To elucidate the mechanisms that lead to severe hypergastrinaemia during long-term omeprazole therapy for gastro-oesophageal reflux disease (GERD).
To elucidate the mechanisms that lead to severe hypergastrinaemia during long-term omeprazole therapy for gastro-oesophageal reflux disease (GERD).
A total of 26 GERD patients were studied during omeprazole maintenance therapy. Twelve patients with severe hypergastrinaemia (gastrin > 400 ng/L) were compared with 14 control patients (gastrin < 300 ng/L). Helicobacter pylori serology and a laboratory screen were obtained in all patients. Gastric emptying was scored by the evidence of food remnants upon endoscopy 12 h after a standardized meal. Gastric antrum and corpus biopsies were analysed for histological parameters, as well as somatostatin and gastrin concentrations. All patients underwent a meal-stimulated gastrin test and the hypergastrinaemia patients also underwent a vagal nerve integrity assessment by pancreatic polypeptide testing (PPT).
Severe hypergastrinaemia patients had a longer duration of treatment (80 vs. 55 months; P = 0.047) and were characterized by a higher prevalence of H. pylori infection (9/12 vs. 2/14, P = 0.004), corpus mucosal inflammation and atrophic gastritis (P < 0.04). This was reflected in lower serum pepsinogen A concentrations (mean ± S.E.M. 53.6 ± 17.9 vs. 137 ± 16.0 mg/L, P = 0.03), pepsinogen A/C ratio (1.8 ± 0.3 vs. 4.1 ± 0.6, P = 0.005) and mucosal somatostatin concentrations (2.75 ± 0.60 vs. 4.48 ± 1.08 mg/g protein, P = 0.038). Two patients in the hypergastrinaemia group had signs of delayed gastric emptying, but none in the normogastrinaemia group did (P = N.S.). In addition, both groups had a normal meal-stimulated gastrin response.
Severe hypergastrinaemia during omeprazole maintenance therapy for GERD is associated with the duration of therapy and H. pylori infection, but not with abnormalities of gastric emptying or vagal nerve integrity.
Omeprazole profoundly suppresses gastric acid secretion due to binding to the proton pump of the parietal cell. A consequence of long-term treatment with proton pump inhibitors is the occurrence of small increases in serum gastrin levels in a number of patients, with levels increasing 2–4 times above the baseline value.1, 2 This may be the result of an increased gastrin production by antral G-cells, due to a loss of the inhibitory effect of acid on these cells during omeprazole therapy. However, ≈ 3% of patients treated with long-term omeprazole develop severe hypergastrinaemia (serum gastrin > 400 ng/L).3 The clinical significance of this observation in man is unknown. We are not aware of any data showing clinically significant side-effects of severe hypergastrinaemia induced by omeprazole.
The pathophysiological mechanism of severe hypergastrinaemia in these patients is unknown. At present several explanations are available. First, the presence of coexisting chronic atrophic gastritis may cause hypergastrinaemia.2–4 Second, omeprazole is thought to delay gastric emptying, which may lead to increased gastrin output.5 Furthermore, somatostatin, produced by gastric D-cells, inhibits gastrin release. Increases in gastrin levels can occur when the negative feedback inhibition on acid secretion by somatostatin is lost.6 Reduced levels of somatostatin during omeprazole administration have been described, although the mechanism for this reduction is unknown.7
Vagal nerve dysfunction is known to be associated with gastro-oesophageal reflux disease and may lead, both by delayed emptying and abnormal gastric neural stimulation, to increased gastrin levels.8 Finally, Helicobacter pylori infection is usually associated with an increased basal and stimulated serum gastrin level.9–11 During acid suppression H. pylori corpus gastritis increases,12 which further impairs acid secretion. This may additionally contribute to increased gastrin levels.
In this study we investigated the influence of the following factors: atrophic gastritis, gastric emptying, gastric mucosal somatostatin and gastrin concentrations, vagal nerve function and H. pylori infection, on the development of severe hypergastrinaemia during omeprazole maintenance therapy for GERD.
Patients on omeprazole maintenance therapy for gastro-oesophageal reflux disease (GERD) were included in this open parallel study. All patients participated in an ongoing long-term follow-up study of treatment for GERD. Baseline serum gastrin values had been normal (< 100 ng/L) or mildly elevated (< 300 ng/L) in all individuals before the start of omeprazole therapy. Mild hypergastrinaemia during therapy was defined as a serum gastrin < 300 ng/L at all occasions assessed at 6–12 month intervals. Severe hypergastrinaemia was defined as a serum gastrin level > 400 ng/L on at least two consecutive occasions assessed at 6–12 month intervals during omeprazole treatment. Patients with liver or kidney disease, suspected or confirmed malignancy, chronic alcoholism, diabetes mellitus, exocrine pancreatic insufficiency, use of prokinetic drugs, Zollinger–Ellison syndrome, or hypochlorhydria due to (autoimmune) atrophic gastritis were excluded from the study. Use of prokinetic drugs was prohibited during the study.
After obtaining informed consent, each patient was subjected to a clinical examination including a questionnaire concerning past medical history, drug use and a physical examination.
Venous blood samples were analysed for Hb, WBC, platelets, serum creatinine, bilirubin, alkaline phospatase, ASAT, ALAT and calcium. Specific IgA and IgG H. pylori antibodies were measured by a modified enzyme linked immunosorbent assay (ELISA) technique.13 Fasting serum gastrin was determined by radioimmunoassay, as described previously.14 Fasting serum pepsinogen A and C levels were determined, using a radioimmunoassay, and the A/C ratio was calculated.15, 16
Exactly 12 h before the scheduled endoscopy, patients took a standard meal consisting of 50 g of raisins and lettuce to assess gastric emptying. These substances were chosen because they are well recognizable during endoscopy. The patient then fasted until the next morning when an upper gastrointestinal endoscopy was performed according to standard procedures using Olympus Q10/20, GIF 100 or IT 100 endoscopes. Presence and grade of oesophagitis, diaphragmal herniation, food retention or any other abnormality was described.
A total of 12 biopsy specimens were obtained at endoscopy. For histopathological examination, four antrum and four corpus biopsies were fixated in buffered formalin. After both haematoxylin–eosin and Giemsa staining they were graded according to the Sydney system for both active and chronic gastritis, atrophy and presence of H. pylori.17 Four additional biopsy specimens (two antrum and two corpus) were immediately frozen on liquid nitrogen for measurement of gastrin and somatostatin concentrations, as described previously.18, 19
After endoscopy, a meal-stimulated gastrin curve was determined in each patient, following consumption of a standard meal consisting of two slices of brown bread, 20 g cheese, one boiled egg and 250 mL whole milk (equivalent to 44 g fat, 31 g protein, 42 g carbohydrates; 2360 kJ). Blood samples were obtained, by use of an indwelling venous catheter, before endoscopy and just before consumption of the test meal. Postprandial blood samples were taken every 15 min for 1 h after finishing the standard meal. A gastrin rise of ≥ 100% from the baseline value was considered a normal response on meal stimulation.
In the hypergastrinaemia patients, a pancreatic polypeptide test was performed, on a separate test day, in order to test vagal nerve integrity.20, 21 After an overnight fast, blood was taken via an indwelling venous catheter, twice before and at 10 min intervals up to 70 min after the injection of 0.1 IU/kg bodyweight of insulin (Actrapid or Velosulin, Novo Pharma bv, Zoetermeer, the Netherlands) intravenously. Plasma glucose levels were measured at 10 min intervals. When blood glucose levels dropped below 2.5 mmol/L, this was corrected with administration of intravenous glucose. Plasma pancreatic polypeptide levels were determined by radioimmunoassay.22 An increment of the pancreatic polypeptide concentration (ΔPPT) of 100 mmol/L or more was considered a normal response consistent with integrity of the vagal nerve.21
The statistical analysis was carried out using Fisher’s exact, Student’s t- or Mann–Whitney U-tests where applicable. A two-sided P-value < 0.05 was considered significant.
The study was approved by the scientific and ethical committee of the Free University Hospital.
Of a total population of ≈ 160 patients on long-term treatment with omeprazole for GERD, 12 patients met our criteria for severe hypergastrinaemia and 14 patients with mild hypergastrinaemia were included as controls (‘normogastrinaemia’ group). All patients were treated on an outpatient basis. The characteristics of both groups are depicted in Table 1. In the normogastrinaemia group, all patients had had fasting serum gastrin levels on treatment below 300 ng/L prior to study entry, thus matching the inclusion criteria for the mild hypergastrinaemic control group. However, when measured as part of the study, three of these patients had gastrin levels of 313, 331 and 372 ng/L, respectively. In the severe hypergastrinaemia group, two patients had previously undergone a highly selective vagotomy, and these two patients had had a mildly elevated serum gastrin before the start of omeprazole therapy (160 and 280 ng/L, respectively). All other hypergastrinaemia patients had normal serum gastrin levels before the start of omeprazole therapy. The normogastrinaemics and the severe hypergastrinaemia group thus did not differ with respect to mean pre-treatment fasting serum gastrin levels (65 vs. 95 ng/L, Table 1), nor did they differ with respect to age (60 vs. 66 years) or mean maintenance omeprazole dose. The mean duration of therapy was 80 months (severe hypergastrinaemics) vs. 55 months (normogastrinaemics) (P = 0.047).
Laboratory screening (haematology and chemistry) was normal in all patients (data not shown). Fasting serum gastrin levels during omeprazole therapy differed, as expected (mean gastrin (range) in hypergastrinaemia vs. normogastrinaemia patients: 1509 (700–2900) vs. 191 (43–372) ng/L; P < 0.0001; Mann–Whitney U-test) ( Table 1).
Serum pepsinogen A levels were lower in patients with hypergastrinaemia compared to normogastrinaemics (53.6 ± 17.9 vs. 136.9 ± 16.0 mg/L, P = 0.03) ( Table 2). Mean pepsinogen C levels did not differ significantly between patients with hypergastrinaemia or normogastrinaemia: 25.3 ± 2.8 vs. 34.2 ± 5.0 mg/L (P = 0.12). Pepsinogen A/C ratio for normogastrinaemia patients was 4.1 vs. 1.8 for hypergastrinaemia patients (P = 0.005).
In all patients, previous oesophagitis was healed as a result of the omeprazole maintenance therapy. Food retention after a 12 h fast was observed in two patients with hypergastrinaemia and in none of the normogastrinaemia patients (P = N.S.). In patients with hypergastrinaemia a hiatus hernia was seen in six patients and a Barrett oesophagus in two. No gastric polyps were seen in this patient group.
A hiatus hernia was present in 11 out of 14 patients with normogastrinaemia (P = 0.10, Fischer’s exact test), and in four a Barrett oesophagus (P = 0.65). Corpus polyps were observed in two patients (P = 0.48), with both polyps found to be fundic gland cysts upon histological examination. H. pylori infection, as documented by positive histology and serology, was observed more often in patients with hypergastrinaemia (9/12) than in normogastrinaemia patients (2/14) (P = 0.004). Histopathological examination of gastric antrum biopsies showed no difference in active and chronic inflammation or atrophy between the study populations ( Table 3). However, the corpus biopsies of the hypergastrinaemia patients showed both a more pronounced active inflammation (median Sydney scores: 1 vs. 0, respectively, P = 0.037), chronic inflammation (2 vs. 1, P = 0.009) and atrophic gastritis (1 vs. 0, P = 0.028) in comparison to the normogastrinaemia patients ( Table 3). These findings are in agreement with the higher prevalence of H. pylori in this group. Two patients had positive serology and histological signs of chronic active inflammation. Although the pathologist did not observe H. pylori in the biopsy specimens, we considered these patients to be H. pylori-positive. None of the patients had negative serology and positive histology.
The concentation of somatostatin in corpus biopsies was significantly lower in hypergastrinaemia patients compared to controls (mean ± S.E.M. 2.75 ± 0.60 vs. 4.48 ± 1.08 mg/g protein; P = 0.038); ( Table 4). The concentration of gastrin in corpus specimens was very low in both groups, as expected (mean (± S.E.M.) concentration for normogastrinaemia and hypergastrinaemia patients 2.0 (0.5) and 1.7 (0.2) mg/g protein, respectively; P = 0.36). No difference was observed between the groups in the mean somatostatin and gastrin concentrations in antral mucosal biopsies ( Table 4).
Both groups showed an expected rapid rise of serum gastrin levels after meal stimulation ( Figure 1). In the normogastrinaemic patients, the percentage increase in serum gastrin levels and the mean area under the curve (S.E.M.) calculated using the trapezoid rule were 191 ± 28% and 13 701 (2407) ng.min/L, respectively. The hypergastrinaemic patients showed a percentage increase of 86 ± 18%, which is at the lower border of normal. However, the response measured as mean area under the curve (S.E.M.) was very high (34 320 (6033) ng.min/L) and significantly differed from the response observed in the controls (P = 0.006). We therefore concluded that these hypergastrinaemic patients had an intact (normal to increased) serum gastrin response to meal stimulation.
After induced hypoglycaemia the mean (± S.E.M.) peak increment of serum pancreas polypeptide levels in the hypergastrinaemic patients was 125 ± 25 (pmol/L). The two patients with previous highly selective vagotomy had no response on the PPT. Since a PPT was performed only in patients with hypergastrinaemia, a previously described patient group with gastro-oesophageal reflux disease, but without profound acid suppressive therapy was used as a historical control group21 for comparison. In this control group (n = 8) the mean peak increment of pancreatic polypeptide upon insulin-induced hypoglycaemia was 198 ± 42 (pmol/L); this value was not statistically different from the one observed in our population (P = 0.09).
Profound acid suppressive therapy is regularly associated with mild to moderate hypergastrinaemia. However, a small number of patients develop a more severe hypergastrinaemia. In animal studies, severe hypergastrinaemia has been associated with the development of ECL cell hyperplasia and carcinoid tumour induction of the stomach mucosa.23 Although ECL hyperplasia does occur during long-term omeprazole treatment, carcinoid formation in the stomach during omeprazole therapy has not been described in man.24–26
The mechanism responsible for the development of severe hypergastrinaemia during omeprazole therapy is unknown. Factors that may be involved include: loss of the negative feedback inhibition on acid secretion by somatostatin, vagal nerve dysfunction, gastritis and/or atrophy induced by H. pylori infection and a direct effect of omeprazole on gastric emptying. We therefore performed a case-control study, investigating different pathophysiological mechanisms that may contribute to the development of hypergastrinaemia during acid suppressive therapy for GERD.
In patients on long-term omeprazole treatment for GERD, we did not observe a difference in gastric emptying between severely hypergastrinaemic patients and controls, as judged by the retention of remnants of a standardized meal in the stomach 12 h after consumption. Only two patients with hypergastrinaemia had a delayed gastric emptying; none of the normogastrinaemic patients did. We did not use the standard radioisotope measurements, because the method is inconvenient, with the need for both solid and liquid substrates, and more importantly, has very large normal confidence limits. This test therefore has little sensitivity. Although less conventional, we believe that our method of scoring well-recognizable remnants of a standardized meal comes closer to daily clinical practice, is simple to perform and convenient for the patient. Apart from gastric emptying, we also did not observe any differences between the two groups with respect to the prevalence of vagal nerve dysfunction, as judged by pancreatic polypeptide testing. These findings suggest that vagal nerve dysfunction does not play a significant role in the development of severe hypergastrinaemia during proton pump inhibitor therapy.
On meal stimulation a physiological increase in serum gastrin occurred in both groups, even in the hypergastrinaemia patients in whom pre-stimulation gastrin levels were increased to a considerable extent. This demonstrates an intact physiological response of the gastric mucosa during long-term proton pump inhibitor therapy, although the relative response to meal stimulation was diminished in patients with severe hypergastrinaemia in comparison to our normogastrinaemic control group (P = 0.009).
Mean serum pepsinogen A levels and the A/C ratio were significantly lower in patients with hypergastrinaemia than in patients with mild hypergastrinaemia (P = 0.03 and 0.005). No difference was found in serum pepsinogen C concentrations between the groups.
Serum pepsinogen A and C concentrations are dependent on the degree of gastric mucosal inflammation. In non-atrophic gastritis pepsinogen A and C concentrations are increased and the A/C ratio is decreased. In atrophic gastritis pepsinogen A levels decrease, whereas pepsinogen C levels remain relatively stable, thus leading to a further decrease in the pepsinogen A/C ratio.27, 28 Pepsinogen A is produced by the chief cells which are present in the corpus only, whereas pepsinogen C is produced by the mucus neck cells in the duodenum, antrum and corpus. Our findings of a decrease in serum pepsinogen A and A/C ratio in the hypergastrinaemia patients are compatible with the histological signs of an increased atrophy of the corpus mucosa in these patients.
Development of severe hypergastrinaemia was associated with the presence of H. pylori. Nine out of 12 patients with extreme hypergastrinaemia were infected, compared to only two of 14 patients with mild hypergastrinaemia (P = 0.004). H. pylori is known to increase serum gastrin in infected individuals and it has been shown that H. pylori eradication can lead to normalization of serum gastrin levels.29 This is, to our knowledge, the first report of an association between the development of severe hypergastrinaemia during long-term omeprazole therapy for GERD and H. pylori infection. In agreement with this observation is the occurrence of profound active, chronic inflammation and atrophy of corpus mucosa biopsies in hypergastrinaemic patients. No differences were found in antrum biopsies for inflammation and atrophy scores between the two groups.
Somatostatin concentrations in corpus mucosa biopsies were significantly decreased in patients with hypergastrinaemia, while no differences were found in concentrations of somatostatin in antrum biopsies ( Table 4). A decrease in antrum somatostatin concentration, together with an increase in gastrin, has been described in H. pylori infected subjects,27, 30 possibly caused by a decreased ( D-cell mediated) inhibition of G-cells. Moreover, during omeprazole treatment a dose-dependent relationship was found between somatostatin and gastrin. Omeprazole 40 mg daily for 8 days produced a decrease in somatostatin and an increase in gastrin concentration in antrum biopsies.31 This effect was not observed during administration of 20 mg omeprazole. In our study a decrease in somatostatin concentration in corpus biopsies was only observed in hypergastrinaemic patients.
In conclusion, the development of severe hypergastrinaemia in patients treated long-term with omeprazole for GERD is associated with the duration of therapy and H. pylori infection, but not with abnormalities in vagal nerve function or delayed gastric emptying. These patients maintain a normal meal-stimulated gastrin response. The presence of H. pylori-induced gastritis seems to aggravate the increase of basal serum gastrin. It remains to be proven whether eradication of H. pylori can diminish serum gastrin in patients with severe hypergastrinaemia during long-term omeprazole treatment for GERD.
This study was supported by a grant from Astra Pharmaceuticals b.v., Zoetermeer, the Netherlands.