The proportion of duodenal ulcers not associated with Helicobacter pylori infection or the use of non-steroidal anti-inflammatory drugs (NSAIDs) is increasing.
The proportion of duodenal ulcers not associated with Helicobacter pylori infection or the use of non-steroidal anti-inflammatory drugs (NSAIDs) is increasing.
To identify the clinical and endoscopic characteristics of non-H. pylori, non-NSAID duodenal ulcers.
Clinical and endoscopic data and H. pylori status were prospectively collected from consecutive patients who underwent upper endoscopy from 1997 to 1999. Patients with duodenal ulcers were identified, and those with non-H. pylori, non-NSAID duodenal ulcers were analysed further.
A total of 11 717 upper endoscopies were performed in 8344 patients. Of these, 1153 (14%) had duodenal ulcers. Of 599 patients with active ulcers and known H. pylori status, 104 (17%) had ulcers not associated with H. pylori or the use of NSAIDs, 393 (66%) had ulcers associated with H. pylori alone, 51 (8.5%) had ulcers associated with the use of NSAIDs alone and 51 (8.5%) had ulcers associated with both. Multivariate logistic regression analysis revealed that the presence of concomitant diseases (odds ratio=15.0; 95% confidence interval, 8.64–25.9; P < 0.001) and the absence of epigastric pain/discomfort (odds ratio=0.52; 95% confidence interval, 0.29–0.91; P=0.022) were independent predictors for non-H. pylori, non-NSAID duodenal ulcers.
Non-H. pylori, non-NSAID duodenal ulcers exhibit certain distinct clinical and endoscopic characteristics. The presence of concomitant diseases is an important predictive factor.
Peptic ulcer disease continues to be a socioeconomic burden, and results in considerable mortality due to complications such as bleeding and perforation.1 In China, between 1977 and 1986, the average prevalence of duodenal ulcer in patients undergoing upper endoscopy was 23% in the south and 9.7% in the north, with the average rate of duodenal ulcer almost twice that of gastric ulcer.2 Pooled data obtained later demonstrated similar results.3
Helicobacter pylori infection is present in up to 80–100% of patients with duodenal ulcer and 60–95% of patients with gastric ulcer.4, 5 The eradication of H. pylori infection leads to the cure of H. pylori-associated peptic ulcer disease and prevents ulcer complications.4–7 Following widespread anti-H. pylori therapy in both primary practice and the hospital setting and, in particular, the improvement of socioeconomic and living conditions, the prevalence of peptic ulcer disease, especially duodenal ulcers, is decreasing in many countries.5, 8, 9 While overall peptic ulcer disease appears to be declining, the proportion of H. pylori-negative peptic ulcers may be increasing. For example, in Australia, 15 years ago, over 85% of patients with peptic ulcer disease were infected with H. pylori,10, 11 but the rate has now decreased to approximately 55%.12, 13 The use of non-steroidal anti-inflammatory drugs (NSAIDs) may contribute to H. pylori-negative ulcers.13–17 However, there are an increasing number of patients suffering from peptic ulcers that are not associated with H. pylori infection or NSAID use.13, 15–17 These peptic ulcers may be referred to as non-H. pylori, non-NSAID peptic ulcers or idiopathic ulcers.18, 19 The aim of the present study was to identify the clinical and endoscopic characteristics of non-H. pylori, non-NSAID duodenal ulcer disease, in order to provide clues about its aetiology and pathophysiology, and thus help to establish long-term management strategies.
Clinical and endoscopic information and the H. pylori status of consecutive patients who underwent a routine or emergency upper endoscopy at the Department of Medicine, Queen Mary Hospital from January 1997 to December 1999 were collected prospectively using a newly established computerized endoscopy reporting system. This system was specially designed to collect the required information prospectively. Clinical information was recorded on a standard questionnaire before endoscopy, and entered into the reporting system, together with endoscopic findings, after endoscopy. The clinical information included patient age and gender, major indication for upper endoscopy, duration of symptoms, smoking and drinking history, concomitant diseases and use of aspirin and non-aspirin NSAIDs over the past 3 months. The dosage and duration of NSAID use were also included. Endoscopic information included current endoscopic findings, lesion site, size and number if present. H. pylori status was determined using a 24-h rapid urease test on antral biopsy specimens, histological detection of Helicobacter-like organisms on biopsies taken from gastric antrum and body and/or 13C-urea breath test.20, 21 The rapid urease test as a single test for the diagnosis of H. pylori infection has been validated in our centre with sensitivity, specificity and positive and negative predictive values of 99%, 100%, 100% and 99%, respectively.21
As many patients had more than one endoscopy performed during the period of study, the first endoscopy was used as the index endoscopy. However, when a patient had duodenal ulcer(s), the first endoscopy at which duodenal ulcer was diagnosed was used as the index endoscopy. Thus, each patient had one representative endoscopy. Patients with active ulcers (visible duodenal mucosal defect with diameter of ≥ 3 mm and depth ≥ 0.5 mm) and healed ulcers (ulcer scarring and/or duodenal deformity) in the duodenum were defined as having duodenal ulcers. As the main purpose of this study was to determine the clinical and, in particular, endoscopic characteristics of non-H. pylori, non-NSAID duodenal ulcers, it was the clinical and endoscopic information at the index endoscopy of patients with definite active duodenal ulcers that was analysed throughout the paper (Figure 1). Patients who had used NSAIDs for at least 3 days at any dosage within 3 months prior to the index endoscopy were considered to be NSAID users.13 In addition, for active duodenal ulcer patients who were H. pylori-negative as determined by the rapid urease test alone at the index endoscopy, an appointment was made within 3 months after the index endoscopy for a 13C-urea breath test. H. pylori tests (i.e. the rapid urease test, histology and 13C-urea breath test) at previous and subsequent endoscopies were also referred to. H. pylori infection was defined as positive if a patient had a positive result at any endoscopy or at the follow-up 13C-urea breath test (Figure 1). Patients with clinical suspicion of Zollinger-Ellison syndrome will be investigated further.
Because we intended to identify the clinical and endoscopic characteristics of non-H. pylori, non-NSAID duodenal ulcers, we compared the demographic, clinical and endoscopic variables between non-H. pylori, non-NSAID duodenal ulcers and H. pylori-associated duodenal ulcers. Associations amongst these variables were assessed using the chi-squared test (with Yates’ correction if required), or the Fisher’s exact test. Moreover, a multivariate logistic regression analysis was used to identify independent predictor (or risk) factor(s) for non-H. pylori, non-NSAID duodenal ulcers. Odds ratios (OR) and 95% confidence intervals (CI) were estimated where appropriate. The t-test for independent samples was used to determine age differences between patient groups, and the Mann–Whitney U-test was used to determine the difference in disease duration, ulcer size, depth and number between groups of patients with duodenal ulcers. All P values calculated were two-tailed. The alpha level of significance was set at P < 0.05.
Overall, 11 717 endoscopies on 8344 patients were performed during the 3-year period. Of these patients, 1153 (13.8%) had duodenal ulcers, with 645 patients having definite active duodenal ulcers and the others having healed duodenal ulcers (Figure 1).
The H. pylori status was unknown in 46 (7.1%) of the 645 patients. In the remaining 599 patients, 444 (74.1%) patients were positive and 155 (25.9%) patients were negative for H. pylori infection (Figure 1). Overall, 102 (17.0%) of 599 patients had a history of taking NSAIDs within 3 months prior to the index endoscopy, including six with aspirin and 96 with other NSAIDs. None of the patients had taken cyclooxygenase-2 inhibitors. These consisted of 51 (11.5%) of the 444 patients with H. pylori infection and 51 (32.9%) of the 155 patients without the infection (OR=0.27; 95% CI, 0.17–0.41; χ2=37.3; P < 0.001). Therefore, the proportion of patients with non-H. pylori, non-NSAID duodenal ulcers was 17.4% (104/599) (Figure 2). The clinical and endoscopic characteristics of active chronic duodenal ulcers in relation to H. pylori infection and the use of NSAIDs are listed in Table 1.
For the purpose of this study, the clinical and endoscopic characteristics of non-H. pylori, non-NSAID duodenal ulcers were compared with those of duodenal ulcers associated with H. pylori infection alone, i.e. patients taking NSAIDs were excluded from further analysis. Patients with non-H. pylori, non-NSAID ulcers were significantly older than those with H. pylori-associated ulcers (66.3 ± 17.8 years vs. 53.2 ± 17.2 years; t=6.85; P < 0.001) (Table 1). Moreover, the proportion of patients with non-H. pylori, non-NSAID ulcers increased with age: 3.0% (1/33), 14.8% (9/61), 12.8% (14/109), 12.5% (10/80), 20.9% (18/86), 32.4% (24/74) and 51.9% (28/54) in the age groups < 30, 30–39, 40–49, 50–59, 60–69, 70–79 and 80 years or over, respectively (χ2=52.7, d.f.=6, P < 0.001).
Clinically, epigastric pain/discomfort was less common in patients with non-H. pylori, non-NSAID ulcers than in those with H. pylori-associated ulcers as a major indication for upper endoscopy (28.8% vs. 59.8%; OR=0.27; 95% CI, 0.17–0.44; χ2=31.7; P < 0.001). On the other hand, bleeding was more common as an indication in patients with non-H. pylori, non-NSAID ulcers than in those with H. pylori-associated ulcers (50.0% vs. 25.2%; OR=0.97; 95% CI, 1.90–4.64; χ2=23.9; P < 0.001) (Table 1). The prevalence of anaemia was higher, albeit not statistically significant, in patients with non-H. pylori, non-NSAID ulcers than in patients with H. pylori-associated ulcers (12.5% vs. 8.1%; OR=1.61; 95% CI, 0.81–3.20; χ2=1.90; P=0.168) (Table 1). The duration of symptoms ranged from 3 months to 45 years, but did not differ between patients with non-H. pylori, non-NSAID ulcers or H. pylori-associated ulcers (Table 1).
Overall, 105 (26.4%) patients had one or more concomitant diseases: 68 (65.4%) of the 104 patients with non-H. pylori, non-NSAID ulcers and 37 (9.4%) of the 393 patients with H. pylori-associated ulcers (OR=18.2; 95% CI, 10.7–30.8; χ2=154.6; P < 0.001). Severe diseases, such as malignancy, cerebrovascular accident, chronic renal failure, hepatic cirrhosis, ischaemic heart disease, diabetes mellitus and chronic obstructive airway disease, were more common in patients with non-H. pylori, non-NSAID ulcers than in patients with H. pylori-associated ulcers (Table 1).
Endoscopically, the number of active duodenal ulcers ranged from one to nine, with a median of one. The ulcer size (maximum diameter of an ulcer) ranged from 3 to 62 mm, with a median of 6 mm, and the ulcer depth ranged from 0.5 to 5 mm, with a median of 2.0 mm. Patients with non-H. pylori, non-NSAID ulcers appeared to have deeper (Z=2.47, P=0.014) and a larger number of (Z=1.45, P=0.15) ulcers, and were less likely to have ulcers in the anterior part (37.4% vs. 51.1%; OR=0.57; 95% CI, 0.35–0.89; χ2=6.13, P=0.013), compared with patients with H. pylori-associated ulcers (Table 1).
In order to identify independent factors, a logistic regression analysis was performed in a model including age, gender, current smoking and alcohol drinking, the presence of concomitant diseases, indications for endoscopy, duration of dyspeptic symptoms and the number, size and depth of duodenal ulcers. The presence of concomitant diseases (OR=15.0; 95% CI, 8.64–25.9; χ2=93.2; P < 0.001) and the absence of epigastric pain/discomfort (OR=0.52; 95% CI, 0.30–0.91; χ2=5.27; P=0.022) were independent predictors for non-H. pylori, non-NSAID ulcers (Table 2).
Because the presence of concomitant diseases was identified to be an independent predictor for non- H. pylori, non-NSAID ulcers, its clinical and endoscopic significance was further investigated, and the major findings are listed in Table 3.
The present study demonstrates that the prevalence of duodenal ulcers in patients referred for upper endoscopy in Hong Kong between 1997 and 1999 was 14%. This rate was much lower than that between 1977 and 1986 (32%) in the same population.2 In addition, the male/female ratio decreased from 2.83 between 1977 and 1986 to 1.32 in the present study.2 These figures reflect a real change in the pattern of duodenal ulcer disease in Hong Kong over the past 20 years. Eradication therapy of H. pylori infection is probably one of the reasons for the decreased prevalence of duodenal ulcer disease, although other factors, including the improvement of socioeconomic and living conditions, may also be important. Improvement of hygiene also leads to a lower incidence of H. pylori infection in the population. As this continues, the relative importance of H. pylori-negative duodenal ulcers will inevitably increase. Indeed, our results suggest that the relative proportion of H. pylori-negative duodenal ulcers appears to be increasing. In the late 1980s, less than 10% of patients with duodenal ulcers were H. pylori-negative in Hong Kong,22–24 which is significantly lower than the rate (26%) observed in the present study. Moreover, the present study showed that non-H. pylori, non-NSAID ulcers were present in 17% of patients with duodenal ulcers.
Aetiological factors for the development of non- H. pylori, non-NSAID duodenal ulcers remain to be identified. Smoking has been reported to be a possible cause of peptic ulcer disease, and may be responsible for non-H. pylori, non-NSAID duodenal ulcers.25, 26 However, in the present study, no difference was found in smoking history between patients with non-H. pylori, non-NSAID duodenal ulcers and those with H. pylori-associated ulcers, suggesting a limited role of this factor. This is in agreement with previous studies showing that the relapse of duodenal ulcer after anti-H. pylori therapy is associated with a recurrence of H. pylori infection, but independent of smoking.27, 28 Psychological stress has been postulated to be another risk factor for the development of peptic ulcers.29–30 However, a causal link between psychological stress and idiopathic peptic ulcer disease remains to be established. The findings of the present study provide new insights into this issue. Altogether, 65% of patients with non-H. pylori, non-NSAID duodenal ulcers had concomitant diseases. Patients with underlying diseases are likely to develop psychological stress, thus increasing the risk for duodenal ulcers. There was heterogeneity within non- H. pylori, non-NSAID duodenal ulcers. Patients with concomitant diseases had duodenal ulcers clinically and endoscopically similar to those present in patients with a history of NSAID use, while patients without concomitant diseases had ulcers similar to those present in H. pylori-positive patients. We speculated that drugs other than NSAIDs might be responsible for the development of non-H. pylori, non-NSAID duodenal ulcers in some patients, resulting in ‘NSAID-like’ ulcers, while other factors, perhaps infective and chemical factors, might lead to ‘H. pylori-like’ ulcers. However, the underlying significance of such heterogeneity in the management of duodenal ulcers needs to be further investigated. Diet has traditionally been thought to play a role in the development of duodenal ulcers.2, 31 However, previous studies did not control for H. pylori infection, therefore a direct link between diet and duodenal ulcer remains to be clarified. Zollinger-Ellison syndrome is a rare cause of non-H. pylori, non-NSAID duodenal ulcers.14, 15, 19, 32
One of the limitations of this study was that, in 63 (40%) of the 155 H. pylori-negative patients, the diagnosis of H. pylori was based on the rapid urease test alone. This was mainly due to the poor health status of the patients. Fifty-three patients either died within 3 months after the index upper endoscopy or had severe and/or multiple medical problems and were unable to return for follow-up. However, application of the rapid urease test on antral biopsy specimens as a single method to detect H. pylori infection has been evaluated in our centre. Such a diagnostic approach achieved sensitivity, specificity and positive and negative predictive values of over 99% in selected uninvestigated dyspeptic patients.21 The accuracy of the test may have been decreased in the present study where all eligible patients with unknown medication history were included (see below). However, we followed up almost half (57/120) of the patients with a negative rapid urease test at index endoscopy using a 13C-urea breath test within 3 months. Three of these patients were positive, giving an estimated negative predictive value of 94.7% for the rapid urease test. Therefore, even with a negative predictive value of 90%, only six of the 63 patients who were not followed up would have been re-classified as having the infection, and the proportion of H. pylori-negative duodenal ulcers would have remained as high as 25%. This rate is clinically significant in Hong Kong where the incidence of duodenal ulcer is considerably high.
The second limitation of this study was the unknown previous medication history. Triple therapy leads to the cure of H. pylori-associated ulcers in over 90% of cases, and the relapse of duodenal ulcer after triple therapy is associated with a recurrence of the infection.4–6, 33, 34 Thus, it is unlikely that patients with current active duodenal ulcers had received triple therapy. Even if a patient had received triple therapy previously and had a relapsed duodenal ulcer, this patient would have persistent (or recurrent) and detectable H. pylori infection in the stomach. Studies have shown that proton pump inhibitors and antibiotics decrease the density of organisms in the gastric antrum, leading to a negative result for a rapid urease test.35, 36 However, recent medication with proton pump inhibitors normally results in healing of the ulcers, but the previous use of such agents may have little effect on the current H. pylori status. The use of antibiotics without aiming to eradicate H. pylori infection is rare in patients with dyspepsia, and thus the use of antibiotics in a few cases should not have affected the overall results of this study.
The final concern is that some patients may not report or may be unaware of NSAID use. In the present study, many patients with non-H. pylori, non-NSAID ulcers had concomitant diseases that may require NSAIDs; thus, it is likely that some patients who used NSAIDs were incorrectly classified as non-NSAID users. Indeed, Lanas et al. demonstrated that between 13% and 22% of patients with gastrointestinal bleeding and perforation who claimed not to have used aspirin had objective evidence of current aspirin intake.37, 38 If this is the case in the present study, up to 23 of the 104 patients with non-H. pylori, non-NSAID ulcers would need to be re-classified as NSAID users. However, this is unlikely to change the differences in the clinical and endoscopic characteristics between non-H. pylori, non-NSAID ulcers and H. pylori-associated ulcers, and between non-H. pylori, non-NSAID ulcers with and without concomitant diseases.
In conclusion, non-H. pylori, non-NSAID duodenal ulcers exhibit distinct clinical and endoscopic characteristics. The presence of concomitant diseases is an important predictive factor for most non-H. pylori, non-NSAID duodenal ulcers.
We thank Professor C. L. Lai and Drs G. Lau, M. F. Yuen, C. K. Hui, R. Go and O. O. Chan of the Department of Medicine, University of Hong Kong, Queen Mary Hospital, Hong Kong, for performing some of the endoscopies; nurse specialist M. Chong and endoscopy nurses V. S. Y. Tang, D. K. K. Chang and D. M. Y. Lee for assistance; and Mr James Ma of Queen Mary Hospital, Hong Kong, for the computer endoscopy reporting system. This study was supported by the Peptic Ulcer Research Fund and the Simon K. Y. Lee Gastroenterology Research Fund, University of Hong Kong, Hong Kong.
*Contributed equally to this work.