AIM: To investigate the predictive value of narrow-band imaging with magnifying endoscopy (NBI-ME) for identifying gastric intestinal metaplasia (GIM) in unselected patients.

METHODS: We prospectively evaluated consecutive patients undergoing upper endoscopy for various indications, such as epigastric discomfort/pain, anaemia, gastro-oesophageal reflux disease, suspicion of peptic ulcer disease, or chronic liver diseases. Patients underwent NBI-ME, which was performed by three blinded, experienced endoscopists. In addition, five biopsies (2 antrum, 1 angulus, and 2 corpus) were taken and examined by two pathologists unaware of the endoscopic findings to determine the presence or absence of GIM. The correlation between light blue crest (LBC) appearance and histology was measured. Moreover, we quantified the degree of LBC appearance as less than 20% (+), 20%-80% (++) and more than 80% (+++) of an image field, and the semiquantitative evaluation of LBC appearance was correlated with IM percentage from the histological findings.

RESULTS: We enrolled 100 (58 F/42 M) patients who were mainly referred for gastro-esophageal reflux disease/dyspepsia (46%), cancer screening/anaemia (34%), chronic liver disease (9%), and suspected celiac disease (6%); the remaining patients were referred for other indications. The prevalence of Helicobacter pylori (H. pylori) infection detected from the biopsies was 31%, while 67% of the patients used proton pump inhibitors. LBCs were found in the antrum of 33 patients (33%); 20 of the cases were classified as LBC+, 9 as LBC++, and 4 as LBC+++. LBCs were found in the gastric body of 6 patients (6%), with 5 of them also having LBCs in the antrum. The correlation between the appearance of LBCs and histological GIM was good, with a sensitivity of 80% (95%CI: 67-92), a specificity of 96% (95%CI: 93-99), a positive predictive value of 84% (95%CI: 73-96), a negative predictive value of 95% (95%CI: 92-98), and an accuracy of 93% (95%CI: 90-97). The NBI-ME examination overlooked GIM in 8 cases, but the GIM was less than 5% in 7 of the cases. Moreover, in the 6 false positive cases, the histological examination showed the presence of reactive gastropathy (4 cases) or H. pylori active chronic gastritis (2 cases). The semiquantitative correlation between the rate of LBC appearance and the percentage of GIM was 79% (P < 0.01).

CONCLUSION: NBI-ME achieved good sensitivity and specificity in recognising GIM in an unselected population. In routine clinical practice, this technique can reliably target gastric biopsies.

Gastric cancer is the fourth most common cancer and the second leading cause of cancer-related death worldwide[1,2]. However, the mechanism of gastric carcinogenesis is unknown. The intestinal type tumour, according to the Lauren classification, is characterised by the presence of malignant cells forming glandular structures; in contrast to diffuse type carcinomas, these tumours are generally thought to be preceded by a sequence of precursor lesions. In this multistep model of gastric carcinogenesis, described several years ago by Correa et al[3], Helicobacter pylori (H. pylori) causes chronic inflammation of the gastric mucosa, which slowly progresses through the premalignant stages of atrophic gastritis, intestinal metaplasia and dysplasia, intramucosal carcinoma and invasive neoplasia[4].

Intestinal metaplasia (IM) is generally considered as the “field cancerization” in the gastric mucosa. At the cell level, intestinalized glands provide the cellular substrate that allows the gastric non-invasive neoplasia to develop. The reported progression rates of IM and dysplasia to cancer vary greatly from 0% to 10% per year for IM and from 0% to 73% per year for dysplasia[1]. The prevalence of this condition in Europe varies by country[5-9]; for instance, a Swedish group reported an IM prevalence of 23% in the general population[10], while the prevalence has been estimated to be up to 25% in the Italian population, mainly occurring in the gastric antrum[7,11-13].

In Western countries, premalignant gastric lesions and early gastric cancer are generally diagnosed upon histologic examination of random biopsies, whereas in Asian countries, especially in Japan, the presence and extension of these lesions are frequently established during an endoscopy[14-16]. These remarkable differences are explained by different training and attitudes towards the inspection of the stomach in countries with a high gastric cancer incidence.

The narrow-band imaging system (NBI) is based on the modification of optical filter spectral characteristics in the light source, which improves the visibility of mucosal structures. NBI may be combined with magnification endoscopy to obtain a clear visualisation of surface and vascular patterns[1,2]. Earlier studies using the NBI system with magnification endoscopy (NBI-ME) in the gastric mucosa showed that the appearance of a light blue crest (LBC) in the mucosa is a distinctive endoscopic finding that suggests an increased likelihood of detecting IM in the stomach. More precisely, blue-whitish patchy areas are often observed in NBI images of the antrum in patients with gastric IM. Uedo et al[16] tested NBI-ME on 34 patients with atrophic gastritis and found that the LBC appearance was frequently observed in gastric antrum as blue-white lines visible on the epithelial surface. These authors demonstrated that the appearance of LBCs correlated with histological evidence of IM, with a sensitivity of 89% (95%CI: 83-96), a specificity of 93% (95%CI: 88-97), a positive predictive value of 91% (95%CI: 85-96), a negative predictive value of 92% (95%CI: 87-97) and an accuracy of 91% (95%CI: 88-95).

Despite these promising findings, the added value of NBI-ME in the stomach, and more precisely in detecting premalignant and malignant gastric lesions, requires confirmation from controlled longitudinal trials. Furthermore, a uniform and validated classification system is needed. Thus, the aim of our study was to define the value of NBI-ME for identifying gastric IM in an unselected population through the visualisation of LBC appearance. A secondary aim was to provide a semiquantitative evaluation of LBC appearance and to verify a correlation with the histological measurement of IM percentage.

The study included 100 consecutive patients (58 F/42 M, mean age 67 ± 12 years) who presented at our endoscopy unit between December 2010 and February 2012. The demographic and clinical characteristics of the patients are shown in Table 1. The patients were referred to our endoscopy unit mainly because of gastro-oesophageal reflux disease and/or dyspepsia symptoms (46%); 34% underwent the endoscopic procedure for cancer screening, anaemia, a positive faecal occult blood test or suspicion of a peptic ulcer; 9% for chronic liver disease; 6% for suspected celiac disease and the remaining patients for other indications. The prevalence of H. pylori infection on the biopsies was 31%; in addition, 67% of the patients used proton pump inhibitors, 16% were current smokers, and 25% of the patients were alcohol consumers.

Accuracy of LBC presence for diagnosing gastric intestinal metaplasia

For the diagnosis of IM, compared to histological assessment, the NBI-ME findings had an accuracy of 93% (95%CI: 90-97), a sensitivity of 80% (95%CI: 67-92), a specificity of 96% (95%CI: 93-99), a positive predictive value of 84% (95%CI: 73-96), and a negative predictive value of 95% (95%CI: 92-98).

Interestingly, in our study, the NBI-ME examination overlooked the presence of gastric IM in 8 cases (false negative cases), and 7 out of the 8 cases had an IM that was less than 5%. Moreover, in the 6 false positive cases, the histological examination showed the presence of reactive gastropathy (4 cases) or H. pylori active chronic gastritis (2 cases).

The semiquantitative correlation between the grades of LBC appearance and the percentage of histological IM is shown in Table 2 and Figure 2; the correlation index was 79% (P < 0.01).

Intestinal metaplasia	LBC (n)				Total (n)
	Negative	+	++	+++
0%	59	6	0	0	65
≤ 5%	7	8	2	0	17
5%-45%	1	6	5	2	14
45%-80%	0	0	2	1	3
≥ 80%	0	0	0	1	1

LBC: Light blue crest.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Graphic representation of the correlation between the degree of light blue crest appearance during narrow-band imaging with magnifying endoscopy examination and the percentage of gastric intestinal metaplasia detected by histological examination. LBC: Light blue crest.

Intestinal metaplasia has been widely studied as the main mucosal background for the development of oesophageal and gastric adenocarcinomas. However, to date, no endoscopic patterns that clearly define IM or dysplasia in the stomach have been found. Kaminishi et al[19] evaluated the accuracy of endoscopic findings for diagnosing chronic gastritis and found that ash-coloured nodular changes were highly specific (98% ± 99%) but had a very poor sensitivity (6% ± 12%) for detecting histological IM; the authors concluded that conventional endoscopy is not really useful for diagnosing gastric IM. Later, Dinis-Ribeiro et al[20] observed that magnifying chromoendoscopy with methylene blue staining had good sensitivity (76%) and specificity (87%) for detecting gastric IM. Moreover, methylene blue chromoendoscopy has been found to be useful for assessing the extent of IM[21]. However, this latter technique is time consuming and has some limitations, including the need for preparation with mucolytic agents, dye spraying and vigorous irrigation of the mucosal surface; furthermore, a recent report suggests that methylene blue chromoendoscopy induced questionable oxidative DNA damage in a patient with Barrett’s mucosa[22]. Therefore, collecting biopsies for subsequent histological evaluation remains the current gold standard for diagnosing IM. However, the fact that premalignant gastric lesions may occur multifocally represents a limitation of random biopsy sampling.

Several new imaging techniques have been developed recently to improve the diagnostic yield of endoscopy. The NBI-ME system has the practical advantage of not requiring complex preparation procedures, the use of any staining technique or a long duration. The additional value of NBI-ME in the detection of gastric pre-malignant lesions is still unclear, especially in countries with a low gastric cancer incidence[7].

Recently, Capelle et al[23] compared the yield of NBI alone to that of white light endoscopy (WLE) in the surveillance of 43 patients with IM and dysplasia. The authors concluded that NBI alone considerably increased the detection of premalignant gastric lesions compared to routine WLE and appeared to be superior to WLE for the surveillance of patients with these advanced gastric lesions. However, a combination of NBI and magnification is likely to provide the best alternative to current endoscopic practice. Previous studies have already demonstrated a high correlation between the microvascular patterns found with NBI-ME and the diagnosis of gastric cancer[24,25]. Moreover, the combined optical technology of NBI with ME has been shown to identify with great accuracy the presence of gastric IM through the visualisation of LBC appearance on the epithelial surface[16]. However, the main limitations of the above-mentioned studies are that they were all conducted in patients with already known diagnoses of intestinal metaplasia/dysplasia/gastric cancer or in a population at a high risk for developing gastric cancer. Other limitations included the small sample sizes and the limited assessment of the gastric antrum and angulus with the exclusion of the gastric body.

Therefore, no prospective data are available on the diagnostic utility and accuracy of NBI-ME for detecting IM in an unselected population presenting at an endoscopy unit in routine clinical practice. Our study showed that NBI-ME allowed us to detect gastric IM areas with an accuracy of 93%, a sensitivity of 80%, a specificity of 96%, a positive predictive value of 84% and a negative predictive value of 95%. These results are similar to those recently published by Pimentel-Nunes et al[26], who found a sensitivity of 68% and a specificity of 87% using the same method and a simplified classification system for NBI in the diagnosis of gastric lesions. In our investigation, NBI-ME underestimated the presence of IM in 8 patients; for 7 of these patients, metaplasia was histologically estimated to be less than 5%. The fact that NBI-ME has its major limitation in identifying gastric IM lower than 5% when performed by experienced endoscopists represents a new finding about the diagnostic yield of this technique; at this stage, we can assume that this value constitutes the sensitivity limit of this technique. The false positive cases (6 patients) presented with a histological diagnosis of reactive gastropathy (4 cases) or H. pylori active chronic gastritis (2 cases). Both of these conditions may sometimes represent a confounding factor because they show endoscopic features similar to those of gastric IM.

The prevalence of histologically observed IM in our cohort (35%) was slightly higher than the prevalence estimated in the general Western population (25%)[5-9] and in other studies on the Italian population (20%-30%)[5,8,11-13]. A possible explanation for our findings may be related to the higher mean age and prevalence of H. pylori infection in our patients, as it is well known that H. pylori infection, increased age and smoking (> 20 cigarettes/daily) are the main risk factors for the development of premalignant gastric lesions[27]. However, the prevalence of IM in the general population is reported to be highly variable. Additionally, the majority of data come from studies on patients with dyspeptic symptoms and/or suspected H. pylori infections, while information derived from unselected patients (i.e., consecutive patients enrolled independently of GI symptoms or clinical status) is limited.

The semiquantitative evaluation of LBC appearance in our study and its correlation with the histological measurement of IM percentage shows that the agreement was high (79%). Previously, Uedo et al[16] also found a significant correlation between the diffuse positivity of LBC (+++) mucosa and the presence of histological markers of IM, such as immunohistochemical staining for CD10 and Alcian blue/high iron diamine. However, these authors did not assess the correlation between LBC appearance and the percentage of histologic IM. Therefore, to the best of our knowledge, our study is the first to show a good correlation between the degree of IM detected by NBI-ME and the severity of IM identified by histological measurement in a series of consecutive patients without specific indications to perform an upper GI endoscopy.

In contrast to Uedo et al[16] who used NBI-ME and Capelle et al[23] who performed NBI alone, we opted to evaluate not only the antrum and the angulus of the stomach but also the body; including the body was recently emphasised in the ESGE guidelines[28], despite the significantly longer examination time. This measurement was included because our intention was to demonstrate the potential application of NBI-ME in a population-based screening and surveillance setting, and we thought that gastric assessment of only one region of the stomach was not adequate and fully informative. However, IM is predominantly found in the antrum, which was confirmed in our study. Therefore, limiting the endoscopic evaluation to the antrum permits an accurate observation and a shorter duration of the examination. Moreover, it has been shown that three biopsies (two antral and one angular) appear to be the best compromise between acceptable accuracy and ease of use in the detection of IM by endoscopy[29].

Our study had some limitations. First, the study included only a single centre. Further research in larger multicentre prospective studies is necessary to evaluate whether NBI-ME yields adequate results in the detection and grading of gastric IM. Second, the endoscopic procedure of WLE and NBI-ME was performed by the same endoscopist; thus, detection of IM by NBI-ME could possibly be biased by the previous WLE observations. Third, as a learning curve must still be defined, regular training is mandatory to improve our findings. Indeed, despite our prolonged training, we obtained a value of sensitivity that was lower than the value obtained by Uedo et al[16]. This difference was likely the result of differences in training between Japanese and Western gastroenterologists. Due to the higher gastric cancer incidence, Japanese endoscopists are trained to scrutinise gastric mucosal areas that are compatible with atrophy and early cancer and spend more time on a thorough mucosal examination than in Western countries. In addition, uniform criteria for this technique must be adopted in large controlled trials in Western or Eastern countries.

In conclusion, NBI-ME was shown to be a valid method for IM detection, except in those cases with IM lower than 5%, which may not be clinically relevant. In routine clinical practice, this technique can reliably target which patients should be biopsied to evaluate IM and those who do not need biopsies. Moreover, a semi-quantitative evaluation of LBC appearance was feasible as there was a good correlation with the histological assessment of IM percentage.