Changes of gastric mucosal architecture during long-term omeprazole therapy: results of a randomized clinical trial


Dr L. Lundell, Gastrocentrum, Karolinska University Hospital, Huddinge, S-141 86 Stockholm, Sweden.


Background  The impact of long-term acid suppression on the gastric mucosa remains controversial.

Aim  To report further observations on an established cohort of patients with gastro-oesophageal reflux disease, after 7 years of follow-up.

Methods  Of the original cohort randomized to either antireflux surgery or omeprazole, 117 and 98 patients remained in the medical and surgical arms, respectively. Gastric biopsies were taken at baseline and throughout the study.

Results  Fifty-three antireflux surgery and 39 omeprazole-treated patients had Helicobacter pylori infection at randomization. Eighty-three omeprazole-treated and 60 antireflux surgery patients remained H. pylori negative over the 7 years, and no change was observed in mucosal morphology except for a change in endocrine cell population (linear and diffuse hyperplasia, P = 0.03). During the 7-year study many patients, who were initially H. pylori infected, had the infection eradicated leaving only 13 omeprazole and 12 antireflux surgery patients still infected. In these patients, omeprazole induced a deterioration of the mucosal inflammation scores (P = 0.01) with a numerical increase of glandular atrophy.

Conclusions  Long-term omeprazole therapy does not alter the exocrine oxyntic mucosal morphology in H. pylori-negative patients, but mucosal endocrine cells appear to be under proliferative stimulation; in H. pylori-positive patients there are changes in mucosal inflammation and atrophy.


Long-term proton pump inhibitory (PPI) therapy has gained widespread acceptance as an effective and safe therapy for gastro-oesophageal reflux disease (GERD). Despite the overall favourable safety profile of this class of drugs, one significant concern has arisen during the recent years. This relates to the interaction with and an eventual facilitation of the Helicobacter pylori (Hp)-driven gastric mucosal processes, eventually leading to glandular atrophy.1–3 No doubt acid inhibition therapy induces a shift towards a corpus predominant gastritis with a concomitant improvement in inflammatory parameters of the antral mucosa of Hp-infected subjects.4–6 Cohort-based studies have suggested that the risk of developing glandular atrophy in these situations are increased many fold compared to controls.3, 7–10 When these issues have been explored under controlled randomized trial conditions, no supportive evidence was gained.11

Profound acid inhibition therapy is, however, regularly followed by bacterial overgrowth of the gastric juice.12–14 The clinical significance of this has also been debated ever since the introduction of the histamine H2-receptor antagonists. One novel aspect of the potential harmful effects of this bacterial flora is to either facilitate or induce gastric mucosal inflammatory responses, which might add to the adverse effects as alluded to above.15 As part of a randomized controlled clinical trial comparing omeprazole (ome) with antireflux surgery (ARS) we had the opportunity to address these safety aspects on inflammation and atrophy developments of the gastric mucosa after 7 years of follow up.

Material and methods

The primary aim of the study was to compare the clinical efficacy and safety of treatment with ome and ARS in patients with GERD. This is the second interim analysis where the first one was performed after 3 years.11 Patients with symptoms of chronic GERD and documented oesophagitis at endoscopy, and who also were considered suitable for ARS, were eligible for inclusion into the study. These patients were initially given 20–40 mg of ome daily to control symptoms and to heal the oesophagitis, usually for 4–8 weeks, but according to the protocol a treatment period of 4 months at the most was allowed. More than 8 weeks of active therapy was only exceptionally required. Valid data on pre-entry consumption of antisecretory drugs other than PPI were not captured.

Initially 344 patients were enrolled but during the study period 34 patients had an incomplete response to different doses of ome daily. Thus, these patients, according to the protocol, were offered and accepted antireflux surgery. Consequently 310 patients were randomized of whom 155 were allocated to ome, 154 finally treated (138 had 20 mg and 16 patients 40 mg daily) and 155 to ARS.

Nine patients initially allocated to ARS, eventually refused to be operated on and one patient had a bronchial carcinoma diagnosed after randomization and thus the operation was cancelled. One patient was excluded because no oesophagitis had been documented, leaving 144 patients in the surgical therapy arm for follow up. All but 24 patients in the surgery group had the operation done within 3 months from the randomization (median 62 days), during which time they were kept symptom free on ome. One patient, who had an uneventful post-operative course, died 3 months after the operation due to myocardial infarction. During the 7 years of follow-up, in total 46 patients were withdrawn from the ARS group, three due to unacceptable adverse events, two due to non-compliance with the study protocol, six were lost to follow-up, one patient did not comply with the follow-up procedures, one due to other reason, and 33 did not continue after the first 5 years, leaving 98 patients who completed the 7-year follow-up. In the medical arm 1 patient was never treated with ome and among the remaining patients the respective withdrawal figures were 11, 2, 4, 5, 1 and 18, leaving 113 patients who were adequately followed. The details of the operative procedures have been given elsewhere.11

Relevant demographic information in the two study groups is detailed in Table 1, showing that there were no apparent differences between the two study groups. In the ome arm 30 patients were Hp positive at baseline compared to 36 among those treated by surgery. During the 7 year follow-up 10 ome-treated patients converted to become negative and 11 among those who had ARS.

Table 1.  Number of patients by baseline characteristics and Helicobacter pylori (Hp)-status at baseline and 7 years, patients followed for 7 years
Hp neg–negHp neg–posHp pos–negHp pos–posHp unknownTotalHp neg–negHp neg–posHp pos–negHp pos–posHp unknownTotal
Age (years)
Smoking habits (cigarettes/day)
Alcohol consumption (drinks/week)

At the time of entry into the study (i.e. after initial oesophagitis-healing therapy) and during the subsequent course, endoscopy was performed on an outpatient basis in patients fasted overnight. At each visit, the macroscopic appearance of the oesophageal mucosa was assessed according to predefined criteria.16 Simultaneously a complete macroscopic investigation of the whole stomach and proximal duodenum was also performed. According to the protocol endoscopic investigations were scheduled 1, 3, 5 and 7 years after randomization.


At each endoscopic investigation, six biopsy specimens were taken from the midportion of the corpus mucosa along the greater curvature, approximately 10 cm distal to the gastro-oesophageal junction. The details of the further processing of the biopsy specimens have been described previously.11 One series of sections with haematoxylin and eosin (H&E) for general pathological analysis, the second was stained with Sevier–Munger silver staining technique17 for argyrophil (enterochromaffin-like) cells and the third series was impregnated with Steiner silver for detection of Hp. Complementary staining was used in the form of Grimelius’ silver staining18 and chromogranin A immunostaining for endocrine cells. To firmly establish the presence or absence of Hp infection, an immunohistochemical technique was used with polyclonal antibodies against Hp (Dakopatts, Copenhagen, Denmark) in dilutions 1:100 and 1:500 for primary and secondary antibodies respectively. The presence of Hp was also assessed according to the Sydney system.19

In the section stained with H&E, the corpus mucosa was assessed for the following variables: inflammation, activity, atrophy and intestinal metaplasia.19 These variables were scored as either normal, mild, moderate or severe. Intestinal metaplasia was graded quantitatively in the same way as atrophy. Intestinal metaplasia was checked when needed with alcian blue (pH 2.5)/periodic acid–Schiff for acid and neutral sialomucins and with high-iron diamine for acid-sulfated mucins.

The distribution and character of the endocrine cell population were assessed according to the system described in detail by Solcia et al.20 The scores range from normal, to diffuse, linear and micronodular hyperplasia. Eventual dysplastic and neoplastic changes had to be recorded but not scored.

The same pathologist at AstraZeneca, Södertälje, Sweden (N.H.), who was unaware of the group affiliation of the patient as well as the time during the study when actual biopsy specimens had been taken, assessed all biopsy specimens.

Statistics and ethics

Statistical evaluations of the data were performed by use of cross-tabulation and descriptive statistics. In addition, the difference between ARS and ome therapy, with respect to the changes from baseline to the last value of gradings of the inflammation and the point prevalences of glandular atrophy within the corpus mucosa, were assessed with the Wilcoxon rank sum test. The same test was used for evaluation of differences from baseline to 7 years on therapy within respective group.

Before inclusion into the trial, informed consent to participate was obtained from all subjects. The local ethics committees approved the study protocol.


Exocrine oxyntic mucosa

Hp-negative patients

Sixty patients who initially had an antireflux operation and 83 ome-treated patients were Hp negative throughout the study period. We were unable to demonstrate any significant changes in the histomorphology of the gastric corpus mucosa, neither between the groups nor within respective groups. (Tables 2 and 3).

Table 2.  Number of patients by severity of the corpus mucosal inflammation at baseline and last visit (omeprazole or antireflux surgery) in Helicobacter pylori non-infected GERD patients
BaselineLast value
 None652  67
 Mild142  16
 All794  83
 None492  51
 Mild62  8
 Moderate 1 1
 All555  60
Table 3.  Number of patients by severity of the glandular atrophy of the oxyntic mucosa of Helicobacter pylori non-infected patients at baseline and after 7 years of therapy
BaselineLast value
 None81   81
 Mild2  2
 All83   83
 None59   59
 Moderate1  1
 All60   60

Hp-infected patients

Depending on the quite substantial number of patients who for, various reasons, converted to become Hp negative, we had only 13 ome and 12 ARS patients who remained Hp-infected throughout the 7 years of therapy. There was a slight but consistent change in the inflammatory score (density of lymphoplasmo cellular elements) in the mucosa of ome-treated patients (Table 4), without corresponding change in the operated patients (P = 0.01, compared to ARS). Regarding corpus mucosal activity (degree of infiltration with neutrophils) we recorded a stable situation in both study groups (Table 5).

Table 4.  Gastric mucosal inflammation on the oxyntic mucosa at baseline and after 7 years in persistently Helicobacter pylori-infected patients
BaselineLast value
 Mild 1427
 Moderate  134
 Severe   22
 All 15713
 None   11
 Mild 2  2
 Moderate 1416
 Severe 1113
 All 45312
Table 5.  Number of patients by severity of activity of the corpus mucosa of Helicobacter pylori-infected patients at baseline and after 7 years of therapy
BaselineLast value
 None 1  1
 Mild 42 6
 Moderate 22 4
 Severe  112
 All 75113
 None  1 1
 Mild12  3
 Moderate 2 35
 Severe 1113

Assessment of corpus glandular atrophy revealed two ome-treated patients who developed severe atrophy from having had none at baseline and two went from none to mild atrophy at the 7-year follow-up (Table 6). No single one of the 12 Hp-infected ARS patients developed severe atrophy but three were scored as having mild atrophy at the last follow-up visit from having had a normal score at baseline. These numerical differences between groups did not reach statistical significance. Very few cases (in total four) of intestinal metaplasia were recorded but only of type I with no change over time.

Table 6.  Glandular atrophy of the oxyntic mucosa in Helicobacter pylori-infected patients at baseline and after 7 years of therapy
BaselineLast value
 None72 211
 Mild1 1 2
 None83  11
 Moderate1   1
 All93  12

Oxyntic endocrine cells

In Hp-negative patients treated with ome for 7 years we recorded a slight increase in the point prevalence of diffuse and linear hyperplasia (six vs. two cases) (ns). One patient developed micronodular hyperplasia from having a normal finding in the baseline biopsies, whereas the reverse was true in another patient. In the surgical group, two patients developed a similar diffuse and linear changes respectively. Also in the permanently Hp-infected patients a similar pattern emerged. However, in these subjects a tendency to develop micronodular hyperplasia over time seemed to be more apparent in the ome treated patients (P = 0.03) (Table 7a and b).

Table 7.  Argyrophil cell characteristics (hyperplasia) of the oxyntic mucosa at baseline and after 7 years of therapy in (a) Helicobacter pylori (Hp)-non-infected and (b) Hp-infected GERD patients
BaselineLast value
 None7532 80
 Diffuse  112
 Micronodular1   1
 None5811 60
 All5811 60
 Normal62 412
 Micronodular   11
 All62 513
 None9  211
 Micronodular1   1
 All10  212


This study addresses some aspects on morphological changes in the exocrine and endocrine components of the oxyntic mucosa during long-term profound acid inhibition therapy. We tried to add further information to the question whether PPI therapy facilitates the Hp-induced processes, which ultimately lead to destruction of gastric-oxyntic glands ? Secondly, the basic study design offered the opportunity to elucidate the question of whether long-term acid inhibition, which by necessity induces upper gastro intestinal bacterial overgrowth, might even affect the morphology of the oxyntic mucosa in Hp-negative subjects?

In the present study, covering a treatment period of 7 years, we observed a successive increase in the severity of mucosal inflammation in the Hp-infected ome-treated GERD patients compared to those having only ARS. This inflammatory response, driven by the Hp-infection, appeared to be a slow process with small but significant differences between our study groups. Studies have repeatedly demonstrated that short-term PPI therapy induces a worsening in the grade of the co-existing gastritis in Hp-infected subjects.21–23 When gastric pH is increased, Hp density increases in the corpus area, the consequences of which are more pronounced inflammation.6, 24, 25 On the contrary, the degree of activity of the corpus mucosa was relatively stable over time. This is an important observation, not the least since granulocytes are considered to be the source of the reactive oxygen metabolites and therefore may be involved in the processes leading to damage of the surface epithelium.26–29

The current long-term study was adversely affected by the fact that from the original cohort of Hp-positive patients we subsequently found that a substantial number of those had been eradicated from the infection during the course of the study. This is probably due to unintentional antibiotic treatment for other infectious diseases, since our ambition was to maintain the study cohort intact for the entire duration of the study period. As a consequence, 12 and 13 patients, in respective study group, remained infected when gastric biopsies were obtained 7 years after randomization, thereby severely restricting our possibility to address the atrophy issue in the ome-treated Hp-positive subjects. There seemed, however, to be a trend towards more cases of glandular atrophy to develop in the medical treatment arm. In fact the annual incidence of glandular atrophy, based on our own data, was in the same magnitude as previously calculated by other investigators.30–32 Importantly enough we were unable to detect any case with type III intestinal metaplasia, irrespective of therapy.33–36 A recent controlled randomized study showed that not only the corpus predominant pangastritis, developing during long-term PPI therapy, was eliminated by Hp eradication, but also that the glandular atrophy seemed to regress.37 Similar observations are of key importance suggesting reversibility of corresponding epithelial changes, giving support to the proposal that Hp eradication should be recommended when embarking on long-term PPI therapy in similar cases.38

The acidic environment within the stomach is considered to be the most important barrier against colonization and infection.12–14, 39 Hence it is well established that states of low acid secretory capacity (pH ≥ 4) all predispose to colonization of the stomach by aerobic and to a minor degree, by anaerobic, Gram-positive and Gram-negative bacteria.40, 41 An intense discussion has been going on and is still vivid about the potential adverse consequences of these phenomena: increased susceptibility for enteric infections, alterations in nutrient absorptions and formation of potentially carcinogenic N-nitroso compounds coupled to the pro-inflammatory, inflammatory processes in progress within the gastric mucosa and depletion of vitamin C levels.42, 43

A unique database on the incidence of general gastro-intestinal symptoms, on long-term ome therapy, comes from the comparison with ARS where the 5-year follow-up data have been given in detail elsewhere.44 These data clearly showed equality between the two treatment groups regarding the profile of adverse gastrointestinal events, suggesting an unchanged susceptibility to symptomatic enteric infections at least in the Western world.

An additional aspect on the role of non-Hp bacterial flora in the gastric juice would be to act as a cofactor in the development of chronic inflammation and ultimately to gland destruction.45, 46 Data have been presented to show that in Hp-infected subjects treated with acid suppression, the presence of non-Hp bacteria preceded by time the development of corpus atrophic gastritis. Secondly, the magnitude and ranking of the cytokine increase were associated with the risk of having atrophic gastritis.47 It is also possible that non-Hp bacteria and/or their by-products can act as persistent antigenic stimuli that induce an inflammatory response per se and/or to enhance the inflammatory cascade initiated by the Hp-infection.41 In our GERD patients being devoid of Hp-infection, chronic acid suppression did not induce any mucosal changes suggestive of pro-inflammatory or inflammatory drive, when studied over 7 years. Consequently, we can conclude that it is most unlikely that the non-Hp bacteria exert detrimental effects on the human oxyntic gland mucosa.

The stomach is a powerful endocrine organ as well. In the corpus mucosa the dominating endocrine cells are the histamine-producing enterochromaffine-like cells (ECL-cells).48, 49 Intragastric acidity regulates gastrin release from the antral G cells through a negative feedback mechanism. Gastrin exerts a powerful trophic effect on ECL-cells in all species studied.20, 50 Consequently, it has been shown that maintained acid suppression is followed by an increase in the prevalence of linear-micronodular ECL-cell hyperplasia, which was more prominent in Hp-infected subjects than in those who were non-infected.2, 51 The present results confirm and extend these observations although we were able to statistically substantiate this exaggerated process only in Hp-infected subjects. This more clear difference in the ECL-cell morphology seen in Hp-infected subjects may be depending on the diminution of surrounding cells as a consequence of the destruction of glands.52, 53 Theoretically it can, however, be argued that in these subjects the endocrine cell population is exposed to a more pronounced proliferative stimulation due to the synergistic effect of acid suppression therapy and Hp-driven process on gastrin release from the antral G cells. From a purely methodological viewpoint, the long-term effects of hypergastrinaemia per se are better elucidated in the Hp-negative subjects. Consequently, we can conclude that the chronically elevated gastrin levels seem to exert a continuous proliferative drive on the endocrine cell population of the oxyntic gland area in PPI-treated patients. Over time the point prevalence of any type of ECL-cell hyperplasia increased from 3/83 at baseline to 7/83 after 7 years on therapy. Do these figures give rise to concern regarding safety? This is most likely not the case due to two factors. First, the corresponding figures in the controls were 0 vs. 2/60 and secondly until now we have not observed any dysplastic or neoplastic changes in the endocrine cell population. Recent data have been presented to show that serum chromogranin A levels may reflect the dynamics of ECL cells of the mucosa54, 55 and that this marker expresses a gradual increase over time in Hp-infected subjects treated long-term with PPI therapy. Based on the current knowledge, it is therefore important to further study the morphology of endocrine cell population of the human stomach when patients have been exposed to acid inhibitory drug therapy for ≥10 years.


No external funding was received for this study.