Increased risk of fundic gland polyps during long-term proton pump inhibitor therapy


Dr J. H. Kleibeuker, Department of Gastroenterology and Hepatology, University Medical Center Groningen, PO Box 30.001, 9700 RB Groningen, the Netherlands.




It is controversial whether proton pump inhibitor use leads to fundic gland polyp development.


To determine whether fundic gland polyp development is due to proton pump inhibitor use and to investigate mechanisms involved.


Proton pump inhibitor use and the presence of fundic gland polyps were assessed in consecutive patients undergoing oesophagogastroduodenoscopy. Biopsies from fundic gland polyps and gastric mucosa were taken. Dysplasia was graded as negative, low or high grade. Prevalence of parietal cell hyperplasia and parietal cell protrusions and the proportional cystic area were assessed.


599 patients participated, 322 used proton pump inhibitors, 107 had fundic gland polyps. Long-term proton pump inhibitor use was associated with an increased risk of fundic gland polyps (1–4.9 years use: OR 2.2, 95% CI: 1.3–3.8; ≥5 years: OR 3.8, 95% CI: 2.2–6.7) while short-term therapy (<1 year) was not (OR 1.0, 95% CI: 0.5–1.8). Low-grade dysplasia was found in one fundic gland polyp. Fundic gland polyps associated with long-term proton pump inhibitor use had a larger proportional cystic area and higher frequency of parietal cell hyperplasia and parietal cell protrusion.


Long-term proton pump inhibitor use is associated with an up to fourfold increase in the risk of fundic gland polyps. Risk of dysplasia is negligible. Aetiologically, these polyps seem to arise because of parietal cell hyperplasia and parietal cell protrusions resulting from acid suppression.


Fundic gland polyps (FGPs) are the most common gastric polyps.1 They are found in up to 1.9% of the general population (known as sporadic FGPs) and in up to 84% of patients with familial adenomatous polyposis (FAP), where they are known as syndromic FGPs.2–4 FGPs are characterized histologically by distorted glandular architecture consisting of microcysts, mostly lined with chief and parietal cells.5 Interestingly, FGPs occur almost exclusively in patients without Helicobacter pylori infection.4 FGPs have always been regarded as benign lesions, with at most low-grade dysplasia (intraepithelial neoplasia). There have, however, been case reports of FGPs harbouring severe dysplasia or even gastric adenocarcinoma, particularly when associated with FAP, but also in one sporadic case.6

The potential association of FGPs with the use of proton pump inhibitors (PPIs) has been a topic of debate as these drugs were first prescribed.7 PPIs, which are currently widely used to inhibit gastric acid secretion, bind irreversibly to the hydrogen/potassium pump, leading to inhibition of acid secretion by the parietal cells. In 1992, Graham reported three cases of FGPs which developed after 1 year of treatment with the PPI omeprazole.8 As this initial observation there have been numerous case reports and small series describing the occurrence of FGPs in association with long-term use of PPIs.9–14 In several of these cases regression of FGPs after cessation of PPI therapy was reported.10, 13 Two studies in larger series of patients have shown conflicting results regarding the prevalence of FGPs in patients on long-term PPI therapy when compared with a control group (Table 1).15, 16 Another four studies, published as abstract or letters, but never as full paper, indicate that there could be an association between FGPs and PPI use (Table 1).17–20 A case–control study in FAP patients suggested that the risk of dysplasia in FGPs increases during PPI use.21 Thus, although PPIs are widely regarded as safe, the discussion whether the association between FGPs and PPI use exists continues. It has even been suggested that routine endoscopies should be performed in patients on long-term PPI therapy to monitor FGP development9, 22 and the question whether there is a contraindication for long-term PPI therapy in FAP patients remains unanswered. To settle this issue, it is necessary to assess the prevalence of FGPs in patients on long-term PPIs in comparison with patients not using PPIs. The relatively low prevalence of FGPs makes a case–control study most feasible.23

Table 1.   Summary of case–control studies regarding the association between PPIs and FGPs
AuthorStudy typePPI useNo PPI useP-value
NMonthsFGPs (%)NFGPs (%)
  1. * Baseline values for same patients.

  2. † Fisher's exact test.

  3. ‡ Chi-squared test.

  4. § Reported as full paper.

  5. ¶ Only reported as abstract or letter.

  6. N, number of patients; N.S., not significant; PPI, proton pump inhibitor; FGP, fundic gland polyp.

Choudhry et al.15§Retrospective2310–98520720.3<0.001‡
Vieth and Stolte16§Retrospective2251>1528 0965N.S.
Collins and Tydd18Retrospective5412–603554*0*<0.001‡
Mogadam and Houk19Prospective2718–48261332<0.001‡
Reekmans et al.20Prospective115Long928661<0.0001†

The aims of this case–control study were to determine whether FGP development is related to PPI use. We hypothesized that longer PPI use is associated with a higher prevalence of FGPs and that the acid suppression is directly involved in the development of the FGPs. We collected data regarding PPI use in FAP and other patients before undergoing diagnostic oesophagogastroduodenoscopy and looked specifically for FGPs. When FGPs were found, biopsies were collected for histological verification and further investigation.

Materials and methods

Patient characteristics

All patients undergoing diagnostic oesophagogastroduodenoscopy at the endoscopy department of the University Medical Center Groningen between November 2002 and March 2005 were asked to participate in the study and to complete a questionnaire regarding the use of PPIs. The type of PPI used, the dosage used and the length of use were recorded. The reported data were verified and completed using the hospital patient records where possible and, when necessary, missing information was obtained from the general practitioner. Intermittent PPI use was evaluated as no PPI use. PPI use was graded into three categories: no reported use, <1 year PPI use and >1 year PPI use. The category >1 year PPI use was further divided into 1–4.9 years and 5 or more years PPI use. For each patient, the sex and age were also recorded. A database containing information about FAP patients (n = 50) treated at the University Medical Center Groningen from 1970 to 2005 was available. In each patient, the diagnosis FAP had been established by the presence of hundreds of adenomatous polyps found at colonoscopy at an early age, histopathological examination and, where possible, APC gene mutation analysis. This database was used to identify patients with FAP in our study population. During the study period 33 FAP patients had one or more oesophagogastroduodenoscopies.

Collection of FGPs

All oesophagogastroduodenoscopies were performed by five experienced gastroenterologists who had been additionally instructed in the recognition of FGPs and the method of data collection. When FGPs were found, the following characteristics of the FGPs were registered: localization in corpus and/or fundus, number of FGPs, estimated maximal size and any unusual characteristics. At least two biopsies were taken from each of the following: the largest corpus FGP, the largest fundus FGP, normal looking antrum mucosa and normal looking corpus mucosa. The biopsies were fixed in formalin, embedded in paraffin and stained with haematoxylin and eosin (H & E) for routine histology.

Histological examination

The H & E slides were retrieved from the files of the Department of Pathology, University Medical Center Groningen. The histological classification of FGPs was revised by one pathologist (JW). The diagnosis FGP was confirmed when histological examination of the endoscopically identified polyp revealed fundic mucosa with cystically dilated glands as defined by Lee and co-workers.24, 25 Criteria for the diagnosis of dysplasia (intraepithelial neoplasia) included nuclear enlargement or pleomorphism, stratification and hyperchromatism.26 Dysplasia was graded as negative, low grade or high grade, according to Wu et al.27 In the antral and corpus biopsies, the presence of H. pylori infection was noted for each patient using both H & E and Giemsa staining. In FGPs and biopsies of normal looking corpus mucosa from 15 patients from each PPI-use group additional histological analysis was performed. The proportional cystic area of each polyp was measured using photographs. First, the total area of the FGP tissue segment was measured followed by the total area of the cysts within the FGP to determine the area of cysts as a percentage of the total area of the polyp (Image-pro plus, version 4.6, MediaCybernetics, Silver Spring, MD, USA). The presence of parietal cell hyperplasia (PCH) was determined by estimating the percentage of parietal cells in the total cell population of the FGP tissue segment semiquantatively as absent (0%), <50% parietal cells and >50% parietal cells. Parietal cell protrusions (PCP) were defined as prominent intraluminal protrusions of parietal cell cytoplasm resulting in a serrated glandular lumen and quantified as absent, moderately present (<50% of cells) or extensively present (>50% of cells) in the FGP.28

Statistical analysis

spss for Windows software (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses. Group sample sizes of n = 300 PPI users and n = 300 non-PPI users achieve 81% power to detect a difference of 8%, assuming the proportion of FGPs in the group non-PPI users is 10% (α = 5%, tested two-sided). The odds ratios and 95% confidence intervals (CI) for the presence of FGPs during PPI treatment were calculated. Differences between patient groups were tested using the Mann–Whitney U-test and the chi-square test. Multivariate logistic regression analysis was used to determine influence of both age and PPI use on the prevalence of FGPs. P-values of <0.05 were considered significant.


Patient characteristics

Table 2 summarizes the characteristics of patients described in this study. Patients with FGPs were older than patients without FGPs (60 ± 12 years vs. 52 ± 16 years, respectively, mean ± s.d., P < 0.001). As shown in Table 3, the mean age of non-FAP patients on long-term (>1 year) PPI therapy (58 ± 13, mean ± s.d.) was higher than that of patients not using PPIs (51 ± 17, mean ± s.d., P < 0.001) and patients on short-term (<1 year) PPI therapy (53 ± 15, mean ± s.d., P = 0.001). About 33 of the 50 FAP patients underwent oesophagogastroduodenoscopy during the study period. As expected the mean age of these 33 patients was lower than that of non-FAP patients (44 ± 16 vs. 53 ± 16, mean ± s.d., P < 0.001). None of the patients with FGPs had H. pylori or signs of gastritis.

Table 2.   Patient characteristics
 Non-FAP patientsPatients with FAP
  1. FAP, familial adenomatous polyposis; PPI, proton pump inhibitor; FGP, fundic gland polyp.

Male/female (%)42/5846/54
Mean age (range)53 (17–88)44 (15–92)
PPI use, % (N)
 None46 (277)82 (27)
 <1 year27 (162)6 (2)
 >1 year27 (160)12 (4)
FGP, % (N)
 Absent82 (492)42 (14)
 Present18 (107)58 (19)
Table 3.   FGPs in patients with no, <1 year and >1 year of PPI use
 No PPI use<1 year PPI>1 year PPI
  1. PPI, proton pump inhibitor; FGP, fundic gland polyp.

Male, % (N)38 (105)46 (75)44 (70)
Age, mean (range)51 (17–88)53 (21–83)58 (22–86)
FGPs, % (N)12 (32)11 (18)36 (57)

FGPs are associated with long-term PPI use

The prevalence of FGPs in non-FAP patients without PPI therapy and on short- and long-term PPI therapy is shown in Table 3. PPI therapy was associated with an increased risk of FGPs (OR 2.3, 95% CI: 1.5–3.6). Subgroup analysis showed that short-term PPI therapy (<1 year) was not associated with an increased risk of FGPs (OR 1.0, 95% CI: 0.5–1.8). Long-term PPI use (>1 year) was associated with an increased risk of FGPs (OR 2.8, 95% CI: 1.8–4.5). Within the group of long-term PPI users the risk was higher in patients with 5 or more years of PPI therapy (OR 3.8, 95% CI: 2.2–6.7) than for patients with 1–4.9 years of PPI therapy (OR 2.2, 95% CI: 1.3–3.8). These results are illustrated in Figure 1. Multivariate logistic regression analysis showed that both age and PPI use were independent predictors of the presence of FGPs (P < 0.001 for both). There was no difference in FGP risk in patients treated with omeprazole when compared with other PPIs, indicating that the effect is a class effect and not related specifically to omeprazole.

Figure 1.

 Odds ratios (with corresponding 95% confidence intervals) for fundic gland polyps (FGPs) in patients using proton pump inhibitors (PPIs). ‘Total’ shows the risk of all PPI users vs. patients who had never used PPIs. <1 year, >1 year, 1–5 years and >5 years show the risk of FGPs in patients who had used PPIs for these time periods when compared with patients who had never used PPIs.

Endoscopic and histological characteristics of FGPs

There were no differences in the localization, size or number of FGPs found in patients without PPI use, patients with short-term (<1 year) PPI use and patients with long-term (>1 year) PPI use (Table 4). In the series of non-FAP FGPs dysplasia was very rare. In one of 107 FGPs there was a focus of low-grade dysplasia. In the remaining non-FAP FGPs no dysplasia was found. Dysplasia was much more common in FAP-associated FGPs: 46% of these FGPs were dysplastic, 8% had high-grade dysplasia and 38% low-grade dysplasia.

Table 4.   Characteristics of non-FAP FGPs
 No PPI useShort-term PPI useLong-term PPI use
  1. PPI, proton pump inhibitor; FGP, fundic gland polyp; FAP, familial adenomatous polyposis.

Number of patients321857
Localization (%)
Number (%)
Size (%)
 ‘small’ undefined38633
 1–4 mm282217
 5–9 mm192218
 >10 mm62211

In FGPs, PCH and PCP were more frequent in both the FGPs and the biopsies of normal gastric corpus mucosa from patients with long-term (>1 year) PPI use compared to patients with no reported PPI use (Figures 2 and 3). The proportion of the FGP consisting of cysts was higher in the FGPs from patients with long-term (>1 year) PPI use (15% ± 2%, mean ± S.E.M.) than in FGPs from patients without reported PPI use (9% ± 2%, mean ± S.E.M., P = 0.035; Figure 3).

Figure 2.

 Haematoxylin and eosin-stained tissue sections of fundic gland polyps (FGPs). (a) FGP from a patient who had never used proton pump inhibitors (PPIs) showing a few small cysts (arrow), original magnification – 40×. (b) FGP from a patient with more than 1 year of PPI use showing a large number of cysts (arrow), original magnification – 40×. (c) FGP from a patient with more than 1 year of PPI use showing parietal cell hyperplasia and parietal cell protrusions (arrow), original magnification – 100×. (d) FGP from a patient with more than 1 year of PPI use showing parietal cell hyperplasia and parietal cell protrusions (arrow), original magnification – 400×.

Figure 3.

 Parietal cell hyperplasia in fundic gland polyps (FGPs) (a) and normal looking gastric corpus mucosa (c) expressed depending on length of proton pump inhibitor (PPI) use. Differences between groups were significant (P < 0.01). Parietal cell protrusions in FGPs (b) and normal looking gastric corpus mucosa (d) expressed depending on duration of PPI use. Differences between groups were significant (P < 0.01). (e) Mean percentage of surface area made up of cysts in FGPs from patients without PPI use and from those with <1 or >1 year of PPI use. Mean percentage of cysts is higher in patients on >1 year PPI therapy* compared with the mean percentage in FGPs from patients with no PPI use (P = 0.035).

PPI use is not related to FGP development in FAP patients

Three FAP patients with and three FAP patients without FGPs were on PPI therapy. Two FAP patients had a large adenomatous polyp, both with low-grade dysplasia, in a bed of FGPs. Endoscopy 3 years earlier in both patients had shown FGPs, but no adenomas, in one patient and no abnormalities in the other patient. The first of these two patients used PPIs, but only for 1 month. There was also one FAP patient with high-grade dysplasia in a FGP, but this patient was not on PPI therapy either. Of the remaining patients (n = 16), 30% had low-grade dysplasia in the FGPs, with only one of these patients using PPIs.


In this study, we show that long-term use of PPIs is associated with a fourfold increase in the risk of developing FGPs. Furthermore, we show that the risk of FGPs increases with longer PPI use. Microscopically, FGPs from patients treated with PPIs were characterized by more and/or larger cysts and more frequent PCH and PCP in both the FGPs and the normal looking corpus mucosa compared with FGPs and normal mucosa from patients without PPI use. This data supports an aetiology related to acid-suppressive therapy.

Although numerous case reports and small series have described FGPs developing during long-term PPI use, studies, designed to prove a firm association, have shown conflicting results. The incidence of FGPs after (long-term) PPI therapy has been reported to be between 1% and 36%.15–20 In the largest retrospective study, with over 30 000 patients, Vieth and Stolte showed no difference in the prevalence of FGPs between controls and patients after at least 4 weeks of PPI therapy.16 This is not surprising considering that in our study and all other positive studies an association was only found after at least 12 months of PPI therapy. In accordance with this, in the retrospective study by Choudry et al., a higher risk of FGP development was found after a mean of 37 months of PPI therapy.15 Four other studies which have only been reported in abstract form or as a letter have all shown an increased risk of FGP development after at least 12 months of PPI therapy.17–20 Our study is the first to demonstrate that the risk of FGPs continues to increase with longer PPI use. Data on length and frequency of PPI use was based on patient recall, so this may have introduced a bias. However, the data for the majority of the patients could be confirmed using hospital records and general practitioner records, thereby reducing the chance that this inaccuracy significantly influenced the results.

Dysregulation of the Wnt-APC-β-catenin pathway, as well as CpG island methylation and cellular distribution of tuberin and the glucocorticoid receptor, have been shown to play a role in the development of both non-FAP and FAP-associated FGPs, independently of PPI use.2, 24, 29, 30, 31 The mechanisms through which FGPs develop during PPI therapy are, however, largely unknown. The development of hyperplasia and protrusions of the parietal cells (PCH and PCPs) is thought to be an initial step.32 PCPs are defined as hypertrophic parietal cells showing tongue-like protrusions of the apical membrane into the lumen of corpus glands. Cats et al. found that the prevalence of PCP increased during PPI therapy.33 The largest increase was seen in the first 3 months of PPI use and this was related to a rise in serum gastrin levels.33 It is thought that the morphological changes leading to PCP occur because the secretory canaliculi are filled with hydrochloric acid, additionally stimulated by higher levels of gastrin, but active secretion of the acid is inhibited by the PPI therapy.33 A second step in the formation of FGPs is thought to be fundic gland cyst (FGC) formation.33 FGCs are intramucosal cysts which are believed to form from glands which are dilated because of increased intraglandular pressure. This increase in pressure could be caused by increased resistance to outflow from the gland due to blockage of the isthmus by PCPs.34 Cats et al. described FGC prevalence to increase from 8% at baseline to 35% after 12 months of PPI therapy.33 FGCs can present endoscopically as FGPs when they become larger.

Interestingly, FGPs develop almost exclusively in H. pylori-negative patients and FGCs also develop more often in H. pylori-negative patients.35 The absence of H. pylori infection in our patients with FGPs is in accordance with these previous observations. It has been suggested that enzymatic degradation of gastric mucus by H. pylori protease may facilitate the glandular outflow and thus protect against retention and cystic dilation.33 This does not take place in H. pylori-negative patients thus increasing the risk of FGP development. In summary, present data indicate that PCPs develop in the majority of patients during the first months of PPI therapy in association with an increase in serum gastrin level. During the first year of PPI therapy a subgroup of patients without H. pylori infection develop FGCs which are not visible endoscopically but can be identified in biopsies of normal looking gastric mucosa.33 As also described in the present study, an increased prevalence of macroscopic FGPs can be seen after at least 12 months of PPI therapy.17–20 These data suggest that an aetiology of PPI-associated FGPs differs from that of FGPs which develop without PPI use.

Based on our data, the development of FGPs and/or dysplasia in FGPs in FAP patients does not seem to be related to PPI therapy. Attard et al. found dysplasia in FGPs in two of two paediatric FAP patients on long-term PPI therapy when compared with three of seven paediatric FAP patients not using PPIs.21 We did not find this association in our adult population and it has not been reported elsewhere suggesting that the observation by Attard et al.21 was based on chance or is specific for the paediatric population.

Although interesting from a mechanistic point of view, there seem to be few clinical implications of the association between long-term PPI therapy and FGP development as there is no increased risk of cancer developing in these polyps. In conclusion, long-term use of PPIs is associated with a fourfold increase in the risk of developing FGPs. The risk of dysplasia in these polyps is not increased suggesting that additional endoscopic evaluation of these polyps is not indicated.


We would like to thank N. Zwart for assisting with the data collection. Furthermore, we would like to thank the staff at the endoscopy center, University Medical Center Groningen, Groningen, the Netherlands for help in the data collection. No financial support has been received for this study.