Autoimmune Gastropathy in Type 1 Diabetic Patients With Parietal Cell Antibodies: Histological and clinical findings

We studied 88 type 1 diabetic patients (51 men and 37 women), comprising 47 PCA⁺ and 41 PCA^- subjects matched for age, sex, duration of diabetes, and metabolic control. They were consecutively recruited according to PCA status irrespective of symptoms from 399 patients (75 PCA⁺ and 324 PCA^-) attending the outpatient Antwerp University diabetes clinic. All together, 63% (47 of 75) of all PCA⁺ patients and 37% (41 of 110) of all eligible, matched PCA^- patients agreed to undergo gastroscopy. Every patient fulfilled the criteria of type 1 diabetes (31). No patient used antibiotics, anti-inflammatories, prokinetics, or proton pump inhibitors or had Crohn’s disease, colitis ulcerosa, celiac disease, systemic lupus erythematosus, or rheumatoid arthritis. The study was approved by the local ethics committee. Each subject gave informed consent, in accordance with the Helsinki Declaration.

Islet cell antibodies (ICAs, normal level <12 Juvenile Diabetes Foundation units) were determined by indirect immunofluorescence on cryosections of human donor pancreas, GAD-65 antibodies (GADAs) (normal level <2.6% tracer bound), and tyrosine phosphatase (IA2 antigen, normal level <0.5% tracer bound) by liquid-phase radiobinding assay using Centricon-purified recombinant human ³⁵S-GAD65 and the ³⁵S-labeled intracellular domain of IA2 as tracer (32), respectively. PCAs were detected using indirect immunofluorescence on rat gastric mucosa (Medical Diagnostics California, Carlsbad, CA) (normal level <1/20 dilution) (3). This assay correlated well with the enzyme immunoassay for H⁺/K⁺ATPase antibodies (Varelisa, Pharmacia and Upjohn, Freiburg, Germany) (normal level <10 units/ml) (n = 175, r = 0.85, P < 0.0001). Antibodies to intrinsic factor (AIF) were measured by radiobinding assay (Diagnostic Products, Los Angeles, CA) (normal level <1.1). Iron deficiency anemia was defined as microcytic hypochromic anemia with a transferrin saturation ≤20% or a decreased iron (male < 8.95, female < 7.16 μmol/l) or ferritin level (male < 20, female < 12 μg/l) (13). Pernicious anemia was defined as macrocytic anemia, subnormal vitamin B12 levels, with positive AIF and/or PCA. Serum gastrin was measured by radioimmunoassay technique (Euro-Diagnostics, Malmö, Sweden) (normal level <110 ng/l), vitamin B12 (normal level 120–715 pmol/l), and erythrocyte folate levels (normal level 270-1950 nmol/l) using Simultrac-SNB RIA kit (ICN, Orangeburg, NY). HLA-DQ typing was performed as previously described (33).

Serum was assayed for HP IgG antibodies by enzyme-linked immunosorbent assay (Roche Diagnostics, Brussels, Belgium). ¹³C Urea breath tests were performed as previously described (34). Antrum, corpus, and fundus biopsies were examined for HP colonization using modified Giemsa and/or immunostaining (35). HP infection was diagnosed if any test was positive. The period between urea breath test and gastroscopy was maximally 14 days.

Gastric acid secretion studies were performed in 42 patients (15 PCA⁺ and 27 PCA^-) after a 12-h overnight fast using pentagastrin-stimulated acid output (6 μg/kg s.c. Pentagastrin Injection BP; Cambridge Laboratories, Newcastle upon Tyne, U.K.) (35). Hypochlorhydria was defined as a maximal acid output <15 mmol H⁺/h.

At upper gastrointestinal endoscopy (Olympus Videoscope GIF/Q140; Olympus, Melville, NY), at least two biopsies from fundus, corpus, antrum, and descending duodenum were taken and evaluated by two investigators, unaware of the patient’s clinical and laboratory data to minimize interobserver variation. The visual analog scale of the updated Sydney system was used to evaluate inflammation (chronic infiltrate), activity (acute infiltrate), atrophy (glandular loss), intestinal metaplasia, and HP colonization, graded as follows: 0 = absent, 1 = mild, 2 = moderate, and 3 = severe (27). Furthermore, the distribution of lymphocytic infiltration (1 = diffuse, 2 = focal, and 3 = diffuse and focal), degree of lymphocytic destruction of oxyntic glands, and status of oxyntic glands (1 = partial atrophy and 2 = partial atrophy with preserved islets of oxyntic mucosa) (28), and the presence of ECL cells were evaluated (20). Preatrophic autoimmune gastritis includes diffuse lymphocytic infiltration of the lamina propria, focal lymphocyte-mediated destruction of oxyntic glands, and parietal cell pseudohypertrophy (28). Parietal cells were quantified as the percentage of total cells per gland, counted on four sections (fundus/corpus) stained by hematoxylin and eosin, modified Giemsa, and immunohistochemistry. Twenty glands per biopsy were evaluated (magnification ×390). Twelve sections were counted from each patient, enabling intraobserver variation to be determined. The result per patient represents the mean of these values, and the result of different groups (PCA⁺ and PCA^-) represents an average of these means.

Immunostaining was performed for HP (polyclonal rabbit HP IgG antibodies, dilution 1/200; Dako, Glostrup, Denmark) (35), B-cells (CD20) (L26, M0755, 1/80; Dako), T-cells (CD45RO) (UCHL1, M0742, 1/80; Dako), chromogranin A (DAK-A3, M0869, 1/200; Dako), and H⁺/K⁺ATPase. Antigen retrieval by microwave pretreatment using an EDTA buffer (pH 9) was applied for immunostaining of HP and B- and T-cells. Incubation steps for H⁺/K⁺ATPase immunostaining were as follows: monoclonal mouse H⁺/K⁺ATPase (β-subunit) antibody (IgG1) (Clone 2,611, MA3-923, 1/4,000; Affinity Bioreagents, Golden, CO); biotinylated goat anti-mouse immunoglobulin (E0433, 1/400; Dako), streptavidine-biotinylated horseradish peroxidase (K0377, 100 μl/coupe, 20′; Dako), and 3,3′-diaminobenzidine tetrahydrochloride (HK 153–5K; BioGene, San Ramon, CA). MA3-923 binds to an epitope within amino acids 1–13 or 15–28 located on the cytoplasmic side of the β-subunit of H⁺/K⁺ATPase and inhibits its enzymatic activity (36).

Results were analyzed using SPSS (SPSS, Chicago, IL). Distributions of continuous data were tested for normality by Kolmogorov-Smirnov test. The unpaired t test, Mann-Whitney-U test, or ANOVA was used to determine differences between groups. Bonferroni adjustments for multiple comparisons were made. Pearson’s or Spearman’s rank correlation test was used. Differences in distributions of categorical data were evaluated by χ² or Fisher’s exact test. Stepwise forward logistic regression analysis was done to assess the strength and independency of associations. A two-tailed P < 0.05 was considered significant.

We biopsied 47 PCA⁺ and 41 matched PCA^- type 1 diabetic patients with a mean age of 42 ± 13 years, a disease duration of 20 ± 11 years, and an HbA_1c level of 7.8 ± 1.1%. ICAs were positive in 34%, IA2As in 40%, and GADAs in 68%. This group comprised 18 PCA^- HP^-, 23 PCA⁺ HP^-, 23 PCA^- HP⁺, and 24 PCA⁺ HP⁺ subjects. Intra- and interobserver variations in evaluating histological sections were minimal: 2.4 and 5.1%, respectively.

PCAs were equally common in male and female subjects, and in particular GADA⁺ patients were prone to exhibit PCAs (OR 2.9 [1.1–7.4], P = 0.038). Iron deficiency anemia (3.4 [1.1–10.3], P = 0.04) and pernicious anemia (12.2 [1.5–99.5], P = 0.005) were more frequent in PCA⁺ than in PCA^- subjects (Table 1). Women were more prone to have iron deficiency anemia than men (4.6 [1.5–13.5], P = 0.005). In the group of 399 patients, from which these 88 were recruited, iron-deficiency anemia (21 vs. 9%, P = 0.0035) and pernicious anemia (15 vs. 0.3%, P < 0.0001) were more prevalent in PCA⁺ than in PCA^- patients. Gastrin levels were higher (P = 0.027) and hypergastrinemia was more common in PCA⁺ patients (3.1 [1.2–8.0], P = 0.025), but vitamin B12 and folate levels were similar. Basal (P = 0.012), maximal (P = 0.004), and peak acid output (P = 0.011) were lower and hypochlorhydria was more frequent (12.1 [2.7–54.3], P = 0.0008) in PCA⁺ subjects. Maximal (1.7 ± 2.9 vs. 14.4 ± 10.1 mmol H⁺/h, P = 0.048) and peak acid output (2.2 ± 3.8 vs. 21.2 ± 13.7 mmol H⁺/h, P = 0.03) were lower in those with than without pernicious anemia.

Autoimmune atrophic gastritis was present in 57% of PCA⁺ and 10% of PCA^- subjects (OR 12.5 [3.8–40.7], P < 0.0001). Corpus atrophy was more severe (P = 0.02) and there were less parietal cells (P < 0.0001) in those with PCA. Interestingly, the percentage of parietal cells in glands correlated inversely with PCA titer (r = −0.45, P = 0.004) and H⁺/K⁺ATPase antibody level (r = −0.51, P = 0.001). Signs of preatrophic autoimmune gastritis were more frequent in PCA⁺ patients, as documented by a more pronounced lymphocytic infiltration in corpus mucosa and partial atrophy of oxyntic glands. Concentrations of B-cells and T-cells were greater in those with PCA, with B-cells predominating. Corpus intestinal meta/dysplasia and ECL cell hyper/dysplasia were more common in PCA⁺ patients.

HP infection was diagnosed in 47 patients with a similar prevalence in PCA⁺ and PCA^- patients (Table 1). Two-way ANOVA with PCA and HP status as fixed factors showed that HP infection did not influence iron, gastrin, and H⁺/K⁺ATPase antibody levels, parietal cell mass, corpus or antrum atrophy and metaplasia, and ECL cell hyperplasia. Parameters of preatrophic autoimmune gastritis were also similar in HP⁺ and HP^- patients. Antral inflammation (P = 0.005) was more pronounced in HP-infected subjects.

Stratifying patients according to the presence of AG showed no differences in sex, age, duration of diabetes, HLA-DQ type, or thyroid peroxidase antibody positivity (aTPO) between these two groups (Table 2). However, H⁺/K⁺ATPase antibody (P = 0.001) and gastrin levels (P = 0.001) were higher in AG⁺ patients. Concentrations of gastrin correlated positively with H⁺/K⁺ATPase antibody levels (r = 0.73; P < 0.0001) and inversely with the percentage of parietal cells per gland (r = −0.43, P = 0.002). Logistic regression showed that PCA status (β = 1.44; P = 0.04) and gastrin levels (β = 0.01; P = 0.026), but not HP, age, diabetes duration, sex, or HLA-DQ type, were independent risk factors for autoimmune gastritis. Iron deficiency anemia (OR 3.9 [1.4–10.9], P = 0.015), pernicious anemia (4.6 [1.3–16.9], P = 0.022), and hypochlorhydria (OR 20.0, P = 0.0002) were more frequent in the AG⁺ group.

There were less parietal cells (P < 0.0001) and corpus inflammation (P = 0.021) was more severe in patients with autoimmune gastritis. Concentrations of B-cells (P = 0.03) were higher in those with autoimmune gastritis. Antral morphology was similar between the two groups. Corpus intestinal metaplasia was present in 11 AG⁺ patients, including one PCA⁺ patient with linitis plastica, compared with one AG^- subject (OR 30.8 [3.7–254.1], P < 0.0001). ECL cell hyperplasia was found in seven PCA⁺ patients with autoimmune gastritis, of whom one showed a carcinoid tumor, but in no PCA^- subjects. Six of seven patients with ECL hyperplasia presented with hypergastrinemia (13.4 [1.5–117.6], P = 0.007). A combination of corpus intestinal metaplasia and ECL cell hyperplasia was present in six PCA⁺ patients with autoimmune gastritis. Thus, (pre)malignant lesions were found in 12 of 47 (26%) PCA⁺ and 39% of PCA⁺ type 1 diabetic patients with autoimmune gastritis.

Type 1 diabetic, particularly GADA⁺ (5), patients have a high prevalence of PCA. PCA positivity and hypergastrinemia were risk factors for autoimmune gastritis. PCA⁺ subjects are prone to develop iron deficiency anemia, pernicious anemia, autoimmune gastritis, intestinal metaplasia, and ECL hyperplasia.

PCAs target gastric H⁺/K⁺ATPase. We and others (6,35) have showed that acid output was lower in PCA⁺ patients and that fasting pH correlated with iron levels, supporting evidence that hypochlorhydria may impair iron absorption and cause iron deficiency anemia (11). Gastrin levels were higher in our PCA⁺ patients and correlated with H⁺/K⁺ATPase antibody levels and inversely with the percentage of parietal cells. Others (12,37,38) noted that gastrin levels correlated with peak acid output and corpus atrophy.

Pernicious anemia, present in 2.6–4% of type 1 diabetic subjects (3,16), was 10 times more common in our PCA⁺ than PCA^- patients. PCA and AIF may impair vitamin B12 absorption and cause pernicious anemia, which appears in later stages of autoimmune gastritis (39). Pernicious anemia affects both sexes equally (3,15,16).

PCA⁺ subjects had a higher prevalence of autoimmune gastritis with (2,3,9) or without (28) total atrophy, higher mucosal B- and T-cell concentrations with B- cells predominating (9,40), and less parietal cells in oxyntic glands (20 vs. 29%) than PCA^- patients. In normal stomachs parietal cells occupy 28–34% of oxyntic glands (30).

We confirmed a correlation between PCA titer and the severity of corpus atrophy (38). This suggests that humoral mechanisms involving cytotoxic autoantibodies play a role in mediating mucosal damage in autoimmune gastritis (9,41). PCA can lyse parietal cells in vitro (41), but it is unlikely that they are pathogenic in vivo, because gastric H⁺/K⁺ATPase is not accessible to circulating autoantibodies (10). However, PCA in gastric secretions might have direct access to the apical surface of parietal cells (6). Recently, CD4⁺ T-cells recognizing H⁺/K⁺ATPase were suggested to mediate autoimmune gastritis (10). Burman et al. (42) observed that T-cells predominated in autoimmune gastritis. Moreover, not every patient with autoimmune gastritis has PCA. Possible explanations for seronegative cases are exhaustion of the autoimmune response as the parietal cell autoantigens are depleted (43), an immunological reaction restricted to a cellular response, unrecognized autoantibodies, and the development of autoimmune gastritis before the development of PCAs (9).

At least four biopsies were studied to document autoimmune gastritis, thereby reducing sampling error. Autoimmune gastritis (57%) and (pre)malignant gastric lesions (26%) were frequent in our PCA⁺ subjects. Results are based on a representative number of PCA⁺ and matched PCA^- patients. Not all eligible patients agreed to undergo gastroscopy, and this might, however, have introduced some bias. We estimate that at least one-third of all PCA⁺ type 1 diabetic patients have autoimmune gastritis. Its prevalence in the total diabetic population would be 5–10%. Nearly 40% of our PCA⁺ patients with autoimmune gastritis exhibited (pre)malignant gastric lesions; 11 patients had intestinal metaplasia, comprising 1 with an adenocarcinoma, and 7 showed ECL cell hyperplasia, including 1 subject with a gastric carcinoid tumor. The prevalence of gastric carcinoid tumors, developing on a background of ECL cell hyperplasia induced by chronic hypergastrinemia, ranges from 4–7% in patients with autoimmune gastritis/pernicious anemia and is 13 times more prevalent than in control subjects (17,19,20). Autoimmune gastritis with hypo/achlorhydria may not only result in hypergastrinemia but also in overgrowth of bacteria producing mutagenic N-nitroso compounds, which can lead to intestinal metaplasia-dysplasia, precancerous conditions (44). Patients with atrophic gastritis/pernicious anemia have a three- to fivefold increased gastric cancer risk, ranging from 1 to 9% (17,18), although this has not been shown in all studies (23).

Longitudinal studies suggest that HP may induce early stages of autoimmune gastritis by stimulating granulocytes to produce autodestructive oxygen radicals, which are mutagenic and ultimately lead to corpus atrophy (45) and HP self-destruction due to intestinal metaplasia or achlorhydria (46). Models for the pathogenesis of HP-associated autoimmunity are molecular mimicry and/or T-helper 1-induced expression of HLA class II and costimulatory molecules on gastric epithelial cells (24–26). However, we and others (35,46,47) have found no or a negative relation between HP infection and autoimmune gastritis. Nevertheless, eradicating HP and doing a rebiopsy 6 months later to note its effect on histology is recommended.

Manifestations of autoimmune gastropathy (Fig. 1) are sufficiently prevalent, and the benefits of early diagnosis and treatment (supplementation of iron or vitamin B12 or removal of [pre]malignant gastric lesions) are such that all type 1 diabetic patients should be screened. It seems prudent to test PCA status at onset of diabetes and then yearly for 3 years, then at 5 years and 5-yearly thereafter, or at any time if there are clinical indications, because the test may later become positive (43). At yearly intervals gastrin, iron, vitamin B12 levels, and a complete blood count should be performed. We also suggest performing at least once endoscopy with biopsy in PCA⁺ patients, particularly in those with high gastrin levels. Patients with mild to moderate mucosal dysplasia should be followed endoscopically every 5 years (17). Polyps should be removed, and adenocarcinoma should be excised. Gastric carcinoid tumors are rare and have a far better outcome than carcinoma. Endoscopic surveillance at 5-year intervals has been proposed for ECL hyperplasia (19). For gastric carcinoid tumors associated with autoimmune gastritis, <1 cm and/or fewer than three, expectant therapy or endoscopic removal of accessible tumors is proposed (48). Alternative options include antrectomy (49) and treatment by octreotide (50).

PCA⁺ type 1 diabetic patients are at increased risk for iron deficiency anemia, pernicious anemia, hypochlorhydria, autoimmune gastritis, and (pre)malignant gastric lesions. Therefore, all PCA⁺ patients should be thoroughly examined, particularly those with hypergastrinemia, and gastroscopy with biopsies should be performed.

This study was partly presented at the 36th Annual Meeting of the European Association for the Study of Diabetes, Jerusalem, 17–21 September 2000 and supported by a grant from the European Foundation for the Study of Diabetes.

We thank Dr. F. Peiffer and Dr. G. Moorkens (Department of Endocrinology-Diabetology and Internal Medicine) who contributed to the recruitment of patients. We gratefully acknowledge Dr. C. Van Campenhout and K. Van Cotthem (laboratory of Immunology and Protein Chemistry), the technical staff of the laboratory of Pathology, D. Callens (laboratory of Gastroenterology), Apr. M. Martin (laboratory of Hormonology), and Dr. B. Manuel y Keenoy, J. Vertommen, P. Aerts, S. Schrans, and M. Vinckx (laboratory of Endocrinology-Diabetology) of the University of Antwerp.