High Incidence of Mortality and Recurrent Bleeding in Patients With Helicobacter pylori Negative Idiopathic Bleeding Ulcers

GASTROENTEROLOGY 2009;137:525 531 High Incidence of Mortality and Recurrent Bleeding in Patients With Helicobacter pylori Negative Idiopathic Bleeding Ulcers GRACE LAI HUNG WONG, VINCENT WAI SUN WONG, YAWEN CHAN, JESSICA YUET LING CHING, KIM AU, ARIC JOSUN HUI, LARRY HIN LAI, DOROTHY KAI LAI CHOW, DANNY KA FAI SIU, YAN NI LUI, JUSTIN CHE YUEN WU, KA FAI TO, LAWRENCE CHEUNG TSUI HUNG, HENRY LIK YUEN CHAN, JOSEPH JAO YIU SUNG, and FRANCIS KA LEUNG CHAN Institute of Digestive Disease, The Chinese University of Hong Kong; Hong Kong SAR, China BACKGROUND & AIMS: The long-term prognosis of peptic ulcers associated with neither Helicobacter pylori nor nonsteroidal anti-inflammatory drugs (NSAIDs) is unknown. METHODS: This 7-year prospective cohort study recruited patients with bleeding ulcers from January to December 2000. H pylori negative idiopathic bleeding ulcers were defined as having tested negative for H pylori, having no exposure to aspirin or analgesics within 4 weeks before endoscopy, and having no other identifiable causative factors. After ulcers healed, patients were divided into 2 groups: patients with prior H pylori negative idiopathic bleeding ulcers (H pylori negative idiopathic ulcer cohort; n 120) and those with H pylori positive, NSAID-negative bleeding ulcers who received eradication therapy (H pylori ulcer cohort; n 213). Both groups were followed for 7 years without gastroprotective therapy. The primary endpoints were recurrent ulcer bleeding and mortality. RESULTS: The 7-year cumulative incidence of recurrent ulcer bleeding was 42.3% (95% CI, 36.5% 48.1%) in the H pylori negative idiopathic ulcer cohort and 11.2% (95% CI, 8.8% 13.6%) in the H pylori ulcer cohort (a difference of 31.1%; 95% CI, 27.7% 34.5%; P.0001). Significantly more patients died in the H pylori negative idiopathic ulcer cohort (87.6%; 95% CI, 83.0% 92.2%) than in the H pylori ulcer cohort (37.3%; 95% CI, 34.0% 40.6%) with a difference of 50.3% (95% CI, 49.0% 51.6%; P.0001). CONCLUSIONS: Patients with history of H pylori negative idiopathic bleeding ulcers have a high risk of recurrent ulcer bleeding and mortality. With the declining prevalence of Helicobacter pylori infection in many countries, the proportion of patients with idiopathic peptic ulcers has been increasing. Studies in North America have shown that between 11% and 44% of peptic ulcers are not associated with H pylori infection or the use of nonsteroidal anti-inflammatory drugs (NSAIDs). 1 4 A meta-analysis of 7 US trials found that 20% of patients with H pylori associated ulcers had recurrent ulcers within 6 months, despite successful cure of H pylori infection and no reported use of NSAIDs. 5 In a pooled analysis of 6 clinical trials with a total of 2900 patients, 27% of duodenal ulcers were not associated with NSAID use or H pylori infection. 6 The emerging problem of H pylori negative idiopathic peptic ulcers is not only limited to Western countries. Previously, H pylori negative idiopathic peptic ulcers accounted for 5% of peptic ulcers in Asia. 7,8 A recent Korean study reported that the proportion of peptic ulcers not associated with H pylori infection or NSAID use was 20%. 9 The hypothesis that the incidence of H pylori negative idiopathic peptic ulcers is increasing has been controversial. It has been argued that the increase in diagnosis of H pylori negative idiopathic ulcers merely reflects a falling incidence of H pylori ulcers. Graham 10 postulated that, if the absolute number of H pylori negative idiopathic ulcers remained constant, whereas the incidence of H pylori ulcers decreased, a greater proportion of H pylori negative idiopathic ulcers would be diagnosed. Contrary to Graham s hypothesis, we previously have demonstrated that the true incidence of H pylori negative idiopathic bleeding ulcers had increased by 4-fold from 1997 to 2000. 11 It is uncertain whether patients with a history of H pylori negative idiopathic bleeding ulcers require longterm acid suppressive therapy because the natural history of this condition is unknown. In a 1-year interim analysis of a prospective cohort study, we previously reported that 13% of patients with a history of H pylori negative idiopathic bleeding ulcers had recurrent bleeding or symptomatic ulcers compared with 2.5% of patients with prior H pylori bleeding ulcers. 11 In this final analysis, we reported the 7-year outcome of these 2 cohorts of patients with regard to recurrent ulcer bleeding and mortality. Abbreviations used in this paper: ASA, American Society of Anesthesiology; H pylori, Helicobacter pylori; NSAID, nonsteroidal antiinflammatory drug; PPI, proton pump inhibitor. 2009 by the AGA Institute 0016-5085/09/$36.00 doi:10.1053/j.gastro.2009.05.006

526 WONG ET AL GASTROENTEROLOGY Vol. 137, No. 2 Materials and Methods Study Population This was a 7-year, single-center, prospective cohort study conducted at the Prince of Wales Hospital, which serves a local population of 1.2 million people in Hong Kong. We screened consecutive patients with a clinical diagnosis of upper gastrointestinal bleeding from January 2000 to December 2000. All patients provided informed written consent. Patients were excluded if they were in the intensive care unit, had terminal diseases, had malignancy, or had previous gastrectomy. The study was approved by the Joint Chinese University of Hong Kong New Territories East Cluster Clinical Research Ethics Committee. All patients who were diagnosed to have upper gastrointestinal bleeding underwent endoscopy within 24 hours of onset of bleeding. An ulcer was defined as a mucosal break with an apparent depth and a diameter of 5 mm. Biopsy specimens were taken from the antrum (2 biopsies), corpus (2 biopsies), and ulcer edge (4 quadrant biopsies) to exclude H pylori infection and other pathologic conditions such as neoplasms, Crohn s disease, cytomegalovirus, and herpes infection. H pylori infection was diagnosed with the use of a biopsy urease test (CLO test; Delta West, Bentley, Western Australia, Australia) and histology with hematoxylin and eosin stain and Giemsa stain if necessary. H pylori infection was considered to be present if either the biopsy urease test or histology was positive for the bacterium. We repeated these diagnostic tests for H pylori in the absence of acid suppressive therapy on follow-up endoscopy. Patients underwent gastric acid secretion tests if suspected to have Zollinger Ellison syndrome as previously described. 11 Drug History All patients were scrutinized for any recent use of aspirin, NSAIDs, or traditional Chinese medicine. Patients who denied taking any analgesics were further screened for musculoskeletal pain or other pain disorders, including headache, dysmenorrhea, and pharyngitis. For patients with a painful condition who denied taking analgesics, a collateral drug history was obtained from family members and primary care physicians. We retrieved over-the-counter drugs and prescriptions from patients, family members, and primary care physicians. A territory-wide electronic database was used to identify NSAIDs, aspirin, acid suppressants, antibiotics, and other drugs such as potassium supplements, calcium channel blockers, and antidepressants before the onset of gastrointestinal bleeding. Exposure to NSAIDs or aspirin was defined as 1 dose of the drug taken within the 4 weeks before hospitalization. Patients who had used traditional Chinese herbal medicine were also considered as possible NSAID users because previous reports indicated that these drugs frequently contained NSAIDs. 12,13 Definition of Study Cohorts Three prerequisites must be met before the diagnosis of H pylori negative idiopathic ulcers was made. First, there must be no exposure to aspirin, NSAIDs, or drugs of unknown nature, including traditional Chinese medicine, within the 4 weeks before hospitalization. Second, biopsies taken during endoscopy must be negative for both the urease test and histology for H pylori in the absence of acid suppressive therapy. Third, no other causes of ulceration must be found. H pylori ulcers were defined as having a positive test for H pylori in any of the biopsy specimens and also no exposure to NSAIDs, aspirin, over-the-counter analgesics, or traditional Chinese medicine within 4 weeks before the bleeding episode. Patients who used one of these drugs within 4 weeks before ulcer bleeding were classified to have NSAID ulcers regardless of their H pylori status. The finding of intestinal metaplasia or gastric atrophy or both without the bacteria was regarded as past H pylori infection. Grading of overall health and comorbidity was performed according to the American Society of Anesthesiology (ASA) classification as previously described. 11 Follow-Up Patients with H pylori negative idiopathic ulcers received an 8-week course of proton pump inhibitor (PPI), and patients with H pylori ulcers received a 1-week PPI-based triple therapy. In the H pylori negative idiopathic ulcer group, follow-up endoscopy was scheduled for patients when they had stopped PPI therapy for 1 week. We checked compliance to medication before follow-up endoscopy. Biopsies were taken from both the antrum and the corpus for histologic evidence of H pylori. In the H pylori ulcer group, follow-up endoscopy was performed on all patients with gastric ulcers and those with duodenal ulcers who had poor drug compliance or who remained symptomatic 8 weeks after completion of eradication therapy to retest for H pylori status. After ulcer healing, the H pylori negative idiopathic ulcer cohort and the H pylori ulcer cohort were followed at regular intervals without prophylactic gastroprotective therapy for 7 years. Drugs, including PPIs, histamine-2 receptor antagonists, misoprostol, sucralfate, bismuth, NSAIDs, and aspirin, were prohibited during the study period. Regular use of prohibited drugs was defined by 50% exposure during the follow-up period. During each follow-up visit, patients were assessed for dyspeptic symptoms, musculoskeletal pain, symptoms of recurrent bleeding, drug violation, and hemoglobin level. Antacids and simple analgesics, including paracetamol, dextropropoxyphene, or tramadol, were prescribed for dyspepsia and musculoskeletal pain, respectively. Endoscopy was repeated if the patients experienced recurrent hematemesis, melena, a drop of hemoglobin level 2 g/dl, or dyspeptic symptoms not relieved by antacids. Biopsy urease test and

August 2009 H PYLORI NEGATIVE IDIOPATHIC BLEEDING ULCERS 527 histologic specimens from antrum, corpus, and ulcer margin were repeated. Endpoints The primary outcomes were recurrent ulcer bleeding and overall mortality. Recurrent ulcer bleeding was defined as hematemesis or melena with ulcers confirmed by endoscopy or a decrease in the hemoglobin level 2 g/dl in the presence of endoscopically proven ulcers. The secondary outcome was symptomatic ulcers. Statistical Analysis The patients baseline characteristics were presented as descriptive data. We used Student s t test to compare means, Mann Whitney U test to compare medians, and Pearson chi-square test to compare categorical data. The Kaplan Meier method was used to estimate the likelihood of reaching the primary outcomes of recurrent ulcer bleeding and mortality within 7 years. The log-rank test was used to compare time-to-event curves between the H pylori negative idiopathic ulcer cohort and the H pylori ulcer cohort. A Cox-proportional hazards model with backward stepwise regression was used to identify possible covariates as significant predictors of recurrent ulcer bleeding and mortality which included age ( 70 years), ASA grade ( 3), location of ulcer (duodenal versus nonduodenal), the presence or absence of past H pylori infection, concomitant use of NSAIDs, and patients who bled while in the hospital. All statistical tests were 2-sided. Statistical significance was taken as P.05. Statistical analysis was performed by Statistical Package for Social Science (SPSS version 11.5; SPSS Inc, Chicago, IL). Results Baseline Characteristics From January 2000 to December 2000, 675 patients had bleeding ulcers confirmed by endoscopy. Thirty-seven patients (5.5%) were excluded: 14 patients had previous gastrectomy and 23 patients did not have biopsies for H pylori. One hundred twenty patients (18.8%) had H pylori negative idiopathic bleeding ulcers. Three hundred five patients (47.8%) used low-dose aspirin, NSAIDs, overthe-counter analgesics, or traditional Chinese medicine. Three hundred thirty-three patients (52.2%) were nonusers. Among the nonusers, 213 patients (33.4%) were positive for H pylori. Compared with patients with H pylori ulcers, patients with H pylori negative idiopathic ulcers were older, had a higher proportion of gastric ulcers, had higher ASA grades, and had a higher proportion of patients who bled while in hospital (Table 1). Among patients with H pylori negative idiopathic bleeding ulcers, 3 received calcium channel blockers, 1 received potassium supplement, and 1 took a tricyclic antidepressant. No patient used acid suppressants or Table 1. Baseline Clinical Characteristics H pylori negative idiopathic ulcer cohort (n 120) H pylori ulcer cohort (n 213) P value Age at baseline, 73 (22 95) 61 (15 97).001 median (range), y Male, n (%) 72 (60.0) 141 (66.2).26 Hemoglobin, median 9.0 (4.7 15.9) 9.3 (4.0 16.6).39 (range), g/dl Ulcer location, n (%).006 Gastric ulcer 63 (52.5) 79 (37.1) Duodenal or 57 (47.5) 134 (62.9) gastroduodenal ulcer Ulcer diameter, median 8 (5 50) 9 (5 30).47 (range), mm ASA grade 3, n (%) a 61 (50.8) 49 (23.0).001 Patients who bled while in hospital, n(%) 28 (23.3) 19 (8.9).001 a ASA classification: grade 1, normal healthy patients; grade 2, mild systemic illness; grade 3, severe but incapacitating systemic illness; and grade 4, life-threatening illness. cocaine before the onset of bleeding. Thirty-four patients (28.3%) had histologic evidence of past H pylori infection as previously reported. 11 One specimen (0.8%) showed cytomegalovirus inclusion bodies, but the patient did not have viral antigenemia. None had herpes infection or evidence of Crohn s disease. Gastric acid secretion tests were not performed because none of the patients fulfilled the screening criteria for Zollinger Ellison syndrome as previously described. 11 Follow-Up After ulcer healing, 120 patients were enrolled in the H pylori negative idiopathic ulcer cohort and 213 in the H pylori ulcer cohort. The median follow-up was 30.1 months (range, 1 89.2 months) in the H pylori negative idiopathic ulcer cohort and 79.4 months (range, 1 89.7 months) in the H pylori ulcer cohort (P.001). The difference in follow-up duration between the 2 cohorts was due to the significantly higher rate of mortality in the H pylori negative idiopathic ulcer cohort (Table 2). Ten patients (8.3%) in the H pylori negative idiopathic ulcer cohort who defaulted in the first year were later identified and continued to be followed until the end of the study. 11 No patient in the H pylori ulcer cohort was lost to follow-up. Histamine-2 receptor antagonists were found to be regularly used in 3 patients (2.5%) from the H pylori negative idiopathic ulcer cohort and 1 patient (0.5%) from the H pylori ulcer cohort (P.10). Regular use of other prohibited drugs, including PPIs, misoprostol, sucralfate, bismuth, NSAIDs, and aspirin, was not found in both cohorts. Recurrent Ulcer Bleeding A total of 26 patients in the H pylori negative idiopathic ulcer cohort underwent endoscopy for sus-

528 WONG ET AL GASTROENTEROLOGY Vol. 137, No. 2 Table 2. Causes of Death H pylori negative idiopathic ulcers (n 120) H pylori ulcers (n 213) P value Patients deceased, n (%) 68 (56.7) 58 (27.2).001 Malignancy, n (%) 24 (20.0) 18 (8.5).002 Gastric cancer 0 (0) 6 (2.8).06 Esophageal cancer 1 (0.8) 0 (0).36 Colorectal cancer 1 (0.8) 1 (0.5) 1.00 Pancreatic cancer 1 (0.8) 0 (0).36 Cholangiocarcinoma 1 (0.8) 1 (0.5) 1.00 Gallbladder cancer 1 (0.8) 0 (0).36 Hepatocellular carcinoma 6 (5.0) 4 (1.9).18 Lung cancer 3 (2.5) 3 (1.4).67 Lymphoma/leukemia 4 (3.3) 0 (0).02 Renal cell carcinoma 1 (0.8) 1 (0.5) 1.00 Prostate cancer 1 (0.8) 0 (0).36 Ovarian cancer 1 (0.8) 0 (0).36 Melanoma 1 (0.8) 0 (0).36 Nasopharyngeal 0 (0) 1 (0.5) 1.00 carcinoma Adenocarcinoma of 2 (1.7) 1 (0.5).30 unknown primary Pneumonia, n (%) 10 (8.3) 11 (5.2).25 Sepsis (other than 10 (8.3) 4 (1.9).005 pneumonia), n (%) Cardiovascular event, n (%) a 6 (5.0) 10 (4.7).90 Renal failure, n (%) 10 (8.3) 4 (1.9).005 Gastrointestinal bleeding, 2 (1.7) 4 (1.9).89 n(%) Other, n (%) b 6 (5.0) 7 (3.3).44 a Cardiovascular event was defined as acute coronary syndrome or acute pulmonary edema related to congestive heart failure. b Other causes of death included liver failure, chronic obstructive pulmonary disease, fall from height, and uncertain causes. Figure 1. Kaplan Meier estimates of 7-year cumulative probabilities of recurrent ulcer bleeding in the H pylori negative idiopathic ulcer cohort and the H pylori ulcer cohort (absolute difference, 31.1%; 95% CI, 27.7% 34.5%; P.0001 by log-rank test). pected rebleeding; 25 were confirmed to have bleeding from recurrent ulcers. Nine patients (36.0%) with recurrent ulcer bleeding used concomitant NSAIDs or lowdose aspirin; none of the recurrent bleeding ulcers were positive for H pylori. Nineteen patients in the H pylori ulcer cohort underwent endoscopy for suspected rebleeding, 18 were confirmed to have recurrent ulcer bleeding. Among them, 8 patients (44.4%) used concomitant NSAIDs or low-dose aspirin, 4 (22.2%) had relapse of H pylori infection, 1 (5.6%) had both concomitant NSAID use and relapse of H pylori infection, and, finally, 5 (27.8%) had no cause identified for their recurrent bleeding ulcers. The 7-year cumulative incidence of recurrent ulcer bleeding was 42.3% (95% CI, 36.5% 48.1%) in the H pylori negative idiopathic ulcer cohort and 11.2% (95% CI, 8.8% 13.6%) in the H pylori ulcer cohort (difference, 31.1%; 95% CI, 27.7% 34.5%, P.0001 by log-rank test; Figure 1). Patients who had evidence of past H pylori infection in the H pylori negative idiopathic ulcer cohort did not have increased risk of recurrent ulcer bleeding (P.19 by log-rank test). After excluding recurrent ulcer bleeding associated with concomitant NSAIDs or relapse of H pylori infection, the 7-year probability of recurrent ulcer bleeding was 25.2% (95% CI, 20.1% 30.3%) in the H pylori negative idiopathic ulcer cohort and 3.0% (95% CI, 1.7% 4.3%) in the H pylori ulcer cohort (difference, 22.2%; 95% CI, 18.4% 26.0%; P.0001 by log-rank test; Figure 2). Three patients in the H pylori ulcer cohort and none in H pylori negative idiopathic ulcer cohort had nonbleeding symptomatic ulcers. In the Cox-proportional hazards model, concomitant NSAID use was an independent factor predicting recur- Figure 2. Kaplan Meier estimates of 7-year cumulative probabilities of recurrent ulcer bleeding after excluding recurrent ulcer bleeding associated with concomitant NSAIDs or relapse of H pylori infection in the H pylori negative idiopathic ulcer cohort and the H pylori ulcer cohort (absolute difference, 22.2%; 95% CI, 18.4% 26.0%; P.0001 by log-rank test).

August 2009 H PYLORI NEGATIVE IDIOPATHIC BLEEDING ULCERS 529 Table 3. Analysis of Possible Factors Predicting Recurrent Ulcer Bleeding With the Cox-Proportional Hazard Model in the H pylori Negative Idiopathic Ulcer Cohort (n 120) Hazard ratio 95% CI P value Age 70 y 1.4 0.5 3.6.48 ASA grade 3 2.6 1.0 6.7.05 Location of ulcer (duodenal vs 1.9 0.6 5.4.36 nonduodenal) Past H pylori infection 1.6 0.7 3.8.23 NSAID use 2.7 1.5 5.1.002 Patients who bled while in hospital 0.7 0.2 2.7.61 rent ulcer bleeding in the H pylori negative idiopathic ulcer cohort (Table 3). After adjusting for possible confounding covariates, including age, location of ulcer, ASA grade, and in-patient bleeder in the Cox-proportional hazards model, H pylori negative idiopathic ulcer remained an independent risk factor associated with recurrent ulcer bleeding (hazard ratio, 4.0; 95% CI, 2.1 7.5; P.001). Mortality The overall rate of mortality was significantly higher in the H pylori negative idiopathic ulcer cohort (87.6%; 95% CI, 83.0% 92.2%) than in the H pylori ulcer cohort (37.3%; 95% CI, 34.0% 40.6%; absolute difference 50.3%; 95% CI, 49.0% 51.6%; P.0001; Figure 3). In the Cox-proportional hazards model, ASA grade 3 was an independent risk factor predicting rate of mortality in the H pylori negative idiopathic ulcer cohort (Table 4). H pylori negative idiopathic ulcer remained an independent risk factor associated with mortality (hazard ratio, 1.7; 95% CI, 1.2 2.4; P.004) after adjusting for age, location Figure 3. Kaplan Meier estimates of 7-year cumulative probabilities of all-cause mortality in the H pylori negative idiopathic ulcer cohort and the H pylori ulcer cohort (absolute difference, 50.3%; 95% CI, 49.0% 51.6%; P.0001 by log-rank test). Table 4. Analysis of Possible Factors Predicting Overall Mortality With the Cox-Proportional Hazard Model in the H pylori Negative Idiopathic Ulcers (n 120) Hazard ratio 95% CI P value Age 70 y 1.4 0.8 2.4.21 ASA grade 3 3.3 1.9 5.7.001 Past H pylori infection 0.6 0.3 1.1.09 NSAID use 0.8 0.3 1.8.57 Patients who bled while in hospital 1.3 0.7 2.2.43 of ulcer, ASA grade, and in-patient bleeder in the Coxproportional hazards model. More patients in the H pylori negative idiopathic ulcer cohort died of malignancy, sepsis (other than pneumonia), and renal failure than did patients in the H pylori ulcer cohort (Table 2). Only a few patients died within 30 days of the index episode of bleeding: 4 in the H pylori negative idiopathic ulcer cohort and 3 in the H pylori ulcer cohort. Six patients in the H pylori ulcer group, all of whom had failed eradication of H pylori, died of gastric cancer; but none of the patients with H pylori negative idiopathic ulcers had gastric cancer. They initially presented as H pylori related bleeding gastric ulcers that responded to PPI therapy on follow-up endoscopy. After enrollment, they developed recurrent anemia, epigastric pain, or weight loss. Repeated endoscopy showed gastric cancers that were confirmed by histology. The rate of mortality because of uncontrolled gastrointestinal bleeding in both groups was low: 2 (1.7%) with H pylori negative idiopathic ulcers and 4 (1.9%) with H pylori ulcers. Discussion We set out to study the long-term clinical outcome of patients with H pylori negative idiopathic bleeding ulcers and to compare them with those who had H pylori bleeding ulcers. In this 7-year prospective cohort study, we found that patients with a history of H pylori negative idiopathic bleeding ulcers had an almost 4-fold increased risk of recurrent ulcer bleeding (42.3%) than did patients with prior H pylori bleeding ulcers (11.2%). Our findings indicated that patients with H pylori negative idiopathic bleeding ulcers have a high tendency of recurrent complications. Little data are available on the long-term outcome of H pylori negative idiopathic ulcers in the literature. A Danish study of 32 patients with uncomplicated H pylori negative duodenal ulcers also reported a high recurrence rate of 35% in 2 years, 14 which is consistent with our current observation. Can the high incidence of recurrent bleeding in the H pylori negative idiopathic ulcer cohort be attributable to ethnic difference or advanced comorbidity? Ethnic heterogeneity probably exists because our cohort consisted of predominantly gastric ulcers, whereas case series from

530 WONG ET AL GASTROENTEROLOGY Vol. 137, No. 2 Western countries were largely duodenal ulcers. 6,14,15 The high proportion of patients with advanced comorbidities in our cohort might also contribute to more gastric ulcers than duodenal ulcers. However, neither the location of ulcer nor the advanced comorbidity predicted recurrent ulcer bleeding. We found that H pylori negative idiopathic bleeding ulcer remained an independent risk factor for recurrent ulcer bleeding after adjusting for possible confounders. Patients with H pylori negative idiopathic bleeding ulcers also had a higher overall rate of mortality. Although advanced comorbidity significantly contributed to the high rate of mortality, note that H pylori negative idiopathic bleeding ulcer per se also predicted mortality. The role of acid in the pathogenesis of H pylori negative idiopathic ulcers remains controversial. The study by McColl et al 15 identified 6 patients with H pylori negative idiopathic duodenal ulcers who have acid hypersecretion and rapid gastric emptying; 4 of them did not respond to histamine 2 receptor antagonists. However, a subsequent Japanese study found that only one-third of patients with idiopathic ulcers had acid hypersecretion. 7 In the current study, more than half of the patients in the H pylori negative-idiopathic ulcer cohort had gastric rather than duodenal ulcers and approximately 30% had chronic gastritis. Although we did not measure acid secretion, our finding suggested that acid hypersecretion was unlikely to be a major factor in the pathogenesis of H pylori negative idiopathic ulcers, at least among our patients. Thus, we did not consider long-term prophylaxis with a PPI to be mandatory in our patients with H pylori negative idiopathic ulcers when this study was designed. Future studies are needed to determine the efficacy of PPIs or other gastroprotective agents such as misoprostol in the prevention of ulcer recurrence. H pylori negative idiopathic ulcers are often overdiagnosed because of false-negative tests for H pylori.in our H pylori negative idiopathic ulcer cohort, histologic evidence of past H pylori infection was found in approximately 30% of patients. We used both urease test and histology taken from multiple sites on repeated occasions to minimize false-negative tests for H pylori. The subgroup of patients with histologic evidence of past H pylori infection did not have increased risk of recurrent ulcer bleeding. In regions where H pylori infection is prevalent, a substantial proportion of the population showing features of past H pylori infection is not unexpected. Some may develop hypochlorhydria as a consequence of longstanding mucosal inflammation. Interestingly, a Japanese study found that only one-third of patients with H pylori negative idiopathic ulcers had acid hypersecretion. 7 We did not use molecular techniques to detect occult H pylori infection because its clinical significance is doubtful. Recently, one study reported that isolated duodenal colonization by H pylori was detected in a small series of patients with idiopathic duodenal ulcers. 16 However, some of them still had ulcer recurrence despite successful eradication of H pylori. The significance of duodenal colonization by H pylori in these patients remains uncertain. Another common reason for overdiagnosing H pylori negative idiopathic ulcer is the underreporting of NSAID use. We did not measure serum salicylate level and in vitro platelet function test. It might be possible that some patients with surreptitious use of NSAIDs or aspirin were misclassified into the cohort of H pylori negative idiopathic ulcers. To overcome this potential pitfall, we adopted a systematic and meticulous approach to identify drug use as previously described. 11 Patients who had prior exposure to any unknown drugs, including Chinese herbs, were also excluded because we could not exclude the possibility of NSAIDs in these drugs. Our strategy should have minimized the chance of missing surreptitious use of aspirin or NSAIDs. Although ulcer recurrence was uncommon in the H pylori ulcer cohort, we found that approximately 28% of recurrent ulcer bleeding had no identifiable cause. This finding echoed the meta-analysis of 7 US trials that 20% of patients with a history of H pylori associated ulcers had recurrent ulcers despite successful eradication of H pylori and no concomitant use of NSAIDs. 5 It is possible that H pylori might be an innocent bystander in some patients with peptic ulcers. In conclusion, patients with a history of H pylori negative idiopathic bleeding ulcers had a considerable risk of recurrent ulcer bleeding and mortality. Long-term prophylaxis with a gastroprotective agent is probably needed. References 1. Kurata JH, Nogawa AN. Meta-analysis of risk factors for peptic ulcer (nonsteroidal anti-inflammatory drugs, Helicobacter pylori, and smoking). J Clin Gastroenterol 1997;24:2 17. 2. Jyotheeswaran S, Shah AN, Jin HO, Potter GD, Ona FV, Chey WY. Prevalence of Helicobacter pylori in peptic ulcer patients in greater Rochester, NY: is empirical triple therapy justified? Am J Gastroenterol 1998;93:574 578. 3. Sprung DJ, Apter MN. What is the role of Helicobacter pylori in peptic ulcer and gastric cancer outside the big cities? J Clin Gastroenterol 1998;26:60 63. 4. Schubert M, McGuire VA, Dewitt JM, Taylor CA. Prospective evaluation of the prevalence of H pylori in duodenal and gastric ulcer (is its role overstated?). Gastroenterology 1999;116: A305. 5. Laine L, Hopkins RJ, Girardi LS. Has the impact of Helicobacter pylori therapy on ulcer recurrence in the United States been overstated? A meta-analysis of rigorously designed trials. Am J Gastroenterol 1998;93:1409 1415. 6. Ciociola AA, McSorley DJ, Turner K, Sykes D, Palmer JB. Helicobacter pylori infection rates in duodenal ulcer patients in the United States may be lower than previously estimated. Am J Gastroenterol 1999;94:1834 1840. 7. Nishikawa K, Sugiyama T, Kato M, et al. Non-Helicobacter pylori and non-nsaid peptic ulcer disease in the Japanese population. Eur J Gastroenterol Hepatol 2000;12:635 640.

August 2009 H PYLORI NEGATIVE IDIOPATHIC BLEEDING ULCERS 531 8. Chan HL, Wu JC, Chan FK, et al. Is non-helicobacter pylori, non-nsaid peptic ulcer a common cause of upper GI bleeding? A prospective study of 977 patients. Gastrointest Endosc 2001; 53:438 442. 9. Jang HJ, Choi MH, Shin WG, et al. Has peptic ulcer disease changed during the past ten years in Korea? A prospective multicenter study. Dig Dis Sci 2008;53:1527 1531. 10. Graham DY. Large US clinical trials report a high proportion of H pylori-negative duodenal ulcers at study entry as well as a high recurrence rate after cure of the infection: have we all been wrong? Gastroenterology 1998;114:A17. 11. Hung LC, Ching JY, Sung JJ, et al. Long-term outcome of Helicobacter pylori-negative idiopathic bleeding ulcers: a prospective cohort study. Gastroenterology 2005;128:1845 1850. 12. Forster PJG, Calverley M, Hubball S, McConkey B. Chuei-Fong- Tou-Geu-Wan in rheumatoid arthritis. BMJ 1979;2:308. 13. Gertner E, Marshall PS, Filandrinos D, Potek AS, Smith TM. Complications resulting from the use of Chinese herbal medications containing undeclared prescription drugs. Arthritis Rheum 1995;38:614 617. 14. Bytzer P, Teglbjaerg PS. Helicobacter pylori-negative duodenal ulcers (prevalence, clinical characteristics, and prognosis results from a randomised trial with 2-year follow-up). Am J Gastroenerol 2001;96:1409 1416. 15. McColl KE, el-nujumi AM, Chittajallu RS, et al. A study of the pathogenesis of Helicobacter pylori negative chronic duodenal ulceration. Gut. 1993;34:762 768. 16. Pietroiusti A, Forlini A, Magrini A, Galante A, Bergamaschi A. Isolated H pylori duodenal colonization and idiopathic duodenal ulcers. Am J Gastroenterol 2008;103:55 61. Received March 13, 2009. Accepted May 5, 2009. Reprint requests Address requests for reprints to: Francis K.L. Chan, MD, Department of Medicine and Therapeutics, 9/F Prince of Wales Hospital, 30-32 Ngan Shing Street, Shatin, Hong Kong. e-mail: fklchan@cuhk.edu.hk; fax: (852) 2637-3852. Conflicts of interest The authors disclose the following: Dr Henry L.Y. Chan is a member of the advisory boards of Novartis and Schering-Plough. Dr Joseph J.Y. Sung received consulting fees from the National Health Research Institutes of Taipei, The Hong Kong Police Force, Lippincott Williams & Wilkins, and the Hong Kong College of Physicians and paid lecture fees by AstraZeneca Hong Kong Limited, GSK Pharmaceuticals International, and the American Society for Gastrointestinal Endoscopy. Dr Francis K.L. Chan received an independent research grant and a consulting fee from Pfizer and paid lecture fees by Pfizer, Takeda, and AstraZeneca. All authors participated in patient recruitment and follow up. F.K.C. and J.Y.C. contributed to the design of the study; G.L.W., J.Y.C., K.A., Y.C., and L.C.H. were responsible for data collection, entry, and analysis; K.T. was responsible for the histologic assessment of the biopsy samples; G.L.W., V.W.W., A.J.H., and F.K.C. participated in the preparation of the report.