Longitudinal Study of Influence of Helicobacter pylori on Current Risk of Duodenal Ulcer Relapse M. RYE CLAUSEN, M. B. FRANZMANN, C. HOLST, T. I . A. SQ)RENSEN,P. CHRISTOFFERSEN, P. MATZEN, E. KRAG & THE HVIDOVRE ULCER PROJECT GROUP* Depts. of Medical and Surgical Gastroenterology and Dept. of Pathology, Hvidovre University Hospital, and Biostatistical Research Unit, University of Copenhagen, Copenhagen, Denmark
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
Rye Clausen M, Franzmann MB, Holst C, Sarensen TIA, Christoffersen P, Matzen P, Krag E , The Hvidovre Ulcer Project Group. Longitudinal study of influence of Helicobacter pylori on current risk of duodenal ulcer relapse. Scand J Gastroenterol 1992;27:421-426. Seventy-four patients with duodenal ulcer were followed up longitudinally for 2 years after initial ulcer healing. Endoscopy including biopsy of the antral mucosa was performed every 3rd month and whenever clinical symptoms of relapse occurred. The presence of Helicobacter pylori in the biopsy specimens was scored as 0 (none), 1 (sporadic occurrence), 2 (clusters), and 3 (numerous bacteria found diffusely in the mucus layer). The incidence rates of ulcer relapse per patient-month, grouped in accordance with these scores, were (with 95% confidence intervals) 0.073 (0.048-0.111), 0.083 (0.052-0.133), 0.123 (0.096-0.157), and 0.069 (0,041-0.116), respectively. No significant differences in incidence rates across H . pylori scores were observed when taking into account the observation period after healing of the first ulcer, number of ulcer recurrence (lst, 2nd, 3rd), sex, age, smoking habits, peak acid output, time of healing of the preceding ulcer, treatment of the present ulcer (cimetidine, antacids, or no treatment), or type and degree of gastritis. Thus, although H . pylori is prevalent in patients with duodenal ulcer disease, the present study indicates that H . pylori does not have a substantial note in the precipitation of active duodenal ulcer. Key words: Duodenal ulcer; Helicobacfer pylori; longitudinal study; relapse
T. I . A . Smrensen, M . D . , Dept. of Gastroenterology 261, Huidoure Hospital, DK-2650 Copenhagen, Denmark
Helicobacterpylori may be a cause of active chronic gastritis (1,2), and a role in the pathogenesis of duodenal ulcer has also been suggested (3-6). H . pylori is frequently present in patients with duodenal ulcer disease (7-9). In clinical trials drug-induced eradication of H . pylori is followed by a decreased ulcer relapse rate (5,6, lo). However, these findings do not prove a causal relationship between H . pylori and the development of duodenal ulcer (11). Such a relationship would be further supported if it could be shown that the risk of ulcer relapse in the course of time is related to occurrence of H. pylori and its possible changes over time (10). We have addressed this question by examining antral biopsy specimens collected at repetitive endoscopies in a previous longitudinal study of duodenal ulcer patients who had not been treated with drugs influencing H . pylori ( 1 2 ) . PATIENTS AND METHODS
Patients Ninety-one patients admitted with duodenal ulcer to the Medical and Surgical Depts. of Gastroenterology, Hvidovre * Mernbers of The Hvidovre Hospital Ulcer Project Group: 0. Bonnevie, P. M. Christiansen, E. Krag, E. Kragelund, P. Madsen, P. Matzen.
University Hospital, from May 1978 through January 1980 were enrolled in the previous study (12). The present study includes 74 of these patients, from whom biopsy specimens from the antral mucosa were obtained. Thirty-six were women and 38 men, and the median age was 57 years (range, 21-82 years). The criteria for inclusion in the original study were presence of an ulcer with a diameter of more than 5 mm localized in the duodenal bulb, the pyloric canal, or the prepyloric region within 1 cm from the pyloric ring. This corresponds to the extension of the duodenal mucosa with glands of Brunner (13). Criteria for exclusion were complications requiring surgical intervention, concomitant gastric ulcer, previous gastric surgery, gallstones, pregnancy, age below 18 years, or psychiatric disorders. Furthermore, patients who within the preceding month had taken ulcer-healing drugs, non-steroidal anti-inflammatory drugs, or corticosteroids were excluded. The study was performed in accordance with the Helsinki I1 declaration, and informed consent was obtained from all patients.
Design At the initial endoscopy ulcer diameters were measured using an open biopsy forceps (6 mm) as reference, and the area of the ulcer was calculated. Gastric acid secretion was
422
M . Rye Clausen et al.
Table I. Relative distribution of Helicobacrer pylori scores in a longitudinal study of recurrent duodenal ulcers
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
H . pylori score
Primary Active Healed
74 74
24 18
18 25
43 40
15 17
100 100
1st recurrence Active Healed
57 57
10 8
14 24
56
51
20 17
100 100
2nd recurrence 39 Active Healed 39
6
9
3
28
61 53
24 16
100 100
* Active ulcer - the first endoscopy showing the ulcer present; healed ulcer = the first normal endoscopy after an endoscopy showing the ulcer present.
measured after stimulation with histamine diphosphate, 0.04 mg/kg body weight, and expressed as peak acid output (PAO). The patients wcre observed for 2 weeks without treatment but were allowed free use of a inagnesium~aluminium-containing antacid solution neutralizing SO mmol hydrochloric acid per 10 ml. After 2 weeks endoscopy was repeated. If the ulcer area was reduced by less than SO%, patients were randomized to cimetidine or intensive antacid therapy. Patients in whom thc ulccr was still present but the area of the ulcer was reduced by more than 50%) were followed up with endoscopy every 2nd week without treatment. Patients whose ulcers had not healed after a total of 6 weeks of observation were randomized to the same two types of therapy. All patients with healed ulcers were observed for 2 years with endoscopy, including antral and duodenal biopsies, and a clinical follow-up every 3rd month or sooner if required by recurrence of symptoms. The surveillance program of endoscopically verified recurrences was the same as at enrollment to the study. Patients were excluded from further study if they had a third recurrence.
Biopsies At each endoscopy a biopsy specimen was taken from the antral mucosa less than 2 cm from the pylorus and from the duodenal bulb. The endoscope and the biopsy forceps were disinfected in 2% glutaraldehyde between procedures.
solved by subsequent consensus on each biopsy. Gastritis was classified by type (none; acute; chronic; active chronic) and degree (none; mild; moderate; severe) (15). The duodenal specimens were assessed for gastric metaplasia (15). The occurrence of typical Helicobacter bacteria was scored as 0 = none; 1 = sporadic occurrence; 2 = clusters seen in most microscopic fields at high power (x400); 3 = numerous microorganisms seen diffusely in the mucus layer in all fields. To validate the criteria applied for the microscopic examination for H . pylori, we studied biopsy specimens from another 20 patients consecutively referred to endoscopy because of dyspepsia. Two specimens were obtained from the prepyloric antral mucosa, one for microscopy by the same pathologists and one for an immediate urease test (CLO test, Delta West Ltd., Canning Vale, Australia). In all but one case we found complete agreement between the histology and the outcome of the CLO test. Our findings are in agreement with the extensive study by McNulty et al. (16).
Estimation of incidence The incidence rate of ulcer recurrence (per month) was assessed by the number of recurrences per month of observation accumulated for all patients at risk of recurrence-that is, with healed ulcers. For each patient the observation time between any two endoscopies was assigned the H . pylori score obtained at the endoscopy performed at the beginning of the interval. The incidence rate of ulcer recurrence for each H . pylori score was then estimated by dividing the total number of recurrences observed at the end of the intervals assigned that particular score by the total observation time assigned the same score. By this estimation procedure it is assumed that the risk of recurrence does not change over the observation time included. Possible confounding of the relationship between the H . pylori score and the risk of recurrence by changes in risk over observation time and other patient characteristics or by the treatment applied was taken into account by stratification. Statistical analysis Confidence limits of the incidence rates were estimated under the assumption that the distribution of the logarithm
Tablc 11. Helicobacier pylori score between any subsequent endoscopies among patients with healed ulcers
Histology Biopsy specimens were fixed in 10% formalin and embedded in paraffin. Sections were stained with haematoxylin and
H. pylori score at one endoscopy
eosin for general cytology and histology and with Levaditi's silver method (14) to visualize H . pyfori. All sections were evaluated independently by two pathologists who had no information about the clinical data. Initial disagreement was
0 1 2 3
H. pylori score at subsequent endoscopy 0
1
2
3
31
6 12 11 2
10 20 67 19
1 6 23 31
9 4 1
Duodenal Ulcer Relapse and H. pylori
423
of the incidence is asymptotic normal with a variance equal to the reciprocal of the observation time (17). Differences between incidence rates and trends in incidence rates across classes of characteristics were statistically evaluated by using the statistical package GLIM, by likelihood ratio tests of the change in fit between the following models (27):
A: In Aij = qj(unspecified with regard to effects of variable i and j) B: In Aij = q bj (additive effects of variable i and j) C: In Aii = icu + bj (additive effects of variable i and j with trend in variable i) D: In Aij = /3, (no effect of variable i)
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
+
where Aii is the incidence rate specified with regard to characteristics i = 1, , . . , n (for example, H . pylori score 0, 1, 2, and 3) and j = 1, . . ., n (for example, male = 1and female = 2), and cu and p are parameters of the models as specified. Significant difference between model A and model B indicates that there is a significant interaction between the two characteristics; that is, the influence of either characteristic on incidence of recurrence is dependent on the other characteristic. Significant difference between model B and model C indicates significant departure from a trend across classes of variable ‘i’when stratification by variable ‘j’ is maintained. Significant difference between model C and model D indicates significant trend across the classes of variable ‘i’.
RESULTS Antral H. pylori scores Table I shows that more than 75% of all patients with active ulcers had an H. pylori score greater than 1. The same distribution of H . pylori scores was found at the first endoscopy performed after ulcer healing. During the course the ranges of H. pylori scores in patients with healed ulcers were 0-0 in 4 patients, 0-1 in 3, 0-2 in 13, 0-3 in 22, 1-2 in 10, 1-3 in 11, and 2-3 in 11 patients. No patients had score 3 throughout the entire course. Table I1 compares the H. pylori scores among patients with healed ulcers at any two subsequent endoscopies. The score increased in 46 endoscopies, decreased in 66, and remained unchanged in 141. The number of recurrences was 116, the time at risk of recurrence was 1225 months, and, correspondingly, the overall incidence of ulcer recurrence, irrespective of occurrence of H . pylori, was 0.095 per patient-month at risk (95% confidence limits, 0.079-0.114). The incidence of ulcer recurrence in relation to H . pylori scores of 0, 1, 2, and 3 was 0.073 (0.04&0.111), 0.083 (0.052-0.133), 0.123 (0.096-0.157), and 0.069 (0.041-0.116), respectively (Fig. 1). The differences were borderline significant (p = 0.06), but this was apparently due to the isolated high incidence at H. pylori score 2. The incidence rates showed a significant
0
1
2
3
H.pylori score Fig. 1. Incidence of ulcer recurrence in relation to Helicubacter pyluri score. The bars indicate 95% confidence intervals. N = the number of ulcer recurrences; T = time (months) at risk of recurrence. The broken line indicates the overall risk of recurrence (irrespective of H . pylori score).
departure from trend of H . pylori scores (p = 0.04), and there was no significant trend (p = 0.35). There was a decreasing incidence of first recurrence over time ( p = 0.04) and an increasing incidence from first through third recurrence ( p = 0.06) irrespective of stratification by H. pyloriscore (columns of table 111). The analysis showed, however, that the relation between the H . pylori score and the incidence of the first ulcer recurrence did not depend on period of observation after healing of the first ulcer (test for interaction, p = 0.98). Similarly, the relation between the score and incidence of recurrence did not depend on number of recurrence (lst, 2nd, or 3rd) (test for interaction, p = 0.93). These analyses indicate that the relation between H . pylori score and risk of ulcer recurrence can be validly assessed over the total observation period. When the relation between the H . pylori score and the incidence of the first ulcer recurrence alone (first row of Table 111) was evaluated, there was no significant departure from a trend ( p = 0.14) and no significant trend ( p = 0.63). When taking into account the observation period after the first ulcer, the relation between the H . pylori score and incidence of first recurrence was compatible with a trend mode1 ( p = 0.19), but there was no uniform significant trend as such ( p = 0.41) (rows of Table 111). The same kind of analysis with stratification by first, second, and third recurrence showed a significant departure from trend ( p = 0.04), but there was no significant trend across the H. pylori scores ( p = 0.39). Rather than using the H . pylori score, the mere presence
424
M . Rye Clausen et a1
Table 111. Incidence of ulcer recurrence in relation to observation time, number of recurrence, and Helicobacter pylori score Incidence of recurrence per month NO.
or
patients 1st recurrence (total) &4 months*
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
4-8 months* 8+ months* 2nd recurrence 3rd recurrence
57 27 15
15 39 20
Observation time
H . pylori score
(months)
Total
0
1
2
3
706 261 160 285 362 157
0.08 0. 10 0.09 0.05 0.11 0.13
0.07 0.08 0.06 0.05 0.08
0.07 0.11 0.05
0.11 0.13
0.05 0.08 0.09 0.03 0.08 U.13
0.10
0.04
0.14 0.07
0.12 0.06
0.13 0.18
* After healing of initial ulcer.
(score 1-3) versus absence (score 0) of H . pylori may be related to the risk of recurrence. In this analysis there were no significant differences (no significant interaction) either in the H. pylori occurrence versus risk of recurrence across the observation periods after healing of the first ulcer ( p = 0.84) or across the number of recurrences ( p = 0.96). This enables an analysis of the total observation period with stratification by observation period after healing of the first ulcer or by number of recurrence. These analyses showed no significant differences in the incidence of recurrence whether H . pylori was present or absent at the preceding endoscopy ( p = 0.15 and 0.17, respectively). The relationship between the incidence of recurrence and the H . pylori score was also assessed after stratification by sex, age, smoking habits, peak acid output, time of healing of the preceding ulcer, and type of treatment (Table Iv) Considering the results of the preceding analyses and the actual sample size, we did not take into account period of observation after first ulcer or number of recurrence in these analyses. Except for age ( p = 0.01) and smoking habits ( p = 0.001). all characteristics could be analysed on the basis of a trend model (allp > 0.08). However, none of the analyses indicated significant trends across H . pylori scores (all p > 0.20). The departure from the trend model when stratifying by age and smoking habits was not ascribable to consistent differences in incidence rates (Table IV). Active chronic gastritis preceded 84 of the 116 recurrences, but the incidence of recurrence was not significantly related to type of gastritis (no gastritis, 0.06; acute, 0.08; chronic, 0.11; active chronic, 0.10; p = 0.32). Nor was the incidence significantly related to degree of inflammation (no inflammation, 0.06; mild, 0.12; moderate, 0.09; severe, 0.08; p = 0.15). Both type and degree of inflammation enabled analysis of the relation between H. pylori scores and incidence of recurrence ( p = 0.51 and p = 0.37, respectively) by trend models, but there were no significant trends ( p = 0.84 and p = 0.58, respectively). Duodenal H. pylori scores Among the 60 patients in whom duodenal biopsy speci-
mens were obtained, no H . pylori was found at 93.5% of the endoscopies. Score 1 was seen in 4.3%, score 2 in 2.0%, and score 3 in only 0.2%. The rare occurrence of H. pylori precluded estimation of the relation to the risk of ulcer relapse. None of the patients had gastric metaplasia of the duodenal mucosa.
DISCUSSION Experimental evidence supporting a causal role of H. pylori in the development of duodenal ulcer is not available, since deliberately exposing subjects to H . pylori (Koch’s third postulate) for obvious reasons is impossible and because no suitable animal model exists (11). The reduction in relapse rate of duodenal ulcer observed in controlled clinical trials after initial healing with drugs eradicating H . pylori (colloid bismuth subcitrate with or without antibiotics) lends support to a causal relationship (5,6, 10). However, it is possible that other effects of colloid bismuth subcitrate (18,19) or antibiotics (20) could be responsible for the reduced relapse rates. Another possibility is that eradication of the bacteria is most effective in patients who for other reasons have the lowest risk of relapse. A longitudinal study like the present one may provide an alternative to the experimental evidence. In this kind of study the influence of the current status of mucosal infection on the risk of ulcer relapse in the near future may be assessed. There were three main findings in the present study. First, healing of duodenal ulcers was not related to change in occurrence of H. pylori. Second, the current incidence of recurrence of duodenal ulcer did not increase by increasing H . pylori score. Third, this finding was valid even when taking into account the period of observation after the first ulcer, the number of recurrence (first, second, third) and patient characteristics such as sex, age, smoking habits, peak acid output, time of healing of the preceding ulcer, treatment, and type and degree of gastritis. The finding that the prevalence of H. pylori was similar whether the ulcer was in an active stage or had just healed
Duodenal Ulcer Relapse and H. pylori
425
Table 1V. Incidence of ulcer recurrence in relation to patient characteristics, treatment, and Heficobacter pylori score
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
Incidence of recurrence per month
H. pylori score
No. of recurrences
Observation time (months)
Total
0
1
2
3
Sex Male Female
68 48
556 669
0.12 0.07
0.08 0.06
0.11 0.06
0.15 0.10
0.12 0.03
Age (years) -49 50-64 65+
37 47 32
373 381 471
0.10 0.12 0.07
0.04 0.14 0.08
0.11 0.13 0.01
0.15 0.12 0.11
0.11 0.12 0.03
Smoking Yes No
88 28
933 292
0.09 0.10
0.05 0.19
0.09 0.05
0.14 0.09
0.07 0.05
P A 0 (rnmol/l) - 30 31-44 45 +
30 46 40
412 463 350
0.07 0.10 0.11
0.06 0.10 0.07
0.05 0.05 0.16
0.11 0.13 0.15
0.03 0.12 0.08
Healing time I month
44 72
646 579
0.07 0.12
0.04 0.10
0.05 0.11
0.10 0.15
0.05 0.10
Treatment No Antacid Cirnetidine
63 29 24
842 191 192
0.07 0.15 0.13
0.06 0.14 0.07
0.07 0.13 0.07
0.10 0.20 0.19
0.04 0.14 0.19
suggests that absence of the bacteria is not necessary for ulcer healing. This is in accordance with findings by Sipponen et al. (21). There is little doubt that H . pylori occurs more frequently among patients who have duodenal ulcer disease, but our study suggests that the bacteria are not involved in the precipitation of active duodenal ulcers among these patients. The higher frequency of the bacteria can be due to a secondary infection in an otherwise damaged mucosa with diminished resistance towards bacterial invasion. The wellestablished associations between duodenal ulcer disease and active chronic gastritis (22-24), and between H . pylori and active chronic gastritis (7,25-28) may then reflect that these bacteria cause this type of gastritis (1,2). Furthermore, our finding that the incidence of recurrence was not related to the H . pylori score when gastritis, particularly active chronic gastritis. had been diagnosed suggests that the bacteria are not pushing the disease process from inflammation to ulceration. The likelihood that our study has yielded a false-negative result for statistical or methodologic reasons should be considered but seems to be rather low for the following reasons. First, in view of the confidence limits of the incidence of recurrence at each H . pylori score (Fig. l), the steepest possible relationship is an increase in incidence from 0.07 month-' at score 0 to 0.11 month-' at score 3, which is not particularly strong. Even though the mere presence of the bacteria rather than the degree of infestation is considered
crucial, our results were essentially negative. This implies that mucosal infection, as assessed here, may not be a prerequisite of ulcer recurrence. It seems reasonable to consider the relatively high incidence at H . pylori score 2 (Fig. 1) an outlier-the deviation of the risk estimate at score 3 from a consistent upward trend is much greater than the deviation of score 2 from no trend, as also indicated by the formal statistical analysis for trend. Second, the H . pylori score of the individual patients did change in the intervals between the endoscopies, but we cannot exclude the possibiliy that an increase in the score occurred just before ulcer recurrence. On the other hand, the follow-up endoscopies were carried out at least every 3 months. Thus, if a relationship between the score and the incidence does exist, it implies-to be compatible with our findings-that a change in scoring in either direction would be followed by a corresponding change in risk of recurrence within days or a few weeks. If the relationship could only be demonstrated by much more frequent endoscopies, it would not only be out of the scope of the clinical management, but doubt would arise about whether increase in scores is primary or secondary to the process of development of the recurrence. Third, there is some uncertainty in the H . pylori scoring because of biopsy sampling error, specimen storage and processing, and evaluation of the individual specimens. However, there is no obvious reason for postulating a systematic scoring bias that could have eliminated a positive
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
426
M . Rye Clausen el a1
relationship between the score and the incidence of recurrence. W e have not used the urease assay or culturing for assessment of H . pylori infection, but it is noteworthy that the microscopic scoring was well in accordance with the urease test in another patient series. Moreover, the frequency of bacteria in our study was of the same order of magnitude as i n other studies ( 3 , 4 , 7 , 2 6 , 2 9 ) . Further, t h e relationship between H. pylori and risk of recurrence could be less obvious in the crude figures if t h e bacteria were less frequent in patients with characteristics otherwise predisposing to risk of recurrence. We considered sex, age, smoking habits, peak acid output, healing time, and treatment and found no consistent evidence of confounding. Only rarely did we find H. pylori and gastric metaplasia in the duodenum in the periods between t h e relapses of ulcers. Information about occurrence of H. pylori and gastric metaplasia in such patients is still rather scarce (30). In conclusion, the present study does not support a substantial pathogenic role of H . pylori in precipitation of duodenal ulcer relapse.
ACKNOWLEDGEMENTS T h e authors thank Per Kragh Andersen, lic. scient., for advice on the statistical analysis and Anna-Lise Petersen for data recording.
REFERENCES 1. Marshall B, Armstrong J, McGechie D, Glancy R. Attempt to
fulfil Koch’s postulates for pyloric campylobacter. Med J Aust 1985;142:436-9. 2. Morris A, Nicholson G . Ingestion of Campylobacter pyloridis causes gastritis and raised fasting gastricpH. Am J Gastroenterol 1987;82:192-9. 3. Price AB, Levi J, Dolby JM, et al. Campylobacter pyloridis in peptic ulcer diseasc: microbiology, pathology, and scanning electron microscopy. Gut 1985;26:1183-8. 4. Humphreys H , Bourke S, Dooley C, et al. Effect of treatment on Campylohacter pylori in peptic disease: a randomised prospective trial. Gut 1988;29:27%83. 5. Marshall BJ, Goodwin CS, Warren JR, et al. Prospective double-blind trial of duodenal ulcer relapse after eradication of Campylobacter pylori . Lancet 1988;2:1437-42. 6. Rauws EAJ, Tytgat GNJ. Cure of duodenal ulcer associated with eradication of Helicohacter pylori. Lancet 1990;335: 1233-5. 7. Marshall BJ, Warren JR.Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet 1984;l: 1311-5. 8. Marshall BJ, Royce H, Annear DI, et al. Original isolation of Campylobacter pyloridis from human gastric mucosa. Microbios Letters 1984;25:83-8. Received 12 April 1991 Accepted 16 November 1991
9. McNulty CAM, Watson DM. Spiral bacteria of the gastric antrum. Lancet 1984;l:106%9. 10. Coghlan JG, Gilligan D, Humphries H, et al. Campylobacter pylori and recurrence of duodenal ulcer-a 12-month follow-up study. Lancet 1987;2:1109-11. 11. Peterson WL. Heliobacter pylori and peptic ulcer disease. N Engl J Med 1991;324:1043-8. 12. Frederiksen HJB, Matzen P, Madsen P, et al. Spontaneous healing of duodenal ulcers. Scand J Gastroenterol 1984;19: 417-21. 13. Landhoe-Christensen E. The duodenal glands of Brunner in man, their distribution and quantity. Acta Pathol Microbiol Scand 1944;Suppl 52. 14. Wartkin AS, Stany AC. A more rapid and improved method of demonstrating spirochaetes in tissues. Am J Syphilis 1920; 4: 97-9. 15. Whitehead R. Mucosal biopsy of the gastrointestinal tract. In: Benning JL, editor Major problems in pathology. Vol. 111. 2nd ed. Philadelphia: WB Saunders, 1975:15. 16. McNulty CAM, Dent JC, Uff JS, Gear MWL, Wilkinson SP. Detection of Campylobacter pylori by the biopsy urease test: an assessment in 1445 patients. Gut 1989;30:1058-62. 17. Aitkin M, Clayton D. The fitting of exponential, Weibull and extreme value distributions to complex censored survival data using GLIM. Appl Statist 1980;29:156-63. 18. Hall DWR. Review of the modes of action of colloidal bismuth subcitrate. Scand J Gastroenterol 1989;24 Suppl 1 5 7 5 6 . 19. Sercombe J, Pounder RE. Bismuth accumulates in the body during treatment with tripotassium dicitrato bismuthate. Aliment Pharmacol Ther 1989;3:21-8. 20. Satoh H, Guth PH, Grossman MI. Role of bacteria in gastric ulceration produced by indomethacin in the rat: cytoprotective action of antibiotics. Gastroenterology I983;84: 483-9. 21. Sipponen P,Varis K, Cederberg A, et al. Campylobacter pylori is associated with chronic gastritis but not with active peptic ulcer disease. APMIS 1988;96:84-8. 22. Schrager J, Spink R, Mitra S. The antrum in patients with duodenal and gastric ulcers. Gut 1967;8:497-508. 23. Dooley CP, McKenna D, Humphreys H, et al. Histological gastritis induodenal ulcer: relationship to Campylobacter pylori and effect of ulcer therapy. Am J Gastroenterol1988;83:278-82. 24. Varis K, Sipponen P, Cederberg A, Salmi H. Chronic gastritis and Le3-phenotype in peptic ulcer disease and in non-ulcer dyspepsia. Hepatogastroenterol 1987;34:46-7. 25. Rauws EAJ, Langenberg W, Houthoff HJ, Zanan HC, Tytgat GNJ. Campylobacter pyloridis-associated chronic active antral gastritis-a prospective study of its prevalence and the effects of antibacterial and antiulcer treatment. Gastroenterology 1988;94:33-40. 26. Jiang SJ, Lui DY, Zhang DZ, et al. Campylobacter-like organisms in chronic gastritis, peptic ulcer and gastric carcinoma. Scand J Gastroenterol 1987;22:SS3-8. 27. Jones DM, Lessels AM, Eldridge J. Campylobacter like organisms on the gastric mucosa: culture, histological, and serological studies. J Clin Pathol 1984;37:1002-6. 28. Borsch G, Schmidt G, Wegener M, et al. Campylobacter pylori: prospective analysis of clinical and histological factors associated with colonization of the upper gastro-intestinal tract. Eur J Clin Invest 1988;18:133-8. 29. Graham DY,Klein PD, Opekun AR, et al. Epidemiology of Campylohacter pyloridis infection. Gastroenterology 1987;92: 1411. 30.Axon AR. Duodenal ulcer: the villain unmasked? Eradicating Helicohacter pylori will cure most patients. Br Med J 1991; 302;919-20.
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
Rye Clausen M, Franzmann MB, Holst C, Sarensen TIA, Christoffersen P, Matzen P, Krag E , The Hvidovre Ulcer Project Group. Longitudinal study of influence of Helicobacter pylori on current risk of duodenal ulcer relapse. Scand J Gastroenterol 1992;27:421-426. Seventy-four patients with duodenal ulcer were followed up longitudinally for 2 years after initial ulcer healing. Endoscopy including biopsy of the antral mucosa was performed every 3rd month and whenever clinical symptoms of relapse occurred. The presence of Helicobacter pylori in the biopsy specimens was scored as 0 (none), 1 (sporadic occurrence), 2 (clusters), and 3 (numerous bacteria found diffusely in the mucus layer). The incidence rates of ulcer relapse per patient-month, grouped in accordance with these scores, were (with 95% confidence intervals) 0.073 (0.048-0.111), 0.083 (0.052-0.133), 0.123 (0.096-0.157), and 0.069 (0,041-0.116), respectively. No significant differences in incidence rates across H . pylori scores were observed when taking into account the observation period after healing of the first ulcer, number of ulcer recurrence (lst, 2nd, 3rd), sex, age, smoking habits, peak acid output, time of healing of the preceding ulcer, treatment of the present ulcer (cimetidine, antacids, or no treatment), or type and degree of gastritis. Thus, although H . pylori is prevalent in patients with duodenal ulcer disease, the present study indicates that H . pylori does not have a substantial note in the precipitation of active duodenal ulcer. Key words: Duodenal ulcer; Helicobacfer pylori; longitudinal study; relapse
T. I . A . Smrensen, M . D . , Dept. of Gastroenterology 261, Huidoure Hospital, DK-2650 Copenhagen, Denmark
Helicobacterpylori may be a cause of active chronic gastritis (1,2), and a role in the pathogenesis of duodenal ulcer has also been suggested (3-6). H . pylori is frequently present in patients with duodenal ulcer disease (7-9). In clinical trials drug-induced eradication of H . pylori is followed by a decreased ulcer relapse rate (5,6, lo). However, these findings do not prove a causal relationship between H . pylori and the development of duodenal ulcer (11). Such a relationship would be further supported if it could be shown that the risk of ulcer relapse in the course of time is related to occurrence of H. pylori and its possible changes over time (10). We have addressed this question by examining antral biopsy specimens collected at repetitive endoscopies in a previous longitudinal study of duodenal ulcer patients who had not been treated with drugs influencing H . pylori ( 1 2 ) . PATIENTS AND METHODS
Patients Ninety-one patients admitted with duodenal ulcer to the Medical and Surgical Depts. of Gastroenterology, Hvidovre * Mernbers of The Hvidovre Hospital Ulcer Project Group: 0. Bonnevie, P. M. Christiansen, E. Krag, E. Kragelund, P. Madsen, P. Matzen.
University Hospital, from May 1978 through January 1980 were enrolled in the previous study (12). The present study includes 74 of these patients, from whom biopsy specimens from the antral mucosa were obtained. Thirty-six were women and 38 men, and the median age was 57 years (range, 21-82 years). The criteria for inclusion in the original study were presence of an ulcer with a diameter of more than 5 mm localized in the duodenal bulb, the pyloric canal, or the prepyloric region within 1 cm from the pyloric ring. This corresponds to the extension of the duodenal mucosa with glands of Brunner (13). Criteria for exclusion were complications requiring surgical intervention, concomitant gastric ulcer, previous gastric surgery, gallstones, pregnancy, age below 18 years, or psychiatric disorders. Furthermore, patients who within the preceding month had taken ulcer-healing drugs, non-steroidal anti-inflammatory drugs, or corticosteroids were excluded. The study was performed in accordance with the Helsinki I1 declaration, and informed consent was obtained from all patients.
Design At the initial endoscopy ulcer diameters were measured using an open biopsy forceps (6 mm) as reference, and the area of the ulcer was calculated. Gastric acid secretion was
422
M . Rye Clausen et al.
Table I. Relative distribution of Helicobacrer pylori scores in a longitudinal study of recurrent duodenal ulcers
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
H . pylori score
Primary Active Healed
74 74
24 18
18 25
43 40
15 17
100 100
1st recurrence Active Healed
57 57
10 8
14 24
56
51
20 17
100 100
2nd recurrence 39 Active Healed 39
6
9
3
28
61 53
24 16
100 100
* Active ulcer - the first endoscopy showing the ulcer present; healed ulcer = the first normal endoscopy after an endoscopy showing the ulcer present.
measured after stimulation with histamine diphosphate, 0.04 mg/kg body weight, and expressed as peak acid output (PAO). The patients wcre observed for 2 weeks without treatment but were allowed free use of a inagnesium~aluminium-containing antacid solution neutralizing SO mmol hydrochloric acid per 10 ml. After 2 weeks endoscopy was repeated. If the ulcer area was reduced by less than SO%, patients were randomized to cimetidine or intensive antacid therapy. Patients in whom thc ulccr was still present but the area of the ulcer was reduced by more than 50%) were followed up with endoscopy every 2nd week without treatment. Patients whose ulcers had not healed after a total of 6 weeks of observation were randomized to the same two types of therapy. All patients with healed ulcers were observed for 2 years with endoscopy, including antral and duodenal biopsies, and a clinical follow-up every 3rd month or sooner if required by recurrence of symptoms. The surveillance program of endoscopically verified recurrences was the same as at enrollment to the study. Patients were excluded from further study if they had a third recurrence.
Biopsies At each endoscopy a biopsy specimen was taken from the antral mucosa less than 2 cm from the pylorus and from the duodenal bulb. The endoscope and the biopsy forceps were disinfected in 2% glutaraldehyde between procedures.
solved by subsequent consensus on each biopsy. Gastritis was classified by type (none; acute; chronic; active chronic) and degree (none; mild; moderate; severe) (15). The duodenal specimens were assessed for gastric metaplasia (15). The occurrence of typical Helicobacter bacteria was scored as 0 = none; 1 = sporadic occurrence; 2 = clusters seen in most microscopic fields at high power (x400); 3 = numerous microorganisms seen diffusely in the mucus layer in all fields. To validate the criteria applied for the microscopic examination for H . pylori, we studied biopsy specimens from another 20 patients consecutively referred to endoscopy because of dyspepsia. Two specimens were obtained from the prepyloric antral mucosa, one for microscopy by the same pathologists and one for an immediate urease test (CLO test, Delta West Ltd., Canning Vale, Australia). In all but one case we found complete agreement between the histology and the outcome of the CLO test. Our findings are in agreement with the extensive study by McNulty et al. (16).
Estimation of incidence The incidence rate of ulcer recurrence (per month) was assessed by the number of recurrences per month of observation accumulated for all patients at risk of recurrence-that is, with healed ulcers. For each patient the observation time between any two endoscopies was assigned the H . pylori score obtained at the endoscopy performed at the beginning of the interval. The incidence rate of ulcer recurrence for each H . pylori score was then estimated by dividing the total number of recurrences observed at the end of the intervals assigned that particular score by the total observation time assigned the same score. By this estimation procedure it is assumed that the risk of recurrence does not change over the observation time included. Possible confounding of the relationship between the H . pylori score and the risk of recurrence by changes in risk over observation time and other patient characteristics or by the treatment applied was taken into account by stratification. Statistical analysis Confidence limits of the incidence rates were estimated under the assumption that the distribution of the logarithm
Tablc 11. Helicobacier pylori score between any subsequent endoscopies among patients with healed ulcers
Histology Biopsy specimens were fixed in 10% formalin and embedded in paraffin. Sections were stained with haematoxylin and
H. pylori score at one endoscopy
eosin for general cytology and histology and with Levaditi's silver method (14) to visualize H . pyfori. All sections were evaluated independently by two pathologists who had no information about the clinical data. Initial disagreement was
0 1 2 3
H. pylori score at subsequent endoscopy 0
1
2
3
31
6 12 11 2
10 20 67 19
1 6 23 31
9 4 1
Duodenal Ulcer Relapse and H. pylori
423
of the incidence is asymptotic normal with a variance equal to the reciprocal of the observation time (17). Differences between incidence rates and trends in incidence rates across classes of characteristics were statistically evaluated by using the statistical package GLIM, by likelihood ratio tests of the change in fit between the following models (27):
A: In Aij = qj(unspecified with regard to effects of variable i and j) B: In Aij = q bj (additive effects of variable i and j) C: In Aii = icu + bj (additive effects of variable i and j with trend in variable i) D: In Aij = /3, (no effect of variable i)
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
+
where Aii is the incidence rate specified with regard to characteristics i = 1, , . . , n (for example, H . pylori score 0, 1, 2, and 3) and j = 1, . . ., n (for example, male = 1and female = 2), and cu and p are parameters of the models as specified. Significant difference between model A and model B indicates that there is a significant interaction between the two characteristics; that is, the influence of either characteristic on incidence of recurrence is dependent on the other characteristic. Significant difference between model B and model C indicates significant departure from a trend across classes of variable ‘i’when stratification by variable ‘j’ is maintained. Significant difference between model C and model D indicates significant trend across the classes of variable ‘i’.
RESULTS Antral H. pylori scores Table I shows that more than 75% of all patients with active ulcers had an H. pylori score greater than 1. The same distribution of H . pylori scores was found at the first endoscopy performed after ulcer healing. During the course the ranges of H. pylori scores in patients with healed ulcers were 0-0 in 4 patients, 0-1 in 3, 0-2 in 13, 0-3 in 22, 1-2 in 10, 1-3 in 11, and 2-3 in 11 patients. No patients had score 3 throughout the entire course. Table I1 compares the H. pylori scores among patients with healed ulcers at any two subsequent endoscopies. The score increased in 46 endoscopies, decreased in 66, and remained unchanged in 141. The number of recurrences was 116, the time at risk of recurrence was 1225 months, and, correspondingly, the overall incidence of ulcer recurrence, irrespective of occurrence of H . pylori, was 0.095 per patient-month at risk (95% confidence limits, 0.079-0.114). The incidence of ulcer recurrence in relation to H . pylori scores of 0, 1, 2, and 3 was 0.073 (0.04&0.111), 0.083 (0.052-0.133), 0.123 (0.096-0.157), and 0.069 (0.041-0.116), respectively (Fig. 1). The differences were borderline significant (p = 0.06), but this was apparently due to the isolated high incidence at H. pylori score 2. The incidence rates showed a significant
0
1
2
3
H.pylori score Fig. 1. Incidence of ulcer recurrence in relation to Helicubacter pyluri score. The bars indicate 95% confidence intervals. N = the number of ulcer recurrences; T = time (months) at risk of recurrence. The broken line indicates the overall risk of recurrence (irrespective of H . pylori score).
departure from trend of H . pylori scores (p = 0.04), and there was no significant trend (p = 0.35). There was a decreasing incidence of first recurrence over time ( p = 0.04) and an increasing incidence from first through third recurrence ( p = 0.06) irrespective of stratification by H. pyloriscore (columns of table 111). The analysis showed, however, that the relation between the H . pylori score and the incidence of the first ulcer recurrence did not depend on period of observation after healing of the first ulcer (test for interaction, p = 0.98). Similarly, the relation between the score and incidence of recurrence did not depend on number of recurrence (lst, 2nd, or 3rd) (test for interaction, p = 0.93). These analyses indicate that the relation between H . pylori score and risk of ulcer recurrence can be validly assessed over the total observation period. When the relation between the H . pylori score and the incidence of the first ulcer recurrence alone (first row of Table 111) was evaluated, there was no significant departure from a trend ( p = 0.14) and no significant trend ( p = 0.63). When taking into account the observation period after the first ulcer, the relation between the H . pylori score and incidence of first recurrence was compatible with a trend mode1 ( p = 0.19), but there was no uniform significant trend as such ( p = 0.41) (rows of Table 111). The same kind of analysis with stratification by first, second, and third recurrence showed a significant departure from trend ( p = 0.04), but there was no significant trend across the H. pylori scores ( p = 0.39). Rather than using the H . pylori score, the mere presence
424
M . Rye Clausen et a1
Table 111. Incidence of ulcer recurrence in relation to observation time, number of recurrence, and Helicobacter pylori score Incidence of recurrence per month NO.
or
patients 1st recurrence (total) &4 months*
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
4-8 months* 8+ months* 2nd recurrence 3rd recurrence
57 27 15
15 39 20
Observation time
H . pylori score
(months)
Total
0
1
2
3
706 261 160 285 362 157
0.08 0. 10 0.09 0.05 0.11 0.13
0.07 0.08 0.06 0.05 0.08
0.07 0.11 0.05
0.11 0.13
0.05 0.08 0.09 0.03 0.08 U.13
0.10
0.04
0.14 0.07
0.12 0.06
0.13 0.18
* After healing of initial ulcer.
(score 1-3) versus absence (score 0) of H . pylori may be related to the risk of recurrence. In this analysis there were no significant differences (no significant interaction) either in the H. pylori occurrence versus risk of recurrence across the observation periods after healing of the first ulcer ( p = 0.84) or across the number of recurrences ( p = 0.96). This enables an analysis of the total observation period with stratification by observation period after healing of the first ulcer or by number of recurrence. These analyses showed no significant differences in the incidence of recurrence whether H . pylori was present or absent at the preceding endoscopy ( p = 0.15 and 0.17, respectively). The relationship between the incidence of recurrence and the H . pylori score was also assessed after stratification by sex, age, smoking habits, peak acid output, time of healing of the preceding ulcer, and type of treatment (Table Iv) Considering the results of the preceding analyses and the actual sample size, we did not take into account period of observation after first ulcer or number of recurrence in these analyses. Except for age ( p = 0.01) and smoking habits ( p = 0.001). all characteristics could be analysed on the basis of a trend model (allp > 0.08). However, none of the analyses indicated significant trends across H . pylori scores (all p > 0.20). The departure from the trend model when stratifying by age and smoking habits was not ascribable to consistent differences in incidence rates (Table IV). Active chronic gastritis preceded 84 of the 116 recurrences, but the incidence of recurrence was not significantly related to type of gastritis (no gastritis, 0.06; acute, 0.08; chronic, 0.11; active chronic, 0.10; p = 0.32). Nor was the incidence significantly related to degree of inflammation (no inflammation, 0.06; mild, 0.12; moderate, 0.09; severe, 0.08; p = 0.15). Both type and degree of inflammation enabled analysis of the relation between H. pylori scores and incidence of recurrence ( p = 0.51 and p = 0.37, respectively) by trend models, but there were no significant trends ( p = 0.84 and p = 0.58, respectively). Duodenal H. pylori scores Among the 60 patients in whom duodenal biopsy speci-
mens were obtained, no H . pylori was found at 93.5% of the endoscopies. Score 1 was seen in 4.3%, score 2 in 2.0%, and score 3 in only 0.2%. The rare occurrence of H. pylori precluded estimation of the relation to the risk of ulcer relapse. None of the patients had gastric metaplasia of the duodenal mucosa.
DISCUSSION Experimental evidence supporting a causal role of H. pylori in the development of duodenal ulcer is not available, since deliberately exposing subjects to H . pylori (Koch’s third postulate) for obvious reasons is impossible and because no suitable animal model exists (11). The reduction in relapse rate of duodenal ulcer observed in controlled clinical trials after initial healing with drugs eradicating H . pylori (colloid bismuth subcitrate with or without antibiotics) lends support to a causal relationship (5,6, 10). However, it is possible that other effects of colloid bismuth subcitrate (18,19) or antibiotics (20) could be responsible for the reduced relapse rates. Another possibility is that eradication of the bacteria is most effective in patients who for other reasons have the lowest risk of relapse. A longitudinal study like the present one may provide an alternative to the experimental evidence. In this kind of study the influence of the current status of mucosal infection on the risk of ulcer relapse in the near future may be assessed. There were three main findings in the present study. First, healing of duodenal ulcers was not related to change in occurrence of H. pylori. Second, the current incidence of recurrence of duodenal ulcer did not increase by increasing H . pylori score. Third, this finding was valid even when taking into account the period of observation after the first ulcer, the number of recurrence (first, second, third) and patient characteristics such as sex, age, smoking habits, peak acid output, time of healing of the preceding ulcer, treatment, and type and degree of gastritis. The finding that the prevalence of H. pylori was similar whether the ulcer was in an active stage or had just healed
Duodenal Ulcer Relapse and H. pylori
425
Table 1V. Incidence of ulcer recurrence in relation to patient characteristics, treatment, and Heficobacter pylori score
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
Incidence of recurrence per month
H. pylori score
No. of recurrences
Observation time (months)
Total
0
1
2
3
Sex Male Female
68 48
556 669
0.12 0.07
0.08 0.06
0.11 0.06
0.15 0.10
0.12 0.03
Age (years) -49 50-64 65+
37 47 32
373 381 471
0.10 0.12 0.07
0.04 0.14 0.08
0.11 0.13 0.01
0.15 0.12 0.11
0.11 0.12 0.03
Smoking Yes No
88 28
933 292
0.09 0.10
0.05 0.19
0.09 0.05
0.14 0.09
0.07 0.05
P A 0 (rnmol/l) - 30 31-44 45 +
30 46 40
412 463 350
0.07 0.10 0.11
0.06 0.10 0.07
0.05 0.05 0.16
0.11 0.13 0.15
0.03 0.12 0.08
Healing time I month
44 72
646 579
0.07 0.12
0.04 0.10
0.05 0.11
0.10 0.15
0.05 0.10
Treatment No Antacid Cirnetidine
63 29 24
842 191 192
0.07 0.15 0.13
0.06 0.14 0.07
0.07 0.13 0.07
0.10 0.20 0.19
0.04 0.14 0.19
suggests that absence of the bacteria is not necessary for ulcer healing. This is in accordance with findings by Sipponen et al. (21). There is little doubt that H . pylori occurs more frequently among patients who have duodenal ulcer disease, but our study suggests that the bacteria are not involved in the precipitation of active duodenal ulcers among these patients. The higher frequency of the bacteria can be due to a secondary infection in an otherwise damaged mucosa with diminished resistance towards bacterial invasion. The wellestablished associations between duodenal ulcer disease and active chronic gastritis (22-24), and between H . pylori and active chronic gastritis (7,25-28) may then reflect that these bacteria cause this type of gastritis (1,2). Furthermore, our finding that the incidence of recurrence was not related to the H . pylori score when gastritis, particularly active chronic gastritis. had been diagnosed suggests that the bacteria are not pushing the disease process from inflammation to ulceration. The likelihood that our study has yielded a false-negative result for statistical or methodologic reasons should be considered but seems to be rather low for the following reasons. First, in view of the confidence limits of the incidence of recurrence at each H . pylori score (Fig. l), the steepest possible relationship is an increase in incidence from 0.07 month-' at score 0 to 0.11 month-' at score 3, which is not particularly strong. Even though the mere presence of the bacteria rather than the degree of infestation is considered
crucial, our results were essentially negative. This implies that mucosal infection, as assessed here, may not be a prerequisite of ulcer recurrence. It seems reasonable to consider the relatively high incidence at H . pylori score 2 (Fig. 1) an outlier-the deviation of the risk estimate at score 3 from a consistent upward trend is much greater than the deviation of score 2 from no trend, as also indicated by the formal statistical analysis for trend. Second, the H . pylori score of the individual patients did change in the intervals between the endoscopies, but we cannot exclude the possibiliy that an increase in the score occurred just before ulcer recurrence. On the other hand, the follow-up endoscopies were carried out at least every 3 months. Thus, if a relationship between the score and the incidence does exist, it implies-to be compatible with our findings-that a change in scoring in either direction would be followed by a corresponding change in risk of recurrence within days or a few weeks. If the relationship could only be demonstrated by much more frequent endoscopies, it would not only be out of the scope of the clinical management, but doubt would arise about whether increase in scores is primary or secondary to the process of development of the recurrence. Third, there is some uncertainty in the H . pylori scoring because of biopsy sampling error, specimen storage and processing, and evaluation of the individual specimens. However, there is no obvious reason for postulating a systematic scoring bias that could have eliminated a positive
Scand J Gastroenterol Downloaded from informahealthcare.com by University of British Columbia on 12/11/14 For personal use only.
426
M . Rye Clausen el a1
relationship between the score and the incidence of recurrence. W e have not used the urease assay or culturing for assessment of H . pylori infection, but it is noteworthy that the microscopic scoring was well in accordance with the urease test in another patient series. Moreover, the frequency of bacteria in our study was of the same order of magnitude as i n other studies ( 3 , 4 , 7 , 2 6 , 2 9 ) . Further, t h e relationship between H. pylori and risk of recurrence could be less obvious in the crude figures if t h e bacteria were less frequent in patients with characteristics otherwise predisposing to risk of recurrence. We considered sex, age, smoking habits, peak acid output, healing time, and treatment and found no consistent evidence of confounding. Only rarely did we find H. pylori and gastric metaplasia in the duodenum in the periods between t h e relapses of ulcers. Information about occurrence of H. pylori and gastric metaplasia in such patients is still rather scarce (30). In conclusion, the present study does not support a substantial pathogenic role of H . pylori in precipitation of duodenal ulcer relapse.
ACKNOWLEDGEMENTS T h e authors thank Per Kragh Andersen, lic. scient., for advice on the statistical analysis and Anna-Lise Petersen for data recording.
REFERENCES 1. Marshall B, Armstrong J, McGechie D, Glancy R. Attempt to
fulfil Koch’s postulates for pyloric campylobacter. Med J Aust 1985;142:436-9. 2. Morris A, Nicholson G . Ingestion of Campylobacter pyloridis causes gastritis and raised fasting gastricpH. Am J Gastroenterol 1987;82:192-9. 3. Price AB, Levi J, Dolby JM, et al. Campylobacter pyloridis in peptic ulcer diseasc: microbiology, pathology, and scanning electron microscopy. Gut 1985;26:1183-8. 4. Humphreys H , Bourke S, Dooley C, et al. Effect of treatment on Campylohacter pylori in peptic disease: a randomised prospective trial. Gut 1988;29:27%83. 5. Marshall BJ, Goodwin CS, Warren JR, et al. Prospective double-blind trial of duodenal ulcer relapse after eradication of Campylobacter pylori . Lancet 1988;2:1437-42. 6. Rauws EAJ, Tytgat GNJ. Cure of duodenal ulcer associated with eradication of Helicohacter pylori. Lancet 1990;335: 1233-5. 7. Marshall BJ, Warren JR.Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet 1984;l: 1311-5. 8. Marshall BJ, Royce H, Annear DI, et al. Original isolation of Campylobacter pyloridis from human gastric mucosa. Microbios Letters 1984;25:83-8. Received 12 April 1991 Accepted 16 November 1991
9. McNulty CAM, Watson DM. Spiral bacteria of the gastric antrum. Lancet 1984;l:106%9. 10. Coghlan JG, Gilligan D, Humphries H, et al. Campylobacter pylori and recurrence of duodenal ulcer-a 12-month follow-up study. Lancet 1987;2:1109-11. 11. Peterson WL. Heliobacter pylori and peptic ulcer disease. N Engl J Med 1991;324:1043-8. 12. Frederiksen HJB, Matzen P, Madsen P, et al. Spontaneous healing of duodenal ulcers. Scand J Gastroenterol 1984;19: 417-21. 13. Landhoe-Christensen E. The duodenal glands of Brunner in man, their distribution and quantity. Acta Pathol Microbiol Scand 1944;Suppl 52. 14. Wartkin AS, Stany AC. A more rapid and improved method of demonstrating spirochaetes in tissues. Am J Syphilis 1920; 4: 97-9. 15. Whitehead R. Mucosal biopsy of the gastrointestinal tract. In: Benning JL, editor Major problems in pathology. Vol. 111. 2nd ed. Philadelphia: WB Saunders, 1975:15. 16. McNulty CAM, Dent JC, Uff JS, Gear MWL, Wilkinson SP. Detection of Campylobacter pylori by the biopsy urease test: an assessment in 1445 patients. Gut 1989;30:1058-62. 17. Aitkin M, Clayton D. The fitting of exponential, Weibull and extreme value distributions to complex censored survival data using GLIM. Appl Statist 1980;29:156-63. 18. Hall DWR. Review of the modes of action of colloidal bismuth subcitrate. Scand J Gastroenterol 1989;24 Suppl 1 5 7 5 6 . 19. Sercombe J, Pounder RE. Bismuth accumulates in the body during treatment with tripotassium dicitrato bismuthate. Aliment Pharmacol Ther 1989;3:21-8. 20. Satoh H, Guth PH, Grossman MI. Role of bacteria in gastric ulceration produced by indomethacin in the rat: cytoprotective action of antibiotics. Gastroenterology I983;84: 483-9. 21. Sipponen P,Varis K, Cederberg A, et al. Campylobacter pylori is associated with chronic gastritis but not with active peptic ulcer disease. APMIS 1988;96:84-8. 22. Schrager J, Spink R, Mitra S. The antrum in patients with duodenal and gastric ulcers. Gut 1967;8:497-508. 23. Dooley CP, McKenna D, Humphreys H, et al. Histological gastritis induodenal ulcer: relationship to Campylobacter pylori and effect of ulcer therapy. Am J Gastroenterol1988;83:278-82. 24. Varis K, Sipponen P, Cederberg A, Salmi H. Chronic gastritis and Le3-phenotype in peptic ulcer disease and in non-ulcer dyspepsia. Hepatogastroenterol 1987;34:46-7. 25. Rauws EAJ, Langenberg W, Houthoff HJ, Zanan HC, Tytgat GNJ. Campylobacter pyloridis-associated chronic active antral gastritis-a prospective study of its prevalence and the effects of antibacterial and antiulcer treatment. Gastroenterology 1988;94:33-40. 26. Jiang SJ, Lui DY, Zhang DZ, et al. Campylobacter-like organisms in chronic gastritis, peptic ulcer and gastric carcinoma. Scand J Gastroenterol 1987;22:SS3-8. 27. Jones DM, Lessels AM, Eldridge J. Campylobacter like organisms on the gastric mucosa: culture, histological, and serological studies. J Clin Pathol 1984;37:1002-6. 28. Borsch G, Schmidt G, Wegener M, et al. Campylobacter pylori: prospective analysis of clinical and histological factors associated with colonization of the upper gastro-intestinal tract. Eur J Clin Invest 1988;18:133-8. 29. Graham DY,Klein PD, Opekun AR, et al. Epidemiology of Campylohacter pyloridis infection. Gastroenterology 1987;92: 1411. 30.Axon AR. Duodenal ulcer: the villain unmasked? Eradicating Helicohacter pylori will cure most patients. Br Med J 1991; 302;919-20.
