Advances in the endoscopic diagnosis and treatment of Barrett’s neoplasia

Cytosponge

The cytosponge is a sponge contained within a capsule that is attached to a string. The capsule is ingested with water and dissolves in the stomach after 3 to 5 minutes. The string then is pulled to retrieve the sponge and cells collected from the oesophagus^6,7. The cells are analysed with the biomarker trefoil factor 3 to make a diagnosis of BO.

An initial study of the cytosponge⁶ demonstrated a 3% pickup rate of BO in a primary care setting with a majority of patients (82%) reporting low levels of anxiety, making it a potential tool for mass screening. More recently, Fitzgerald et al. followed up on their work with the cytosponge in a large case-control study⁸. In total, 1,110 patients were recruited: 463 patients with symptoms of dyspepsia and reflux and 647 patients with a prior diagnosis of BO underwent gastroscopy following cytosponge examination; 93.9% of patients had a successful cytosponge examination. Overall, cytosponge sensitivity for detecting BO was 79.9%, increasing to 87.2% in patients with more than 3 cm of BO. The specificity for diagnosing BO was 92.4%. Further trials on the cytosponge device are ongoing but these data suggest that it is an acceptable and accurate device.

Transnasal endoscopy and oesophageal capsule endoscopy

A recent randomised controlled trial compared the use of unsedated TNE versus sedated gastroscopy for BO screening. Two hundred and nine patients were recruited to standard gastroscopy (surveillance oesophago-gastro-duodenoscopy, or sOGD), unsedated TNE in a mobile research van (muTNE) or a hospital outpatient endoscopy suite (huTNE). Uptake was greater in the unsedated TNE group: 47.5% for muTNE and 45.7% for huTNE versus 40.7% for standard gastroscopy. Complete evaluation of the oesophagus was similar between the groups: 99% muTNE, 96% huTNE and 100% sOGD⁹.

Chak et al.¹⁰ examined the acceptability of TNE versus OCE in a randomised controlled trial. They found that uptake for screening examination was low: 15.2% of patients (n = 1,210). Effectiveness of screening for the detection of BO was similar for both technologies. A meta-analysis¹¹ of studies investigating OCE as a screening modality for BO in patients with reflux symptoms demonstrated an overall sensitivity of 77% and a specificity of 86%.

Although technology is advancing in the field of BO screening, there are insufficient data about its cost-effectiveness. These novel approaches appear to be acceptable to patients but more data are required to target whom and when to offer screening.

High-definition white light endoscopy

With the advent of charge-coupled device (CCD) chips, high-definition white light (HDWL) endoscopes are able to capture and display high-definition images with pixel densities of more than 10 million pixels, making standard definition (pixel density of 100,000 to 400,000) endoscopes obsolete²⁵.

The sensitivity and specificity of HDWL endoscopy in detecting Barrett’s neoplasia are 40%–64% and 98%–100%, respectively^26,27. BSG and American Gastroenterological Association (AGA) guidelines recommend the use of high-resolution endoscopes for Barrett’s surveillance^1,28.

Virtual chromoendoscopy

The enhancement of mucosal surface and vascular patterns using optical and digital filter technologies has added to the arsenal of the advanced endoscopist in the quest to improve dysplasia detection. Narrow band imaging (NBI) (Olympus, Tokyo, Japan) and blue laser imaging (BLI) (Fujifilm, Tokyo, Japan) use a filter located in front of the light source. This technology filters white light and limits the wavelength of the light projected to 415 to 540 nm²⁹. When projected onto a mucosal surface, this ‘narrow band’ of light appears blue and green. The blue light penetrates the superficial layer of the mucosa, thereby enhancing the view of superficial capillaries and the crypt patterns in the mucosal surface. In contrast, technologies such as i-Scan (Pentax, Tokyo, Japan) and Fujinon intelligent chromoendoscopy (FICE) (Fujinon, Tokyo, Japan) employ complex proprietary algorithms to digitally reproduce a narrow-spectrum image at the push of a button on the endoscope. BLI³⁰ is a new technology with white and blue lasers that produce narrow-band light.

Narrow band imaging. NBI is the most studied optical imaging technology thus far and has a sensitivity and specificity of 47%–100% and 72%–100%, respectively, for detecting Barrett’s neoplasia^27,31–34. NBI selectively enhances mucosal vascular patterns by narrowing the spectrum of light, reducing the amount of red light in the displayed image whilst narrowing the spectrum of blue and green, making blood vessels appear dark against the background mucosa³⁵.

A majority of these studies were conducted in tertiary referral centres by expert endoscopists evaluating an enriched population with a high index of suspicion of neoplasia. The technology has yet to be validated in non-expert hands or in a surveillance population. Therefore, we would suggest that training in the use of this technology and data in a surveillance population be required prior to adoption in routine clinical practice.

i-Scan. i-Scan uses post-processor technology that reconstructs the image transmitted from the endoscope by using a computer-based algorithm which is able to accentuate both surface patterns and vasculature³⁵. A randomised control trial found that the yields of acetic acid-guided versus i-Scan-guided biopsies in detecting specialised columnar epithelium were comparable³⁶. Verna et al. found that dysplasia detection rate using this technology was inferior to that using the standard four-quadrant biopsy technique³⁷. However, these studies were poorly designed and have small sample sizes. More robust studies on the utility of i-Scan in the detection of dysplasia in BO are required.

Fujinon intelligent chromoendoscopy. FICE uses a CCD in the endoscope to capture spectral reflectance data. A matrix processing circuit found in the video processor then receives the data. The reflectance spectra of corresponding pixels that make up the conventional image are mathematically estimated. From this information, a single-wavelength virtual image is reconstructed. Three such single-wavelength images can be selected and assigned to the red, green and blue monitor inputs to display a composite colour-enhanced multiband image in real time. This can be used like NBI to remove data from the red part of the waveband and narrow the green and blue spectra³⁵.

There is a paucity of research evaluating the utility of FICE in detection of dysplasia in Barrett’s. The sole published study to date is a prospective pilot study carried out in a tertiary centre. It was found that the dysplasia detection rate of FICE, when used in conjunction with acetic acid, is 86%³⁸.

Autofluorescence imaging. Autofluorescence imaging (AFI) is based on a principle that a specific light wavelength can cause fluorescence of endogenous biomolecules such as collagen, nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), and porphyrins. These molecules can accumulate in dysplastic oesophageal mucosa³⁹. A randomised cross-over multi-centre trial on an enriched population found a marginal gain of AFI over quadrantic biopsies which did not reach statistical significance⁴⁰. Endoscopic trimodal imaging (ETMI) systems which integrate AFI with HDWL and NBI have not been shown to be superior to standard resolution white light endoscopes^41,42. With a lack of evidence of its efficacy and high false-positive rates, the use of AFI and ETMI at present remains in the domain of endoscopic research^39–41.

Chromoendoscopy with white light endoscopy

Methylene blue chromoendoscopy. Three randomised cross-over trials found that the diagnostic accuracy of methylene blue 0.5%-directed biopsies is higher than random 2 cm quadrantic biopsies^43–45. However, a meta-analysis which included data from six trials found that overall methylene blue chromoendoscopy was not superior to random biopsies in the detection of specialised intestinal metaplasia or dysplasia⁴⁶.

Acetic acid chromoendoscopy. Acetic acid is a weak acid that causes acetowhitening of the oesophageal mucosa. Over a period of seconds to minutes, dysplastic tissue will start to lose the acetowhitening effect before the surrounding non-dysplastic Barrett’s tissue. Differential loss of acetowhitening highlights the neoplastic focus as a red spot on a white background (Figure 1). This is an extremely promising technique with high sensitivity, universal applicability and negligible cost.

Figure 1. Intramucosal carcinoma (left) and high-grade dysplasia (right) highlighted by acetic acid.

Two large cohort studies have demonstrated effectiveness of acetic acid used at concentrations of 2.5% and 1.5%, respectively^47,48, in the detection of dysplasia within Barrett’s in high-risk populations. The two reported similar results, with sensitivities for dysplasia between 90% and 95% and specificities between 75% and 85%. A further study using 2.5% acetic acid found that the number of biopsies needed to detect neoplasia could be significantly reduced if acetic acid targeted biopsies were used in place of mapping biopsies, thereby reducing pathology-related costs by 97%⁴⁹. Tholoor et al.⁵⁰ reported the use of 2.5% acetic acid in a surveillance population and were able to demonstrate a threefold increase in neoplasia detection as compared with conventional protocol-guided biopsies. However, this was a non-randomised trial and an ongoing randomised trial, the ABBA study⁵¹, will answer this question soon. The ABBA study is a randomised, crossover, tandem endoscopy study comparing standard quadrantic biopsy protocol versus acetic acid targeted biopsies, in a Barrett’s surveillance population.

Cross-sectional optical imaging. Optical coherence tomography (OCT) and confocal laser endomicroscopy (CLE) are emerging technologies that are able to obtain micro-anatomical images of the oesophageal mucosa. OCT technology measures the difference between the backscatter of near-infrared low-coherence light below the tissue surface and a reference beam⁵². Using this information, it is able to reconstruct the microanatomy of the superficial mucosal layer⁵³. CLE, on the other hand, involves the integration of a confocal laser microscope in the distal tip of a conventional video endoscope⁵⁴. Although both technologies have sensitivities of 68% to 86% and specificities of 73% to 83%^55,56, the setup costs are high and special training is required for image interpretation. This currently limits its use in routine clinical practice. Table 1 summarises the performance of current imaging technologies in the diagnosis of Barrett’s neoplasia.

Endoscopic mucosal resection

Experience of endoscopic resection (ER) for early oesophageal adenocarcinoma began in the early 1990s in Asia; since then, techniques have significantly progressed. Initial experience came from the strip biopsy technique, which was further refined by the suck-and-cut and multi-band ligator techniques (Figure 2)⁶¹. Ell et al.⁶² reported the first (n = 64) series of ER for early Barrett’s cancer, demonstrating a 97% remission rate. Despite a short follow-up period (mean of 12 months), there was a significant rate of recurrence of 14%.

Figure 2. EMR of Barrett’s HGD.

(a) An area of Barrett’s high-grade dysplasia. (b) The same area demonstrating acetowhitening effect. (c) The same lesion as viewed down a multi-band ligator. (d) Pseudopolyp created by the band ligator. (e) Resection defect following endoscopic mucosal resection.

In the early years, a high rate of metachronous and recurrent lesions hampered the apparent technical success of ER with reported recurrence rates of up to 35%⁵². This led to a refinement of techniques incorporating both endoscopic mucosal resection (EMR) and ablative therapies (discussed below). A randomised trial⁶³ (The APE Study) comparing EMR + argon plasma coagulation (APC) ablation versus EMR + observation demonstrated a significantly reduced risk of metachronous lesions in the ablation arm, 3% versus 37%, such that ablation following ER is now standard treatment.

Pech et al.⁵² reported their outcome data of 1,000 patients with early Barrett’s cancer treated by ER, demonstrating a long-term complete remission rate of 93.8%. There was a 14.5% recurrence rate, and a majority of recurrences were treated endoscopically. Their serious complication rate was 1.5%, and no mortality was reported.

Endoscopic submucosal dissection

The main limitation of the EMR technique is that en bloc resection is possible only for lesions of less than 15 mm; lesions larger than this require piecemeal resection, making adequate histological assessment difficult. Figure 3 demonstrates the steps involved in the endoscopic submucosal dissection (ESD) technique. Experience with ESD in Japanese studies of early oesophageal squamous cell cancer has demonstrated improved outcomes over EMR. To date, three European studies have reported outcomes of ESD for neoplastic Barrett’s. Neuhaus et al.⁶⁸ reported on 30 patients with early neoplastic lesions up to 30 mm: en bloc resection was achieved in 90%, the complete neoplasia eradication rate was 96.4%, and there were no reported complications. More recently, Chevaux et al.⁶⁹ reported on their outcomes of 75 consecutive patients; ESD was performed on lesions of more than 15 mm, achieving an en bloc resection rate of 90% and a neoplasia eradication rate of 92%, and oesophageal strictures developed in 60% of patients. Probst et al.⁷⁰ reported on 87 patients with early oesophageal adenocarcinoma achieving a 95.4% en bloc resection rate with a recurrence rate of 2.4%; 11.7% of patients had stricturing, and no perforations were reported. Disappointingly, however, these studies have reported low R0 resection rates (38.5% to 48.5%).

Figure 3. ESD of Barrett’s IMC.

(a) pT1a/M3 intramucosal cancer arising in Barrett’s oesophagus. (b) The same lesion following acetic acid. Note the differential early loss of acetowhitening. (c) Edges of the lesion marked with endoscopic submucosal dissection (ESD) knife under virtual chromoendoscopy. (d) Submucosal injection. (e) Mucosal incision with ESD knife. (f) Resection base following ESD. (g) Resection specimen.

ESD appears to be a promising addition to current treatment techniques, especially for larger lesions. Our own data on 51 ESDs, on selected patients with Barrett’s cancer, have demonstrated an R0 resection rate for cancer of 88%, a recurrence rate of 3% and no complications⁷¹. A recently reported randomised controlled trial of EMR (n = 20) versus ESD (n = 31) demonstrated superior en bloc, R0 and curative resection rates for ESD; however, there was no difference in clinical outcome for either technique⁷². In our experience, large nodular lesions have a high risk of containing cancer and therefore we believe they should be removed in an en bloc fashion. Early European data have shown the feasibility and safety of ESD in Barrett’s neoplasia but have not proven its superiority over EMR. This has to be addressed in a well-designed study for a select group of patients to identify the exact role of ESD in Barrett’s neoplasia.