Improve outcomes from oesophageal cancer through innovative screening and surveillance tests

The future plans can be divided into two main components. The first is to extend the application of the Cytosponge collection device to the early detection of squamous cell cancer of the oesophagus (ESCC), which accounts for 50% of oesophageal cancers in the UK and is responsible for 90% of oesophageal cancer cases in the developing world. If a screening test for oesophageal cancer was introduced into the UK it would be ideal for this to detect the precursors for both of the main subtypes of oesophageal cancer, not just Barrett's oesophagus. The second is to take the proof of concept ex vivo work on molecular imaging, designed to highlight areas of dysplasia for biopsy and endoscopic therapy in Barrett's oesophagus, into the clinical setting. These two areas will synergise with the programme of work on early cancer detection in my laboratory. Ongoing research is focussed on addressing more basic research questions: i) the molecular triggers for the development of Barrett's metaplasia; ii) characterising the genomic aberrations in Barrett's dysplasia and adenocarcinoma as a prelude to identifying prognostic and therapeutic targets and iii) discovering and performing retrospective validation of risk stratification biomarkers for Barrett's oesophagus. These discoveries need to feed into clinical studies so that their potential for patient benefit can be realised. The ongoing BEST2 Cytosponge study described earlier is a good example of laboratory research being moved into a clinical study. The two projects described here are other examples of work which needs to be done to enable translation into the clinical setting and this NIHR Research Professorship would be an ideal platform in which to maximise pull through into the clinical arena. 1) A novel approach to screening and diagnosis for squamous cell cancer using the Cytosponge and associated assays Background There have been a number of previous efforts to develop non-endoscopic screening for ESCC in the high incidence areas of China since the 1950s and 1960s. However the analysis has relied on liquid-based cytology techniques in which cells are stained, and examined microscopically for epithelial atypia, similar to the evaluation of a Pap smear, (reviewed in Nature Review Clinical Oncology Lao-Sirieix and Fitzgerald 2012). The success of the Cytosponge as a diagnostic test for Barrett's oesophagus hinges on the combination of a successful cell sampling device which is acceptable to patients, coupled with a sample processing technique in which the cells are processed to a pseudobiopsy which still retains tissue architecture. This sample can then be assayed for biomarkers, either a single marker or multiple biomarkers, which enables an objective, quantifiable measure of the disease stage. Based on this positive experience with developing an approach to population based screening for Barrett's Oesophagus the purpose of the current proposal is to develop a non-endoscopic primary screening method for squamous cell dysplasia and asymptomatic ESCC, so that we can screen for asymptomatic ESCC using similar methods. Preliminary data We have developed a biomarker pipeline to ascertain suitable biomarkers for ESCC similar to that which proved successful for elucidating TFF3 as a diagnostic biomarker for BE (Gut Lao-Sirieix 2009). It comprises 4 steps: (i) Publically available expression array data is mined to identify mRNA species differentially expressed in normal squamous tissue (NE) and ESCC tissue, (ii) The top putative biomarkers are validated in mRNA extracted from snap frozen specimens of NE and ESCC, (iii) Biomarkers successfully validated in ii) are then validated at the protein level in paraffin-embedded specimens from NE and normal gastric mucosa compared with ESCC and squamous dysplasia (iv) Validated protein markers from iii) are then taken forward to samples, collected with the Cytosponge from known ESCC patients and controls. Using this algorithm there are 80 potential targets with differential expression between disease states of sufficient magnitude to point to clinical potential. Data for 6/10 targets which have validated so far is shown in Figure 1 and immunohistochemical assays are available for all of these. Proteomics would be another discovery platform but at the moment there are no publically available datasets. From the BEST2 Barrett's trial we also have experience in assaying for specific DNA mutations as an alternative biomarker approach and this will be a possibility for ESCC if required in addition to immunohistochemical approaches.Hence the principal aims of this proposed study are: 1) To validate putative markers at the protein level on paraffin-embedded specimens from the normal squamous oesophagus-dysplasia-carcinoma sequence and to use existing Cytosponge specimens from individuals previously recruited into the BEST clinical trial without ESCC to ensure high specificity (lack of false positives). 2) To assess the sensitivity and specificity of putative markers in Cytosponge samples for identifying asymptomatic individuals with high grade oesophageal squamous dysplasia (HGD) or ESCC in an ongoing endoscopic screening programme in a high risk population in China as a prelude to a clinical trial in the UK. For aim 1 markers validated by qRT-PCR will be validated at the protein level using immunohistochemistry (existing or custom generated antibodies as required for TFF3) in FFPE tissue from normal mucosa, LGD, HGD and ESCC (n=80). Samples are available from Addenbrooke's Hospital tissue bank for this purpose. We will then apply the most promising targets to a larger cohort of ESCC samples (n=270) which will be obtained as part of collaborative bio-resource from Dr Dawsey, NIH Washington, USA. Strict automated protocols will be used for immunohistochemical staining and scoring to ensure applicability to routine clinical NHS laboratories. Markers solely expressed in the stromal compartment will be discarded since the Cytosponge samples only the epithelium. Biomarkers with promising protein expression will then be applied to a cohort of Cytosponge specimens (N=500) collected as part of the BEST study from patients without any declared squamous cell cancer to ensure that the antibody is sufficiently specific (low false positives). Markers with a specificity of >90% (from Cytosponge samples) and sensitivity of >80% (from the retrospective cohort study) for detecting ESCC or HGD compared to the normal oesophagus will be taken forward to the prospective clinical study. For aim 2 the field testing of the identified and validated protein biomarker(s) will be performed as a research component added on to an ongoing endoscopic screening program in a high-risk population in China. Dr. Wang is the Principal Investigator of the Chinese National Early Detection and Early Treatment of Esophageal and Cardia Cancer (EDETEC) Program, which now screens over 50,000 asymptomatic adults in high-risk areas of China every year. The ongoing activities in this program, supported by the Chinese government, include a short questionnaire (patient identification and demographics), a Lugol’s chromoendoscopy examination, pathology processing and reading of the biopsies, and endoscopic therapy of all lesions diagnosed histopathologically as severe squamous dysplasia or superficial invasive esophageal squamous cell carcinoma (SESCC). The Cytosponge test will be performed prior to the screening endoscopy. Processing and immunohistochemical staining of the Cytosponge cell block and selected endoscopic biopsies will be performed utilising antibodies to the protein biomarkers identified in the validation step(s) described above. The screening characteristics (sensitivity, specificity, positive predictive value, and negative predictive value) of the protein biomarker(s) and histopathological assessment of the H&E from the Cytosponge specimen will be evaluated versus the gold standard of the worst endoscopic biopsy diagnosis in each patient. From these 500 individuals, we expect approximately 75 low-grade (mild) dysplasia (LGD), 75 HGD, and 5 ESCC patients, based on data from Dr Dawsey obtained from previous screening studies in China (Pan et al., Acta cytol, 2008; Dawsey et al., Acta Cytol, 1994). This number of individuals will provide good estimates of sensitivity and specificity for use of the putative biomarker(s) as a primary screening test in a realistic population setting in a cost effective manner. If the diagnostic accuracy of the test appears promising, additional larger field studies will be undertaken (with separate funding from NIHR or HTA) to further evaluate its clinical utility in the UK as part of a screening algorithm to detect precursors for both squamous and adenocarcinoma by applying more than one biomarker assay. Such a successful primary screening test, combined with endoscopic confirmation and endoscopic therapy, could have a tremendous impact on the global morbidity and mortality of oesophageal cancer in the UK and globally which in 2008 accounted for 404,000 deaths worldwide. 2) Development of an in vivo molecular imaging to highlight areas of dysplasia for biopsy and endoscopic therapy in Barrett's oesophagus. The proof of concept ex vivo work performed in my laboratory has shown that the lectin Wheat Germ Agglutinin (WGA) has the potential to be used as a molecular imaging tool for Barrett's dysplasia (Nat Med 2012). However, further work is required to determine the optimal panel of lectins and to refine the endoscopic imaging technology in order to minimise autofluorescence. Furthermore, toxicity studies are required prior to in vivo use. Hence, the specific aims of this proposal are to perform laboratory and clinical studies outlined in aim i) which will enable the design of a future prospective trial (aim ii). Specifically the aims are: ia) Identification of further molecular imaging glycan targets from already available lectin array and gene expression array data. ib) Perform ex vivo validation and subsequent toxicity studies on lectins, alone and in combination, which have already been identified as interesting. ic) Adapt a novel multi-spectral imaging system developed by Professor Vasilis Ntziachristos for use with a standard endoscope for the detection of fluorophores in vivo that minimises interference from tissue autofluorescence. This will be a collaborative project and this has already been established through our mutual collaborator Professor Kevin Brindle. id) Perform a pilot molecular imaging study in patients immediately prior to oesophagectomy to establish feasibility and enable more detailed power calculations.ii) To establish a prospective in vivo molecular imaging trial in the BE surveillance population using one or more lectins to determine their sensitivity and specificity as molecular imaging tools. Experimental plans In brief, our previous lectin array data provided binding data for 78 lectins. We initially took 3 lectins which demonstrated an epithelial binding pattern forwards to histochemical studies. Additional lectins also showed differential binding in dysplasia and may also have potential as glycan probes. We therefore plan to confirm these findings on tissues obtained from Barrett’s metaplasia, dysplasia and cancer. While a single labelled lectin might be sufficient to detect dysplasia in general we expect a combination of biomarkers to give the highest sensitivity and specificity due to molecular heterogeneity of the tissue; however this requires that there is not significant competition between the lectins for binding glycan epitopes. Lectin histochemistry using lectins in combination and alone on sequential sections will therefore be undertaken on mucosal tissue from fresh oesophagectomy specimens (n=10) and the overall level of lectin binding (tagged with different fluorochromes) will be assessed using confocal microscopy prior to validation at a macroscopic level on ex vivo specimens using the IVIS 200 camera. For the whole biopsy study, fluorescently labelled lectins (Alexa 680), alone or in combination as determined from the studies above, will be applied to biopsies taken from 40 patients with varying degrees of dysplasia and binding determined using the IVIS 200 camera. The degree of lectin binding will be correlated with the degree of dysplasia from the histopathological assessment of the same biopsy. Lectins that show promise at this stage will be progressed to whole organ (at oesophagectomy) ex-vivo proof-of-principle studies to allow transition to in vivo studies. Some of the lectins are edible foodstuffs e.g. WGA. For other lectins in order to provide the necessary safety data to the MHRA toxicity studies will be performed. Increasing concentrations of lectin will be applied to an existing panel of oesophageal cell lines and the effects on morphology, proliferation, apoptosis and metabolic activity will be determined. In vivo toxicity studies would be performed in collaboration if required and Professor Duncan Jodrell, who heads the Pharmacology and Drug Development Group at the CRUK Cambridge Research Institute. Adaptation of a novel multi-spectral imaging system will be required for in vivo lectin imaging studies. Current endoscopes use cameras placed at the tip of the instrument, which limits performance (sensitivity, spectral acquisition). Professor Vasilis Ntziachristos (Director of the Institute of Biological and Medical Imaging in Munich) has developed a system that provides simultaneous colour and fluorescence images, for accurate co-registration, performs all necessary corrections by spectral decomposition, is configurable for any fluorescence range, including near-infrared and can be interfaced to a conventional clinical endoscope. Real time spectral decomposition of the fluorochrome signal is used to separate out signal from tissue auto-fluorescence and to distinguish individual fluorophores with high sensitivity and specificity. The system would be tested first on ex vivo oesophagectomy specimens. Calibration experiments would include: (i) The fluorescence of fresh (BE) biopsies (n=30, range of disease stages) following lectin labelling will be assessed using the multispectral endoscope and values compared with those measured using the IVIS 200 and using a standard endoscope. Application of the fluorochrome alone, in the absence of the lectin, will serve as a control for background fluorescence. The signal-to-background ratios for lectin binding, as measured by each of the 3 techniques, will be determined. In the oesophagectomy model biopsies will be taken from areas of low and high fluorescence (determined using the multi-spectral endoscope) and these will be sent for histological analysis to determine the presence of dysplasia. Imaging in patients immediately prior to oesophagectomy will then be performed to establish feasibility and to generate pilot data to inform design of a larger clinical study. We will recruit BE patients (n=20) undergoing oesophagectomy. In theatre, prior to oesophagectomy and while under general anaesthetic, fluorescein-conjugated WGA (5µg/ml) +/_ other lectins alone or in combination will be applied to the mucosa for 10 minutes prior to washing with lectin buffer. Binding will be determined using a standard endoscope (Lucera GIF-Q260) as well as with the multispectral endoscope. Oesophageal biopsies will be taken from areas with the highest and lowest fluorescence intensities. The areas from which these biopsies are taken will be marked on a grid to allow for subsequent correlation with the histological mapping of the oesophagus and fluorescence measured (with subtraction of signal from auto-fluorescence) using the IVIS 200 camera prior to fixation, paraffin embedding, sectioning, H+E staining and histopathological assessment. Glucosamine (1g in 10ml), which competes with WGA for binding to GlcNAc, will then be applied to remove WGA binding and a further measurement of mucosal fluorescence taken. (If other lectins are used then the specific monosaccharide which competes with their binding would be used). The oesophagus will then be photographed, fixed and sectioned into mega-blocks to allow accurate histological mapping of the entire BE mucosa. The primary endpoint will be detection of an increased incidence of dysplasia in biopsies taken from areas of low WGA binding. Secondary endpoints will be assessment (from histological mapping) of whether all areas of dysplasia were detected. Following the pilot and during the duration of the Research Professorship, we will apply for funding for a multicentre molecular imaging study in BE surveillance. If multiple lectins complete this pilot phase, we will take forward the lectin with the greatest signal-to-background ratio, or, depending on the results of our preliminary work, a group of lectins that can be applied in combination. The primary endpoint will be to ascertain whether targeting biopsies in this manneris associated with a higher rate of dysplasia detection than the standard protocol and to determine the sensitivity and specificity of the targeted technique. The sample size and duration would be informed by the studies described above. This trial would enable a clinical evaluation of the molecular imaging technique, with the goal if successful, of introducing it into clinical endoscopic surveillance programmes for high risk patients with Barrett's oesophagus. LaySummary>