Clinical and technical evidence

A literature search was carried out for this briefing in accordance with the interim process and methods statement. This briefing includes the most relevant or best available published evidence relating to the clinical effectiveness of the technology. Further information about how the evidence for this briefing was selected is available on request by contacting mibs@nice.org.uk.

Published evidence

Three studies are summarised in this briefing including 1,185 people with confirmed or suspected Barrett's oesophagus and a further 330 images obtained with narrow band imaging (NBI).

There are also 2 meta-analyses that are applicable to NBI. Thosani et al. (2016) reported pooled sensitivity, negative predictive value, and specificity for NBI of 94.2% (95% confidence interval [CI], 82.6 to 98.2), 97.5% (95% CI, 95.1 to 98.7), and 94.4% (95% CI, 80.5 to 98.6), respectively (reference standard not reported). Qumseya et al. (2013) reported that advanced imaging techniques (such as NBI) increased the diagnostic yield for detection of dysplasia or cancer by 34% (95% CI, 20% to 56%; p<0.0001) in comparison with non-advanced endoscopy.

Table 1 summarises the clinical evidence as well as its strengths and limitations.

Overall assessment of the evidence

The evidence for NBI is of good quality and provides useful information as to how NBI could be used in the NHS. The evidence includes a randomised crossover trial comparing NBI (targeted biopsy) with standard white-light endoscopy (random biopsy), a feasibility study that reports diagnostic accuracy and a report on the development and validation of a new criteria for optical diagnosis of dysplasia intended for use with NBI. The research was done in specialist centres and endoscopies were done by highly trained endoscopists so may not be reproducible in all NHS settings.

Table 1 Summary of selected studies

Beg et al. (2018)

Study size, design and location

1,022 people with Barrett's oesophagus in a before‑and‑after study.

Location: UK.

Intervention and comparator(s)

NBI.

Comparators: acetic acid chromatography and white-light endoscopy.

During the prospective period data were obtained for 560 gastroscopies.

Endoscopies happening after April 2011 were done using white-light endoscopy and NBI. If the length of Barrett's epithelium was greater than 3 cm, acetic acid chromatography was also used.

Key outcomes

During 2007 to 2010 dysplasia was detected in 11.0% of patients, low-grade dysplasia in 7.7% and high-grade dysplasia or cancer in 3.3%.

During 2011 to 2014 dysplasia was detected in 11.3% of patients, low-grade dysplasia in 9.4% and high-grade dysplasia or cancer in 1.9%.

Strengths and limitations

This study showed that using NBI and acetic acid chromatography did not result in an increased detection rate of dysplasia.

This study has a before‑and‑after design and introduces a change in practice between the prospective period and the study period, which may have confounded results.

Nogales et al. (2017)

Study size, design and location

Evaluation of 100 images of Barrett's oesophagus taken using NBI.

Location: Spain.

Intervention and comparator(s)

Images obtained using non-magnifying NBI in the Evis Exera III endoscope.

Interpretation of images compared with clinical assessment using white-light endoscopy and histopathology.

Key outcomes

Dysplasia prediction accuracy for NBI was 81.1%, sensitivity 48.4%, specificity 91%, positive predictive value 61.4% and negative predictive value 85.5%. Intraobserver concordance for dysplasia was weak, K=0.4.

Strengths and limitations

This study used the BING classification system to assess images. Diagnosis was made using a photographic image taken during endoscopy, which may have confounded results.

Sharma et al. (2016)

Study size, design and location

Development and validation of a classification system to identify high-grade dysplasia and oesophageal adenocarcinoma with NBI.

230 images were reviewed during the development and validation process.

Location: international (Barrett's international NBI group, BING).

Intervention and comparator(s)

NBI.

No comparator.

Key outcomes

The classification system was developed and agreed during a meeting of the BING. The classification system was then independently validated by experts.

The BING criteria identified patients with dysplasia at 85% overall accuracy, 80% sensitivity, 88% specificity, 81% positive predictive value, and 88% negative predictive value, compared with histology results. When dysplasia was identified with a high level of confidence, these values were 92%, 91%, 93%, 89%, and 95%, respectively, compared with histology results.

Agreement between validating experts was high (K=0.681).

Strengths and limitations

Although this study does not compare NBI with any other method of endoscopy it provides useful guidance on how to classify images obtained by NBI.

This study was funded by the company.

Sharma et al. (2013)

Study size, design and location

123 people with Barrett's oesophagus in a randomised crossover trial.

Location: international.

Intervention and comparator(s)

NBI and high-definition white-light endoscopy.

During high-definition white-light endoscopy biopsies were taken every 2 cm as well as additional biopsies in areas with visible lesions.

During NBI endoscopy only targeted biopsies were obtained.

All patients received both endoscopy methods.

Key outcomes

Both high-definition white-light endoscopy and NBI detected intestinal metaplasia in 92% of patients. Use of NBI resulted in significantly fewer biopsies compared with high-definition white-light endoscopy (3.6 compared with 7.6 per person, p<0.0001). NBI detected a higher proportion of areas with dysplasia (30% compared with 21%, p=0.01).

Endoscopists performing the procedures noted that irregular mucosal and vascular patterns were present in all areas of high-grade dysplasia and that non-dysplastic tissue appeared normal.

Strengths and limitations

The study is of good quality and is published in a leading journal. The study was done in 3 centres, 2 in the US and 1 in Netherlands, and therefore may not be representative of NHS practice. In this study results were analysed per lesion rather than by patient. This suggests NBI is able to detect more lesions but not necessarily more people with high-grade dysplasia.

Singh et al. (2013)

Study size, design and location

40 people with Barrett's oesophagus in a preliminary feasibility study.

Location: Australia.

Intervention and comparator(s)

Optical diagnosis using NBI with biopsy of target lesions, NBI dual focus mode (magnification up to 70 times) was also used.

NBI optical diagnosis results were compared with final histopathological diagnosis.

Key outcomes

NBI: sensitivity 100%, specificity 93.8%, positive predictive value 68.6%, negative predictive value 100%.

NBI dual focus: sensitivity 100%, specificity 86.2%, positive predictive value 73.6%, negative predictive value 100%.

Strengths and limitations

The authors specified that the dual focus mode was used in areas that appeared normal after NBI examination. It is not clear if this mode is available in all endoscope models.

Abbreviations: NBI, narrow band imaging.

Recent and ongoing studies