**Box 1. Biopsying Barrett's oesophagus**

Dysplasia is often macroscopically invisible. In patients with Barrett's oesophagus, endoscopists are therefore advised to follow a rigorous biopsy protocol. The British Society of Gastroenterologists recommends the following:


**NB/** Even with strict adherence to this policy <5% of oesophageal mucosa is sampled.

Endoscopic Detection and Eradication of Dysplastic Barrett's Oesophagus 153

Biopsies should be taken (as per the BSG guidelines described in Box 1) and should start distally and work proximally to minimise any obstruction to the endoscopic view caused by bleeding. A description of the Barrett's segment should then be recorded using the Prague C&M classification (see section 2) and positions of random biopsies and suspicious areas recorded meticulously. Where possible, visible macroscopic neoplasia should also be

Visible macroscopic early neoplastic lesions in Barrett's oesophagus are classified using the Paris classification. A description of the superficial (0) lesions is detailed below.

Several techniques have been developed to improve endoscopic recognition of dysplasia and intramucosal carcinoma in Barrett's oesophagus. These aim to minimise sampling randomness and also facilitate targetted endoscopic resection in patients with histologically confirmed HGD / IMC. In addition, they aim to improve assessment of disease extent and

Modern high resolution endoscopes which generate up to one million pixel images (compared to the 300,000 pixel images of traditional scopes) have been shown to have a higher sensitivity for the detection of early Barrett's neoplasia provided they are used by expert endoscopists.(Kara et al., 2005a; Kara et al., 2005c) These high definition endoscopes should be used in conjunction with a high definition television to further enhance the projected image quality and enable projection onto a larger screen without loss of image

Studies have shown that up to 80% of patients referred to specialist units with biopsy proven HGD without visible abnormality will be found to have 1 or more visible abnormalities when endoscopy is repeated by an expert endoscopist using a high resolution endoscope. (10,12 from endoscopic work-up) (Kara et al., 2005a; Curvers et al., 2008d)

Chromoendoscopy utilises stains which bind selectively to different oesophageal mucosa and so can enable discrimination between non-dysplastic Barrett's oesophagus and HGD /

classified according to the Paris classification (Box 2).

0 **Superficial lesions**

 0-Ip Pedunculated 0-Is Sessile lesions

 0-IIa Slightly elevated 0-IIb Completely flat 0-IIc Slightly depressed

**Box 2. Classification of visible early Barrett's neoplasia**

0-I Protruding / polypoid lesions

0-III Excavated / ulcerated lesions

0-II Non-protruding / non-excavated lesions

Most dysplastic Barrett's lesions are of superficial type (0-II).

minimise the incidence of missed synchronous tumours.

**4.1 High resolution endoscopy by expert endoscopists** 

resolution.

**4.2 Chromoendoscopy** 

Endoscopic recognition of gross mucosal abnormalities such as ulceration, nodularity, and erythema is relatively straightforward. The problem is that early neoplastic lesions are frequently flat and often have little or no visible mucosal abnormality. Only 50-70% of HGD can be identified by experienced endoscopists using white light endoscopy. This figure is lower for non-specialists and is considerably lower for the detection of LGD. In addition, more than 20% of intramucosal cancers may be missed endoscopically, even in specialist units. This is particularly concerning when it is considered that routine biopsy protocols used for Barrett's surveillance have been shown to miss up to 57% of early neoplastic lesions. (Vieth et al., 2004)

As early neoplasia in Barrett's oesophagus is a relatively rare finding, the lack of familiarity of most endoscopists with its typical appearances is a significant limiting factor in its detection. Knowledge of the appearance of these early lesions is therefore key to their early recognition. Figure 1 illustrates a range of mucosal abnormalities within segments of Barrett's oesophagus which are consistent with early neoplastic change.

Fig. 1a (left) illustrates a nodular area of what proved to be HGD in a tongue of Barrett's oesophagus. 1b (right) illustrates multifocal nodular Barrett's neoplasia. In this example clinicians must have a very high level of suspicion for the presence of invasive malignancy

Another factor critical to endoscopic detection of dysplastic Barrett's oesohagus is a systematic approach to mucosal inspection. The oesophagus may be cleaned using water or 1% acetylcysteine to remove saliva and gastric refluxate and the oesophagus must be adequately distended by inflation. Special care must be taken in patients with a hiatus hernia as in these cases the distal extent of the Barrett's segments can be difficult to identify meaning dysplasia at the oesophagogastric junction can be missed. In addition, clinicains should be aware that the majority of neoplastic lesions are located between 12 and 6 o'clock in the endoscopists view. (Curvers et al., 2008b) Importantly clinicians should also commit to investing considerable time to endoscopic inspection (as well as the time required for multiple biopsies), and endoscopy lists should be planned accordingly when patients with known Barrett's oesophagus are attending for surveillance.

Endoscopic recognition of gross mucosal abnormalities such as ulceration, nodularity, and erythema is relatively straightforward. The problem is that early neoplastic lesions are frequently flat and often have little or no visible mucosal abnormality. Only 50-70% of HGD can be identified by experienced endoscopists using white light endoscopy. This figure is lower for non-specialists and is considerably lower for the detection of LGD. In addition, more than 20% of intramucosal cancers may be missed endoscopically, even in specialist units. This is particularly concerning when it is considered that routine biopsy protocols used for Barrett's surveillance have been shown to miss up to 57% of early neoplastic

As early neoplasia in Barrett's oesophagus is a relatively rare finding, the lack of familiarity of most endoscopists with its typical appearances is a significant limiting factor in its detection. Knowledge of the appearance of these early lesions is therefore key to their early recognition. Figure 1 illustrates a range of mucosal abnormalities within segments of

Fig. 1a (left) illustrates a nodular area of what proved to be HGD in a tongue of Barrett's oesophagus. 1b (right) illustrates multifocal nodular Barrett's neoplasia. In this example clinicians must have a very high level of suspicion for the presence of invasive malignancy Another factor critical to endoscopic detection of dysplastic Barrett's oesohagus is a systematic approach to mucosal inspection. The oesophagus may be cleaned using water or 1% acetylcysteine to remove saliva and gastric refluxate and the oesophagus must be adequately distended by inflation. Special care must be taken in patients with a hiatus hernia as in these cases the distal extent of the Barrett's segments can be difficult to identify meaning dysplasia at the oesophagogastric junction can be missed. In addition, clinicains should be aware that the majority of neoplastic lesions are located between 12 and 6 o'clock in the endoscopists view. (Curvers et al., 2008b) Importantly clinicians should also commit to investing considerable time to endoscopic inspection (as well as the time required for multiple biopsies), and endoscopy lists should be planned accordingly when patients with

known Barrett's oesophagus are attending for surveillance.

Barrett's oesophagus which are consistent with early neoplastic change.

lesions. (Vieth et al., 2004)

Biopsies should be taken (as per the BSG guidelines described in Box 1) and should start distally and work proximally to minimise any obstruction to the endoscopic view caused by bleeding. A description of the Barrett's segment should then be recorded using the Prague C&M classification (see section 2) and positions of random biopsies and suspicious areas recorded meticulously. Where possible, visible macroscopic neoplasia should also be classified according to the Paris classification (Box 2).


Several techniques have been developed to improve endoscopic recognition of dysplasia and intramucosal carcinoma in Barrett's oesophagus. These aim to minimise sampling randomness and also facilitate targetted endoscopic resection in patients with histologically confirmed HGD / IMC. In addition, they aim to improve assessment of disease extent and minimise the incidence of missed synchronous tumours.

#### **4.1 High resolution endoscopy by expert endoscopists**

Modern high resolution endoscopes which generate up to one million pixel images (compared to the 300,000 pixel images of traditional scopes) have been shown to have a higher sensitivity for the detection of early Barrett's neoplasia provided they are used by expert endoscopists.(Kara et al., 2005a; Kara et al., 2005c) These high definition endoscopes should be used in conjunction with a high definition television to further enhance the projected image quality and enable projection onto a larger screen without loss of image resolution.

Studies have shown that up to 80% of patients referred to specialist units with biopsy proven HGD without visible abnormality will be found to have 1 or more visible abnormalities when endoscopy is repeated by an expert endoscopist using a high resolution endoscope. (10,12 from endoscopic work-up) (Kara et al., 2005a; Curvers et al., 2008d)

#### **4.2 Chromoendoscopy**

Chromoendoscopy utilises stains which bind selectively to different oesophageal mucosa and so can enable discrimination between non-dysplastic Barrett's oesophagus and HGD /

Endoscopic Detection and Eradication of Dysplastic Barrett's Oesophagus 155

63 patients with Barrett's oesophagus using white light endoscopy and confocal microscopy. Intravenous fluorescein was administered to generate vascular contrast and at sites of neoplasia could be seen to disperse within the lamina propria due to irregular neovascularisation. Accuracy of CM was found to be 97.4% (sensitivity 93%, specificity 98%). (Kiesslich et al., 2006) In another study by Dunbar et al, CM was shown to help target biopsies to areas of neoplasia, doubling diagnostic yield per biopsy taken, and avoiding the need for biopsy in two thirds of patients undergoing surveillance. However, no overall increase in neoplasia was identified when CM targetted biopsying was compared to random

Confocal microscopy is an expensive technique and requires the use of exogenous contrast. It has already demonstrated potential in early diagnosis of Barrett's neoplasia although the excellent results reported by some studies have not been universally matched. (Pohl et al., 2008) Further studies are required before this technique can be reccommended for

Molcular biomarkers associated with neoplastic cells can be labelled using a specifically targetted probe molecule which has been tagged with a visual agent such as a fluoresecent dye. (Pierce et al., 2008) (Thekkek et al., 2011) The probe molecule selectively binds to the biomarker so that areas of neoplasia can be visualised with a high signal to noise ratio. Lu et al identified a cell surface peptide specific to adenocarcinoma which they labelled using a fluorescein-tagged antibody delivered topically. The oesophagus was then washed to remove any unbound antibody and a fluorescence endoscope was used to visualise

Other similar studies have used a range of potential biomarkers with similar effect. This is clearly a very promising technique for the detection of early neoplasia but further on-going work is necessary to identify novel molecular targets in order to improve sensitvity and

Raman spectroscopy is an optical diagnostic technique which has shown considerable potential for early diagnosis of a variety of malignant disease states including oesophageal neoplasia. Raman spectroscopy measures the molecular-specific, inelastic scattering of laser light within tissue in order to generate a unique molecular 'fingerprint'. Normal, dysplastic and cancerous tissues have differing biochemical cellular components leading to characteristic spectral differences which can be analysed. Laboratory based Raman spectrometers are capable of discriminating between eight pathological groups in the distal oesophagus (including Barrett's metaplasia, HGD and adenocarcinoma) with sensitivities between 73% and 100%. (Kendall et al., 2003) Several groups are currently investigating the potential for endoscopic Raman spectroscopy using a fibre-optic Raman probe. Fibreoptic Raman sectroscopy has already demonstrated encouraging results following in vivo trials in the stomach, bladder and cervix. Although some way off clinical implementation in the oesohagus, in vivo and ex vivo results are promissing and this technique may become widely available in the short to medium term to enable instant endoscopic diagnosis of dyslasia (without the need for biopsy) and to facilitate immediate, targetted

specificity before widespread implementation of the technique can be contemplated.

quadrantic random biopsies every 2cm.

neoplastic disease. (Lu and Wang, 2008)

widespread use.

**4.7 Labelling of biomarkers** 

**4.8 Raman spectroscopy** 

endotherapy.

adenocarcinoma. Staining and lesion defining agents utilised include methylene blue, indigo carmine, and acetic acid. Results from studies utilising this technique have been mixed citing problems such as an inability to uniformly coat the oesophageal mucosa with the stain, and excessive time necessary for stain spraying as particular concerns.(Shaheen and Richter, 2009; Lim et al., 2006; Ragunath et al., 2003) None of these techniques has been shown to consistently out-perform high resolution endoscopy in the detection of early neoplastic lesions.(Curvers et al., 2008c) Chromoendoscopy is often both labour-intensive and operator-dependent and therefore although it may have a role when used in specialist centres by expert users, it is unlikely to develop a wider role in routine clinical practice.

### **4.3 Narrow band imaging (NBI)**

NBI filters white light into blue and green wavelengths (at the push of a button) giving more accurate images of the mucosal and vascular patterns in the oesophageal lining. This increased superficial imaging of the oesophagus (without the need for staining) can be used to identify dysplastic lesions within Barrett's segments.(Kara et al., 2006a) In the hands of experienced users the technique has shown promise however, results have been mixed.(Sharma et al., 2006a; Curvers et al., 2008a) A recent trial from Holland shows no diagnostic benefit from either NBI or chromoendoscopy.(Curvers et al., 2008a) However, data on the accuracy of NBI is still inconclusive and results of ongoing mulitcentre randomised controlled trials are awaited.

#### **4.4 Autofluorescence imaging (AFI)**

Following excitation with short wavelengths of light many endogenous tissues emit fluorescence radiation which can be measured using fluorescence spectroscopy. Metaplastic and dysplastic Barrett's oesohagus have been shown to emit slightly different fluorescence spectra enabling the technique to be used as a mechanism to discriminate between the two pathologies. AFI appears to improve the detection of early Barrett's neoplasia when used in combination with high resolution endoscopy, although the false positive rate is relatively high.(Curvers et al., 2008b; Kara et al., 2005b; Kara et al., 2006b) Further studies are clearly indicated to truely assess the potential long-term role for AFI.

#### **4.5 Optical coherence tomography (OTC)**

OCT is analogous to ultrasound but can produce higher quality images as it relies on scattering of near infrared light as opposed to reflection of sound waves. OCT can obtain cellular images of sub-epithelial tissue through differences in their light scattering properties and avoids the need for exogenous contrast material.

In a study of 55 patients with Barrett's oesophagus, OCT was shown to delineate between HGD and oesophageal adenocarcinoma with a sensitivity of 83% and a specificity of 75%.(Evans et al., 2006) Similarly, a study of 33 patients demonstrated a diagnostic accuracy of 78% for the identification of dysplastic Barrett's oesophagus but with considerable user discrepency (56% to 98%). (Isenberg et al., 2005) Further clinical evaluation is required to fully assess the performance of OCT and assess the feasibility of introducing this promising diagnostic tool into routine clinical practice.

#### **4.6 Confocal microscopy (CM)**

CM magnifies the mucosa by more than 1000 fold producing images with 1-2 µm spatial resolution and allowing real time visualisation of cellular structures. Kiesslich et al studied 63 patients with Barrett's oesophagus using white light endoscopy and confocal microscopy. Intravenous fluorescein was administered to generate vascular contrast and at sites of neoplasia could be seen to disperse within the lamina propria due to irregular neovascularisation. Accuracy of CM was found to be 97.4% (sensitivity 93%, specificity 98%). (Kiesslich et al., 2006) In another study by Dunbar et al, CM was shown to help target biopsies to areas of neoplasia, doubling diagnostic yield per biopsy taken, and avoiding the need for biopsy in two thirds of patients undergoing surveillance. However, no overall increase in neoplasia was identified when CM targetted biopsying was compared to random quadrantic random biopsies every 2cm.

Confocal microscopy is an expensive technique and requires the use of exogenous contrast. It has already demonstrated potential in early diagnosis of Barrett's neoplasia although the excellent results reported by some studies have not been universally matched. (Pohl et al., 2008) Further studies are required before this technique can be reccommended for widespread use.
