Alimentary Pharmacology & Therapeutics

Risk factors for progression to gastric neoplastic lesions in

patients with atrophic gastritis
L. VANNELLA*, E. LAHNER*, J. OSBORN , C. BORDIà, M. MIGLIONE*, G. DELLE FAVE* &
B. ANNIBALE*

*Department of Digestive and Liver SUMMARY

Disease, Sant’Andrea Hospital, II
School of Medicine, University Sapi- Background
enza of Rome, Rome, Italy;  Depart-
Atrophic gastritis, involving the gastric body mucosa, predisposes to
ment of Public Health Sciences,
University Sapienza of Rome, Rome, gastric neoplastic lesions (GNL). However, regular gastroscopic-histolog-
Italy; àDepartment of Pathology, Uni- ical follow-up for GNL is not recommended for patients with atrophic
versity of Parma, Parma, Italy gastritis.
Correspondence to: Aim
Dr B. Annibale, Department of
Digestive and Liver Disease,
To evaluate risk factors for the progression to GNL in a cohort of
Sant’Andrea Hospital, Via di patients with atrophic gastritis.
Grottarossa 1035, 00189 Rome, Italy.
E-mail: bruno.annibale@uniroma1.it Methods
A total of 300 patients with atrophic gastritis [205 women, aged 54
Publication data (18–78) years] underwent gastroscopy with six gastric antrum and body
Submitted 16 November 2009 biopsies. All patients had at least one follow-up gastroscopy ⁄ histology
First decision 2 December 2009 at an interval of at least 1 year after the atrophic gastritis diagnosis.
Resubmitted 2 February 2010
Baseline clinical and histological features were analysed as risk factors
Accepted 15 February 2010
Epub Accepted Article 18 February for the development of GNL by Cox-regression.
2010
Results
During a median follow-up of 4.3 (1–16.5) years, 15 GNL were detected
in 14 of the 300 patients with atrophic gastritis: three were gastric can-
cer, whereas 12 were non-invasive neoplasia. The annual incidence for
GNL was 1%. Cox-regression analysis identified the following risk fac-
tors: age over 50 years (HR 8.8, 95%CI 1.2–68.4), atrophic pangastritis
(HR 4.5, 95% CI 1.5–14.1) and severe intestinal metaplasia in the gastric
body (HR 4.0, 95% CI 1.3–11.8).

Conclusions
Atrophic pangastritis, severe body intestinal metaplasia and ⁄ or age over
50 years increase the risk for developing GNL in patients with atrophic
gastritis. In this subset of patients, an endoscopic-histological follow-up
for GNL surveillance may be worthwhile.

Aliment Pharmacol Ther 31, 1042–1050

1042 ª 2010 Blackwell Publishing Ltd

doi:10.1111/j.1365-2036.2010.04268.x
 R I S K O F N E O P L A S I A I N A T R O P H I C G A S T R I T I S 1043

were informed about the increased risk of GC associ-

INTRODUCTION
ated with AG and about the need to undergo regular
Gastric cancer (GC) is the fourth most common cancer follow-up by gastroscopy.
worldwide and the second cause of cancer-related Inclusion criteria were the presence of AG and at
death.1, 2 Nowadays, it is accepted that the carcinogen- least one follow-up investigation by gastroscopy with
esis involves a multistep progression from Helicobacter biopsies at an interval of at least 1 year after the diag-
pylori-related chronic inflammation of the gastric nosis of AG. As shown in Figure 1, 103 (25.6%)
mucosa, to atrophic gastritis (AG), intestinal metapla- patients of the baseline population dropped out during
sia, dysplasia and, finally, intestinal-type GC.3 Other follow-up and the final follow-up population com-
authors have proposed the concept of ‘gastritis of the prised 300 patients (205 women, median age 54 years,
carcinoma phenotype’ showing that the corpus-pre- range 18–78 years).
dominant gastritis is ‘per se’ a condition at increased
risk for the development of GC.4–6 When the gastritis
Diagnosis of atrophic gastritis and gastric
involves the gastric body, pathophysiological changes,
neoplastic lesions
such as increased pH, reduced ascorbic acid and
scavenging of nitrites and other potential carcinogenic The diagnosis of AG was based on the histological
substances, potentially lead to the development of confirmation of gastric body atrophy and the pres-
gastric neoplastic lesions (GNL).7, 8 ence of fasting gastrin above upper normal values.15
Atrophic gastritis, involving the gastric body All patients underwent gastroscopy with three biop-
mucosa, is considered a precursor condition for GC sies taken from the gastric antrum (within 3 cm of
and non-invasive neoplasia (NiN).3, 9 Previous studies the pyloric ring, lesser and greater curve, anterior or
reported a varying progression rate of AG to GC up to posterior wall) and three biopsies from the midbody
2% per year, while data on progression of AG to NiN along the greater curve.16 In case of lesion, addi-
are scarce.10–12 However, these studies were mostly tional biopsies were obtained. The degree of gastritis
performed on small series of patients and based on was assessed according to the updated Sydney
older histological classifications. System.17 Atrophy of the gastric body mucosa was
In patients with AG, regular gastroscopic follow-up defined as focal or complete replacement of oxyntic
for GC surveillance is not considered cost-effective in glands by metaplastic pyloric or intestinal glands.
Western countries and not recommended.13 As sug- Atrophy of the antral mucosa was defined as focal
gested in a recent review, to evaluate the exact value or complete disappearance of antral glands or their
of surveillance in patients with AG and establish fol- replacement by intestinal metaplastic epithelium.15
low-up frequencies, more precise data, preferably The definitions of GNL (NiN and GC) were based on
obtained in large prospective studies with adequate the Padova International Classification and WHO
follow-up, are needed.14 Thus, the aim of the present guidelines.18, 19 All biopsies were examined indepen-
study was to evaluate in a prospective cohort of AG dently by two experienced pathologists (CB, MM),
patients risk factors for the progression to gastric neo- unaware of the clinical data of the patients. In the
plastic lesions. event of disagreement, the biopsies were re-examined
simultaneously by both pathologists until agreement
was reached.15
METHODS
In all patients, fasting gastrin levels and Pepsinogen I
were measured. Gastrin levels were evaluated by a spe-
Patients and study design
cific radioimmunoassay (RIA: normal values £40
Between January 1992 and June 2008, as the result of pg ⁄ mL); Pepsinogen I was measured using a commer-
a screening programme for atrophic body gastritis, cial RIA kit (Pepsik; Sorin, Saluggia, Italy; normal
we diagnosed this condition in 403 consecutive values 20–80 ng ⁄ mL).15 The presence of anaemia was
patients (269 women, median age 55 years, range evaluated in all AG patients: pernicious anaemia was
18–95 years).11, 15 At diagnosis of AG, all patients defined as low haemoglobin concentration, MCV>100 fl
underwent a clinical interview based on a structured together with low cobalamin levels and iron deficiency
questionnaire to assess personal and clinical data, anaemia was defined as low haemoglobin concen-
family history for GC and smoking habit. All patients tration, MCV < 80 fl and ferritin < 30 ng ⁄ mL. 20, 21

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd
1044 L . V A N N E L L A et al.

n = 403 patients with atrophic gastritis referred to the In the case of H. pylori eradication treatment, a gas-
Department of digestive and liver diseases between troscopy ⁄ histology was performed at 6 months to
January 1992 and June 2008 evaluate the efficacy of cure and then patients were
monitored by the general follow-up protocol.
If NiN was detected, an extensive bioptic sampling
of at least ten biopsies of the gastric mucosa was taken
Patient not eligible for follow-up because he:
n=1 after 6 and 12 months for low-grade NiN and after 3,
• Had gastric cancer at 1st gastroscopy
6 and 12 months for high-grade NiN.23 If NiN was not
confirmed at two following gastroscopies, patients
were followed up by the general follow-up protocol.
n = 402 patients with atrophic gastritis eligible for When gastric polyps were detected, polypectomy
follow-up
was performed. In the case of a histological diagnosis
of GC, the indication for mucosectomy and ⁄ or surgery
was evaluated.
Patients lost at follow-up because they: At any time, when patients referred the onset of
• Died for other causes n=4 alarm symptoms, such as dysphagia, vomiting, dys-
• Refused any follow-up or not traceable n = 98
pepsia, and ⁄ or recent onset or worsening of iron defi-
ciency anaemia, gastroscopy was performed
immediately.
n = 300 (74.4%) patients with atrophic gastritis included
in the follow-up population of the study with follow-up
of at least 1 year Data analysis and statistical evaluation
The occurrence of GNL in the baseline population
Figure 1. Study population. and in the follow-up population of AG patients was
evaluated and data were expressed as prevalence and
incidence rate (person-year). Standard descriptive sta-
tistics were expressed as median and range or, when
Assessment of Helicobacter pylori status and
appropriate, absolute counts and percentages. Inter-
eradication
vals between initial AG diagnosis and GNL diagnosis
Helicobacter pylori infection was considered positive were analysed by Kaplan–Meier analysis. Sydney sys-
when the bacterium (Giemsa stain) and ⁄ or the presence tem scores from AG patients who developed GNL
of neutrophil granulocytes were detected at histology during follow-up and those who did not were analy-
and ⁄ or on the basis of a positive IgG titre for sed using Fisher’s exact test. Histological data were
H. pylori.11 Helicobacter pylori IgG antibodies were analysed also according to OLGA System,24 which
determined using ELISA commercial kit (GAP test IgG, describes the extension of atrophy with or without
Biorad, Milan, Italy). Helicobacter pylori positive AG metaplasia in both antral and body gastric mucosa.
patients received triple therapy (bismuth-based triple The scores of atrophy from antral and body gastric
regimens). The absence of H. pylori and neutrophil mucosa are combined using a scale ranging from 0
granulocytes at histology and a decrease by at least (absence of atrophy and metaplasia) to IV (atrophy
50% in the initial titre of H. pylori IgG was considered involving antrum and gastric body). Univariate and
as a criterion for the successful cure of infection.16 multivariate Cox-regression analysis was performed
Overall, 192 (64%) of patients were treated and to identify risk factors for progression to GNL. The
successful cure of infection was achieved in 62.5%. incident GNL were coded with value 1 in the binary
dependent variable. The following characteristics were
considered as covariates: age over 50 years, gender,
Follow-up protocol
presence of atrophic pangastritis defined as the con-
In general, the first follow-up gastroscopy was sched- comitant presence of atrophy in the gastric antrum
uled at intervals between 2 and 4 years after diagnosis and body, presence of severe intestinal metaplasia in
of AG, during which the same baseline bioptic sam- the gastric body, presence of intestinal metaplasia in
pling was repeated.22 the gastric antrum, presence of stage IV of OLGA

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd
 R I S K O F N E O P L A S I A I N A T R O P H I C G A S T R I T I S 1045

system, presence of pernicious anaemia, presence of follow-up, whereas the two female patients had a pre-
H. pylori. For each covariate, the hazard ratio (HR) vious diagnosis of low-grade NiN 1 year before diag-
and 95% CI are reported. Cox-regression analysis was nosis of GC. The stage of the GC was T2N1M0 in the
run using MedCalc version 10.1.2, Mariakerke, Bel- male patient and T1N0M0 in the two female patients.
gium. A P value of less than 0.05 was considered In the first case, follow-up gastroscopy was performed
statistically significant. because of the new onset of iron deficiency anaemia
and dyspepsia; in the other two cases, the diagnosis of
GC was yielded at 12 months after the diagnosis of
RESULTS
low-grade NiN (one present at the time of diagnosis
of AG and the other diagnosed at 4-year-follow-up).
Baseline
After gastrectomy, the male patient underwent chemo-
At the time of diagnosis of AG, ten GNL (2.5%) therapy and died of neoplastic disease progression,
were detected in ten of the 403 AG patients [6 one female patient died of complications after surgery
women, median age 62 (range 49–77) years]. In one and one female patient is alive and free of neoplasia
patient (0.2%), a GC was diagnosed and the remain- to date.
ing nine patients had a diagnosis of NiN (2.3%), Of the 12 NiN, eight were located in the gastric
which was low-grade in all but one. In detail, the antrum and four in the body, all but one were low-
patient (male, 63 years of age) with intestinal-type grade, four were polyps and eight were detected in
adenocarcinoma (T2N0M0) had presented with iron apparently normal mucosa. Seven low-grade NiN on
deficiency anaemia and an ulcerative lesion in the apparently normal gastric mucosa were confirmed at
gastric body in the context of a OLGA stage 3 AG following controls, and all patients are alive and free
with severe intestinal metaplasia. The patient under- of neoplasia at the longest follow-up available. After
went gastrectomy and is alive. As regards NiN, eight polypectomy, one low-grade and one high-grade
patients had a low-grade NiN: in six of them, the NiN recurred in two patients; also these patients are
NIN were detected in apparently normal mucosa in alive and free of neoplasia at the longest follow-up
the gastric antrum and in two patients, adenomas available.
with low-grade dysplasia (<1 cm) were found in the
gastric body. One patient had an adenoma with
Risk factors
high-grade dysplasia (3 cm) located in the gastric
antrum. Figure 3 shows the comparison of baseline atrophy
and intestinal metaplasia scores according to Sydney
System between AG patients who developed GNL
Follow-up
During an overall median follow-up of 4.3 years
(range 1–16.5 years), 15 (5.0%) GNL were detected in 100
Gastric neoplastic lesions (%)

14 of the 300 AG patients [7 female, median age 57

(range 51–77 years)]: three GNL were GC, whereas the 80
Proportion free of

remaining 12 GNL were NiN. Thus, an annual inci-

60
dence rate (person-year) of 1% for all GNL and of
0.2% for GC was observed. Figure 2 shows the Kap- 40
lan–Meier curves indicating the intervals between ini-
tial AG diagnosis and GNL diagnosis. 20

As shown in the Table 1, all three GC were intesti- 0

nal-type adenocarcinomas located in the gastric 0 5 10 15 20
antrum and were diagnosed 8 years, 5 years and Years of follow-up
1 year after diagnosis of AG in one male (aged
51 years) and two female patients (aged 51 and Figure 2. Kaplan–Meier analysis: Intervals between initial
77 years) respectively. The male patient had a negative diagnosis of atrophic gastritis and diagnosis of gastric
neoplastic lesions. Proportion of patients with atrophic
follow-up gastroscopy 2 years before diagnosis of
gastritis who remained free of gastric neoplastic lesions.
GC and had had no evidence of GNL during previous

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd
 Table 1. Patients with atrophic gastritis who developed neoplastic lesions during the follow-up
1046 L . V A N N E L L A et al.

Age ⁄ gender Macroscopic Neoplastic lesions

patients lesions Localization Histology Intervention during follow-up Outcome

51 F Normal mucosa Body Low-grade dysplasia Monitoring Confirmed dysplasia Free of neoplastic lesions
56 M Multiple polyps Body Low-grade dysplasia Monitoring Not recurrence Free of neoplastic lesions
of 0.5 cm
66 F Normal mucosa Antrum Low-grade dysplasia Monitoring Confirmed dysplasia Free of neoplastic lesions
65 M Normal mucosa Antrum Low-grade dysplasia Monitoring Confirmed dysplasia Free of neoplastic lesions
57 M Normal mucosa Antrum Low-grade dysplasia Monitoring Confirmed dysplasia Free of neoplastic lesions
51 M Ulcer Antrum Intestinal Gastrectomy
adenocarcinoma
68 F Normal mucosa Antrum Low-grade dysplasia Monitoring Not recurrence Free of neoplastic lesions
68 M Normal mucosa Antrum Low-grade dysplasia Monitoring Confirmed dysplasia Free of neoplastic lesions
55 M Normal mucosa Body Low-grade dysplasia Monitoring Confirmed dysplasia Free of Neoplastic lesions
71 F Normal mucosa Antrum Low-grade dysplasia Monitoring Confirmed dysplasia Free of Neoplastic lesions
51 F Multiple polyps Body Low-grade dysplasia Polypectomy Intestinal adenocarcarcinoma*
of 0.2–0.4 cm
57 M Single polyp of 3 cm Antrum High-grade dysplasia Polypectomy Recurrence dysplasia Free of neoplastic lesions
54 F Single polyp of 0.7 cm Antrum Low-grade dysplasia Polypectomy Recurrence dysplasia Free of neoplastic lesions
77 F Normal mucosa Antrum Low-grade dysplasia Monitoring Progression Intestinal adenocarcinoma 

* Adenocarcinoma and two areas of high-grade dysplasia were found on antral mucosa. Patient performed gastrectomy and is alive.
  A focus of adenocarcinoma and two areas of high-grade dysplasia on normal mucosa were found at 12 months. Patient performed gastrectomy and died of complica-
tions post-surgery.

ª 2010 Blackwell Publishing Ltd

Aliment Pharmacol Ther 31, 1042–1050
 R I S K O F N E O P L A S I A I N A T R O P H I C G A S T R I T I S 1047

during follow-up (n = 14) and those who did not

DISCUSSION
(n = 286). AG patients with GNL had more frequently
severe intestinal metaplasia in the gastric body This study found three independent risk factors for the
(Figure 3b; P = 0.001) and the concomitant presence development of GNL in AG patients: age over
of antral atrophy and intestinal metaplasia (Fig- 50 years, the presence of atrophic pangastritis and of
ure 3c,d; P < 0.001). Moreover, AG patients who severe body intestinal metaplasia. A previous study
developed GNL at follow-up had more frequently reported that patients aged >45 years had an about
OLGA stage IV compared with AG patients who did twofold increased risk of intestinal metaplasia progres-
not (15% vs. 3%, P = 0.02). The baseline values of sion than younger subjects.25 In a recent large cohort
gastrin and pepsinogen I were not different between study, it was shown that the increasing age at initial
AG patients who developed GNL and who did not diagnosis was associated with the progression to more
[gastrin 500 (40–2500) pg ⁄ mL vs. 500 (40–2700) advanced lesions and to GC.26 These findings suggest
pg ⁄ mL respectively; pepsinogen I 10 (5–41) ng ⁄ mL vs. that age could be considered as a criterion to select a
10 (3–70) ng ⁄ mL respectively]. No difference was subgroup of AG patients for GNL surveillance as rec-
found between two groups with regard to smoking ommended for other GI tumours such as colorectal
habit and family history for GC (data not shown). neoplasia screening.
As shown in the Table 2, Cox-regression analysis We found that the presence of atrophic pangastritis
identified the following significant risk factors for increased the risk to GNL progression by 4.5 times.
the development of GNL: age over 50 years (HR 8.8), This finding confirms the result of previous
the presence of atrophic pangastritis (HR 4.5) and the works.6, 11, 27 In particular, in our 2005 study, we
presence of severe intestinal metaplasia in the gastric found that the presence of atrophic pangastritis was
body (HR 4.0). Other features such as the gender, Stage strongly associated with the presence of GNL.11 In the
4 of OLGA system, the presence of pernicious anaemia, present study, this finding is even strengthened by an
as well as the positivity to H. pylori, were not predic- increased sample size of the AG cohort, a longer fol-
tive of the development of GNL. Among patients with low-up period and a more accurate statistical evalua-
pernicious anaemia (n = 129), 53 (41%) were H. pylori tion. On the contrary, this finding may be in contrast
positive, among those without pernicious anaemia with older papers, in which corpus-restricted atrophy
(n = 171), 139 (81.3%) were H. pylori positive. associated with pernicious anaemia, was considered at

(a) (b)
Atrophy of the body Intestinal metaplasia of the body
100
90 11
100
90 80 40*
80 28
46 70
70 55
60
60 50
50 25 34
40 25 40
Figure 3. h absent mild 25 30
30 15
moderate severe. Compari- 20 20
29 10 20 27
son of baseline atrophy and 10 20
intestinal metaplasia scores 0 0
AG pts with GNL AG pts AG pts with GNL AG pts
according to Sydney System without GNL without GNL
between AG patients who (c) (d)
developed GNL and who did Atrophy of the antrum Intestinal metaplasia of the antrum
100 5 6
1

not. a. Atrophy of the body; b. 90 100 5 1

5
15 13 90 10 8
Intestinal Metaplasia of the 80
80
70 70
body, *P = 0.0016 vs. other 60 35
45** 60
group; c. Atrophy of the 50 50
antrum *P = 0.0001 vs. other 40 80 40 86
group, **P = 0.001 vs. other 30 30
20 20 50*
35*
group; d. Intestinal Metaplasia 10 10
of the antrum *P = 0.0004 vs. 0 0
AG pts with GNL AG pts AG pts with GNL AG pts
other group.
without GNL without GNL

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd
1048 L . V A N N E L L A et al.

Overall progression to gastric neoplastic lesions Table 2. Risk factors for

progression for gastric
HR 95% CI HR 95% CI neoplastic lesions in univariate
Baseline univariate multivariate and multivariate Cox
regression analysis
Gender
Male 1.0 0.9–5.0 1.0 0.6–3.9
Female 2.1 1.5
Age
<50 years 1.0 1.9–102.3 1.0 1.2–68.4
>50 years 14.0 8.8
Pattern of gastric atrophy
Corpus-restricted atrophy 1.0 2.8–13.4 1.0 1.5–14.1
Atrophic pangastritis 6.1 4.5
Gastric body intestinal metaplasia
Absent, mild and moderate 1.0 1.8–11.0 1.0 1.3–11.8
Severe 4.5 4.0
Gastric antrum intestinal metaplasia
Absent 1.0 2.4–14.2 1.0 0.4–4.0
Present (mild, moderate, severe) 5.9 1.2
OLGA system staging
Stage I, II, III 1.0 1.6–18.8 1.0 0.2–3.7
Stage IV 5.5 0.8
Pernicious anaemia
Absence 1.0 0.5–2.7 1.0 0.2–2.0
Presence 1.1 0.6
Helicobacter pylori infection
Negative 1.0 0.5–2.7 1.0 0.4–2.6
Positive 1.1 1.0

Significant risk factors are in bold typeface.

higher risk for GC.28, 29 However, the comparison with However, the Kaplan Meier curve shows that the
older studies may be difficult for methodological dif- majority of AG patients observed for a maximum of
ferences and because the extension of atrophy in the 16.5 years did not develop GNL, suggesting that a sur-
stomach has not been systematically evaluated in pre- veillance programme for all AG patients may not be
vious literature. cost-effective. A subset of AG patients at higher risk
Finally, this study confirms the presence of severe for GNL may be easily identified combining the cut-
intestinal metaplasia in the gastric body as a risk factor off of age together with two gastric histological
(HR 4) for progression to GNL. In the literature, the parameters available at the time of AG diagnosis,
intestinal metaplasia extension was already related to a potentially allowing to select those AG patients in
higher risk of GC.27 Indeed, an extensive replacement of whom the gastroscopic ⁄ histological surveillance may
normal gastric mucosa with intestinal metaplasia is be warranted.
often considered as a hallmark of severity of AG. Regarding the role of H. pylori infection on progres-
This study observed an incidence rate person-year sion to GNL in AG patients, we used the Giemsa stain
of 0.2% for GC and of 1% for all GNL. The reported and the serology for the determination of the bacte-
rate of annual incidence for GC in the general Italian rium. In fact, other non-invasive tests, such as C-urea
population is estimated to be 0.01%.30 Thus, the breath test and stool antigen test, seem to have a low
results of this study are in keeping with previous diagnostic accuracy in this particular clinical setting.31
reports confirming that AG is a condition at increased AG patients with incident GNL were not different from
risk for gastric neoplasms.14 those without for the positivity of H. pylori at baseline

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd
 R I S K O F N E O P L A S I A I N A T R O P H I C G A S T R I T I S 1049

and the rate of eradication (data not shown). From our risk factor with that among patients without the risk
data, the effect of H. pylori on progression rate to factor. As the present study is a cohort study, the
GNL in AG patients does not seem important. This power of these comparisons was not calculated when
finding is consistent with a previous large prospective the study was designed because there was little evi-
study, which had demonstrated that H. pylori eradica- dence to suggest the likely magnitude of the effects
tion may be beneficial in arresting the progression to of the risk factors or their prevalence. A posteriori,
GNL only in patients without AG or intestinal meta- on the basis of our observed data, the study power
plasia.32 However, we did not design a randomized for age >50 years, atrophic pangastritis and severe
prospective study to evaluate the role of H. pylori body intestinal metaplasia could be calculated as
eradication on prevention of GC, and so we cannot 73.6%, 71.4% and 13% respectively. This low power
know if the obtained rate of GNL was influenced by for intestinal metaplasia as risk factor could be
eradication therapy. explained by the fact that this feature was only
In the present cohort study, the AG patients were found in less than half of the GNL patients, so a sur-
referred to a single-centre, the protocol used for gas- veillance programme based on solely this entry crite-
troscopy was the same for each patient, gastric biop- rion would miss more than half the cases. External
sies were systematically taken in the same location validation of these risk factors on prospective case-
and were evaluated by updated Sydney System. series would be needed.
Moreover, each patient was carefully informed about In conclusion, this study shows that AG patients
the need for endoscopic follow-ups, and the partici- who are over 50 years of age or who have atrophic
pation rate of AG patients at follow-up was as high pangastritis or severe body intestinal metaplasia seem
as 74%. While the results of this study seem to be to be at higher risk for developing gastric neoplastic
generally valid, they should be interpreted with cau- lesions. This finding suggests that in this subset of AG
tion. The overall number of observed case-findings of patients, an endoscopic-histological follow-up for GC
GNL was small, thus limiting the power of statistical surveillance may be worthwhile.
analysis. The relative large confidence intervals
observed in our Cox-regression are indicators of this
ACKNOWLEDGEMENTS
problem. The feasibility of longitudinal cohort studies
involving a larger number of case-findings is difficult Declaration of personal interests: None. We thank Prof.
because of the low overall prevalence of AG in the Luca Tardella and Dr Antonella Cuteri for the revision
general population and the relatively low annual of statistical data. Declaration of funding interests: This
incidence rate of GNL. Regarding the concern about work was supported by grants from the Italian Minis-
the study power, significance tests were performed to try for University and Research, PRIN 2007 and from
compare the prevalence of GNL in the presence of a University Sapienza of Rome, Italy.

REFERENCES 4 Uemura N, Okamoto S, Yamamoto S, 7 Waring AJ, Drake IM, Schorah CJ, et al.
et al. Helicobacter pylori infection and Ascorbic acid and total vitamin C concen-
1 Ferlay J, Bray F, Pisani P, et al. GLOBO- the development of gastric cancer. N Engl trations in plasma, gastric juice, and gas-
CAN 2002: Cancer Incidence, Mortality and J Med 2001; 345: 784–9. trointestinal mucosa: effects of gastritis
Prevalence Worldwide. IARC Cancer Base 5 Meining A, Morgner A, Miehlke S, et al. and oral supplementation. Gut 1996; 38:
n.5 version 2.0, IARC Press, Lyon, 2004. Atrophy-metaplasia-dysplasia-carcinoma 171–6.
2 Hundahl SA, Menck HR, Mansour EG, sequence in the stomach: a reality 8 Annibale B, Capurso G, Lahner E, et al.
et al. The National Cancer Data Base or merely an hypothesis? Best Pract Concomitant alterations in intragastric pH
report on gastric carcinoma. Cancer 1997; Res Clin Gastroenterol 2001; 15: 983– and ascorbic acid concentration in
80: 2333–41. 98. patients with Helicobacter pylori gastritis
3 Correa P. Human gastric carcinogenesis: a 6 Miehlke S, Hackelsberger A, Meining A, and associated iron deficiency anaemia.
multistep and multifactorial process. First et al. Severe expression of corpus gastritis Gut 2003; 52: 496–501.
American Cancer Society Award Lecture is characteristic in gastric cancer patients 9 Haber MM. Histologic precursors of
on cancer epidemiology and prevention. infected with Helicobacter pylori. Br J gastrointestinal tract malignancy. Gastro-
Cancer Res 1992; 52: 6735–40. Cancer 1998; 78: 263–6. enterol Clin North Am 2002; 31: 395–419.

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd
1050 L . V A N N E L L A et al.

10 Whiting JL, Sigurdsson A, Rowlands DC, updated Sydney System. International mised trial on Helicobacter pylori eradica-
et al. The long term results of endoscopic Workshop on the Histopathology of Gas- tion. Gut 2004; 53: 1244–9.
surveillance of premalignant gastric tritis, Houston 1994. Am J Surg Pathol 26 de Vries AC, van Grieken NC, Looman
lesions. Gut 2002; 50: 378–81. 1996; 20: 1161–81. CW, et al. Gastric cancer risk in patients
11 Lahner E, Bordi C, Cattaruzza MS, et al. 18 Rugge M, Correa P, Dixon MF, et al. Gas- with premalignant gastric lesions: a
Long-term follow-up in atrophic body tric dysplasia: the Padova international nationwide cohort study in the Nether-
gastritis patients: atrophy and intestinal classification. Am J Surg Pathol 2000; lands. Gastroenterology 2008; 134: 945–
metaplasia are persistent lesions irre- 24: 167–76. 52.
spective of Helicobacter pylori infection. 19 Stanley RH, Lauri AA, et al. Pathology 27 Cassaro M, Rugge M, Gutierrez O, et al.
Aliment Pharmacol Ther 2005; 22: 471– and Genetics of Tumours of Digestive Sys- Topographic patterns of intestinal meta-
81. tem. Lyon, France: IARC Press, 2002: 46– plasia and gastric cancer. Am J Gastro-
12 Dinis-Ribeiro M, Lopes C, da Costa-Pere- 8. enterol 2000; 95: 1431–8.
ira A, et al. A follow up model for 20 Marignani M, Delle Fave G, Mecarocci S, 28 Kokkola A, Sjöblom SM, Haapiainen R,
patients with atrophic chronic gastritis et al. High prevalence of atrophic body et al. The risk of gastric carcinoma and
and intestinal metaplasia. J Clin Pathol gastritis in patients with unexplained carcinoid tumours in patients with perni-
2004; 57: 177–82. microcytic and macrocytic anemia: a pro- cious anaemia. A prospective follow-up
13 Hirota WK, Zuckerman MJ, Adler DG, spective screening study. Am J Gastroen- study. Scand J Gastroenterol 1998; 33:
et al. Standards of Practice Committee, terol 1999; 94: 766–72. 88–92.
American Society for Gastrointestinal 21 Annibale B, Lahner E, Bordi C, et al. Role 29 Strickland RG, Mackay IR. A reappraisal
Endoscopy. ASGE guideline: the role of of Helicobacter pylori infection in perni- of the nature and significante of chronic
endoscopy in the surveillance of prema- cious anaemia. Dig Liver Dis 2000; 32: atrrophic gastritis. Am J Dig Dis 1973;
lignant conditions of the upper GI tract. 756–62. 18: 426–40.
Gastrointest Endosc 2006; 63: 570–80. 22 Lahner E, Caruana P, D’Ambra G, et al. 30 WHO. Global InfoBase: The Impact of
14 de Vries AC, Haringsma J, Kuipers EJ. First endoscopic-histologic follow-up in Cancer in Your Country. Available at:
The detection, surveillance and treatment patients with body-predominant atrophic http://apps.who.int/infobase/report.aspx?
of premalignant gastric lesions related to gastritis: when should it be done? Gastro- rid=153&iso=ITA. Accessed July 2009.
Helicobacter pylori infection. Helicobacter intest Endosc 2001; 53: 443–8. 31 Lahner E, Vaira D, Figura N, et al. Role
2007; 12: 1–15. 23 Rugge M, Farinati F, Baffa R, et al. Gastric of noninvasive tests (C-urea breath test
15 Annibale B, Marignani M, Azzoni C, et al. epithelial dysplasia in the natural history and stool antigen test) as additional tools
Atrophic body gastritis: distinct features of gastric cancer: a multicenter prospective in diagnosis of Helicobacter pylori
associated with Helicobacter pylori infec- follow-up study. Interdisciplinary Group infection in patients with atrophic
tion. Helicobacter 1997; 2: 57–64. on gastric epithelial dysplasia. Gastroen- body gastritis. Helicobacter 2004; 9:
16 Lahner E, Bordi C, Di Giulio E, et al. Role terology 1994; 107: 1288–96. 436–42.
of Helicobacter pylori serology in atro- 24 Rugge M, Genta RM. Staging and grading 32 Wong BC, Lam SK, Wong WM, et al.
phic body gastritis after eradication treat- of chronic gastritis. Hum Pathol 2005; China Gastric Cancer Study Group.
ment. Aliment Pharmacol Ther 2002; 16: 36: 228–33. Helicobacter pylori eradication to prevent
507–14. 25 Leung WK, Lin SR, Ching JY, et al. Fac- gastric cancer in a high-risk region of
17 Dixon MF, Genta RM, Yardley JH, et al. tors predicting progression of gastric China: a randomized controlled trial.
Classification and grading of gastritis. The intestinal metaplasia: results of a rando- JAMA 2004; 291: 187–94.

Aliment Pharmacol Ther 31, 1042–1050

ª 2010 Blackwell Publishing Ltd