Correlation between the endoscopic and histologic diagnosis of gastritis
Article Outline
Abstract
Our aims were to determine the rate of concordance between endoscopic and pathologic diagnoses of gastritis and to determine if there was any common factor in discordant cases. A retrospective analysis of data from 400 patients was performed. The endoscopic diagnoses were compared with the pathologic diagnoses, and histologic slides from discordant cases were reviewed. Of the 400 patients, there was discordance between endoscopy and histology in 136 (34%; κ statistic, 0.31). These discordant cases comprised 56 with normal endoscopy but abnormal histology and 80 with abnormal endoscopy but normal histology. In 13 patients, there was normal histology, although erosions had been diagnosed endoscopically. No consistent histologic features were found in the discordant cases. These findings show that standard endoscopy is a poor predictor of pathologic changes. Biopsies are required for accurate diagnosis of gastritis.
Keywords: Diagnosis, Endoscopy, Gastritis, Helicobacter, Stomach
1. Introduction
Gastroenterologists make the diagnosis of gastritis using features they observe through the endoscope. However, the literature shows a very variable degree of concordance with the histologic diagnosis if biopsies are taken. Most authors report poor concordance between endoscopic and pathologic diagnosis of gastritis, with typical rates of concordance between 54% and 63% [1], [2], [3], [4], [5], [6]. Belair et al [7] found that endoscopy was no better than chance as a test for gastritis or Helicobacter pylori infection. In contrast, one study found a concordance of 97% between histologic and endoscopic diagnoses of gastritis in patients with endoscopically abnormal mucosa and suggested that biopsy is not needed in every case of gastritis [8].
Given the variable but generally poor concordance between endoscopy and histology in this field, the aim of our study was to determine the rate of concordance between endoscopic and histologic diagnosis of gastritis in an unselected series of patients in a general hospital setting. It was designed as a retrospective audit.
2. Materials and methods
The records of patients referred to Southampton General Hospital, United Kingdom, for upper gastrointestinal endoscopy in 2009 and 2010 were retrieved. Patients in whom biopsies were available were eligible for inclusion; cases of polyp or neoplasia were excluded. Details of the endoscopic diagnoses and the patients' symptoms were obtained from the patient records. The endoscopes used were Olympus models GIF-XQ240 and GIF-XQ260 (Olympus, Southend-on-Sea, UK).
Standard histologic criteria were used in making a diagnosis [9]. A case was classified as discordant if the endoscopist recorded normal appearances but the pathologist found an abnormality, or vice versa. The histologic slides in discordant cases were retrieved from the archives and reviewed microscopically. Hematoxylin and eosin–stained slides were available in all cases. Warthin-Starry or modified Giemsa stains were used to identify Helicobacter. The study was given ethical approval by the audit committee of the hospital.
The κ statistic was used to assess the level of agreement, and Fisher exact test was used to compare groups to determine whether any feature of the cases was more likely to be associated with discordant diagnoses. The calculations were performed with GraphPad Software (GraphPad Software Inc., La Jolla, CA).
3. Results
Table 1 shows data of the 400 cases included in the study, comprising 209 females (mean age, 65.5 years) and 191 males (mean age, 64.8 years). The indications for endoscopy were as follows: abdominal pain/dyspepsia, 139 cases; iron deficiency, 86 cases; dysphagia, 42 cases; vomiting/hematemesis, 37 cases; follow-up of previous ulcer, 34 cases; weight loss, 33 cases; melena, 19 cases; follow-up of Barrett esophagus, 13 cases; and other indication, 46 cases (these figures include 49 patients with multiple indications).
Table 1. Demographic data of the patients and the endoscopic and histologic findings
| Total number | Age (y) | Age range (y) | Females | Males | Histology normal | Histology abnormal | Endoscopy normal | Endoscopy abnormal | |
|---|---|---|---|---|---|---|---|---|---|
| All cases | 400 | 65.2 | 2-98 | 209 | 191 | 185 | 215 | 161 | 239 |
| Discordant cases | 136 (34%) | 64.5 | 11-98 | 74 (35%) | 62 (32%) | 80 (43%) | 56 (26%) | 56 (35%) | 80 (33%) |
| Concordant cases | 264 (66%) | 65.5 | 2-95 | 135 (65%) | 129 (68%) | 105 (57%) | 159 (74%) | 105 (65%) | 159 (67%) |
Endoscopically, 239 cases were diagnosed with gastritis. Of these, 37 were said to have erosions, 60 were said to have an ulcer, and 10 were said to have both.
Comparison with histology showed 136 discordant cases, representing 34% of the total (Table 1). The κ statistic for this level of agreement is 0.31. The discordant cases comprised 56 with normal endoscopy but abnormal histology and 80 with gastritis diagnosed endoscopically but normal histology; these 80 cases included 13 (16%) in which erosions had been diagnosed endoscopically.
There was no statistical association between the rate of discordance and the presence of Helicobacter, chemical gastritis, intestinal metaplasia, or any of the indications for endoscopy. Review of the histologic slides showed no consistent features in the positive or negative discordant cases.
4. Discussion
We examined data from 400 patients, which represents one of the largest series addressing the question of concordance between endoscopy and histology published in the literature. We found the concordance to be poor, with a κ statistic of 0.31. Because “substantial agreement” is generally considered to be associated with κ greater than 0.6, the value of 0.31 reflects a poor correlation in this context [10].
In 13 of our cases, the histology was normal although erosions had been diagnosed endoscopically. This observation is in keeping with the study by Gad [11], who found that only 78 (42%) of 172 of lesions diagnosed as gastroduodenal erosions by endoscopy had erosions confirmed histologically. Findings such as these have led to the proposal that terms such as erosion and gastritis should not be used by endoscopists, who should use descriptive terms instead [12].
We found no significant correlation between specific histologic features and the endoscopic findings. This is consistent with the results of other authors [2], [7]. However, high-resolution magnifying endoscopy has a better correlation with histologic findings and might allow definitive diagnoses to be made without biopsy [13], [14].
It could be argued that if the presence or absence of Helicobacter is the only clinical concern, then a urease test may suffice without biopsy. However, because there are other causes of gastric inflammation, a complete picture cannot be obtained by endoscopy and urease testing alone; histologic assessment is also required.
In conclusion, our findings show that gastritis cannot be reliably diagnosed by endoscopy, assuming histology to be the gold standard. This is consistent with most studies in this field, and we agree with other authors that histology is mandatory for accurate diagnosis [4], [5], [6], [7], [11], [15]. If the diagnosis of gastric inflammation is of clinical relevance, biopsies should be performed irrespective of the endoscopic appearances.
References
- . Correlation of gross gastroscopic findings with gastroscopic biopsy in gastritis. N Engl J Med. 1956;254:641–644
- A study of the correlation between endoscopic and histological diagnoses in gastroduodenitis. Am J Gastroenterol. 1987;82:749–753
- The clinical relevance of endoscopic and histological inflammation of the gastroduodenal mucosa in dyspepsia of unknown origin. Scand J Gastroenterol. 1989;24:385–395
- . A study of the concordance between endoscopic gastritis and histological gastritis in an area with a low background prevalence of Helicobacter pylori infection. Singapore Med J. 2002;43:90–92
- Relationship of gastroscopic features to histological findings in gastritis of Helicobacter pylori infection in a general population sample. Endoscopy. 2003;35:946–950
- Endoscopy in the diagnosis of gastritis. Diagnostic value of endoscopic criteria in relation to histological diagnosis. Endoscopy. 1984;16:101–104
- Receiver operator characteristic analysis of endoscopy as a test for gastritis. Dig Dis Sci. 1997;42:2227–2233
- Is biopsy needed in every gastritis found during endoscopy?. Pak J Med Sci. 2009;25:849–851
- Classification and grading of gastritis. The updated Sydney System. International Workshop on the Histopathology of Gastritis, Houston 1994. Am J Surg Pathol. 1996;20:1161–1181
- . Understanding interobserver agreement: the kappa statistic. Fam Med. 2005;37:360–363
- . Erosion: a correlative endoscopic histopathologic muticenter study. Endoscopy. 1986;18:76–79
- . Erosion—an endoscopic-histological puzzle. Endoscopy. 1986;18:75
- High-resolution magnification endoscopy can reliably identify normal gastric mucosa, Helicobacter pylori–associated gastritis, and gastric atrophy. Endoscopy. 2007;39:202–207
- Comparison of high resolution magnifying endoscopy and standard videoendoscopy for the diagnosis of Helicobacter pylori gastritis in routine clinical practice: a prospective study. Helicobacter. 2009;14:12–21
- Correlation between endoscopic features of gastric antrum, histology and Helicobacter pylori infection in adults. Ital J Gastroenterol Hepatol. 1999;31:359–365
PII: S1092-9134(11)00108-0
doi:10.1016/j.anndiagpath.2011.08.002
© 2012 Elsevier Inc. All rights reserved.
