of
The Association of Coloproctology of Great Britain and Ireland
Advancing the cure and treatment of bowel disease /
Research Project Annual Report
Principal investigator / Professor R ArasaradnamCo-I : Mr B Singh, Mr C Harmston & Dr J Covington / Institution / University Hospital Coventry & Warwickshire/ University of Warwick
Project Title / Streamlining colonic imaging in those with suspicion of bowel cancer (through TWW pathway) with testing of urinary volatile organic compounds.
Start date / 1st February 2015 / Finish date / 30th January 2017
Lay Summary
(max 500 words) / The two week wait (TWW) pathway has been in place for over a decade with the intention to diagnose bowel conditions (mainly cancer) at an earlier stage. Whilst this has been helpful, a large proportion of patients undergo invasive testing (colonoscopy) which is often normal. What is required is a simple non invasive (e.g. urine testing) to help prioritise those that indeed have serious disease and thus need to undergo an urgent colonoscopy. Of note only ~ 6% of patients attending the TWW clinic actually have bowel cancer.
In our previous work, we have shown that urine testing (through smell detection) of volatile compounds/ gases in urine can separate those with bowel cancer and even inflammatory bowel disease with 85% sensitivity compared with healthy controls. We thus propose to incorporate urine testing as an added measure for those attending the TWW clinic.
We hope to demonstrate that detection of certain volatile compounds/ gases in urine, will help determine those at higher risk of bowel disease. Thus it is this subgroup that should be prioritised to receive an urgent colonoscopy. This will translate to earlier treatment/surgery and improved patient outcomes.
Background (purpose for project) / Our concept is to use an “Electronic Nose”, an instrument that artificially mimics the biological olfactory system. These instruments employ an array of different broadly tuned, gas phase, chemical sensors. This broad tuning is to a range of different gases and vapours, but importantly no two sensors are tuned to the same chemical groups. When a sample is presented to the e-nose, each sensor responds to the complex mix of chemicals within that sample. As each sensor is different their responses will be unique within the array. By taking the responses as a whole, it is possible to create a chemical ‘fingerprint’ of a smell – or as proposed here colorectal disease. Thus, when the e-nose is presented with the same sample it is able to recognise the smell by its fingerprint. It is unique in that, like a human nose, it identifies odours through their multi-component nature. Hence, it can identify similarities and differences between samples, without the need for identifying the individual components.
The strength of the electronic nose is that it is able to identify overall changes in profile, in real-time and in a way that is easily understood. The electronic nose attempts to map a new smell profile (in our case a urine sample) onto a learnt database, thus giving a user a classification and a confidence in that classification. Such instruments have the additional advantages of small form factor, low cost, ease of use and low maintenance compared with traditional gas analysis instrumentation.
The main aim of this large study is to determine if we could triage those when seen at the TWW clinic using a urine test to risk stratify those most likely to have colon cancer. High-risk patients could be fast tracked to have colonic imaging performed within one week. This would enable more rapid diagnosis and earlier staging and definitive treatment in those with colon cancer. It is possible that this will lead to better cancer outcomes and improve the patient journey.
Introduction / The premise of our work is early detection of disease using novel, non-invasive technology that can eventually by used at point of care. Here we have employed a number of e-nose instruments (designed to mimic the human olfactory system) for the detection of colon cancer. We are also comparing these results with other stool based markers (funded separately).
Methods / This was a prospective cohort interventional study. The intervention being analysis of urinary VOCs. All patients referred via the two week wait pathway with a suspicion of bowel cancer in whom colonic imaging is deemed appropriate was recruited.
Results and discussion / Goals Achieved:
This study commenced in February 2015. It is a two centre study between University Hospitals Coventry & Warwickshire (UHCW) as well as University Hospitals Leicester (UHL).
We have recruited 846 at UHCW and 238 at UHL – total of 1084 patients.
Early Results:
1. For the main study, we have shown that urinary volatile organic compound detection provides a sensitivity of 0.69 (95% CI0.39, 0.91) and specificity of 0.69 (95% CI0.65, 0.73) with area under the curve of: 0.80 (95% CI0.69, 0.90) to separate colon cancer from non-cancer. The receiver operator curve with standard deviation (in grey) is shown below.
Whilst the sensitivity and specificity is lower than our pilot study, this is the largest study to date evaluating urine markers for the detection of colon cancer. Further evaluation of the data and chemical interpretation is underway to seek optimisation.
2. Stool marker sub-study – this was externally funded on the back of this study utilising faecal haemoglobin testing (FIT) and faecal calprotectin (FCP). 430 patients with completed colonic investigations were analysed using Faecal immunochemical test for faecal haemoglobin.
The negative predictive value (NPV) using faecal haemoglobin alone or with both markers (faecal haemoglobin and faecal calprotectin) in combination was similar at 99% for colorectal cancer, with a sensitivity and specificity of 84% and 93% respectively. Faecal haemoglobin measurements were significantly higher in left sided colonic lesions compared with the right side; 713 vs 94 ; p = 0.0203). For adenomas, the NPV using faecal haemoglobin alone or with both markers (faecal haemoglobin and faecal calprotectin) was similar at 94% with a sensitivity and specificity of 69% and 56% respectively. (Manuscript in press; APT 2016).
Conclusion / 1. Stool Marker Tests - Undetectable faecal haemoglobin alone is sufficiently sensitive to exclude colorectal cancer, with higher values in left sided lesions. Faecal calprotectin in combination does not appear to provide additional diagnostic information.
2. Urine Volatile organic compound detection provides a sensitivity of 0.69 (95% CI0.39, 0.91) and specificity of 0.69 (95% CI0.65, 0.73) separating colon cancer from non-cancer.
Recommendations for future work / 1. Discussions with local CCGs in implementing stool marker (FIT) as a screening test prior to TWW referral to reduce burden. Further studies to determine the health economic benefits of implementing faecal immunochemical testing for haemoglobin in primary care are required.
2. Urinary VOCs – shows promise; next step is validation in multi-centre clinical trials.
References
(author, title, date of publication) / Title: Original Research Article: Diagnostic accuracy of faecal biomarkers in detecting colorectal cancer and adenoma in symptomatic patients. Widlak M et al.
Aliment Pharmacol Ther. 2017 Jan;45(2):354-363. doi: 10.1111/apt.13865