Alimentary Pharmacology and Therapeutics
Helicobacter-negative gastritis: a distinct entity unrelated to H. pylori infection R. M. Genta*,† & A. Sonnenberg‡
*Miraca Life Sciences Research Institute, Irving, TX, USA. † Departments of Medicine (Gastroenterology) and Pathology, UTSW – Dallas VAMC, Dallas, TX, USA. ‡ Department of Medicine (Gastroenterology) Portland VAMC, OHSU, Portland, OR, USA.
Correspondence to: Dr R. M. Genta, Miraca Life Sciences, 6655 North MacArthur Blvd, Irving, TX 75039, USA. E-mail: robert.genta@utsouthwestern. edu
Publication data Submitted 26 September 2014 First decision 5 October 2014 Resubmitted 7 October 2014 Accepted 10 October 2014 This article was accepted for publication after full peer-review.
SUMMARY Background Helicobacter-negative gastritis is diagnosed when no organisms are detected in a gastric mucosa with typical features of Helicobacter gastritis (Hp-gastritis). If Helicobacter-negative gastritis consisted mostly of ‘missed’ Helicobacter infections, its prevalence should represent a constant percentage of these infections in a population, and their clinico-epidemiological features would overlap. Aim To compare the epidemiologic patterns of Hp-positive and Hp-negative gastritis. Methods From a pathology database, we extracted demographic, clinical and histopathological data from patients with gastric biopsies (1.2008–12.2013). We allocated patients to high (≥12%) and low (≤6%) H. pylori prevalence regions defined by ZIP code-based data. The prevalence of H. pylori-positive and -negative gastritis by sex, age and state were expressed as a per cent of the total study population stratified accordingly. Results Of 895 323 patients, 10.6% had Hp-gastritis and 1.5% Helicobacter-negative gastritis. Hp-gastritis, but not Helicobacter-negative gastritis, was more common in males than females (OR 1.17, 95% CI: 1.16–1.19). While Hp-gastritis was more prevalent in high than in low-prevalence areas (OR 3.65, 95% CI: 3.57–3.74), Helicobacter-negative gastritis was only minimally affected by the underlying H. pylori prevalence (1.7% vs. 1.5%). The age-specific prevalence of Hp-gastritis peaked in the 4th to 5th decades; Helicobacter-negative gastritis exhibited a low and relatively flat pattern. The geographic distribution of H. pylori-positive and -negative gastritis showed no significant correlation. Intestinal metaplasia was found in 13.0% of patients with Hp-gastritis and in 6.1% of those with Helicobacter-negative gastritis (OR 0.43, 95% CI: 0.40–0.47). Conclusion These data suggest that Helicobacter-negative gastritis is, in the vast majority of cases, a nosologically and epidemiologically distinct entity that deserves further investigation. Aliment Pharmacol Ther
ª 2014 John Wiley & Sons Ltd doi:10.1111/apt.13007
1
R. M. Genta and A. Sonnenberg INTRODUCTION Helicobacter pylori infection was identified 30 years ago as the main cause of chronic active gastritis and peptic ulcer disease,1 and shortly thereafter it was recognised as a major risk factor for gastric cancer and primary gastric B-cell lymphoma.2, 3 Although the diagnosis of H. pylori infection can be established through a variety of non-invasive tests, including the urea breath test, the stool antigen test, and the detection of specific serum antibodies, the histopathological demonstration of the characteristic bacilli remains the gold standard to this day.4 Helicobacter gastritis is one of the most straightforward diagnoses in pathology: the gastric mucosa is infiltrated by a mixed population of mononuclear and polymorphonuclear cells, the surface and foveolar epithelium show variable amounts of polymorphonuclear infiltration, and the characteristic curved (‘seagull-shaped’) organisms are seen attached to the epithelial surface or in the mucus.5 Bacteria can often be seen on slides stained by haematoxylin and eosin, but in many cases, a stain that highlights the organisms is needed. Histochemistry has provided a number of stains that over the years have seen their fortune wax and wane, such as modified Giemsa stains,6 silver stains7 and variations on the toluidine blue stains.8 Many of these have now been supplanted by highly specific and sensitive peroxidase-conjugated immunohistochemical stains, which not only allow the visualisation but also the identification of Helicobacter organisms, as they specifically bind to this species.9 When a pathologist recognises the characteristic features of Helicobacter gastritis, but no bacteria can be detected by any of these special stains, the diagnosis of Helicobacter-negative gastritis is made. Other conditions (such as the vicinity of erosions or ulcers,10, 11 ingestion of alcohol,12 certain medications,13 or autoimmune gastritis14) that can cause active inflammation in the gastric mucosa do not mimic Helicobacter gastritis and, therefore, tend to be easily recognised and appropriately diagnosed. The only known association with Helicobacter-negative gastritis (as defined above) is inflammatory bowel disease. Particularly in patients with Crohn’s disease, the gastric mucosa may show focal or sometimes diffuse chronic active changes indistinguishable from those of Helicobacter gastritis, except for the presence of organisms.15–17 Helicobacter-negative gastritis has not acquired the status of an independent condition because it has been assumed that it simply represents cases of Helicobacter gastritis in which, for a variety of reasons, bacteria are
2
missed. Frequently invoked explanations for the absence of bacteria expected to be there include the use of proton pump inhibitors (which may decrease the numbers of bacteria and shift their populations from the antrum to the corpus),18–21 recent use of antibiotics (that may suppress the infection but not reduce the inflammation)22 and sampling error.23 Therefore, the few attempts to investigate the nature of Helicobacter-negative gastritis have focused on the detection of the missing bacteria by using more sensitive methods. Zsikla et al. used nested and quantitative PCR on archived gastric biopsies with gastritis and detected H. pylori signals in about 20% of histologically negative gastric biopsies. This suggested that a portion of the cases might be related to infections with low bacterial loads that can only be detected by methods more sensitive than the visualisation of the organisms. The majority of cases, however, remained unexplained. The first and only systematic clinicopathological study aimed at discovering the cause of H. pylori-negative gastritis was published in 2013.24 The authors used a biopsy mapping protocol that exceeded the guidelines of the Sydney System and collected detailed social, clinical and histopathological data from almost 500 patients. Gastritis, defined for the purposes of this study as ‘at least [Sydney System] grade 1 in at least two gastric sites or at least grade 2 in at least one gastric site’ was detected in 200 patients. In 41 of them (20.5% of all subjects with gastritis and 8.4% of the total study population), H. pylori could not be detected by histology (including histochemical and immunohistochemical staining), serology or culture. After evaluating the patients’ social, clinical and medication history, the authors concluded that the two groups did not differ significantly in past and current use of PPI, self-reported use of histamine-2-receptor antagonists, aspirin, NSAIDs or previous anti-H. pylori therapy. Thus, the causes and implications of H. pylori-negative gastritis remain unknown. We hypothesised that if the majority of cases of Helicobacter-negative gastritis were ‘missed’ Helicobacter infections, their prevalence should represent a relatively constant percentage of the H. pylori infections in a population. If, as in a pilot study presented only in abstract form,25 82% of all cases of chronic active gastritis show bacteria and 18% do not, this ratio should remain relatively constant across different populations, irrespective of their prevalence of H. pylori, provided that the diagnoses are made by a group of observers who use similar criteria and identical detection methods. This study was
Aliment Pharmacol Ther ª 2014 John Wiley & Sons Ltd
Helicobacter-negative gastritis designed to test this hypothesis using data generated by a single group of pathologists who examined and diagnosed gastric biopsy specimens from almost one million patients during the past 6 years.
PATIENTS AND METHODS Study setting and data source We used a large national pathology database of subjects who underwent oesophagogastroduodenoscopy (OGD) with gastric biopsies between January 2013 and June 2014 in endoscopy centres distributed throughout the USA and whose mucosal biopsy specimens were evaluated and reported by a single group of histopathologists. The group consists of 35 pathologists with subspecialty training in gastrointestinal pathology who practice in the same environment, use uniform diagnostic criteria and standardised diagnostic codes, and participate in daily consensus conferences where cases and diagnostic criteria are discussed. The Miraca Life Sciences internal review board determined that this study was entirely performed using the collection of existing data, documents and reports, and that the information was recorded in such a manner that subjects cannot be identified. Therefore, pursuant to 45 CFR 46, section 101 b(4), the study was exempt from 45 CFR 46 regulations and no informed consent was necessary. Study design All unique patients with gastric biopsies were extracted from the database and their demographic, clinical and histopathological data were recorded. If a patient had multiple OGDs, only data from the chronologically first procedure were included. For each patient, the following histopathological parameters were evaluated: H. pylori infection, Helicobacter-negative gastritis and intestinal metaplasia. The prevalence of H. pylori gastritis in each ZIP code for which at least 500 patients were available was then used to define two regions: high prevalence (where the prevalence of H. pylori gastritis was 12% or greater) and low prevalence (6% or lower). Patients were then allocated to two separate groups: group 1 contained patients from ZIP codes within high-prevalence regions and group 2 were patients from ZIP codes within low-prevalence regions. Histopathological criteria More than 90% of gastric biopsies in this laboratory are routinely stained with a specific anti-Helicobacter monoclonal immunochemical stain (Cell Marque, Rocklin, CA, Aliment Pharmacol Ther ª 2014 John Wiley & Sons Ltd
USA); the remainder are stained with a modified methylene blue stain (HP Blue; Anatech, Ltd., Battle Creek, MI, USA); all specimens are also stained with Alcian Blue–Periodic Acid–Schiff to enhance the detection of intestinal metaplasia. Gastric biopsy specimens are diagnosed following the guidelines and terminology of the Updated Sydney System.5 Briefly, Helicobacter infection was diagnosed when the characteristic curved organisms were visualised in a gastric biopsy specimen (Figure 1a). Intestinal metaplasia was diagnosed in the presence of goblet cells in the gastric mucosa. Chronic active Helicobacter-negative gastritis (referred to in this paper as simply Helicobacter-negative gastritis) was diagnosed only when the lamina propria contained mixed populations of lymphocytes, plasma cells and polymorphonuclear cells, and the surface of foveolar epithelium was
(a1)
(a2)
(b1)
(b2)
Figure 1 | (a1) Typical Helicobacter pylori gastritis, with chronic active inflammation and numerous organisms visible in the sections stained with an immunohistochemical stain specific for Helicobacter (a2). (b1) shows a virtually identical type of gastritis, in which no organisms could be demonstrated by several stains, including an immunohistochemical stain specific for Helicobacter (b2). 3
R. M. Genta and A. Sonnenberg infiltrated by neutrophilic polymorphonuclear cells, but no Helicobacter organisms could be visualised by the anti-Helicobacter monoclonal immunochemical stain (Figure 1b). If active gastritis was detected in the presence of erosions or ulcers, or if features of autoimmune atrophic gastritis were present, a diagnosis of chronic active Helicobacter-negative gastritis was not made. Intestinal metaplasia was diagnosed when intestinal-type epithelium with goblet cells was found in the gastric mucosa. Reviews of quality control records for the years of the study and re-staining of cases with active inflammation and no detectable organisms (performed on >300 slides in the design phases of the project) confirmed the high accuracy of the diagnoses used for the study.
Clinical manifestations Information on the patients’ indications for the index OGD as well as other relevant clinical data (including a history or a diagnosis of inflammatory bowel disease) were extracted from the database. Data analysis and statistics The prevalence by sex, age and state was expressed as a per cent of the total study population stratified accordingly by sex, age and state. For statistical comparisons, the 95% CI of each prevalence rate was calculated using the Poisson distribution. Each two rates were considered significantly different if their respective confidence intervals did not overlap. Patients with H. pylori-positive and -negative gastritis were also compared calculating odds ratios and their 95% CI and using logistic regression analysis to adjust for the confounding influences of age and sex. The geographic distributions of H. pylori-positive and -negative gastritis among different states were compared using linear regression analysis. RESULTS General Of 895 323 unique patients with gastric biopsies included in the present analysis, 108 833 had chronic active gastritis. Helicobacter organisms were demonstrated in 95 004 (87.3%) of these patients; in 13 829 (12.7%), no Helicobacter organisms could be demonstrated, and a diagnosis of Helicobacter-negative chronic active gastritis was made (Table 1). The prevalence of Helicobacter-negative gastritis declined each year, from 2.1% in 2008 to 1.1% in 2014, while the prevalence of Helicobacter-positive gastritis declined from 11.2% to 9.9% (Table 2). 4
Table 1 | Distribution of patient population by diagnosis, sex and Helicobacter pylori prevalence. Gender unknown for 38 patients Hp-positive gastritis Female Male Low Hp prevalence area High Hp prevalence area All patients
Hp-negative gastritis
Total
55 613 39 388 10 075
8651 5177 2712
554 907 340 378 185 047
47 515
4557
273 377
95 004
13 829
895 323
Hp, H. pylori.
Distribution by sex Figure 2 shows the stratification of H. pylori-positive and -negative gastritis by sex. Significantly, more males than females harboured H. pylori-positive gastritis (OR adjusted for age and sex = 1.18, 95% CI: 1.16–1.19). However, men and women were equally affected by Helicobacter-negative gastritis (OR adjusted for age and sex = 0.98, 95% CI: 0.94–1.01). Distribution by H. pylori prevalence Areas of low H. pylori prevalence comprised 185 047 patients and high-prevalence areas comprised 273 377 patients. The 436 899 patients from intermediate-prevalence areas (>6% and