Aliment Pharmacol Ther 2003; 17: 1125–1135.
Helicobacter pylori eradication and gastric ulcer healing —
comparison of three pantoprazole-based triple therapies
P. MA LFERTHEINER *, T. KIR CHNER , M. KISTà, A. LEOD OLTER *, U. PEITZ*, S. STR OBELà,
M. BOH USCHK E§, G. GA TZ§ & THE BY K AD VANCED GASTR IC ULCER STUDY (BAGUS) G ROUP
*Otto-von-Guericke-University, Magdeburg, Germany;  University of Erlangen-Nürnberg, Germany; àDepartment of
Medical Microbiology and Hygiene, University Hospital of Freiburg, Germany; §ALTANA Pharma AG, Konstanz, Germany
Accepted for publication 26 February 2003
SUMMARY
Aim: To study the efficacy of three pantoprazole-based
triple therapy regimens for the eradication of Helicob-
acter pylori infection and gastric ulcer healing.
Methods: In an open, multi-centre, randomized study,
519 H. pylori-positive patients with active gastric ulcer
were randomized to receive pantoprazole (40 mg) (P)
and two of three antibiotics: clarithromycin (500 mg)
(C), metronidazole (500 mg) (M) or amoxicillin
(1000 mg) (A). Triple therapy (PAC, PCM, PAM) was
administered twice daily for 7 days, followed by pan-
toprazole until the ulcer had healed. Antrum and
corpus biopsies were taken to determine the pattern of
gastritis, to assess the H. pylori status and to determine
the strain susceptibility to antibiotics, and from the
ulcer margins and base to exclude malignancy. Scores
based on the Sydney system were used to categorize the
gastritis phenotypically.
INTRODUCTION
Helicobacter pylori infection is the pathogenic key to the
development of the majority of peptic ulcers, supported
by strong biological plausibility. Several bacterial factors
and inflammatory mechanisms, resulting from the
infection, damage the gastric mucosal integrity.1 The
Correspondence to: Professor P. Malfertheiner, Department of Gastroen-
terology, Hepatology and Infectious Diseases, Otto-von-Guericke-University
Magdeburg, Leipziger Strasse 44, D-39120 Magdeburg, Germany.
E-mail: peter.malfertheiner@medizin.uni-magdeburg.de

2003 Blackwell Publishing Ltd
doi: 10.1046/j.0269-2813.2003.01560.
Results: The H. pylori eradication rates for the per
protocol (intention-to-treat) analysis were 89% (67%)
for PAC, 83% (68%) for PCM and 76% (60%) for PAM,
with a significant difference between PAC and PAM.
Healing rates after 4 weeks were 91% for PAM, 90% for
PCM and 88% for PAC (per protocol analysis). The
eradication rates were lower in patients in whom strains
resistant to any antibiotic used in the triple therapies
were detected. Successful eradication [odds ratio, 5.2
(3.3; 8.3)] and the ulcer size (< 15 mm) were
significant predictors for healing after 4 weeks. The
regimens showed a comparable safety profile and
compliance.
Conclusions: Pantoprazole-based triple therapies are
effective in the eradication of H. pylori infection in
gastric ulcer patients, as reported in previous similar
sized studies in duodenal ulcer patients. Successful
eradication and an ulcer size of < 15 mm are the best
predictors of gastric ulcer healing after 4 weeks.
final proof of the role of H. pylori, however, was the
therapeutic achievement of a definitive cure of peptic
ulcers through eradication therapy. Relapse rates after
successful eradication are almost nonexistent in gastric
ulcer2, 3 and duodenal ulcer.4–6 If there are relapses,
they are rarely due to re-infection and are normally
caused by non-steroidal anti-inflammatory drugs
(NSAIDs) or acetylsalicylic acid.7
Despite the similarities between gastric ulcer and
duodenal ulcer, the role of H. pylori infection in gastric
ulcer pathogenesis appears to be less well established
than in duodenal ulcer. There are differences between
1125
1126 P. MALFERTHEINER et al.
gastric ulcer and duodenal ulcer, such as a lower
prevalence of H. pylori infection in gastric ulcer
patients of 70% vs. 90% in duodenal ulcer patients.7, 8
Gastric acid secretion is often reduced in gastric ulcer,
whereas it is normal or increased in duodenal ulcer.9
The difference in acid secretion is related to the
different phenotypic expression of gastritis in gastric
ulcer vs. duodenal ulcer.10, 11 One of the most
important differences, however, is the time required
for healing of duodenal ulcer and gastric ulcer. The
reasons for the delayed healing of gastric ulcer
compared to duodenal ulcer are not clear, but an
influence exerted by the different pattern and activity
of gastritis in duodenal ulcer and gastric ulcer is a
possible explanation.12, 13
One-week proton pump inhibitor-based triple therapy
is accepted as the standard treatment for H. pylori
infection in patients with peptic ulcer.14–17 With proper
use, eradication rates of around 90% have been
achieved. For patients with uncomplicated duodenal
ulcer, the 7-day eradication treatment is sufficient for
duodenal ulcer healing if the infection has been
successfully eradicated. However, this has not been
shown for patients with gastric ulcer, in whom the
healing process is usually slower with less gastric ulcers
healed after 4 weeks. Therefore, the continued admin-
istration of proton pump inhibitor until healing is
complete is currently advised.
The aim of this study was to investigate the efficacy
of three proton pump inhibitor-based regimens [pan-
toprazole in combination with amoxicillin and met-
ronidazole (PAM), clarithromycin and metronidazole
(PCM) or amoxicillin and clarithromycin (PAC)] in the
eradication of H. pylori infection, and to determine
whether successful H. pylori eradication resulted in
the better healing of gastric ulcer. In addition, we
studied the efficacy of the eradication regimen as a
function of primary resistance and, in the case of
treatment failure, the development of resistance of
H. pylori to metronidazole, amoxicillin and clarithro-
mycin.
MATERIALS AND METHODS
This open, multi-centre, randomized, parallel-group
comparison between three treatment groups was con-
ducted in 60 centres in Germany in accordance with
Good Clinical Practice and the Declaration of Helsinki.
Fifteen independent local ethics committees approved

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135
the study and written informed consent was obtained
from all patients.
Patients
Five hundred and nineteen patients of at least 18 years
of age, with one or two endoscopically diagnosed active
gastric ulcers with a diameter of 5–20 mm and with a
positive rapid urease test, were randomized into the
study. H. pylori infection was confirmed by at least one
other positive test result (13C-urea breath test, histology,
culture testing). Major exclusion criteria were: ulcer
complications, duodenal and/or intrapyloric ulcers, a
medical history of Zollinger–Ellison syndrome, reflux
oesophagitis (grades II–IV), pyloric stenosis, subtotal
gastrectomy or vagotomy. Patients with known allergy,
especially to any study drug, malignant diseases, other
severe concurrent diseases or malfunctions or a history
of drug or alcohol abuse were also excluded. The intake
of systemic glucocorticoids or NSAIDs on more than two
consecutive days per week during the last 28 days
before the start of the study was forbidden, except for a
daily dose of up to 150 mg acetylsalicylic acid.
Treatment
H. pylori-infected patients were randomly assigned to
one of the three treatment groups: PAC, PCM or PAM for
7 days, with a twice daily dose of pantoprazole (40 mg),
amoxicillin (1000 mg), clarithromycin (500 mg) or
metronidazole (500 mg), followed by pantoprazole
(40 mg) for another 3 weeks. If the ulcer had not
healed at the control endoscopy after 4 weeks, this
antisecretory treatment was continued for another
4 weeks.
Conduct of the study
All patients were treated for a period of 28 days (or
56 days), followed by a 6-week treatment-free phase.
For each patient, a maximum of five visits was
scheduled, with two to three endoscopies carried out
in order to determine the ulcer diameter, the grade of
gastro-oesophageal reflux disease (GERD), H. pylori
status and the histological pattern of gastritis and
malignancy. Standard laboratory investigations were
performed, with demographic data, medical history,
gastrointestinal symptoms and concomitant medication
also being documented. At baseline and after the
 H. PYLORI ERADICATION AND GASTRIC ULCER HEALING
treatment-free phase, the H. pylori status was deter-
mined. Compliance and clinical symptoms were checked
at each follow-up visit, and clinical symptoms, changes
in concomitant medication and possible adverse events
were documented.
Endoscopy
Endoscopies were performed at baseline and at every
follow-up visit after starting monotherapy. The ulcer
size and location were reported exactly. Additional
abnormalities, such as reflux oesophagitis (graded
according to Savary–Miller), were also carefully des-
cribed. For the histological assessment of gastritis and
the determination of H. pylori infection, a number of
biopsies were carried out as follows: two antrum
biopsies were taken 2–4 cm pre-pylorically at the
greater and lesser curvature, two from the corpus at
the lesser curvature, one approximately 4 cm proximal
to the angulus, and one from the greater curvature
approximately 8 cm from the cardia. For the rapid
urease and culture tests, four biopsies were taken, two
each from the antrum and corpus. Malignancy was
determined from additional biopsies of the gastric ulcer.
At the final examination, three biopsies each were taken
from the antrum and corpus for the determination of
the histological parameters and H. pylori status.
Histology and determination of the H. pylori status
The activity of gastritis and the density of H. pylori
colonization were evaluated by scoring the parameters
according to the updated Sydney classification on a
four-point scale: none (0), mild (1), moderate (2), severe
(3).18 The gastritis was classified using haematoxylin
and eosin staining. The presence of H. pylori was
determined using Warthin–Starry staining. The score
for lymphoid follicles/aggregates was defined as 0
(none), 1 (aggregates), 2 (follicles), 3 (aggregates and
follicles) and 4 (not investigated). The grade of intestinal
metaplasia was categorized as none, incomplete or
complete and analysed quantitatively.
Culture and susceptibility testing of H. pylori isolates
were performed at the Institut für Medizinische Mikrobi-
ologie und Hygiene (Universitätsklinikum Freiburg, Ger-
many). H. pylori was determined by selective cultivation,
biochemical characteristics and microscopic examination
after Gram staining, as described previously.19 Suscepti-
bility testing to given antibiotics was performed using the

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135
1127
E-test (AB Biodisk, Solna, Sweden), according to Glu-
pczynski et al.20 modified after Heep et al.21 The 13C-urea
breath test was performed according to a standard
validated protocol and assayed at IMTM GmbH (Magde-
burg, Germany).22 A patient was considered to be
H. pylori positive if the D13C value exceeded 4&.
For the evaluation of the gastrointestinal symptoms of
ulcer pain (day and night), nausea, vomiting, retching,
heartburn, acid eructation and dysphagia, a score was
assigned (none, 0; mild, 1; moderate, 2; severe, 3) to
quantify the severity of each symptom at each visit.
Statistical methods
The number of patients required for the study was
calculated such that a difference of 7% in the eradica-
tion rates between the three study groups could be
detected. Based on an expected mean eradication rate of
85%, a significance level of 5% and a power of 80%, at
least 125 patients per group were required to detect this
difference for the per protocol analysis. Assuming a
drop-out rate of 25%, at least 500 patients were needed
to achieve a total of at least 375 patients.
The primary criterion of this study was the H. pylori
eradication rate. The point estimate was determined
according to the binomial distribution, and the 95%
confidence interval (CI) was determined according to
the approximation to the F-distribution, restricted to
0,100. For the comparison of eradication rates between
the three groups, the chi-squared test was used, with a
significance level of 5%.
Ulcer healing rates after 4 and 8 weeks were evaluated
in a similar manner to the eradication rate. Logistic
regression models were used for the sub-group analyses
of healing rates as a function of the intake of NSAIDs
and ulcer diameter. In addition, for the healing rates,
the odds ratios were calculated with regard to gender,
age (older vs. younger than 57 years), GERD grade I at
study entry (present vs. absent), ulcer size (5–10 mm
vs. 10–20 mm) and eradication status (eradicated vs.
not eradicated).
Dichotomous parameters (gender, smoking behaviour
and alcohol habits) were compared using Fisher’s exact
two-tailed test. Ordinal variables (age, height, weight
and body mass index) were compared using the
Kruskal–Wallis test. The median and 68% range were
calculated.
Analyses of the results were performed in three
populations. The safety population comprised all
1128 P. MALFERTHEINER et al.

screened patients with a positive rapid urease test result Table 1. Baseline characteristics
who took the study medication at least once. Patients
PCM PAC PAM
for whom a carcinoma of the stomach was diagnosed Characteristic (n ¼ 181) (n ¼ 169) (n ¼ 167)
were included in this population. The intention-to-treat
Age (years ± s.d.) 57 ± 13 57 ± 13 56 ± 13
population comprised all patients fulfilling the inclusion
Height (cm ± s.d.) 170 ± 9 169 ± 9 169 ± 8
criteria who had taken the study medication at least Weight (kg ± s.d.) 75 ± 13 75 ± 14 75 ± 14
once. Patients who completed the study according to Body mass index 26 ± 4 26 ± 5 26 ± 4
the protocol and whose H. pylori status could be (kg/m2 ± s.d.)
determined at the end of the study were included in Race, n (%)
the per protocol analysis. Patients who took antacids Caucasian 178 (98) 168 (99) 166 (99)
Other 3 (2) 1 (1) 1 (1)
and/or NSAIDs during a time interval of 28 days before Gender, n (%)
the start of the study were included in the per protocol Male 110 (61) 93 (55) 91 (54)
population, but considered separately. Female 71 (39) 76 (45) 76 (46)
Smoking status, n (%)
Yes 83 (46) 82 (48) 86 (52)
RESULTS PAC, pantoprazole–amoxicillin–clarithromycin; PAM, pantoprazole–
amoxicillin–metronidazole; PCM, pantoprazole–clarithromycin–met-
Patients ronidazole; s.d., standard deviation.
Of the 519 patients enrolled in the study, two did not
take any study medication and, in 53, the H. pylori different efficacy in eradication. Therefore, susceptibility
status could not be confirmed by another test or a tests for the three antibiotics were performed at
carcinoma of the stomach was diagnosed. Thus, 517 baseline and again at the final visit. These tests could
patients were included in the safety population and 464 only be performed in patients who were positively
patients in the intention-to-treat population. All patients tested for H. pylori, resulting in a low number of tests at
who took the medication at least once and completed the the end of the study. At study entry, tests were
study without major protocol violations were included assessable for a total of 354 patients. At this time, the
in the per protocol population, which comprised 313 majority of H. pylori-positive patients were sensitive
patients. The most prominent protocol violation was to clarithromycin and metronidazole (295, 83.3%).
absence from the final visit, and therefore information Resistance at baseline (primary resistance) was
on the H. pylori status at the end of the study was observed in 14.1% of the examined samples to
missed. These patients were categorized as ‘not eradi- metronidazole, in 1.4% to clarithromycin and in
cated’ in the intention-to-treat analysis. No substantial 1.1% to both antibiotics. No resistance to amoxicillin
differences with regard to demographic characteristics was observed (Table 2). Resistance in the three treat-
were found between the treatment groups (Table 1). ment groups was similar: 19 patients in the PCM
group, 18 in the PAC group and 17 in the PAM group
were resistant to metronidazole, and three patients in
Efficacy
each group were resistant to clarithromycin.
The eradication rates were 89% (67%) in the PAC At the end of the study, susceptibility tests were
group, 83% (68%) in the PCM group and 76% (60%) in performed in all patients (n ¼ 34) who remained
the PAM group after the treatment-free phase for per H. pylori positive after therapy. Resistance to clarithro-
protocol (intention-to-treat) analysis. The numerical mycin was observed in 2.9% (1/34), to metronidazole in
difference of about 5% in the hierarchy PAC > PCM > 50% (17/34) and to both in 23.5% (8/34) (Table 2).
PAM was significant between the PAC and PAM groups Resistance to amoxicillin was not found at baseline or at
(P < 0.05). any time point during the study.
A statistically relevant difference in the metronidazole
resistance rates was observed in the 54 cases between
Determination of resistance
males (40.7%) and females (59.3%). This was reflected
Different primary resistance rates to the antibiotics in by significantly lower eradication rates in females (74%)
the treatment groups might be the reason for the compared with males (82%) in the PAM group.

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135

 H. PYLORI ERADICATION AND GASTRIC ULCER HEALING 1129

Table 2. Determination of resistance to clarithromycin (C), met- 19 patients were not included in the following inten-
ronidazole (M) or both (C + M) at baseline and at the end of the tion-to-treat and per protocol analyses (Table 3). In
study. Susceptibility testing could only be performed in Helicob-
patients with strains resistant to at least one given
acter pylori-positive patients (n ¼ number of patients)
antibiotic relevant for the regimen, eradication rates
C M C+M were 54% and 36% in the PCM and PAM groups,
At baseline 1.4% 14.1% 1.1% respectively (per protocol). In the PAC group, only two
(Assessable in 354 patients) n (5/354) (50/354) (4/354) patients were infected with H. pylori strains resistant to
At study end 2.9% 50% 23.5% clarithromycin, and could not be cured.
(Assessable in 34 patients) n (1/34) (17/34) (8/34)

It was of interest to determine whether the observed
resistant strains were present at baseline or were
newly developed (secondary resistance). Secondary
resistance to metronidazole or clarithromycin was
observed in strains of detected in 14 patients who
had fully sensitive strains at baseline: in five patients
in the PCM group (two with new double resistance,
two with additional clarithromycin resistance and one
with additional metronidazole resistance), seven
patients in the PAM group (all to metronidazole) and
one patient in the PAC group (to metronidazole). One
patient in the PAM group with no assessable baseline
tests showed a strain which was resistant to metroni-
dazole.
It was of particular interest in this study to determine
whether eradication rates were lower in patients with
resistant strains than in those with sensitive strains. In
order to obtain useful evidence on this question,
analyses were performed only in those patients with
an antibiotic resistance relevant for their treatment
group. From the 59 patients with resistant strains at
baseline, 17 patients were resistant to metronidazole
but belonged to the PAC group, and two patients were
not included in the intention-to-treat population. These
Ulcer healing
No significant differences in the ulcer healing rates were
found between the three treatment groups (Figure 1),
either in the intention-to-treat (not shown) or per
protocol analysis. A healing rate of 91% in the PAM
group was reached after 4 weeks. The rates increased to
over 97% in all groups after a further 4 weeks, with
PAC reaching the highest rate with 100%. In the sub-
group of patients in whom susceptibility testing was
performed, the influence of H. pylori eradication on
ulcer healing was analysed. It was found that, in
patients in the intention-to-treat population in whom
H. pylori had been eradicated, ulcer healing rates were
clearly higher than in those with eradication failure.
However, in the per protocol population, the healing
rates were similar (Figure 2).
Furthermore, in order to identify additional factors
influencing ulcer healing, odds ratios were determined
with regard to gender, age, GERD status at baseline,
ulcer size and H. pylori status (Table 4). Patients with
successful eradication and an ulcer size of < 15 mm had
a significantly higher chance of ulcer healing compared
with those with unsuccessful eradication.
In 48% of patients, ulcers had a size of 5–10 mm at
baseline. Medium and large ulcers of 10–15 mm and

Table 3. Comparison of the eradication rates in patients with sensitive or resistant Helicobacter pylori strains to at least one given
antibiotic relevant to the treatment group at baseline [intention-to-treat (ITT) and per protocol (PP) populations]

Eradication rates in patients with sensitive strains (%) Eradication rates in patients with resistant strains
[95% CI] (%) [95% CI]

Population PCM PAC PAM PCM PAC PAM

ITT 69/93 (74%) 82/116 (71%) 58/90 (64%) 8/21 (38%) 0/2 (0%) 5/17 (29%)
[64–83%] [62–79%] [54–74%] [18–62%] — [10–56%]
PP 60/70 (86%) 68/76 (90%) 50/61 (82%) 7/13 (54%) * 5/14 (36%)
[75–93%] [80–95%] [70–91%] [25–81%] — [13–65%]

CI, confidence interval; PAC, pantoprazole–amoxicillin–clarithromycin; PAM, pantoprazole–amoxicillin–metronidazole; PCM, pantoprazole–

clarithromycin–metronidazole.
* No patients with resistant strains at baseline.

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135

1130 P. MALFERTHEINER et al.

100% 98% 100%

97% Relapse
90% 91%
88%
Ulcer relapse after stopping antisecretory treatment was
Ulcer healing rates (%)

80%
observed in 16 of 464 patients (3.4%) in the intention-
to-treat population (12/313 in the per protocol popu-
60%
4 weeks lation). With regard to the intention-to-treat analysis,
8 weeks

40%
eradication had a slight influence on the relapse rates,
as 3.96% of patients (12/303) in whom H. pylori had
20% been eradicated experienced ulcer relapse, compared
with a relapse rate of 2.5% in patients with unsuccessful
0% eradication. With regard to the per protocol analysis,
PCM PAC PAM
the eradication of H. pylori had no obvious influence on
Figure 1. Ulcer healing rates after 4 and 8 weeks (per protocol the relapse rates, i.e. 3.9% of patients (10/259) in
population). PAC, pantoprazole–amoxicillin–clarithromycin; whom H. pylori had been eradicated, in contrast with
PAM, pantoprazole–amoxicillin–metronidazole; PCM, pantopraz- 3.7% (2/54) in whom H. pylori had not been eradicated,
ole–clarithromycin–metronidazole. experienced an ulcer relapse within a short observation
98·2% 100·0% 98·9%
period of up to 14 weeks. Relapse rates were slightly
100%
higher in the PAM group (4.9%) than in the PAC
Eradicated
Not eradicated (4.0%) or PCM (2.7%) groups. Long-term investigations
80%
Ulcer healing rates (%)

68·6% 70·7% are ongoing to observe the patients in this study with
57·7%
remission.
60%
Although the intake of NSAIDs was not permitted
40%
during the study, 18 intention-to-treat patients who
took NSAIDs during a time interval of 28 days before
20% the start of the study were considered separately. These
patients were not statistically significantly different, but
0% lower healing rates (78%) were seen in comparison with
PCM PAC PAM patients who did not take NSAIDs (88%) at the end of
Figure 2. Ulcer healing rates in patients with successful and failed the study. The intake of NSAIDs also had no influence
eradication after 4 weeks (per protocol population). PAC, on the ulcer relapse rate or the development of new
pantoprazole–amoxicillin–clarithromycin; PAM, pantoprazole– ulcers, i.e. only one patient with ulcer relapse or a new
amoxicillin–metronidazole; PCM, pantoprazole–clarithromycin– ulcer took NSAIDs before and during the study.
metronidazole.

15–20 mm were observed in 25% and 27% of intention-

to-treat patients, respectively. The distribution of these
ulcer sizes was very similar in all three treatment groups
(data not shown). High healing rates were reached in all
ulcer categories after 8 weeks, clearly demonstrating
that the speed of healing was dependent on the ulcer
size. After 4 weeks, 84% of patients with small ulcers
(5–10 mm) were healed, compared with only 65% of
patients with large ulcers (15–20 mm) (intention-
to-treat) (Table 5). In addition, the ulcers (5–10 and
10–15 mm) of H. pylori-eradicated patients healed faster
than those of patients who were still H. pylori positive
(intention-to-treat: > 90% vs. < 70% after 4 and
8 weeks). This difference was even more prominent in
patients with large ulcers (intention-to-treat: 41% in
H. pylori-positive vs. 74% in H. pylori-negative patients).
Histological examination
The H. pylori density was equal in the antrum and the
corpus and the number of colonies decreased strongly in
both sites after treatment. It was conspicuous that the
improvement of gastritis was better in patients in whom
H. pylori had been eradicated. At baseline, signs of
chronic and active inflammation, determined by the
mean density of lymphoplasmacellular infiltrates and
the density of granulocyte infiltrates, respectively, were
more prominent in the antrum than in the corpus. The
scores for lymphatic follicles/aggregates and the atro-
phy of the gland body were also higher in the antrum
than in the corpus. In addition, the biopsies from the
antrum revealed a higher proportion of intestinal
metaplasia than did biopsies from the corpus. At the

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135

 H. PYLORI ERADICATION AND GASTRIC ULCER HEALING 1131

Table 4. Odds ratios and 95% confidence

ITT [95% CI] PP [95% CI]
intervals (CI) for ulcer healing after 4 weeks
as a function of several parameters Gender
Male/female 0.8635 [0.5554; 1.3425] 0.7717 [0.3661; 1.6264]
N.S. N.S.
Age of 57 years
Younger/older 1.2429 [0.8030; 1.9239] 1.6142 [0.7604; 3.4268]
N.S. N.S.
GERD I at baseline
No GERD/GERD I 1.4621 [0.6199; 3.4487] 1.3178 [0.2831; 6.1337]
N.S. N.S.
Ulcer size
5–10 mm/10–15 mm 1.5042 [0.8571; 2.6398] 1.0301 [0.3019; 3.5140]
N.S. N.S.
5–10 mm/15–20 mm 2.7136 [1.6276; 4.5244] 5.2965 [2.2218; 12.6262]
P < 0.05 P < 0.05
10–15 mm/15–20 mm 1.8229 [1.0308; 3.2236] 5.1415 [1.6863; 15.6764]
P < 0.05 P < 0.05
Eradication status
Eradicated/not eradicated 5.2228 [3.2797; 8.3170] 1.4794 [0.5974; 3.6638]
P < 0.05 N.S.

GERD, gastro-oesophageal reflux disease; ITT, intention-to-treat; PP, per protocol; N.S., not
significant.

Table 5. Ulcer healing rates and ulcer diameter [and 95% confidence intervals (CI)] in patients with successful Helicobacter pylori
eradication or eradication failure

H. pylori status Population 5–10 mm (%) [95% CI] 10–15 mm (%) [95% CI] 15–20 mm (%) [95% CI]

Not eradicated ITT n ¼ 84 n ¼ 43 n ¼ 34

4 weeks 56/84 (67%) [56–77%] 23/43 (54%) [38–69%] 14/34 (41%) [25–59%]
8 weeks 58/84 (69%) [58–77%] 27/43 (63%) [47–77%] 21/34 (62%) [44–78%]
PP n ¼ 31 n ¼ 12 n ¼ 11
4 weeks 28/31 (90%) [74–98%] 11/12 (92%) [61–100%] 8/11 (73%) [39–94%]
8 weeks 28/31 (90%) [74–98%] 11/12 (92%) [61–100%] 11/11 (100%) [71–100%]
Eradicated ITT n ¼ 141 n ¼ 73 n ¼ 89
4 weeks 132/141 (94%) [88–97%] 67/73 (92%) [83–97%] 66/89 (74%) [64–83%]
8 weeks 140/141 (99%) [96–100%] 72/73 (99%) [93–100%] 88/89 (99%) [94–100%]
PP n ¼ 119 n ¼ 65 n ¼ 75
4 weeks 114/119 (96%) [90–99%] 62/65 (95%) [87–99%] 58/75 (77%) [66–86%]
8 weeks 119/119 (100%) [97–100%] 64/65 (98%) [92–100%] 75/75 (100%) [95–100%]

ITT, intention-to-treat; PP, per protocol.

end of the study, the gastritis scores had decreased in
both the antrum and the corpus due to the suppression
of inflammatory processes caused by H. pylori infection.
The degree of both atrophy and metaplasia remained
unaffected by treatment. However, most of the gastritis
parameters improved during the trial period (Table 6),
and it can be concluded that treatment success was
similar for both antrum and corpus gastritis.
In the absence of endoscopic criteria for suspected
neoplasia, gastric malignancy was diagnosed by histol-
ogy at the initial visit in a total of 25 patients (4.8% of
the safety population).
Gastro-oesophageal reflux disease
At baseline, no oesophagitis was observed in 294
patients (per protocol: 94%) and oesophagitis of grade
I in 16 patients (per protocol: 5.1%). During the first
4 weeks of therapy, improvement of GERD grade I to
normal was observed in 11 patients. Only in two

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135

1132 P. MALFERTHEINER et al.

Table 6. Histological evaluation of gastritis

Antrum Mean Corpus Mean
in the 310 patients in the per protocol
Parameter Phase scores ± s.d. scores ± s.d.
population*
H. pylori colonization Pre 2.09 ± 0.91 2.04 ± 0.70
Post 0.20 ± 0.62 0.19 ± 0.55
Chronic inflammation Pre 2.12 ± 0.51 1.60 ± 0.59
Post 1.44 ± 0.57 1.08 ± 0.56
Active inflammation Pre 1.86 ± 0.75 1.43 ± 0.77
Post 0.17 ± 0.49 0.15 ± 0.48
Atrophy of gland body Pre 0.38 ± 0.78 0.12 ± 0.47
Post 0.39 ± 0.77 0.11 ± 0.43
Lymphatic follicles/aggregates Pre 0.87 ± 1.03 0.35 ± 0.69
Post 0.52 ± 0.80 0.24 ± 0.56
Incomplete metaplasia  (%) Pre 7.5 1.4
Post 5.0 1.0
Complete metaplasia  (%) Pre 25.6 8.1
Post 25.4 8.8

s.d., standard deviation.

* Drop-outs were excluded from the analysis.
  Intestinal metaplasia was assessed only qualitatively, and therefore was not scored.

patients, who initially showed no signs of GERD, was

grade I oesophagitis diagnosed after 4 weeks. None of
the patients showed a deterioration of GERD to higher
grades, i.e. grades II–IV. Six weeks after the end of
treatment (final visit), six patients who initially had no
oesophagitis showed GERD grade I, one patient deteri-
orated to GERD grade III and one patient changed from
grade I to II. These eight patients who showed a
deterioration of GERD grade between the initial and
final visit were all H. pylori negative at the end of the
study. In total, 297 patients (per protocol: 95%) showed
no signs of GERD and 11 patients (per protocol: 3.5%)
showed signs of grade I at the final visit. Thus, the
results of this trial indicate that successful H. pylori
eradication therapy does not have a promoting effect on
GERD.
Patients were asked about gastrointestinal symptoms
at each visit. These were evaluated by assigning a
score for each symptom (data not shown). The
symptom with the highest score in all treatment
groups was ulcer pain during the daytime, followed
by ulcer pain at night, nausea and heartburn. The
scores of all investigated symptoms decreased to
nearly zero during the study course, and there were
only small differences between the treatment regi-
mens. There was no association between the deteri-
oration of GERD and specific symptoms, as the eight
patients with deterioration of GERD experienced all
kinds of symptoms.
Adverse events
Safety data were available for 517 patients and 37%
(n ¼ 193) reported adverse events. Most (21%,
n ¼ 111) occurred during triple therapy, with
13% (n ¼ 67) during monotherapy and 9% (n ¼ 49)
during the drug-free period. The majority of adverse
events were considered to be unrelated or unlikely to be
related to the administration of the study medication.
Only 19 symptoms (6%) were considered as likely to be
related to pantoprazole and no symptom as definitely
related to pantoprazole. On the other hand, 105
symptoms (31%) were assessed as likely to be related
to the intake of antibiotics and two symptoms as
definitely related to the intake of antibiotics. Twenty-
five serious adverse events in 21 patients were observed
during the trial, but only four of these were assessed as
likely to be related to the medication. No relevant
differences in the safety profile of the three treatment
groups with regard to the overall incidence of adverse
events were apparent. The most frequently observed
adverse event in the PCM group was influenza-like
symptoms (4%), and diarrhoea in the PAC and PAM
groups (12% and 5%, respectively).
Compliance
Compliance was assessed by tablet count. During the
7-day triple therapy, 92% of patients were compliant,

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135

 H. PYLORI ERADICATION AND GASTRIC ULCER HEALING
with 89% taking the following pantoprazole mono-
therapy, i.e. more than 80% of the study medication.
DISCUSSION
In this study, the H. pylori eradication rates in patients
with gastric ulcer were 76% with PAM, 83% with PCM
and 89% with PAC. These results from our large trial
confirm previous findings observed in patients with
gastric ulcer from smaller studies.2, 3 The eradication
efficacy found in patients with gastric ulcer is no
different from that reported in large trials conducted in
patients with duodenal ulcer.4–6 The combination of a
proton pump inhibitor with clarithromycin and either
amoxicillin or metronidazole is superior to the combi-
nation of a proton pump inhibitor with amoxicillin and
metronidazole.2, 6
Primary resistance was most frequent to metronidazole
(14%) and rare to clarithromycin (1.4%), with the
expected impact on the eradication rates and a clear
therapeutic superiority of PCM vs. PAM. In contrast
with other European countries, primary clarithromycin
resistance is still low in Germany and amoxicillin
resistance has not been observed.24 Following treatment
failure, resistance to metronidazole was detected in
73.5% (25/34) and secondary clarithromycin resist-
ance occurred in 26.5% (9/34). In eight of nine cases,
clarithromycin resistance was accompanied by a sim-
ultaneous resistance to metronidazole. The proportion
of secondary clarithromycin resistance was at the lower
limit of post-treatment clarithromycin resistance rates,
which often have been found to be of the order of 30–
70%.25, 26 In vitro exposure of metronidazole-resistant
H. pylori strains to metronidazole has been reported to
be accompanied by a dramatic increase in their
mutation rate, thus favouring base exchanges, which
are also involved in clarithromycin resistance.27 The
low rate of secondary clarithromycin resistance
observed in this study is likely to be due to the low
proportion of H. pylori strains with primary resistance to
metronidazole.
The ulcer healing rates after 4 weeks in the three
treatment groups were similar. However, patients with
successful H. pylori eradication — independent of the
administered treatment regimen — showed significantly
higher healing rates after 4 weeks than those with
persistent infection. This adds important information to
the healing mechanism of gastric ulcers, as so far
convincing evidence exists only for a tight correlation

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135
1133
between the power of acid suppression and the speed of
ulcer healing. Recently, based on a meta-analysis,
Huang and Hunt have shown an additional effect of
antibiotics on the acceleration of duodenal ulcer
healing, but this has never been demonstrated to be
the case in gastric ulcer healing.28 In this study, all
patients continued to receive strong acid inhibition with
pantoprazole (40 mg/day) following the 7-day eradica-
tion regimen. Therefore, the observed difference in ulcer
healing between those with and without persistent
infection at 4 weeks clearly indicates an additional
positive effect of H. pylori eradication, which is probably
mediated by healing of the mucosal inflammation.
Indeed, active inflammation was completely reversed
in those with successful eradication by week 4.
The second crucial factor for ulcer healing identified
in our study was the ulcer size. The critical ulcer size
predicting delayed healing is a diameter of more than
15 mm. In terms of the clinical consequence of this
observation, gastric ulcers larger than 15 mm should
be considered for prolonged acid inhibitory therapy.
Even in the case of successful eradication, an ulcer size
larger than 15 mm suggests that treatment for a
longer period should be employed. In clinical practice,
ulcer size is easier to use as a predictor of the speed of
ulcer healing than is the success of eradication therapy.
Successful eradication can be determined only later,
after having stopped eradication therapy, and, in
practical terms, is not useful for deciding whether acid
inhibitory therapy should be continued beyond the
phase of eradication.
A comment needs to be dedicated to the phenotypic
expression of gastritis in conjunction with gastric ulcer.
Although, in patients with duodenal ulcer disease,
gastric inflammation is characteristically antrum pre-
dominant, the gastritis pattern in gastric ulcer varies
from corpus predominant to pangastritis, and even a
slightly greater involvement of the antrum has been
reported.2, 7, 12, 13 In this large series of patients with
gastric ulcer, the density of granulocyte infiltrates
was slightly higher in the antrum than in the corpus
mucosa at baseline. This finding indicates that, in
H. pylori-induced gastric ulcer, global involvement of
the gastric mucosa usually takes place, with slightly less
histological damage to the acid-secreting gastric com-
partments. Although active inflammation improved
dramatically or even disappeared in both the antrum
and the corpus from the first control visit onwards, the
elements of chronic gastritis (assessed by the density of
1134 P. MALFERTHEINER et al.
lymphoplasmacellular infiltrates), although reduced,
persisted in all patients during the follow-up period.
In agreement with other authors, our study confirms
that mucosal atrophy and intestinal metaplasia are
more distinctive in the antrum than in the corpus, and
more than twice as frequent in the antrum than in the
corpus mucosa.13 For both parameters, there were no
significant changes after therapy, and similar results
have also been found in duodenal ulcer studies after
treatment with omeprazole-based triple therapy.23 This
suggests that a complete regeneration of atrophic
mucosa, if it occurs, may take a long time.
NSAIDs are another important aetiological factor in
the development of gastric ulcer, and therefore the use
of NSAIDs was not permitted in this study. Some
patients, however, took NSAIDs before or during the
study; the separate analysis of these patients showed an
expected lower ulcer healing rate than in patients with
no NSAID intake.
Patients with malignant ulcers are frequently reported
to have a higher prevalence of incomplete and complete
metaplasia.18 Importantly, 25 gastric neoplasms were
found at the inclusion visit, emphasizing that the benign
appearance of gastric ulcer should never be taken as an
indication of its benign nature. The fact that, during the
study, another four cases of gastric cancer were
diagnosed further emphasizes the need to perform an
additional endoscopy and histological evaluation in
patients with gastric ulcer to ensure ulcer healing and
confirm the benign nature of the lesion.
With regard to symptomatology in the short term,
ulcer pain and the typical symptoms of GERD, such as
heartburn, improved following H. pylori eradication.
This finding needs to be followed over time, as H. pylori
eradication itself has been reported to provoke the
development of GERD in duodenal ulcer patients.29
In conclusion, the lessons learned from this large trial
on gastric ulcer patients are that H. pylori eradication is
highly effective for gastric ulcer healing, and accelerates
the gastric ulcer healing process in comparison with
acid suppressive therapy alone. Treatment with the PAC
regimen was more effective than treatment with PCM,
and both of these regimens had a significantly higher
eradication rate than PAM. Factors determining the
accelerated healing of gastric ulcer are successful
H. pylori eradication and an ulcer size of < 15 mm.
The clinical impact of this experience may direct us to
shorten or prolong the use of proton pump inhibitor
therapy beyond the phase of eradication.

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135
ACKNOWLEDGEMENTS
We are grateful to Altana, Konstanz, Germany, for
support of the study. We thank the IFE Institut für
Forschung und Entwicklung, Witten, Germany, for
monitoring at the study sites, data collection, source
data verification and preparation of the manuscript.
This work was supported by ALTANA Pharma AG,
Konstanz, Germany, who also provided the study
medication.
To all investigators collaborating in the study, we
express our gratitude (alphabetical order): K.-D. Aßmus
(Nürnberg), P. Balig (Freiburg), J. Bauer (Leipzig), B.
Birkner (München), B. Bokemeyer (Minden), H. Bordel
(Ostercappeln), J. Bott (Elmshorn), M. Braschke (Celle),
W. Burmeister (Hamburg), R. Daikeler (Sinsheim), H.
Daus (Ludwigslust), A. Dettmer (München), E. Dom-
browski (Berlin), U. Ehrle (Pfungstadt), T. Eisenbach
(Duisburg), H. Eisold (Mössingen), H. Eschenburg
(Güstrow), J. Fiedler (Magdeburg), R. Fink (Freising),
O. Friedrichs (Duisburg), R. Goes (Ditzingen), M.
Grothoff (Ahlen), A. Halbach (Hannover), J. Hein
(Marburg), G. Heptner (Dresden), K. Hey (Paderborn),
A. Hoffmann (Magdeburg), H. Jürgens (Oelde), W.
Karlas (Nürnberg), H.-U. Kellner (Kassel), C. Klein
(Künzing), B. Knapp (Siegen), A. Kocjan (Lüdenscheid),
S. Kolbe (Bad Berleburg), W. Krämer (Pirmasens), H.-G.
Krezdorn (München), K. Lenz (Isny), L. Lerche (Lübeck),
A. Mares (Frankfurt), B. Marowski (Berlin), A. Petrides
(Bochum), K. Rupp (Ellwangen), A. Ryschka (Berlin),
A. Sammler (Friedrichsthal), J. Schmitz (Fürth),
M. Schumacher (Wolmirstedt), D. Sehland (Rostock),
R. Sievers (Bederkesa), B. Stölzle (Lindau), W. Tenbieg
(Darmstadt), C. Völker (München), E. von Fritsch
(Erlangen), R. Warncke (Elmshorn), J. Wehnert (Dres-
den), J. Weismüller (Koblenz), R. Wendland (Bergneus-
tadt), P. Witzany (Marktheidenfeld), K. Ziegler (Berlin),
W. Zink (Nürnberg).
REFERENCES
1 Penston JG, McColl KE. Eradication of Helicobacter pylori: an
objective assessment of current therapies. Br J Clin Pharmacol
1997; 43: 223–43.
2 Malfertheiner P, Bayerdörffer E, Diete U, et al. The GU-MACH
study: the effect of 1-week omeprazole triple therapy on
Helicobacter pylori infection in patients with gastric ulcer.
Aliment Pharmacol Ther 1999; 13: 703–12.
3 Axon ATR, O’Morain CA, Bardhan KD, et al. Randomised
double blind controlled study of recurrence of gastric ulcer
 H. PYLORI ERADICATION AND GASTRIC ULCER HEALING
after treatment for eradication of Helicobacter pylori infection.
Br Med J 1997; 314: 565–8.
4 Bayerdörffer E, Miehlke S, Mannes GA, et al. Double-blind trial
of omeprazole and amoxicillin to cure Helicobacter pylori
infection in patients with duodenal ulcers. Gastroenterology
1995; 108: 1412–7.
5 Labenz J, Tillenburg B, Peitz U, et al. Efficacy of omeprazole
one year after cure of Helicobacter pylori infection in duodenal
ulcer patients. Am J Gastroenterol 1997; 92: 576–81.
6 Lind T, Veldhuyzen van Zanten S, Unge P, et al. Eradication of
Helicobacter pylori using one-week triple therapies combining
omeprazole with two antimicrobials: the MACH I study.
Helicobacter 1996; 1: 138–44.
7 Malfertheiner P, Leodolter A, Peitz U. Cure of Helicobacter
pylori-associated ulcer disease through eradication. Baillieres
Best Pract Res Clin Gastroenterol 2000; 14: 119–32.
8 Marshall BJ. Helicobacter pylori. Am J Gastroenterol 1994; 89:
5116–28.
9 Ruiz B, Correa P, Fontham ET, Ramakrishnan T. Antral
atrophy, Helicobacter pylori colonization, and gastric pH. Am J
Clin Pathol 1996; 105: 96–101.
10 Dixon MF. Patterns of inflammation linked to ulcer disease.
Baillieres Best Pract Res Clin Gastroenterol 2000; 14(1):
27–40.
11 Sipponen P, Hyvarinen H, Siurala M. H. pylori corpus gastritis
— relation to acid output. J Physiol Pharmacol 1996; 47(1):
151–9.
12 Sipponen P, Seppälä K, Äärynen M, Helske T, Kettunen P.
Chronic gastritis and gastroduodenal ulcer: a case control
study on risk of coexisting duodenal or gastric ulcer in
patients with gastritis. Gut 1998; 30: 922–9.
13 Schultze V, Hackelsberger A, Günther T, Miehlke S, Roessner
A, Malfertheiner P. Differing patterns of Helicobacter pylori
gastritis in patients with duodenal, prepyloric, and gastric
ulcer disease. Scand J Gastroenterol 1998; 33: 137–42.
14 Current European concepts in the management of Helicobacter
pylori infection. The Maastricht Consensus Report. Gut 1997;
41: 8–13.
15 Hunt RH, Thomson ABR. Canadian Helicobacter pylori Con-
sensus Conference. Can J Gastroenterol 1998; 12: 31–41.
16 Lam SK, Talley NJ. Report of the 1997 Asia Pacific Consensus
Conference on the management of Helicobacter pylori infec-
tion. J Gastroenterol Hepatol 1998; 13: 1–12.
17 Malfertheiner P. The Maastricht recommendations and their
impact on general practice. Eur J Gastroenterol Hepatol 1999;
11: S63–7, discussion S73.
18 Dixon MF, Genta RM, Yardley JH, Correa P. Classification
and grading of gastritis. The updated Sydney System.

2003 Blackwell Publishing Ltd, Aliment Pharmacol Ther 17, 1125–1135
1135
International Workshop on the Histopathology of Gastritis,
Houston, 1994. Am J Surg Pathol 1996; 20: 1161–81.
19 Labenz J, Stolte M, Aygen S, Hennemann O, Bertrams J,
Borsch G. Qualitative and semiquantitative invasive and
noninvasive diagnosis of Helicobacter pylori colonization of
gastric mucosa. Z Gastroenterol 1993; 31: 437–43.
20 Glupczynski Y, Labme M, Hansen W, Crokaert F, Youras-
sowsky E. Evaluation of the E test for quantitative antimi-
crobial susceptibility testing of Helicobacter pylori. J Clin
Microbiol 1991; 29: 2072–5.
21 Heep M, Kist M, Strobel S, Beck D, Lehn N. Secondary
resistance among 554 isolates of Helicobacter pylori after failure
of therapy. Eur J Clin Microbiol Infect Dis 2000; 19: 538–41.
22 Leodolter A, Dominguez-Muños JA, von Arnim U, Kahl A,
Peitz U, Malfertheiner P. Validity of a modified 13C-urea breath
test for pre- and post-treatment diagnosis of Helicobacter pylori
infection in the routine clinical setting. Am J Gastroenterol
1999; 94: 2100–4.
23 Veldhuyzen van Zanten SJO, Bradette M, Farley A, et al. The
DU-MACH study: eradication of Helicobacter pylori, ulcer
healing and relapse in duodenal ulcer patients. Omeprazole
and clarithromycin in combination with either amoxicillin or
metronidazole. Gut 1997; 41: A103(Abstract).
24 Megraud F, Lehn N, Lind T, et al. Antimicrobial susceptibility
testing of Helicobacter pylori in a large multicenter trial: the
MACH 2 study. Antimicrobial Agents Chemother 1999; 43:
2747–52.
25 Lind T, Megraud F, Unge P, et al. The MACH2 study: role of
omeprazole in eradication of Helicobacter pylori with 1-week
triple therapies. Gastroenterology 1999; 116: 248–53.
26 Kist M, Strobel S, Foelsch UR, et al. Prospective assessment
of the impact of primary antimicrobial resistances on cure
rates of Helicobacter pylori infections. Gut 1997; 41: A190
(Abstract).
27 Sisson G, Jeong JY, Goodwin A, et al. Metronidazole activation
is mutagenic and causes DNA fragmentation in Helicobacter
pylori and in Escherichia coli containing a cloned H. pylori
RdxA(+) (Nitroreductase) gene. J Bacteriol 2000; 182:
5091–5.
28 Huang J, Hunt RH. Reply — the importance of clarithromycin
dose in the management of Helicobacter pylori infection: a
meta-analysis of triple therapies with a PPI, clarithromycin
and amoxicillin or metronidazole. Aliment Pharmacol Ther
1999; 13(12): 1693–4.
29 Labenz J, Blum AL, Bayerdörffer E, Meining A, Stolte M,
Börsch G. Curing Helicobacter pylori infection in patients with
duodenal ulcer may provoke reflux esophagitis. Gastroenter-
ology 1997; 112: 1442–7.