Development of type I gastric carcinoid in patients with chronic atrophic gastritis


Dr B. Annibale, Dipartimento Medico-Chirurgico di Scienze Cliniche, Tecnobiomediche e Medicina Traslazionale, Sant’Andrea Hospital, School of Medicine, University Sapienza, Via di Grottarossa 1035, Rome 00189, Italy.


Background  Long-term observational studies assessing the incidence of type I gastric carcinoid (typeIGC) in patients with chronic atrophic gastritis are few.

Aim  To evaluate the occurrence of typeIGC at diagnosis and during follow-up and to identify patient features associated with the presence of typeIGC in a cohort of chronic atrophic gastritis patients.

Methods  Three hundred and sixty-seven chronic atrophic gastritis patients [245 women, age 54 (18–79) years] underwent regular follow-up by gastroscopy. The incidence of typeIGC was determined in chronic atrophic gastritis patients with at least 2 years of follow-up (= 214). Baseline clinical and histological features were analysed as factors associated with the presence of typeIGC by univariate analysis.

Results  Type I gastric carcinoid was diagnosed in nine (2.4%) patients at the moment when chronic atrophic gastritis was diagnosed. After 1463 person-years, six patients developed typeIGC with an annual incidence rate (person-year) of 0.4%. Patients with typeIGC had significantly higher levels of gastrin, chromogranin A and more frequently the presence of body polyps and ECL-dysplasia compared with chronic atrophic gastritis patients without typeIGC.

Conclusions  This cohort study shows that typeIGC is a rare complication in patients with chronic atrophic gastritis, and the presence of body polyps and ECL-dysplasia at gastroscopic/histologic evaluation is strongly associated with the presence of typeIGC.


Gastric carcinoids (GC) or gastric neuroendocrine tumours are tumours derived from enterochromaffin-like (ECL) cells which are localised in the gastric fundus and corpus. The main role of ECL-cells is the secretion of histamine that, in turn, stimulates acid secretion by parietal cells.1 GC are classified into three subgroups, type I to type III, with different pathogenic causes and outcomes.2–4 Type I lesions are more frequent, up to 80% of the total and are gastrin-dependent tumours which are associated with atrophic gastritis.5Gastrin, released by G-cells in the gastric antrum, is the main factor that regulates the release of histamine and produces trophic effects upon ECL-cells.6 In chronic atrophic gastritis (CAG), the loss of appropriate glands in the body leads to achlorhydria, and consequently, hypergastrinaemia. The chronic elevation of gastrin stimulates ECL-hyperplasia and sometimes the development of type I gastric carcinoid (typeIGC).7, 8 Given that only a small group of CAG patients develop typeIGC, factors other than gastrin are necessary for the progression of ECL-cells to typeIGC.

Recently, several large studies in different countries have reported an increasing incidence of all gastric neuroendocrine tumours.9, 10 Although older studies reported a prevalence rate of typeIGC between 1 and 12.5% in patients with CAG,11–13 an estimate of typeIGC incidence in patients with this specific condition is unknown. In fact, long-term observational studies assessing the incidence of typeIGC in CAG patients are few.13–15 For this reason, the aim of the present study was to evaluate the occurrence of typeIGC at diagnosis and during follow-up and to identify patient features associated with the presence of typeIGC in a cohort of CAG patients.


Patients and study design

Between 1992 and 2008, we diagnosed the presence of CAG in 367 (245 women, median age 54 years, range 18–79 years) patients referred to our Unit for unexplained anaemia or long-standing dyspepsia, as described in detail elsewhere.16 Patients underwent gastroscopy with biopsies and those with proven histological diagnosis of CAG were informed about the increased risk of gastric neoplasia associated with CAG and about the need for regular follow-up by gastroscopy. Personal and clinical data, family history for gastric neoplasia and smoking habit were recorded for all patients during a clinical interview. Of the initial population, 102 patients dropped-out because four patients died from causes not-related to CAG and 98 patients refused to participate in further follow-up. Thus the study population included 265 patients (177 women, median age 54 years, range 20–79 years) (see Figure 1). The incidence of typeIGC was determined in CAG patients with at least 2 years of follow-up [= 214; median follow-up 6.3 (2–17) years], after the exclusion of prevalent typeIGC, because this interval of time was considered sufficient for the development of a new lesion of typeIGC. Regarding demographic (gender and age) and clinical features (severity of atrophy), there were no significant differences between patients of the initial CAG group (= 367), of the baseline population (= 265) and of follow-up population.

Figure 1.

 Study population.

Diagnosis of CAG was based on the presence of fasting gastrin above upper normal values and the histological confirmation of gastric body (corpus and fundus) atrophy based on three biopsies taken from the gastric antrum (within 3 cm of the pyloric ring, lesser and greater curve, anterior or posterior wall) and three biopsies from the body along the greater curve.16, 17 The degree of gastritis was assessed according to the updated Sydney System.18 Atrophy of the gastric body mucosa was defined as focal or complete replacement of oxyntic glands by metaplastic pyloric or intestinal glands. Atrophy of the antral mucosa was defined as focal or complete disappearance of antral glands or their replacement by intestinal metaplastic epithelium.16

The ECL-cell status was assessed according to the Solcia classification.19 Hyperplasia was defined by ECL cells proliferation <150 μm and classified as: normal pattern, simple, linear, micronodular and adenomatoid. The diagnosis of dysplasia was based on ECL proliferation between 150 and 500 μm, while that of carcinoid on ECL proliferation >500 μm. MEN-I diagnosis was excluded.20, 21 In case of a visible lesion, additional biopsies were obtained.

All biopsies were examined independently by two experienced pathologists (CB, EP), unaware of the clinical data of the patients. In the event of disagreement, the biopsies were re-examined simultaneously by both pathologists until agreement was reached.16

Pernicious anaemia was defined as low haemoglobin concentration, MCV > 100 fl together with low B12 vitamin levels, responding to intramuscular B12 vitamin treatment.22 Iron deficiency anaemia was defined as low haemoglobin concentration, MCV < 80 fl and ferritin<30 ng/mL.22 The diagnosis of autoimmune thyroid disease (AITD) was based on the presence of thyroid autoantibodies and thyroiditis signs at ultrasound evaluation irrespective of thyroid function as previously described.23

Plasma gastrin levels were determined by radioimmunoassay using antiserum 4562, kindly supplied by Prof. J.F. Rehfeld, Copenhagen, Denmark, as previously described (gastrin normal values ≤40 pg/mL).13 Chromogranin A (CgA) levels were assessed by ELISA kit (Dako A/S, Glostrup, Denmark) (CgA normal values ≤98 ng/mL); Pepsinogen I was measured using a commercial RIA kit (Pepsik; Sorin, Saluggia, Italy; normal values 20–80 ng/mL); Parietal cell antibodies (PCA) were measured by immunofluorescence and Helicobacter pylori IgG antibodies using an ELISA commercial kit (GAP test IgG, Biorad, Milan, Italy).16Helicobacter pylori infection was considered positive when the bacterium (Giemsa stain) was detected at histology and/or on the basis of a positive IgG titre for H. pylori.17 Bismuth-based triple regimen eradication therapy was prescribed in the case of H. pylori positivity and, after 6 months, the absence of H. pylori at histology and a decrease by at least 50% in the initial titre of H. pylori IgG was the criterion for the successful cure of infection.17

Follow-up protocol

The first follow-up gastroscopy was scheduled at intervals between 2 and 4 years after diagnosis of CAG and the same baseline bioptic sampling was repeated.24

If typeIGC was detected, an extensive bioptic sampling of at least ten biopsies (two biopsies in antrum, four in gastric body and four in the fundus) of the gastric mucosa was taken after 6 and 12 months.13 If typeIGC was not confirmed at two following gastroscopies, patients were followed by the general follow-up protocol. Polypoid lesions up to 5 mm was removed by forceps, otherwise an electrocautery snare was used. All patients with a diagnosis of typeIGC underwent at least one imaging procedure for the tumour staging (Octreoscan, CT-scan, Magnetic Resonance Imaging).

Data analysis and statistical evaluation

Type I gastric carcinoid at diagnosis and during follow-up in CAG patients was evaluated and prevalence and incidence rates (person-year) were calculated. Quantitative data are presented as medians and ranges or, absolute counts and percentages. Mann–Whitney and Fisher exact test were used when appropriate.

Univariate analysis was performed to identify patient features associated with the presence of typeIGC. The following baseline patient characteristics were analysed: age at CAG diagnosis, gender, presence of atrophic pangastritis defined as the concomitant presence of atrophy in the gastric antrum and body, presence of pernicious anaemia, gastrin, pepsinogen I, CgA, presence of PCA, presence of dyspepsia, presence of gastric polyp, presence of ECL-dysplasia, smoking habits, presence of H. pylori, family history for gastric cancer and presence of AITD. A P value of less than 0.05 was considered statistically significant.


Occurrence of typeIGC

Type I gastric carcinoid were diagnosed in 9 out of 367 (2.4%) patients at the moment of the initial diagnosis of CAG [six women, median age 59 (range 40–72) years]. As shown in Table 1, all typeIGC patients but one presented polyps of the body with a diameter between 0.3 and 3 cm at gastroscopy. In six out of nine (66.7%) patients with typeIGC, the body gastric mucosa presented a severe atrophy, in the remaining cases the atrophy was moderate. The alterations of ECL-pattern associated with typeIGC were micronodular hyperplasia (= 4; 44.5%) and dysplasia (= 5; 55.5%). Further details of CAG patients with typeIGC such as the presence of pernicious anaemia, the levels of gastrin and CgA, the degree of body atrophy are summarised in the Table 1.

Table 1.   Baseline features of 15 CAG patients with prevalent and incident typeIGCs
PatientsGenderAge (years)Gastrin (pg/mL)CgA (ng/mL)Body atrophy*Pernicious anaemiaEndoscopic findingsDetection of typeIGCRecurrence
  1. * Scored accordingly to the Sydney System.

 1F45305602NoBody polyp of 0.5 cmAt diagnosisNo
 2M679751753YesBody polyp of 0.5 cmAt diagnosisAt 4-month follow-up
 3F409751503NoBody polyp of 0.6 cmAt diagnosisNo
 4F405701403NoNormal mucosaAt diagnosisAt 6, 15, 32-month follow-up
 5F666902453NoBody polyp of 0.5 cmAt diagnosisNo
 6F726802503YesBody polyp of 0.5 cmAt diagnosisAt 3, 5-month follow-up
 7M5920001672YesBody polyp of 0.3 cmAt diagnosisNo
 8F6111501503YesBody polyp of 0.5 cmAt diagnosisNo
 9M5511835472NoBody polyp of 3 cmAt diagnosisNo metastasis after total gastrectomy
10M4710901503YesNormal mucosaAt 31-month follow-upAt 35, 46, 64-month follow-up
11M675004052NoBody polyp of 0.5 cmAt 31-month follow-upNo
12M692201203NoBody polyp of 0.5 cmAt 41-month follow-upNo
13F6528003703NoNormal mucosaAt 7-month follow-upAt 13, 18, 50-month follow-up
14F235652303YesBody polyp of 1.2 cmAt 59-month follow-upNo
15F4914752403YesBody polyp of 0.4 cmAt 18-month follow-upAt 30, 37, 61-month follow-up

Of the initial population, after the exclusion of 102 patients who dropped-out, 265 patients remained in the study and of these, 214 patients had at least 2-years follow-up (see Figure 1). After 1463 person-years, six patients developed typeIGC with an annual incidence rate (person-year) of 0.4% [three women, median age 57 (range 23–69) years] (Table 1). Two out of six patients with incident typeIGC presented ECL dysplastic lesions at first diagnosis of CAG. Observed ECL-pattern alterations associated with typeIGC were linear (= 1; 16.7%), micronodular (= 2; 33.3%), adenomatoid (= 1; 16.7%) hyperplasia and dysplasia (= 2; 33.3%).

Patient features associated with typeIGC

To identify patient features associated with the presence of typeIGC, we compared baseline clinical, biochemical and histological characteristics between CAG patients without (= 199) and with typeIGC (= 15). As shown in Table 2, patients with typeIGC presented significantly higher levels of gastrin and CgA with respect to the other group. At the moment of the diagnosis of typeIGC, 7 out of 15 typeIGC patients (46.6%) had a diagnosis of ECL-dysplasia with respect to none of the other group (< 0.000001). Other analysed features were similar between the two groups.

Table 2.   Baseline clinical, biochemical and histological features of 214 CAG patients with and without typeIGC
= 15
= 199
  1. PCA, parietal cell antibodies; CAG, chronic atrophic gastritis; AITD, autoimmune thyroid disease.

  2. Data are expressed as median or proportions (percentage). Mann–Whitney test or Fisher-test were used for the comparison of baseline features between two groups.

  3. Significant p-values are in bold.

  4. * ECL-dysplasia was concurrent at typeIGC in five prevalent and two incident cases.

Age, years (range)59 (23–72)54 (22–77)0.39
Female gender, n (%)9 (60)137 (68.8)0.7
Gastrin, pg/mL, median (range)832.5 (220–2800)500 (50–2700)0.01
Pepsinogen I, ng/mL, median (range)10 (3–28)10 (3–79)0.6
Chromogranin A, ng/mL, median (range)171 (60–405)100 (30–280)0.0001
Pernicious Anaemia, n (%)7 (46.6)98 (49.2)1
Family history for gastric cancer, n (%)3 (20)18 (9)0.13
Smoking habit, n (%)8 (53.3)81 (40.7)0.39
PCA, n (%)14 (93.3)158 (79.3)0.3
Helicobacter pylori positivity, n (%)7 (46.6)84 (42.2)0.79
Dyspepsia, n (%)7 (46.6)43 (21.6)0.05
AITD, n (%)9 (60)91 (46)0.29
ECL-dysplasia*, n (%)7 (46.6)0<0.000001
Atrophic pangastritis, n (%)2 (13.3)44 (22.1)0.5
Body atrophy, n (%)
 Mild022 (11.3)0.37
 Moderate4 (26.6)55 (27.4)1
 Severe11 (73.4)122 (61.3)0.4

Regarding the features of gastric polyps, the gastric location, the numbers and the endoscopic appearance (sessile or pedunculated) of the polyps were not significantly different between CAG patients with and without typeIGC (see Table 3). Body polyps were more frequently found in patients with typeIGC with respect to the other group (= 0.000001). Twelve out of 15 (80%) polyps diagnosed in CAG patients without typeIGC were hyperplastic while the remaining were adenomas (3/15, 20%).

Table 3.   Comparison of gastric polyps in CAG patients with and without typeIGC
Polypoid lesionTypeIGC*
= 15‡
= 199
  1. Data are expressed as proportions (percentage) and Fisher-test was used for the comparison between the two groups.

  2. Significant p-values are in bold.

  3. * Reported polypoid lesions refer to gastroscopy performed at the moment of typeIGC diagnosis.

  4. † Reported polypoid lesions refer to baseline gastroscopy.

  5. ‡ Three patients had intramucosal carcinoids.

Gastric location
 Antrum06 (3)ns
 Body12 (80)7 (3.5)0.000001‡
 Antrum and body02 (1)ns
 <0.5 cm9/12 (75)7/15 (46.7)ns
 0.5–1 cm2/12 (16.7)6/15 (40)ns
 >1 cm1/12 (8.3)2/15 (13.3)ns
 Single5/12 (41.6)8/15 (53.3)ns
 Multiple7/12 (58.3)7/15 (46.7)ns
Endoscopic appearance
 Sessile12/12 (100)12/15 (80)ns
 Pedunculated03/15 (20)ns

Management and follow-up of typeIGC

All patients with polypoid lesions up to 12 mm underwent endoscopic polypectomy. Endoscopic polypectomy failed on the polyp of 3 cm because of its hard consistency. The patient, a man aged 55, underwent total gastrectomy. At histological evaluation, the tumour extended from the mucosa to the submucosa and infiltrated the muscularis propria. Immunohistochemistry showed positivity for CgA and VMAT-2 with Ki67 proliferation index lower than 1%. In three out of 19 local lymph nodes removed during the surgery a metastatic endocrine tumour was found. When typeIGC was diagnosed, CT-scan and Octreoscan were performed and resulted negative.

In 6 out of 15 (40%) patients with typeIGC, there was a recurrence of the tumour, but all patients with lesions that were small (up to 1 cm) underwent endoscopic polypectomy (see Table 1). At the end of the study, all the patients were alive and in good health.

Concomitant GI findings

During an overall median follow-up of 6.3 (2–17) years, three intestinal-type adenocarcinomas were diagnosed [two women; median age 51 (51–77) years]. All three were located in the gastric antrum and were diagnosed 8, 5 and 1 years after diagnosis of CAG. The stage of the gastric cancer was T2N1M0 in the male patient and T1N0M0 in the two female patients. After gastrectomy, two patients died, one for neoplastic disease progression and the other for complications after surgery. At the end of the study, one female patient was alive and free of neoplasia.


In the present work, we followed-up a cohort of CAG patients for 1463 person-years and found an annual incidence rate (person-year) for the typeIGC of 0.4%. Recently, large studies on population-based data in different countries have reported an increasing incidence of GC among all gastric malignancies and an increase of these among all gastrointestinal carcinoid lesions.9, 10 This increase could be partially explained by the increased diffusion of diagnostic gastroscopy25 and correlates temporally with the increasing use of proton pump inhibitors.26 To our knowledge, few studies have evaluated the real incidence of typeIGC in patients with CAG because the surveillance of CAG was not recommended.27 An old study by Kokkola et al. described eight new cases of typeIGC in 416 patient-years which corresponds to an annual incidence of 2%.14 Sjoblom et al. studied 196 patients with pernicious anaemia, but gastroscopy was performed only in 70 of these patients.15 After 1397 patient-years, two new cases of typeIGC were reported on hospital registries among the initial group of 196 patients. This figure should correspond to an annual incidence rate of 0.1%, but in this study not all patients underwent gastroscopy and the incidence rate can only be obtained indirectly. These differences in the typeIGC incidence rates could be explained by the patient selection. CAG can have a wide range of clinical presentations as already reported.28 Patients included in our series had dyspepsia, iron deficiency anaemia or pernicious anaemia. In particular, the latter was present in almost 50% of our patients, while previous studies included exclusively patients with pernicious anaemia. In our previous study, we diagnosed one new case of typeIGC in a smaller sample of CAG patients (= 96) for a shorter period of observation (median time of 2.5 years) which corresponds to an annual incidence of 0.4%.13 Thus, in the present study, increasing both the cohort of CAG patients and follow-up period (median time of 6, 3 years) the annual incidence rate remained constant.

At initial diagnosis of CAG, we observed nine cases of typeIGC which is 2.4% of all patients. Previous studies reported a rate between 1% and 12.5% of typeIGC.11, 13–15, 29, 30 The wide range of the prevalence rates of typeIGC among several studies could be explained by different settings where patients were selected, such as type of hospital (secondary, tertiary centre), symptoms/signs of presentation, type of patient (in-patient, out-patient).

To identify patient features associated with the presence of typeIGC, we compared the baseline features between CAG patients with and without typeIGC. Higher levels of gastrin and CgA were associated with the presence of typeIGC by univariate analysis. A recent review has confirmed that the CgA is the most sensitive biomarker in neuroendocrine tumours and it is more frequently elevated in well-differentiated tumours compared to poorly differentiated ones.31 However, patients with atrophic gastritis present high values of CgA32 and the low specificity of the CgA assay does not identify those with typeIGC among CAG patients as already reported.33 For the same reasons, gastrin does not detect patients with typeIGC as it is increased in all CAG patients. In fact, as shown in Table 2, although patients with typeIGC had a median value of gastrin higher than the other CAG patients, the ranges of gastrin overlapped widely between the two groups. We retain that these assays are clinically useful for identifying patients with CAG diagnosis or to exclude those from patients with typeIIIGC, but they cannot be considered as markers of the presence of typeIGC.

The presence of ECL-dysplasia was associated more frequently with typeIGC. This pattern of ECL-cells is considered to be the true gastric carcinoid precursor lesion and it can represent the sign of a concomitant carcinoid lesion.19 Our previous study showed that this lesion also carries an increased risk for the development of typeIGC during the follow-up.13 Thus, CAG patients with a diagnosis of ECL-cell dysplasia could benefit from a shorter endoscopic follow-up to exclude concomitant typeIGC lesions or to identify new lesions in the gastric mucosa.

Another finding of the present study was that the presence of body polyps increases the risk of typeIGC in CAG patients. Unfortunately, no features of endoscopic appearance of the gastric polyps (size, number, sessile/pedunculated presentation) permitted a differential diagnosis between patients with and without typeIGC. In the group of CAG patients without typeIGC, almost three-fourths of all polyps were hyperplastic and the remaining were adenomas. Hyperplastic polyps are very common among all benign epithelial gastric polyps34, 35 and more frequently associated with CAG.36, 37 Importantly, some authors have reported an increased risk of neoplastic progression also for this type of lesions. Moreover, CAG has an increased risk of gastric adenocarcinoma.14, 15 As a consequence, each polyp found at gastroscopy should be removed and histologically examined.

Different approaches were proposed for the management of typeIGC such as surgery,38 endoscopic polypectomy,1, 4, 12 somatostatin analogues39, 40 and an endoscopic follow-up without any specific treatment.41, 42 We used a conservative management by endoscopic follow-up and performing polypectomy of all visible polypoid lesions. Only one patient, in whom the endoscopic approach failed, underwent gastrectomy. This patient had a deeper neoplastic lesion in the gastric wall, extended to muscularis propria and metastatic involvement of three lymph nodes. Thus, the surgery in this particular case was appropriate. After surgery, imaging procedures (CT-scan and Octreoscan) for the staging did not reveal metastatic involvement of the other organs. The possibility of more invasive lesions among typeIGC is already described in the literature43, 44 and for patients with larger polyps, local surgical tumour resection is recommended.45

Finally, in our study, the rate of recurrence of the typeIGC during follow-up was almost of 40%. This high recurrence rate could be explained by typeIGC misdiagnosis during the previous gastroscopy, by bioptic sampling errors or the continuous trophic effect of hypergastrinemia on the gastric mucosa. Endoscopic follow-up every 6–12 months, as suggested also by ENETS guidelines, allowed us to identify recurrent lesions or new lesions (incidence-cases) when they had a small size and were easily removable by polypectomy without complications.46 This approach was safe, and the 100% survival of all typeIGC within the study period is in keeping with previous works.41, 44

Our study has some limitations. The first is the high number of CAG patients who dropped-out during the follow-up, almost 27%. We do not know the clinical outcomes of these CAG patients and, as a consequence, the incidence rate could be underestimated. However, we know that this group of patients did not present demographic or baseline clinical features different from the CAG patients who remained in the study and we retain that the main findings of this study remain generally valid.

Moreover, the power of the statistical analysis was limited by the small number of case-findings of typeIGC. However, larger number of case-findings could be obtained only by performing studies with larger cohorts and a longer period of observation.

In conclusion, this cohort study implies that typeIGC is a rare complication in patients with CAG and the presence of body polyps at gastroscopy and ECL-dysplasia at histological evaluation are strongly associated with the presence of typeIGC.


Declaration of personal interests: None. Declaration of funding interests: This work was supported by grants from the Italian Ministry for University and Research, PRIN, COFIN 2007 and University Sapienza Roma 2007–2009, Italy.