptom score was calculated by multiplying the severity and frequency of the symptom, and symptom improvement was defined as > 50% reduction in symptom score.
Results : Overall, 68 patients, 36 on lansoprazole and 32 on placebo, completed the trial. The symptom score was reduced significantly in both groups (P < 0.001). In the lansoprazole group, more patients with than without abnormal reflux showed symptom improvement (92% vs. 33%; odds ratio = 22; 95% confidence interval, 2.3–201.8; χ2 = 10.9; P = 0.001), giving a sensitivity, specificity, positive predictive value, negative predictive value and accuracy of 92%, 67%, 58%, 94% and 75%, respectively. In the placebo group, the rates of symptom improvement were similar between those with and without abnormal reflux (33% vs. 35%, P = N.S.).
Conclusions : Treatment with lansoprazole is a useful test in diagnosing endoscopy-negative gastro-oesophageal reflux disease in Chinese patients with non-cardiac chest pain.
Chest pain is a common medical problem.1 Studies have demonstrated that approximately 30–60% of hospitalized patients with chest pain have the symptom without evidence of coronary artery disease: so-called non-cardiac chest pain.2 Moreover, the annual incidence of non-cardiac chest pain in the community is approximately 25–35%.3,4 Non-cardiac chest pain is of major concern for both patients and health care providers, and thus produces a considerable economic burden to society.5–7 The causes of non-cardiac chest pain are diverse and, perhaps, overlapping. Gastro-oesophageal reflux disease is reported to be the most important cause, accounting for up to 60% of patients with non-cardiac chest pain; psychiatric disorders, musculoskeletal disorders and visceral pain hypersensitivity also contribute.5,8–11 Therefore, the origin of the cause must be identified in order to manage non-cardiac chest pain properly.
Gastro-oesophageal reflux disease is usually defined as symptoms (mainly heartburn and acid regurgitation) and/or tissue damage (e.g. oesophagitis) related to oesophageal exposure to gastric contents.12 Upper gastrointestinal endoscopy is commonly used for the diagnosis of gastro-oesophageal reflux disease. However, patients with acid reflux, but without apparent mucosal damage, may be missed by endoscopy; therefore, an accurate diagnosis of gastro-oesophageal reflux disease often requires further tests, including 24-h oesophageal pH monitoring, radiology and acid infusion testing.12,13 Unfortunately, most of these tests are invasive and costly, and some are not readily available from community-based primary care physicians or even in some hospitals. Previous studies have demonstrated that a trial with a proton pump inhibitor (omeprazole), which suppresses acid secretion, is of diagnostic value for gastro-oesophageal reflux disease, especially in those with negative endoscopy.12–17 In addition, a trial of omeprazole (large dose, short course) has a high sensitivity (69–78%) and specificity (75–86%) as a diagnostic tool for the identification of acid reflux-related non-cardiac chest pain.18–21 However, these studies had many drawbacks, e.g. small numbers of patients were included, and most patients had reflux symptoms and/or endoscopic evidence of erosive oesophagitis. Moreover, to our knowledge, there are no data on the potential diagnostic value of another proton pump inhibitor, lansoprazole, in non-cardiac chest pain. In addition, the diagnostic value of proton pump inhibitors in Chinese patients with non-cardiac chest pain has never been verified.
Therefore, this prospective study was carried out to determine whether treatment with lansoprazole could be used to identify abnormal acid reflux in endoscopy-negative Chinese patients with non-cardiac chest pain, by comparing the symptomatic response to the drug with results obtained by a gold standard test: 24-h oesophageal pH monitoring. We hypothesized that there would be a correlation between abnormal acid reflux and the subsequent outcome of lansoprazole therapy, and thus the response to such therapy could help to identify the cause of and better management strategies for non-cardiac chest pain.
Subjects and methods
In this prospective study, consecutive Chinese patients with chest pain for at least 12 weeks in the preceding 12 months, who were referred to the Cardiology Division of Queen Mary Hospital for evaluation by cardiac catheterization between November 1998 and February 2000, were assessed for their suitability to enter the study. Patients who had a normal coronary angiograph, and whose chest pain was considered by a cardiologist to be non-cardiac in origin, were asked to undergo upper endoscopy within 4 weeks of diagnosis. Gastric antral and corpus biopsies were taken for the detection of Helicobacter pylori infection by both the rapid urease test and histological examination after haematoxylin and eosin staining. Those without peptic ulcer and apparent oesophagitis, oesophageal erosions or ulcers, and who were not on antisecretory therapy, were recruited for baseline symptom assessment and 24-h oesophageal pH monitoring on the same day. Symptoms of chest pain, heartburn, acid reflux, dysphagia and dyspepsia over the past month were assessed by locally validated questionnaires.22 Patients with apparent heartburn, acid reflux, dysphagia and dyspepsia were excluded. The severity of chest pain was defined as: 1, absent (no symptom); 2, mild (present but causing little or no discomfort); 3, moderate (annoying but not interfering with daily activities/waking patient up for a few minutes only); 4, severe (causing marked discomfort and interfering a little with daily routine and/or keeping patient awake for 30–60 min); and 5, very severe (causing marked discomfort and interfering considerably with daily routine and/or preventing patient from sleeping). The frequency was defined as: 1, none; 2, few (1–3 times a month); 3, often (at least once a week); and 4, very often (at least once a day). Symptom scores were calculated by multiplying the severity and the frequency recorded on the questionnaire, with the maximum score being 20. A higher score indicates more severe symptoms.
One week after symptom assessment and 24-h oesophageal pH monitoring, patients were divided into two groups in a random and single-blind manner, and reassessed by a gastroenterologist (KCL) blind to the symptom scores and the presence of abnormal gastro-oesophageal reflux. The first group received lansoprazole 30 mg and the second group received placebo, both daily for 4 weeks. Patients were then followed up 1 month after the start of medication, and the symptoms during the month were assessed again as described above (Figure 1). An improvement in chest pain symptoms after treatment was defined as more than a 50% reduction in symptom scores between the two symptom assessments (i.e. over the 5 weeks).18,20
All patients gave written informed consent. The study was approved by the Ethics Committee of the Faculty of Medicine, The University of Hong Kong.
Twenty-four-hour ambulatory pH monitoring
Twenty-four-hour ambulatory pH monitoring was performed on an out-patient basis by a single technician (NYHW) who had no knowledge of the clinical or endoscopic information. Before each procedure, the pH electrode was calibrated using buffers of pH 1.0 and pH 7.0. The antimony pH electrode was passed transnasally, positioned 5 cm above the proximal border of the manometrically identified lower oesophageal sphincter and connected to a Digitrapper (Synectics Medical, Stockholm, Sweden). Gastric acidity was confirmed in all patients by passing the probe into the stomach before its final placement above the lower oesophageal sphincter. Patients were asked to take their usual diet, but to avoid alcohol and food or beverages with pH < 5.0. A diary card and an event recorder button were used to record the time of symptoms during the 24-h ambulatory period. On completion, data were transferred from the digital recorder to an IBM-compatible personal computer for graphical display and numerical analyses using commercially available software (Polygram 2.05 for Windows, Medtronic, Inc., 1997). All recordings were checked visually for technical quality to exclude artefacts, and analysed by a gastroenterologist (WHCH) experienced in pH studies, who was blind to the symptom scores and treatment. A reflux episode was defined as any fall in distal gastro-oesophageal pH below a threshold of 4 pH units for more than 7.5 s. The reflux episode was considered to have terminated if the pH returned to at least pH 4. Six pH parameters were evaluated, as originally proposed by Johnson and DeMeester,23 namely: (i) total number of reflux episodes; (ii) number of reflux episodes with pH < 4 for more than 5 min; (iii) duration of the longest episode; (iv) percentage total time pH < 4; (v) percentage upright time pH < 4; and (vi) percentage recumbent time pH < 4. The criteria defined locally by Hu et al., which were very similar to those defined by Richter et al., were used to establish whether each of the six parameters was normal or not24,25(Table 1). Patients were defined as having abnormal gastro-oesophageal reflux if any of the six parameters was shown to be abnormal.24,26
Table 1. Results of 24-h ambulatory distal gastro-oesophageal pH monitoring of 70 endoscopy-negative patients with non-cardiac chest pain
Patients were defined to have abnormal gastro-oesophageal reflux if any of the six parameters was abnormal.
The differences in the improvement of chest pain between the two treatment groups and between those with and without abnormal 24-h oesophageal pH were assessed using the chi-squared test (with Yates' correction if required) or Fisher's exact test. Odds ratios (OR) and 95% confidence intervals (CI) were estimated where appropriate. The t-test for independent samples was used to determine age differences, and the Mann–Whitney U-test was used to determine the difference in the symptom scores of chest pain between the patient groups. The Wilcoxon signed rank test was used to determine the changes in symptom scores. All tests were performed using the SPSS system (SPSS Inc., Chicago, IL, USA). All calculated P values were two-tailed. The alpha level of significance was set at P < 0.05.
Abnormal gastro-oesophageal reflux in patients with non-cardiac chest pain and its relation to non-cardiac chest pain symptoms
Two hundred consecutive patients with non-cardiac chest pain were screened. Ten patients were excluded because of the presence of structural heart disease. Seventy-eight patients agreed to participate in the study. Upper endoscopy was normal in 70 (90%) patients. Of the remaining eight patients, three had duodenal ulcer, three gastric ulcer and two endoscopic oesophagitis. Therefore, 70 patients with non-cardiac chest pain and normal endoscopy were recruited for baseline symptom assessment and 24-h ambulatory distal gastro-oesophageal pH monitoring. Symptom scores were in the range 4–16, with a median of 4 and a mean ± S.E.M. of 5.72 ± 0.33. Twenty-four (34.3%) patients had at least one abnormal pH parameter (Table 1).
Effect of lansoprazole on non-cardiac chest pain symptoms
Of the 70 patients, 36 were randomized to receive lansoprazole and 34 to receive placebo. All patients in the lansoprazole group completed the medication and returned for the follow-up visit, but two patients in the placebo group dropped out due to reasons unrelated to medication. Thus, 68 patients were included in the analysis of the clinical trial. The demographic and clinical information and results of 24-h ambulatory distal gastro-oesophageal pH monitoring of these patients are shown in Table 2.
Table 2. Demographic and clinical information and results of 24-h ambulatory distal gastro-oesophageal pH monitoring for patients receiving lansoprazole and placebo treatment
Lansoprazole (n = 36)
Placebo (n = 32)
Overall (n = 68)
Age (years) (mean ± s.d.)
59.6 ± 11.0
56.6 ± 8.5
58.2 ± 10.0
< 60 years
≥ 60 years
Current cigarette smoking
Current alcohol consumption
H. pylori infection
Mean (95% CI)
24-h gastro-oesophageal reflux
Symptom scores were reduced significantly in patients treated with lansoprazole (from 5.69 ± 0.48 to 3.64 ± 0.55; P < 0.0001) and placebo (from 5.75 ± 0.46 to 3.75 ± 0.43; P = 0.0001) (Figure 2). However, there was a significant difference in the reduction of the symptom score between patients with normal and abnormal oesophageal reflux in the lansoprazole group (P = 0.001) (Figure 2).
Overall, 52.8% (19/36) of patients receiving lansoprazole showed an improvement in chest pain symptoms, compared with 34.4% (11/32) of those receiving placebo (P = 0.127). Moreover, 15 patients treated with lansoprazole and nine patients treated with placebo did not experience chest pain after treatment (41.7% vs. 28.1%, P = 0.243). Overall, age, sex, smoking, alcohol consumption, H. pylori infection, endoscopic gastroduodenitis and abnormal reflux were not associated with symptom improvement (data not shown).
The predictive value of the ‘lansoprazole test’
In the lansoprazole group, patients with abnormal gastro-oesophageal reflux were more likely to show symptom improvement than those without abnormal reflux (91.7% vs. 33.3%; OR = 22.00; 95% CI, 2.34–201.8; χ2 = 10.9; P = 0.001; Figure 3). Thus, the ‘lansoprazole test’ had a sensitivity, specificity, positive predictive value, negative predictive value and accuracy of 92%, 67%, 58%, 94% and 75%, respectively, for the identification of acid reflux in endoscopy-negative patients with non-cardiac chest pain (Table 3). However, in the placebo group, the rates of symptom improvement were similar between those with and without abnormal reflux (33.3% vs. 35%) (Figure 3). Age, sex, smoking, alcohol consumption, H. pylori infection and endoscopic gastroduodenitis did not affect symptom improvement in the lansoprazole and placebo groups (data not shown).
Table 3. Sensitivity, specificity, positive and negative predictive values and accuracy of ‘lansoprazole test’ in the identification of abnormal acid reflux in endoscopy-negative patients with non-cardiac chest pain
24-h ambulatory oesophageal pH monitoring
> 50% symptom improvement
≤50% symptom improvement
Positive predictive value (%)
Negative predictive value (%)
In this study, for the first time, lansoprazole was evaluated as a diagnostic tool in patients with non-cardiac chest pain but without apparent oesophageal lesions. Our findings clearly demonstrated that a trial with lansoprazole, 30 mg daily for 4 weeks, significantly improved or prevented the onset of chest pain in most patients with abnormal gastro-oesophageal reflux; both lansoprazole and placebo showed similar effects in patients without abnormal gastro-oesophageal reflux. These observations indicate that symptom response to lansoprazole can be used to identify gastro-oesophageal reflux in patients with non-cardiac chest pain. For example, when patients respond to lansoprazole, abnormal gastro-oesophageal reflux is likely to be the cause of non-cardiac chest pain. On the other hand, when patients respond poorly to the drug, abnormal gastro-oesophageal reflux can be excluded from the risk list, and other causes, such as psychological factors, investigated; in this case, corresponding management strategies, including cognitive behavioural therapy, must be sought.10,27,28
Our findings obtained with lansoprazole in Chinese patients with non-cardiac chest pain are consistent with those obtained with omeprazole in Western populations in the presence of gastro-oesophageal reflux.18–21 In a cross-over study, Fass et al. administered 1-week, high-dose (40 mg in the morning and 20 mg in the afternoon) omeprazole in 37 patients with non-cardiac chest pain with and without gastro-oesophageal reflux disease, as determined by both upper endoscopy and 24-h pH monitoring.18 They reported that, of the 23 patients with gastro-oesophageal reflux disease, 78% (n = 18) responded to omeprazole and 22% (n = 5) responded to placebo. In the 14 patients without gastro-oesophageal reflux disease, only 14% (n = 2) responded to omeprazole and 7% (n = 1) to placebo. Thus, the sensitivity and specificity of the omeprazole test were 78% (18/23) and 86% (12/14), respectively.18 However, this study was performed mainly in male subjects (97%, 36/37) and the incidence of erosive oesophagitis was relatively high (43%, 16/37). One should be cautious in extrapolating these results to the large population of patients with non-cardiac chest pain, where there is a predominance of female subjects of almost 2 : 1, with a lower prevalence of oesophagitis and less frequent chest pain, as in our study. In the present study, there was no significant difference in the improvement of symptoms between males and females, indicating that gender does not appear to affect the symptomatic response to lansoprazole in Chinese patients with non-cardiac chest pain.
Decision analysis has shown that diagnostic strategies which begin with the ‘omeprazole test’ result in reduced costs and improved diagnostic certainty, compared with traditional strategies which begin with invasive diagnostic tests.21 Our study evaluated the ‘lansoprazole test’ in patients with endoscopy-negative non-cardiac chest pain, and obtained a higher response rate (92%) in patients with gastro-oesophageal reflux, with sensitivity, specificity, positive and negative predictive values and accuracy of 92%, 67%, 58%, 94% and 75%, respectively. These findings indicate that a negative result of the ‘lansoprazole test’ can exclude abnormal gastro-oesophageal reflux in patients with non-cardiac chest pain, although a positive result may not be able to identify patients with or without abnormal gastro-oesophageal reflux. In other words, in patients with non-cardiac chest pain, the ‘lansoprazole test’ is very useful in identifying those with abnormal gastro-oesophageal reflux, whereas some (33%) patients without abnormal gastro-oesophageal reflux may be misclassified as having the condition. The low specificity was due to the high response rate (33%) in patients with normal 24-h oesophageal pH. Although there are no ‘gold standards’ so far for the diagnosis of gastro-oesophageal reflux disease, symptoms, endoscopy and/or 24-h oesophageal pH monitoring are generally used to establish the diagnosis. It is unlikely that patients with normal 24-h oesophageal pH who responded to lansoprazole ‘truly’ had gastro-oesophageal reflux disease, as all three diagnostic parameters were applied in the present study. Moreover, similar response rates were observed in the placebo control group with or without abnormal gastro-oesophageal reflux (33% and 35%, respectively). A high placebo response rate has also been reported previously by Achem et al.: 44% of patients with non-cardiac chest pain and gastro-oesophageal reflux obtained significant improvement in individual daily pain scores after treatment with placebo.29 We propose that psychological factors, such as stress, depression and anxiety, are commonly present in patients with non-cardiac chest pain, and thus proton pump inhibitors may not be of greater benefit than placebo in improving the symptoms of patients with non-cardiac, non-reflux chest pain. Nevertheless, from the clinical point of view, lansoprazole can still be used in patients with non-cardiac chest pain without abnormal gastro-oesophageal reflux as some (33%) patients respond to treatment. The low specificity of lansoprazole in the diagnosis of gastro-oesophageal reflux disease may be a problem in clinical practice, as a proportion of patients without gastro-oesophageal reflux disease will be misdiagnosed as having the condition. Nevertheless, proton pump inhibitor treatment may still be beneficial in patients with non-cardiac, non-reflux chest pain by virtue of its ‘placebo’ effect.
A low dose (30 mg daily) of lansoprazole was used in this study. From a consideration of the response rate, which is similar or even higher than that obtained with a high dose of omeprazole,18–21 we believe that this dose is adequate. One possible explanation for this may be that the reflux is mild in the endoscopy-negative patients with non-cardiac chest pain investigated in our study. However, a more likely explanation involves genetic and/or ethnic differences. It has been observed that omeprazole plasma concentrations are significantly lower in white subjects than in Chinese subjects after a standard dosage.30 Proton pump inhibitors, such as omeprazole, are mainly metabolized by the enzyme S-mephenytoin 4′-hydroxylase (CYP2C19) in the liver, and their therapeutic effects are largely dependent on the CYP2C19 genotype status, i.e. a higher dose is required for homozygous extensive metabolizers than for heterozygous extensive metabolizers to achieve the same efficacy.31–33 It has been consistently demonstrated that heterozygous extensive metabolizers are more common in Chinese subjects than in Caucasians, and this is believed to be the major cause of the differences in omeprazole metabolism.34–36 Moreover, it has been found that, in Chinese and Caucasian groups with a similar proportion of CYP2C19 heterozygotes, omeprazole metabolism is still decreased in the former, suggesting possible environmental effects, such as diet, and/or other genetic effects on omeprazole metabolism.30,37 It is probable that lansoprazole exhibits a similar phenomenon to omeprazole, as it is also metabolized by CYP2C19; thus, a lower dose may be required in Chinese subjects than in Caucasians to achieve a similar clinical efficacy.
In addition to diagnosis, lansoprazole may have potential long-term therapeutic benefit in patients with non-cardiac chest pain and gastro-oesophageal reflux, as demonstrated using omeprazole and H2-antagonists.38 Chambers et al. reported that, after treatment with omeprazole, 40 mg at night for 6 weeks, 30% of patients with normal coronary anatomy and 10% of those with coronary disease experienced improvement in chest pain symptoms.39 Furthermore, Achem et al. treated patients with non-cardiac chest pain and gastro-oesophageal reflux with omeprazole, 20 mg b.d., and placebo for 8 weeks; overall symptomatic improvement was reported by 81% of patients on omeprazole, but by only 6% of those on placebo.29 This high improvement rate by omeprazole is consistent with that obtained in previous uncontrolled studies using omeprazole or H2-antagonists,38 as well as in the present study of lansoprazole. Moreover, using decision analysis models, Ofman et al. found that the ‘omeprazole test’ achieved a greater proportion of symptom-free patients at 1 year than invasive diagnostic tests (84% vs. 74%),21 and Borzecki et al. concluded that an empirical therapeutic trial with antisecretory agents for patients with non-cardiac chest pain was more cost-effective than initial investigations for gastrointestinal causes over the short term (weeks), with cost savings persisting for over 1 year.40 However, long-term prospective clinical trials are required to confirm the long-term therapeutic value of this empirical strategy.
There were limitations in the present study. Firstly, patients were divided into two treatment groups in a random, single-blind manner, and this may affect the quality of the results. However, the assessment of symptoms before and after the study was performed by a research nurse (NYHW) who was blind to the treatment assignment, and thus the bias, if any, was negligible. Secondly, the sample size in the present study, although larger than those included in previous studies, may still be insufficient to draw a firm conclusion. Unfortunately, the recruitment of patients with non-cardiac chest pain for such a study is relatively difficult, as it took 16 months to achieve the sample size.
In conclusion, treatment with lansoprazole is a useful test in the diagnosis of endoscopy-negative gastro-oesophageal reflux disease in Chinese patients with non-cardiac chest pain.
We thank nurse specialist M. Chong and endoscopy nurses K. W. Wong, V. S. Y. Tang, D. K. K. Chang and D. M. Y. Lee for assistance. This study was supported by the Peptic Ulcer Research Fund and the Simon K. Y. Lee Gastroenterology Research Fund, The University of Hong Kong.