Article Text
Abstract
Aim To assess characteristics of oxyntic gastric atrophy (OGA) in autoimmune gastritis (AIG) compared with OGA as a consequence of Helicobacter pylori infection.
Methods Patients undergoing oesophagogastroduodenoscopy from July 2011 to October 2014 were prospectively included (N=452). Gastric biopsies were obtained for histology and H. pylori testing. Serum gastrin-17 (G17), pepsinogen (PG) I, PGII and antibodies against H. pylori and cytotoxin-associated gene A protein were determined in all patients. Antibodies against parietal cells and intrinsic factor were determined in patients with advanced (moderate to severe) OGA. Areas under the receiver operating characteristic curves (AUCs) were calculated for serum biomarkers and compared with histology.
Results Overall, 34 patients (8.9%) had advanced OGA by histology (22 women, age 61±15 years). Current or past H. pylori infection and AIG were present in 14/34 and 22/34 patients, respectively. H. pylori-negative AIG patients (N=18) were more likely to have another autoimmune disease (OR 6.3; 95% CI 1.3 to 29.8), severe corpus atrophy (OR 10.1; 95% CI 1.9 to 54.1) and corpus intestinal metaplasia (OR 26.9; 95% CI 5.3 to 136.5) compared with H. pylori-positive patients with advanced OGA. Antrum atrophy was present in 39% of H. pylori-negative AIG patients. The diagnostic performance of G17, PG I and PGI/II was excellent for AIG patients (AUC=0.83, 0.95 and 0.97, respectively), but limited for H. pylori-positive patients with advanced OGA (AUC=0.62, 0.75 and 0.67, respectively).
Conclusions H. pylori-negative AIG has a distinct clinical, morphological and serological phenotype compared with advanced OGA in H. pylori gastritis.
- GASTRIC PATHOLOGY
- AUTOIMMUNITY
- HELICOBACTER PYLORI
Statistics from Altmetric.com
Introduction
Gastric atrophy (GA) is a condition characterised by a decreased density or loss of appropriate glands in the stomach and their replacement with interglandular extracellular matrix and/or metaplastic changes of the gastric glands.1The aetiology of GA is restricted to two risk factors: Helicobacter pylori (H. pylori) infection and autoimmune gastritis (AIG). According to the updated Sydney system classification, GA can be scored as absent, mild, moderate or severe.2 GA can involve pyloric glands (gastric antrum), oxyntic glands (gastric corpus) or both (extensive atrophy).3 According to the current evidence, the causes of oxyntic gastric atrophy (OGA) are (1) H. pylori infection (environmental aetiology), (2) AIG (primary autoimmunity) and (3) combined H. pylori-autoimmune aetiology.4–6 OGA independent of its aetiology is a condition associated with hypo- or achlorhydria, and low serum pepsinogen (PG) I concentrations.7 In patients with autoimmune OGA, secondary hypergastrinaemia, antral G cell hyperplasia and corpus enterochromaffin-like (ECL) cell hyperplasia are commonly observed.8 Patients with OGA are at increased risk for developing iron-deficiency anaemia, pernicious anaemia, type 1 neuroendocrine tumour (NET) of the gastric body and gastric cancer (GC).9
Gastrin-17 (G17), PGI and PGI/II ratio are biomarkers that can be measured in serum to screen for advanced (moderate to severe) OGA. PG I is produced by chief cells which are part of cellular components of oxyntic glands whereas PG II is produced in the whole gastric and duodenal mucosa.10 In patients with advanced OGA serum PG I levels are decreased whereas PG II levels are normal or may even be increased. Therefore, low serum PG I levels or a low ratio of PG I to PG II are used as surrogate markers for advanced OGA.11 In OGA the decreased gastric acidity results in loss of the negative pH control effect and consequent increased gastrin release from antral G cells. Thus, increased serum G17 is a marker for hypo hypochlorhydria or achlorhydria.
The clinical usefulness of G17, PG I and PG I/II ratio for the diagnosis of OGA continues to be an issue of debate.12 In this study we sought to assess the characteristics of advanced (moderate to severe) OGA as a consequence of H. pylori infection as compared with advanced OGA in AIG. Furthermore, we determined the diagnostic performance of G17, PGI and PGI/II ratio for the diagnosis of advanced OGA in these two groups of patients.
Materials and methods
Study population
In this prospective single centre study a total of 452 patients (63% women, mean age 55.7±15.7 years, range 18–94, 48% H. pylori-positive), presenting to the Otto-von-Guericke University Hospital from July 2011 to October 2014 for oesophagogastroduodenoscopy (OGD) were enrolled. Inclusion criteria were as follows: patients older than 18 years of age, patients not receiving antibiotic treatment and proton pump inhibitors (PPIs) 2 weeks before the OGD. Patients on long-term PPI therapy where included if PPI therapy was stopped 2 weeks before OGD. Exclusion criteria were as follows: presence of hepatic, renal, lung, metabolic, hematological, or malignant diseases; pregnancy or nursing. Patients with type 1 G1 NET of the stomach were included as these tumours have an excellent prognosis and represent a common finding in patients with OGA. Serum samples were obtained from all patients before undergoing OGD for determination of PG I and PG II, G17, antibodies against H. pylori, cytotoxin-associated gene A protein (CagA), H+/K+-ATPase antigen (antiparietal cell antibodies, APCA) and intrinsic factor (AIFA).
The study was in accordance with the Helsinki Declaration of 1975 as revised in 1983. All patients provided written informed consent.
OGD and biopsy protocol
During OGD biopsies from gastric antrum (N=2), incisura angularis (N=1) and corpus (N=2) were collected for histological examination. One further biopsy from the antrum and corpus each was collected for rapid urease test (RUT).
H. pylori testing
Patients with positive results for at least one test among serology (either H. pylori-IgG or CagA IgG), RUT, histology or reporting eradication therapy in the past were classified as H. pylori-positive. Patients with negative results in all tests and denying a previous H. pylori eradication therapy were considered H. pylori-negative.
Structured questionnaire
Before undergoing OGD patients were interviewed using a structured questionnaire providing information on demographics, medical history and current daily medication after recruitment. Patients with moderate to severe (advanced) OGA diagnosed by histology (ie, Sydney Score 2–3 based on both corpus biopsies) were contacted again later and specifically asked about long-term PPI therapy (beside the 2 weeks before OGD), vitamin B12 substitution and previous H. pylori infection and eradication therapy in a telephone interview.
Determination of PG I, PG II, G17 and IgG antibodies against H. pylori
Serum samples were obtained after an overnight fast from all participants, centrifuged at 2000g for 15 min and stored at −80°C until the assay was performed. Serum concentration of basal G17, PG I, PG II and immunoglobulin G (IgG) antibodies against H. pylori were measured blinded to case-control status in the same batch by a chemiluminescent enzyme immunoassay using commercial kits (BIOHIT, Helsinki). The averages of the duplicate values for G17, PG I and PG II were used for analysis. With respect to G17, for values above 40 pmol/L, we routinely did not perform any further dilution as higher values of G17 do not add any relevant clinical information. For values above 40 pmol/L the value of 40 pmol/L was used for statistical analysis.
CagA determination
CagA was determined using a CagA IgG kit (GD33; Genesis Diagnostics, London, UK), according to manufacturers’ instructions.13 Patients with anti-CagA IgG ≥6.25 U/mL were classified as H. pylori-positive. All serological examinations were carried out in the same laboratory.
Determination of APCA and AIFA
APCAs against H+/K+-ATPase antigen were detected by immunofluorescence test using rat liver, kidney and stomach as a substrate (Generic Assays GmbH, Dahlewitz/Berlin, Germany). Bound IgG was detected using antihuman IgG fluorescein isothiocyanate at a screening dilution of 1 in 20. APCA were reported as negative or positive if there was cytoplasmic staining of parietal cells. Presence of AIFA was assayed using a quantitative ELISA method (Alegria System, ORGENTEC Diagnostika GmbH, Mainz, Germany). The cut-off used was 6 U/mL. Test results were interpreted according to the manufacturer's instructions. All serological examinations were carried out in a blinded fashion in the same laboratory.
Histology
Multiple histology sections (3–5 μm thick) were obtained from each paraffin embedded biopsy. Sections were stained with H&E, periodic acid–Schiff, chromogranin A and modified Giemsa for H. pylori. The sections were examined by light microscopy. At each biopsy sample site, gastritis was distinguished in non-atrophic (ie, no loss of appropriate glands) and atrophic (ie, loss of appropriate glands). Mucosal atrophy was scored according to the updated Sydney system using a visual analogue scale (0=absent; 1=mild; 2=moderate; 3=severe). On the basis of the topographic locations (oxyntic and angular/antral) of the histology changes, the gastritis stage was also assessed according to the OLGA (Operative Link on Gastritis Assessment) staging system.14 ECL-cell hyperplasia was defined as up to five adjacent cells labelled with chromogranin A. Advanced OGA was defined as moderate to severe OGA (ie, Sydney Score 2–3) based on histological evaluation of both corpus biopsies. Two specialised gastrointestinal pathologists (KF, DJ), jointly assessed all the histology specimens and reached a consensus on the atrophy scores.
Definition of AIG
Criteria for defining AIG were the presence of ECL cell hyperplasia diagnosed by immunohistochemical staining for chromogranin A (up to five adjacent cells with chromogranin A labelling) in patients without previous long-term PPI therapy (conditio sine qua non) with or without positivity for APCA and/or AIFA. Indeed, ECL cell hyperplasia is a constant histological finding in patients with AIG even in early stages of the disease.8 The only differential diagnosis in the presence of ECL cell hyperplasia may be long-term PPI changes. Long-term use of PPIs can suppress acid, which stimulates gastrin production and ECL cell hyperplasia. This differential can be resolved by considering medication history. With respect to APCA and AIFA a seroconversion may be observed in late stages of AIG and thus a subset of AIG patients may have negative APCA and/or AIFA serology.15
Statistical analysis
All data were analysed using the SPSS Statistics (V.22), IBM Corporation, New York, USA. The description of the clinical characteristics has the form mean±SD for continuous variables or n (%) for variables with discrete levels.
Distribution of the demographic characteristics and related factors between patients with and without advanced OGA were compared by Student's t test (Welch test, Satterthwaite's approximation to compute the degree of freedom) for metric parameters and by Fisher’s exact test for categorical variables. OR with corresponding 95% CI were also estimated. In the presence of a zero the OR was estimated using the null hypothesis (Peto OR). In the subanalysis comparing clinical features of patients with H. pylori-negative AIG (N=18) and patients with H. pylori-positive advanced OGA (N=10) the two H. pylori-negative patients without AIG features and the four patients with overlapping features (H. pylori-positive AIG patients) were excluded.
Receivers operating characteristic (ROC) curves were used to calculate the overall diagnostic performance of G17, PGI and the PGI/PGII ratio for the diagnosis of advanced OGA, and in the subgroups of patients with AIG and H. pylori infection. In the subgroup analysis, the diagnostic performance of G17, PGI and the PGI/PGII ratio was tested for diagnosing advanced OGA in patients with AIG (N=22) and those with H. pylori-positive advanced OGA (N=10). For all calculations (ie, for advanced OGA and subgroups) patients without advanced OGA (N=332) served as controls. The two H. pylori-negative patients without AIG features were excluded from this subgroup analysis. If the area under the ROC curve (AUC) was ≥0.60, the best cut-off points were assessed using the Youden-Index, and then sensitivity analysis and likelihood ratios were also calculated. According to the AUC the diagnostic performance of the tests was defined as excellent (0.90–1), good (0.80–0.90), fair (0.70–0.80) or limited (0.60–0.70). The diagnostic performance of G17, PGI and the PGI/PGII ratio was assessed against histology (gold standard). For selected cut-off points diagnostic performance was also calculated. All analyses were made two-tailed with a critical probability of α=5% without α-adjustment. To support the interpretation, the corresponding p values are added in text.
Results
Study population characteristics
Figure 1 shows the number of patients included in the study. Of 452 patients that underwent OGD, 69 did not meet the selection criteria whereas in 17 patients sera were missing. Of the 384 eligible patients (mean age 54.7±15.6 years, range 18–94), 34 (8.9%) had advanced OGA by histology (22 women, age 61±15 years). Of the 22 (65%) patients with AIG, 18 were H. pylori-negative and 4 were H. pylori-positive, respectively. With respect to the remaining 12 patients, 10 were H. pylori-positive with no AIG features (29%) whereas 2 patients (6%) had neither AIG nor present or past H. pylori infection. Clinical, serological and histological characteristics of patients with advanced OGA according to H. pylori status and the presence of AIG are shown in table 1. Interestingly, antrum atrophy of any grade was observed in 39% of patients with H. pylori-negative AIG but in none of the four patients with AIG with a positive H. pylori status. Because of the small number, patients with H. pylori-positive AIG (N=4) and H. pylori-negative advanced OGA (N=2) were excluded from statistical analysis.
Comparison of H. pylori-negative AIG to H. pylori-positive advanced OGA
In table 2 clinical and histological differences between patients with H. pylori-negative AIG (N=18) and those with H. pylori-positive advanced OGA (N=10) are shown. No differences were observed between groups with respect to age and gender distribution. According to a priori defined criteria the prevalence of APCA and ECL cell hyperplasia was higher in patients with H. pylori-negative AIG. One patient with AIG had APCA and AIFA. One patient with ECL cell hyperplasia and a NET of the stomach, but neither APCA nor AIFA, nor current or previous H. pylori infection, nor long-term PPI therapy was classified to having AIG.
Ppatients with H. pylori-negative AIG were more likely to have another autoimmune disease in general (OR 6.3; 95% CI 1.3 to 29.8), autoimmune thyroid disease (OR 7.4; 95% CI 1.3 to 42.7), severe corpus atrophy (OR 10.1; 95% CI 1.9 to 54.1) and corpus intestinal metaplasia (OR 26.9; 95%CI 5.3 to 136.5) compared with H. pylori-positive patients with advanced OGA. NETs were observed only in patients with AIG.
Figures 2 and 3 highlight the differences of endoscopic and histological features of advanced OGA between patients with H. pylori gastritis and AIG.
Diagnostic performance of G17, PG I and PG I/II for the diagnosis of advanced OGA
Figure 4 shows the area under the G17 (AUC=0.76), PGI (AUC=0.86) and PGI/II (AUC=0.86) ROC curves for patients with advanced OGA. Overall, the optimal cut-off levels to detect advanced OGA were 17 pmol/L for G17 (sensitivity, 0.74%; specificity, 0.74%), 60 µg/L for PG I (sensitivity, 0.71%; specificity, 0.93%), and 5 pmol/L for PG I/II ratio (sensitivity, 0.74%; specificity, 0.95%), respectively.
The diagnostic performance of G17, PG I and PG I/II considered as a panel for the diagnosis of advanced OGA as compared with histology was assessed according to manufacturer’s recommendations and according to the ROC curves generated from our study population as well.
According to manufacturer’s recommendations, the diagnostic performance of the panel was as follows: sensitivity 41.7% (95% CI 22 to 63%), specificity 99.7% (95% CI 98 to 100%), positive predictive value 7.1% (95% CI 5 to 10%) and negative predictive value 92.9% (95% CI 90 to 95%). According to the ROC curves generated from our study population sensitivity, specificity, positive and negative predictive values of the panel were as follows: sensitivity 58.8% (95% CI 41 to 75%), specificity 98.4% (95% CI96 to 100%), positive predictive value 5.4% (95% CI 3 to 8%) and negative predictive value 94.6% (95% 92 to 97%). Thus, according to the cut-offs generated from the ROC curves there was a clear increase of the sensitivity with a low decrease in specificity of the panel (data not shown).
Figures 5 and 6 show the area under the G17, PGI, and PGI/II curves of patients with H. pylori-negative AIG (N=18) and H. pylori gastritis with advanced OGA (N=10), respectively. Again, because of the small number, patients with H. pylori-positive AIG (N=4) and H. pylori-negative advanced OGA (N=2) were excluded from statistical analysis.
The performance of G17, PG I and PG I/II was good to excellent for patients with H. pylori-negative AIG (AUC=0.83, 0.95 and 0.97, respectively), but low for H. pylori positive patients with advanced OGA (AUC=0.62, 0.75 and 0.67, respectively). Optimal cut-off levels to detect advanced OGA were 19 pmol/l for G17 (sensitivity 0.82%; specificity 0.76%), 50 µg/L for PG I (sensitivity 0.96%; specificity, 0.95%) and 5 µg/L for PG I/II ratio (sensitivity, 0.96%; specificity 0.95%) for patients with AIG and 10 pmol/L for G17 (sensitivity 0.70%; specificity 0.60%), 80 µg/L for PG I (sensitivity 0.60%; specificity 0.80%) and 5 µg/L for PG I/II ratio (sensitivity 0.40%; specificity 0.95%) for H. pylori-positive patients, respectively. In figures 5 and 6 cut-offs for serum PG I, PG II and G17 levels in patients with histological OGA are shown either according to manufacturer′s recommendations (G17 >10 pmol/L, PG I<30 µg/L, PG II/II ratio <3) or optimised as suggested by the ROC curves based on our study population. The means and SDs of G17, PGI and PGI/II ratio in patients with advanced OGA and in the subgroups of patients with H. pylori-negative AIG and H. pylori gastritis with advanced OGA are shown in table 3.
Discussion
In our study we found H. pylori-negative AIG with a distinct morphological and serological phenotype compared with advanced OGA in H. pylori gastritis. Strict prestudy criteria allowed identifying a few features for the differentiation of these two entities.
The first feature is that patients with H. pylori-negative AIG were more likely to have another autoimmune disease (OR 18.8; 95% CI 2.0 to 173.9). As reported in previous studies, the most common autoimmune disease among H. pylori-negative AIG was autoimmune thyroid disease (39%).16 Furthermore, we confirm the association of patients with H. pylori-negative AIG with rheumatoid arthritis and primary biliary cirrhosis, which has been described in few case reports before.17 ,18 Noteworthy, none of the four patients with H. pylori-positive AIG had another autoimmune disease. Indeed, apart from the case of idiopathic thrombocytopenic purpura (ITP), a role of H. pylori infection in the development of extragastrointestinal autoimmune diseases is unlikely.19
Another feature was that patients with H. pylori-negative AIG were more likely to have severe corpus atrophy, and corpus intestinal metaplasia compared with H. pylori-positive patients with advanced OGA.
In line with our previous report, we confirmed the presence of atrophy in the antrum in a subset (39%) of patients with AIG.11 It is of interest to note, that none of the patients with H. pylori-positive AIG had atrophy of the antrum, which however does not preclude the hypothesis that H. pylori infection might in some cases serve as trigger for AIG.
OGA is a precancerous condition, associated with a fourfold to fivefold increased risk for developing GC and therefore screening for OGA is recommended.20 ,21 H. pylori serology combined with increased serum G17 and a low serum PGI and PGI/II ratio are currently the best available non-invasive tests for identification of subjects with advanced OGA at increased risk of GC.22 ,23 Manufacturer’s recommendations suggest using a G17>10 pmol/L, PG I <30 µg/L and PG I/II<3 as reference values for detecting OGA.24 However, these cut-offs may require adaptations in different populations. According to our data G17>17 pmol/L, PG I<60 µg/L and PG I/II<5 represent better cut-offs for detecting OGA in our population, irrespective of their aetiology.
In a recent study of 91 patients with dyspepsia the discriminatory ability of serological biomarkers to detect OGA of any grade according to the updated Sydney classification was reported to be insufficient.12 The higher diagnostic performance observed in our study population is obtained by the exclusion of patients with early stages of OGA. In early stages of OGA (stage 1 according to the updated Sydney classification) serum levels of OGA biomarkers are not expected to be abnormal, as this is known for other conditions in which serum biomarkers become abnormal only in advanced/end-stage disease (ie, creatine levels in chronic kidney disease, albumin or bilirubin levels in liver cirrhosis).25
Serum parameters indicate a more constant severe loss of function reflected by G17, PG I and PG I/II ratio in AIG. Accordingly, in the subgroup of patients with AIG the diagnostic performance of serological markers was good (increased G17) to excellent (low PG I and PG I/II ratio) for the diagnosis of advanced OGA but their diagnostic performance was limited in patients with present or past H. pylori infection. The high diagnostic performance of an increased G17 and a low PG I and PG I/II ratio in patients with AIG is due to extensive atrophic changes of the oxyntic mucosa in these patients. This is different in the H. pylori-induced OGA where the patchy distribution of focal areas of severe atrophy surrounded by less severe mucosal changes may lead to less functional loss. The severity of OGA may also in part explain the differences between the two groups, as patients with AIG had more often severe (grade 3) atrophy when compared with patients with H. pylori induced OGA. Stratification according to the aetiology of OGA is mandatory for a better interpretation of G17, PGI and PGI/II ratio. By increasing the cut-off for PG I and decreasing the cut-off for G17 the diagnostic performance of these parameters for identifying OGA can be increased in H. pylori-induced OGA.
The risk for GC development is not established in patients with AIG. OGA may lead to the development of pernicious anaemia, which was already present in 50% of our patients with AIG compared with 17% of patients with OGA with H. pylori gastritis. In a recent meta-analysis the pooled GC incidence rate among patients with pernicious anaemia was of 0.27% per person-years, with an overall GC relative risk of 6.8 (95% CI 2.6 to 18.1).26 Thus, the magnitude of GC risk for patients with pernicious anaemia does not differ from that of patients with OGA. According to the ‘ABCD method’ (group A: H. pylori antibody negative and normal PG level; group B: H. pylori antibody positive and normal PG level; group C: H. pylori antibody positive and low PG level; group D: H. pylori antibody negative and low PG level) adopted in Japan that stratifies individuals by serology for their GC risk, patients of the group D (ie, low PG I and/or low PG I/ PG II ratio testing negative for H. pylori serology) carry the highest risk for developing GC (HR 120.4).27 ,28 Most of the patients of the group D are patients with AIG and/or patients with pernicious anaemia. Patients resulting positive for serological OGA should undergo OGD with biopsies according to the Sydney classification and OLGA staging system. According to current guidelines, endoscopic follow-up should thereafter be reserved to patients with extensive GA at 3 year intervals in order to diagnose GC at an early stage.29
In summary, H. pylori-negative AIG has a distinct clinical, morphological and serological phenotype compared with advanced OGA in H. pylori gastritis. An increased G17 and a low PG I and PG I/II ratio showed a good diagnostic performance for advanced OGA. According to the aetiology of advanced OGA the diagnostic performance of serological biomarkers was high in patients with AIG and low in patients with present or past H. pylori infection. The performance of serological markers of OGA can be increased by optimising their cut-offs.
Take home messages
The aetiology of oxyntic gastric atrophy (OGA), a precancerous condition associated with a fourfold to fivefold increased risk for developing gastric cancer, is restricted to two risk factors: Helicobacter pylori infection and autoimmune gastritis (AIG).
Among patients with advanced OGA, patients with H. pylori-negative AIG are more likely to have another autoimmune disease, severe OGA (stage 3 according to the updated Sydney classification) and intestinal metaplasia of the corpus.
The diagnostic performance of an increased G17 and a low PGI and PGI/II ratio for the diagnosis of advanced OGA is high in patients with AIG but low in patients with present or past H. pylori infection, and thus stratification according to the aetiology of OGA is helpful for a better interpretation of these serological biomarkers.
Clinical, morphological and serological criteria may help to discern between AIG and H. pylori-induced advanced OGA.
Acknowledgments
The authors thank the endoscopic team for its assistance during the upper GI endoscopies. The authors also thank Marion Holley, Ursula Stolz (Department of Gastroenterology) for their experimental work.
References
Supplementary materials
Abstract in German
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Abstract in German - Online abstract
Footnotes
Handling editor Cheok Soon Lee
Contributors MVe and PM designed the research study. MVe and MVa collected the data. MVe, AL and DR performed the research. MVe, MVa and F-WR performed statistical analysis. KF and DJ performed the histopathological assessment of gastric biopsies. JW supervised endoscopies of all patients included in the study. MVe and PM analysed the data and wrote the paper. All authors approved the final version of the article, including the authorship list.
Funding Supported in part by a research grant from the Bundesministerium für Bildung und Frauen (BMBF-0315905D) to PM in the frame of ERA-NET PathoGenoMics.
Competing interests PM has served as a speaker for Takeda, AstraZeneca and Reckitt Benckiser.
Ethics approval Ethics Committee of the Otto-von-Guericke University Hospital of Magdeburg (registration number 80/11).
Provenance and peer review Not commissioned; externally peer reviewed.