Diagnostic accuracy of faecal biomarkers in detecting colorectal cancer and adenoma in symptomatic patients

Transcription

1 Alimentary Pharmacology and Therapeutics Diagnostic accuracy of faecal biomarkers in detecting colorectal cancer and adenoma in symptomatic patients M. M. Widlak*,, C. L. Thomas, M. G. Thomas, C. Tomkins, S. Smith,N.O Connell*, S. Wurie*, L. Burns*, C. Harmston**, C. Evans**, C. U. Nwokolo*, B. Singh & R. P. Arasaradnam*,, *Department of Gastroenterology, University Hospitals Coventry and Warwickshire, Coventry, UK. Medical School, University of Warwick, Coventry, UK. Department of Biochemistry, University Hospitals Coventry and Warwickshire, Coventry, UK. Medical School, University of Oxford, Oxford, UK. Midlands and North West Bowel Cancer Screening Hub, University Hospitals Coventry and Warwickshire, Coventry, UK. **Department of Colorectal Surgery, University Hospitals Coventry and Warwickshire, Coventry, UK. Department of Colorectal Surgery, University Hospitals of Leicester, Leicester, UK. Applied Biological and Experimental Sciences, University of Coventry, Coventry, UK. Correspondence to: Prof. R. Arasaradnam, Clifford Bridge Road, Coventry CV2 2DX, UK. Publication data Submitted 18 July 2016 First decision 15 August 2016 Resubmitted 5 October 2016 Resubmitted 26 October 2016 Accepted 26 October 2016 EV Pub Online 1 December 2016 The Handling Editor for this article was Professor Roy Pounder, and it was accepted for publication after full peerreview. SUMMARY Background The diagnosis of colorectal cancer (CRC) can be difficult as symptoms are variable with poor specificity. Thus, there is a quest for simple, non-invasive testing that can help streamline those with significant colonic pathology. Aim To assess using faecal immunochemical test for haemoglobin (FIT) or faecal calprotectin (FCP) to detect CRC and adenoma in symptomatic patients referred from primary care. Methods A total of 799 referred for urgent lower gastrointestinal investigations were prospectively recruited. Of these, 430 completed colonic investigations and returned stool samples, and were included in the final statistical analysis. Faecal immunochemical test for haemoglobin was performed on HM- JACKarc analyser (Kyowa Medex, Tokyo, Japan), and FCP by the EliA Calprotectin immunoassay (Thermo Fisher Scientific, Waltham, United States). Results The negative predictive value (NPV) using FIT alone or both markers (FIT and FCP) in combination was similar at 99% for CRC, with a sensitivity and specificity of 84% and 93%, respectively. FIT measurements were significantly higher in left-sided colonic lesions compared with the right side; 713 vs. 94; P = ). For adenoma, the NPV using FIT alone, or both markers (FIT and FCP) in combination, was similar at 94% with a sensitivity and specificity of 69% and 56%, respectively. Conclusions Undetectable faecal immunochemical test for haemoglobin is sufficiently sensitive to exclude colorectal cancer, with higher values in left-sided lesions. FCP in combination does not appear to provide additional diagnostic information. Further studies to determine the health economic benefits of implementing faecal immunochemical test for haemoglobin in primary care are required. Aliment Pharmacol Ther 2017; 45: doi: /apt.13865

2 Faecal biomarkers for detecting colorectal cancer and adenoma INTRODUCTION Colorectal cancer (CRC) is the fourth most common cancer in the UK, and the second leading cause of cancer-related deaths. Bowel cancer risk is strongly associated with age, with 95% of cases occurring in people aged 50 and over. 1 The most common presenting symptoms of bowel cancer include a persistent change in bowel habit, rectal bleeding and abdominal pain. 1, 2 Most patients with symptoms have benign pathology but in an effort not to miss cancer or precancerous lesions, almost all patients undergo invasive colonic investigations. 3 In 2000, the UK Department of Health introduced the 2-week wait (TWW) referral pathway to streamline symptomatic patients suspected of cancer. Patients with red flag symptoms such as rectal bleeding, weight loss, change in bowel habit, etc., presenting to their primary physician are required to be seen by a specialist within 14 days see Appendix 1. This pathway has been subsequently updated in 2011 and 2015, and was developed with the aim of identifying up to 90% of patients with CRC. However, as the symptoms used to guide referrals are common to many other gastrointestinal complaints, many of the referrals inevitably have a normal outcome. A meta-analysis by Jellema et al. 4 showed that of the red flag symptoms, only weight loss showed a high specificity (89%) for detection of CRC but with a low sensitivity (20%). Other symptoms, such as change in bowel habit and abdominal pain, were shown to be poor predictors of serious underlying pathology. Furthermore, a review of the effectiveness of the pathway by Thorne et al. 5 found that only 10.3% of patients referred via the pathway had a final diagnosis of CRC (based on the assumption that all referrals were appropriate), and that less than a quarter of all CRC patients had been identified using the TWW pathway. The National Institute for Health and Care Excellence (NICE) revised the referral guidance for suspected CRC in 2015 (NG12). 6 In view of the difficulty in assessing patients on symptoms alone, there is a new recommendation to use faecal occult blood (FOB) testing to help aid referral decisions. This is controversial because at present the test method available, if at all, for faecal occult blood is the guaiac method. Laboratories in the UK have withdrawn this test from their repertoire because of poor specificity for CRC in symptomatic patients. Guaiac FOB test is currently being used in the Bowel Cancer Screening Programme in England for asymptomatic individuals but will soon be replaced with faecal immunochemical test for haemoglobin (FIT) which has improved sensitivity and specificity. 7, 8 Faecal calprotectin (FCP) has good evidence for detecting inflammatory bowel disease (IBD) 9, 10 but its value in CRC and adenoma in the symptomatic population is less well studied. 11 The combination of FIT and FCP may improve the sensitivity and specificity of detection of CRC and adenoma compared to using the tests individually, due to the tendency of polyps bleeding into the bowel lumen and the inflammatory environment caused by the migration of neutrophils into the neoplastic tissue associated with CRC. Therefore, measuring both markers will potentially improve specificity by ruling out those who have other causes of bleeding not associated with active inflammation, that is, haemorrhoids, and ruling out those with other forms of inflammation not associated with bleeding, such as that seen after infection. Mowat et al. 12 recently reported a study using the combination of FIT and FCP in a Scottish population, but using a single OC-Sensor io analyser (EIKEN Chemical Co., Tokyo, Japan) and quantitative ELISA Calprotectin EK-CAL (B UHLMANN Laboratories, Basel, Switzerland) technologies respectively. It is not known therefore whether their results are transferable to any other test providers for calprotectin and FIT, because results from different manufacturers differ particularly for FCP. There remains a paucity of studies on the performance of FIT and FCP in combination for symptomatic patients. In view of this, and recent NICE guidance, which did not detail which faecal occult blood test should be used to aid patient referral, we aimed to assess the diagnostic accuracy of faecal biomarkers, FIT and FCP, to detect CRC and adenoma in patients referred from primary care presenting with lower gastrointestinal symptoms. To our knowledge, this is the largest study undertaken in an English population. METHODS Study design and study population Single-centre, prospective, blinded study in symptomatic patients referred with suspected CRC through the national TWW pathway Appendix 1. Ethical approval was granted by the Coventry and Warwickshire Research Ethics Committee as part of the FAMISHED (Food and Fermentation using Metagenomics in Health and Disease) study 09/H1211/38. Aliment Pharmacol Ther 2017; 45:

3 M. M. Widlak et al. Participants and sample collection Study period was between January 2015 and March All the referrals were seen in colorectal and dedicated gastroenterology clinics at University Hospitals Coventry and Warwickshire (UHCW) National Health Service (NHS) Trust. A total of 2822 patients were referred during the period of the study. Of these, 174 did not attend or cancelled their appointments and 1284 were seen out of hours or at weekends limiting the numbers that we could approach to participate in the study. Of the remaining 1364 patients, 351 declined to participate in the study, 166 were deemed physically unfit for further investigation by the clinician and 48 were unable to provide valid consent due to a language barrier, visual impairment or illness (i.e. dementia). A total of 799 patients were consented and recruited. Of these, 311 did not provide stool samples; 62% return rate. A further 58 patients were excluded from the final analysis (Figure 1). A total of 430 patients who met the inclusion criteria capacity to consent, complete colonic investigations (colonoscopy or radiological imaging), as well as returned both stool samples (FIT and FCP) prior to investigations were included in the final statistical analysis. Those under the age of 18, pregnant, not meeting the TWW pathway criteria or having limited colonic examinations were not eligible to participate in the study. Study participants were given a pack including a Universal Sterilin 30 ml stool container, a FIT device along with written and pictorial instructions for collections. Stool collection was aided either by a standard stool collecting device or using a Fe Col faecal collecting device (Alpha Laboratories Ltd., Eastleigh, UK), depending on patients preferences. FIT and FCP were measured in all patients who returned stool samples and results were recorded by a dedicated biochemist. Patients referred to endoscopy or radiology were investigated within 2 weeks of clinic assessment. The decision for investigations, that is, colonoscopy or radiological imaging was at the discretion of the clinician following a discussion with the patient. Participating clinicians, endoscopists and radiologists were blinded to the faecal test results. Diagnosis of CRC, adenoma, IBD and microscopic colitis were confirmed histologically. Measurement of FIT and FCP concentrations Faecal immunochemical test for haemoglobin was performed on the HM-JACKarc analyser (Kyowa Medex, Tokyo, Japan), located at the Midlands and North West Bowel Cancer Screening Hub, Rugby, on a weekly basis. FCP samples were analysed daily alongside the routine calprotectin service at UHCW by the EliA Calprotectin immunoassay (Thermo Fisher Scientific, Waltham, MA, USA). The laboratories at UHCW and the Bowel Cancer Screening Hub are the UKAS (The United Kingdom Accredication Service) accredited and also take part in the UK NEQAS (The United Kingdom National External Quality Assessment Service) external quality assessment schemes. FCP was considered positive at >50 lg/g faeces, which is the established cut-off for current IBD testing and has been used by Mowat et al. 12 for CRC. For FIT, any detectable result (detection limit 7 ug Hb/g faeces) was considered positive. FIT and FCP results were compared to the outcome of their colonic investigations. To correlate the biomarker concentrations with the diagnosis, patients were divided into clinical groups: CRC, adenoma, IBD, microscopic colitis, diverticular disease, haemorrhoids and normal findings. Statistical analysis All analysis was carried out using Matlab R2011b (Math Works, Natick, MA, USA). Groups were compared using the Mann Whitney U-test. Results were considered significant if P < 0.05 with Bonferroni correction applied to correct for multiple comparisons. RESULTS A total of 430 patients were included in the final analysis (median age 67 years; range 29 93; IQR 57 76). There was an equal proportion of both sexes participating with no significant differences between the proportion of men and women returning faecal samples. The most common presenting symptoms alone or in combination with other symptoms were altered bowel habit (64%), rectal bleeding (43%), abdominal pain (30%), anaemia (17%) and weight loss (16%). Twenty-two per cent of patients had a family history of CRC (Table 1). The most common findings at colonic investigations were: normal colon in 157 patients, diverticular disease in 112, diverticular disease in combination with other pathology, that is, haemorrhoids or microscopic colitis in 22, adenoma with low grade dysplasia in 42, adenoma with low grade dysplasia in combination with other pathology in 28, adenoma with high grade dysplasia (HGD) in 1 and CRC in 24. IBD was diagnosed in 7 patients, microscopic colitis alone in 10 and haemorrhoids alone were noted in Aliment Pharmacol Ther 2017; 45:

4 Faecal biomarkers for detecting colorectal cancer and adenoma Total referred n = 2822 Did not attend/cancelled; n = 174 Weekend/evening clinics; n = 1284 Inclusion criteria: Referral as per national two week wait guidance bowel symptoms suggestive of colorectal cancer. Declined; n = 351 Unsuitable/unfit; n = 166 No consent/language barrier; n = 48 Approached n = 1364 Recruited n = 799 Adults. Colonic investigations Colonoscopy or CT colonography or CT abdomen/pelvis with contrast plus flexible sigmoidoscopy. (CT = computed tomography). Capacity to consent. Able to provide stool sample faecal immunochemical test for haemoglobin (FIT) and faecal calprotectin (FCP) prior to investigations. Did not provide stool sample; n = 311 Excluded; n = 58 Details of exclusion (n = 58): FIT sample not returned; n = 14 FCP sample not returned; n = 6 Incomplete colonic examination; n = 13 No investigations performed (patient cancelled procedure or did not attend); n = 5 FIT and FCP samples returned after colonoscopy thus invalid; n = 10 FIT sample unsuitable for analysis (insufficient sample, incorrect labelling, old sample); n = 3 Did not meet national two week criteria; n = 2 Extra colonic malignancy; n = 4 Diagnosis unclear (under investigations); n = 1 Final study population n = 430 Figure 1 Study flow diagram of total patients referred, approached and recruited having met the inclusions criteria over 15 months. Aliment Pharmacol Ther 2017; 45:

5 M. M. Widlak et al. Table 1 Baseline demographic data of study participants and presenting clinical symptoms Study participants 430 Median age in years (range; IQR) 67 (29 93; 57 76) Sex Female 220 (51%) Male 210 (49%) Symptoms Change in bowel habit 276 (64%) Rectal bleeding 185 (43%) Abdominal pain 129 (30%) Anaemia 74 (17%) Weight loss 68 (16%) Family history of colorectal cancer 95 (22%) Faecal calprotectin (µg/g) Normal Cancer Cancer plus other Adenoma Figures 2 and 3 show the box plot distribution of FIT and FCP for the various colonic pathologies identified. Tables S1 and S2 in Appendix S2, show performance matrix and statistical significance of FIT and FCP against the various colonic pathologies. Twenty-four patients with CRC and one patient with HGD adenoma were grouped together as cancer. FIT returned a sensitivity of 84% and specificity of 93% for Adenoma plus other Diverticular disease Diverticular disease plus other Inflammatory bowel disease Microscopic colitis Figure 2 Box plot (median with inter quartile ranges) of faecal calprotectin (FCP) results in normal, colorectal cancer, adenoma, diverticular disease, haemorrhoids, inflammatory bowel disease and microscopic colitis. Note logarithmic scale on y axis. Faecal haemoglobin (µg/g) Normal Cancer Cancer plus other Adenoma Figure 3 Box plot (median with inter quartile ranges) of faecal immunochemical test for haemoglobin (FIT) results in normal, colorectal cancer, adenoma, diverticular disease, haemorrhoids, inflammatory bowel disease and microscopic colitis. Note logarithmic scale on y axis. detection of CRC; area under the curve (AUC) of 0.94 (95% CI: ). Negative predictive value (NPV) of FIT was 99% (95% CI: %). FCP had sensitivity of 68% and specificity of 84% for detecting CRC; AUC of 0.82 (95% CI: ). NPV of FCP was 98% (95% CI: 96 99%). The combination of both, FIT and FCP, had no improvement over FIT alone; sensitivity remaining the same at 84% and specificity of 93% with AUC of 0.95 (95% CI: ). NPV for combined faecal biomarkers was 99% (95% CI: %) (Table 3 and Figure 4). For adenoma detection with no other pathology on colonoscopy or CT, FIT returned a sensitivity of 69% and specificity of 56%; AUC 0.70 (95% CI: ). NPV was 94% (95% CI: 92 96%) (Figure 5). FCP showed only a sensitivity of 43% and specificity of 56% with AUC 0.50 (95% CI: ); in essence no benefit compared to not testing. For detection of both cancer and adenoma, FIT again was superior with a sensitivity of 76% and specificity of 70%; AUC of 0.77 (95% CI: ) and FCP showing no benefit compared to not testing. Adenoma plus other Diverticular disease Diverticular disease plus other Inflammatory bowel disease Microscopic colitis 358 Aliment Pharmacol Ther 2017; 45:

6 Faecal biomarkers for detecting colorectal cancer and adenoma Table 2 Presenting symptoms, faecal immunochemical test for haemoglobin (FIT) and faecal calprotectin (FCP) concentrations in patients with colorectal cancer and high grade dysplasia adenoma No Sex Age Symptoms Histology Location FCP (lg/g) FIT (lg Hb/g) 1 F 88 Altered bowel habit, weight loss Adenocarcinoma Caecum F 76 Abdominal pain Adenocarcinoma Ascending F 81 Altered bowel habit Adenocarcinoma Ascending F 85 Normocytic anaemia Adenocarcinoma Ascending F 83 Iron deficiency anaemia Adenocarcinoma Ascending M 58 Altered bowel habit, weight loss Adenocarcinoma Hepatic flexure <15 <7 7 M 55 Abdominal pain, weight loss Adenocarcinoma Descending 226 <7 8 M 52 Abdominal pain Adenocarcinoma Descending F 79 Altered bowel habit, abdominal pain, anaemia, weight Adenocarcinoma Descending 28.3 <7 loss 10 F 77 Iron deficiency anaemia Adenocarcinoma Sigmoid 324 > M 84 Rectal bleed Adenocarcinoma Sigmoid F 85 Altered bowel habit, abdominal pain, rectal bleed Adenocarcinoma Sigmoid F 60 Altered bowel habit, abdominal pain, rectal bleed Adenocarcinoma Sigmoid F 82 Rectal bleed Adenocarcinoma Sigmoid 274 > M 41 Iron deficiency anaemia Adenocarcinoma Sigmoid 709 > F 55 Altered bowel habit, rectal bleed Adenocarcinoma Recto-sigmoid M 86 Altered bowel habit Adenocarcinoma Recto-sigmoid F 58 Altered bowel habit, rectal bleed Adenocarcinoma Recto-sigmoid M 47 Altered bowel habit, abdominal pain, rectal bleed Adenocarcinoma Recto-sigmoid 337 > M 88 Altered bowel habit, rectal bleed & mass, weight loss Adenocarcinoma Rectum < M 61 Rectal bleed & mass Adenocarcinoma Rectum M 83 Rectal bleed & mass Adenocarcinoma Rectum F 38 Altered bowel habit, abdominal pain, weight loss, iron Adenocarcinoma Rectum 644 >1000 deficiency anaemia 24 F 65 Rectal bleed & mass Adenocarcinoma Rectum < F 86 Altered bowel habit, rectal bleed HGD adenoma Rectum Upper limit of detection of faecal immunochemical test for haemoglobin (FIT) is 1000 lg Hb/g faeces with lower limit of 7 lg Hb/g faeces. In the cancer group, 19 lesions were left-sided. While this study was not powered to investigate differences in marker concentrations according to the location of the malignant lesion, FIT concentrations were significantly higher in left-sided lesions compared with right-sided lesions (713 lg Hb/g faeces vs. 94 lg Hb/g faeces; P = ). FCP concentrations were not significantly different in left vs. right colon; 143 lg/g vs. 175 lg/g respectively; P = (Table 2). For IBD detection, the sensitivity and specificity of FIT was 86% and 76% respectively compared with 86% sensitivity and 91% specificity for FCP. Respective NPVs were similar at 100% for both markers (95% CI: %). For microscopic colitis, both tests combined were also equally good returning a NPV of 99% (95% CI: %). Negative predictive values of FIT for CRC, adenoma and IBD at different cut-off levels are shown in Table 4. DISCUSSION This study evaluated the utility of measuring two faecal biomarkers, FIT and FCP, in patients referred from primary care in England (in accordance with the national referral guidelines) with symptoms suggestive of CRC. Apart from screening, there are several studies that have evaluated the diagnostic performance of FIT in symptomatic and high-risk populations, suggesting high sensitivity and specificity for CRC In fact, Cubiella et al. 13 showed that FIT is more accurate in detecting CRC than the NICE or the Scottish Intercollegiate Guidelines Network (SIGN) referral criteria. The study demonstrated that by a cut-off of 20 lg Hb/g faeces, 20% fewer colonoscopies would have been required to detect 42% more CRCs. In other studies investigating the use of FIT for detection of CRC in symptomatic patients, sensitivity has ranged from 67% to 100% and specificity from 71% to 93% dependent on the cut-off used. 8, 12, 17, 18 The results of these studies suggest that Aliment Pharmacol Ther 2017; 45:

7 M. M. Widlak et al. Table 3 Diagnostic accuracy performance of faecal immunochemical test for haemoglobin (FIT), faecal calprotectin (FCP) and both markers combined for main colonic pathology identified Outcome Test Number normal Number abnormal TPR FPR PPV NPV Sensitivity Specificity Cancer + HGD FIT % 99% 84% 93% Cancer + HGD FCP % 98% 68% 84% Cancer + HGD Both % 99% 84% 93% Adenoma FIT % 94% 69% 56% Adenoma FCP % 90% 43% 56% Adenoma Both % 94% 69% 56% Adenoma + FIT % 96% 61% 59% Adenoma + FCP % 95% 50% 62% Adenoma + Both % 96% 61% 59% Inflammatory Bowel Disease FIT % 100% 86% 76% Inflammatory Bowel Disease FCP % 100% 86% 91% Inflammatory Bowel Disease Both % 100% 86% 91% Diverticular disease FIT % 72% 47% 48% Diverticular disease FCP % 74% 54% 46% Diverticular disease Both % 72% 47% 48% Diverticular disease + FIT % 97% 64% 57% Diverticular disease + FCP % 97% 68% 51% Diverticular disease + Both % 97% 64% 57% Microscopic Colitis FIT % 99% 70% 59% Microscopic Colitis FCP % 99% 70% 51% Microscopic Colitis Both % 99% 70% 59% HGD = high grade dysplasia; Adenoma + and Diverticular disease + indicate other contemporary diagnosis, e.g. microscopic colitis or haemorrhoids, etc. Table 4 Negative predictive values (NPV) of faecal immunochemical test for haemoglobin (FIT) for colorectal cancer, adenoma and inflammatory bowel disease at different cut-off levels Normal vs. cancer FIT (lg Hb/g faeces) NPV colorectal cancer NPV adenoma 50 97% 72% 98% % 72% 98% % 71% 98% NPV inflammatory bowel disease the measurement of FIT concentrations has the potential to be a good rule-out test for clinically important colorectal disease and could therefore reduce the number of unnecessary referrals to endoscopy, easing the pressure on already over-subscribed services. McDonald et al. 17 in a population of symptomatic patients referred for endoscopy from primary care showed that patients with CRC, high-risk adenoma, lowrisk adenoma and IBD had significantly higher median FIT concentrations than those with less clinically important findings. Using a cut-off of 10 lg Hb/g faeces, the True positive rate FCP FIT Combined False positive rate Figure 4 Receiver operator curves (ROCs) for faecal immunochemical test for haemoglobin (FIT), faecal calprotectin (FCP) and both markers combined for colorectal cancer detection area under the curve (AUC) 0.94, 0.82 and 0.95 respectively. study achieved a sensitivity and negative predictive value for CRC of 100%. While specificity range was much lower at only 48 59% for high-risk adenoma, low-risk 360 Aliment Pharmacol Ther 2017; 45:

8 Faecal biomarkers for detecting colorectal cancer and adenoma True positive rate Normal vs. adenoma False positive rate FCP FIT 1 Figure 5 Receiver operator curves (ROCs) for faecal immunochemical test for haemoglobin (FIT) and faecal calprotectin (FCP) for adenoma detection area under the curve (AUC) 0.70 and 0.50 respectively. adenoma and IBD, the negative predictive values for these groups were each 94% respectively. It is however important to note that the sample size for this study was small (n = 280) and therefore the number of CRC cases detected was low (n = 6), potentially limiting the significance of the conclusions. Mowat et al. 12 have shown similar findings in a Scottish population but with important differences. Our study protocol was based within secondary care in England, following referral from primary care utilising a nationally agreed protocol. Importantly, we have drawn similar conclusions but used different analytical and manufacturer protocols, which, suggests that these marker measurements could potentially be transferable to different test providers. Two-thirds of this cohort had nonsignificant colonic findings normal bowel (157 patients; 36.5%), and nonsignificant pathology such as haemorrhoids and diverticular disease (161 patients; 37.4%). Using FIT in this cohort would potentially have avoided a referral and subsequent invasive investigations in 74% patients without serious pathology. Such measures could potentially ease pressures on already overburdened colonoscopy and radiology services. However, the ability to identify those who do have significant bowel pathology is more important in terms of patient outcomes than limiting inappropriate referrals. In this context, the negative predictive values of faecal biomarkers, particularly FIT alone (NPV of 99%), showed that it has significant potential as a rule-out test for CRC. Three cases of CRC would be potentially missed using FIT alone (hepatic flexure and descending colon), and in one case FCP was also undetectable, demonstrating that this test cannot wholly replace clinical judgement. Thus, if CRC is still suspected in a patient with a negative FIT and symptoms persist, consider repeating the test using both biomarkers (FIT and FCP) and/or refer for urgent colonic investigations via the TWW colorectal pathway. This study was not powered to investigate differences in FIT concentrations between cancers at different stages though Digby et al. 16 reported higher median FIT concentrations in more advanced tumours. Our data suggest that the concentration of FIT may be related to the location of the cancer in the colon than stage; with left-sided cancers having much higher concentrations than rightsided ones. This could reflect degradation of haemoglobin in the colon for more proximal lesions as blood from lesions in the right side of the colon are in transit for longer periods. Using detectable FIT the positive predictive value to detect CRC was 44% (95% CI: 39 49%) and 21% (95% CI: 17 25%) for FCP. Conversely, the negative predictive value for CRC, adenoma and even IBD was high even at different cut-off levels of FIT (Table 4). This indicates that these stool tests are not appropriate for use as a rule-in test for these conditions but rather as a rule-out. This interpretation is in agreement with previous studies utilising FIT detection in symptomatic 13, 16, 19 populations. FIT did not perform as well against detection of adenoma with a sensitivity of 69% and specificity of 56%; AUC The negative predictive value was 94%. Thus, in using this as a faecal marker alone, one could argue that potential pre-malignant lesions could be missed. However, the National Bowel Cancer Screening Programme has already begun planning the switch from guaiac-based FOB testing to FIT technology, hence the use of FIT in the symptomatic population will provide a more uniform approach to patients who may harbour pre-malignant adenomas and/or who do not seek medical advice about their symptoms. Moreover, the National Health Service bowel scope screening programme will act as a further safety net to identify those who may harbour pre-malignant adenomas not detected using FIT. In summary, the diagnostic accuracy of FIT and FCP indicate that undetectable FIT is a good rule-out test for CRC, with a negative predictive value of 99%. It is also far more accurate than the current referral Aliment Pharmacol Ther 2017; 45:

9 M. M. Widlak et al. pathway based on symptoms alone. Adding FCP testing alongside FIT showed no additional benefit for the added cost of analysing two tests. This is in agreement with the results of Mowat et al., 12 and confirms that the results are transferable irrespective of population or method of testing. Both FIT and FCP also offer a good rule-out option for inflammatory bowel disaese and microscopic colitis; negative predictive value of 100% and 99% respectively. This means that treatable (benign) symptomatic colonic pathology will not be missed either. Further studies are required to demonstrate the health economic benefits of introducing this test and potential cost savings through reduction in referrals and number of colonoscopic and radiological procedures undertaken. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1. The Arden Cancer Network Two Week Referral Services - GI - Lower. Appendix S2. Supplementary Tables S1 and S2. AUTHORSHIP Guarantor of article: R P Arasaradnam. Specific author contributions: RPA initiated the project, led the gaining of support, grant funding and ethic approval. MMW recruited patients, performed the samples handling and documentation, analysed data, wrote and reviewed the manuscript. RPA, CE, CH, CUN and BS reviewed the manuscript for intellectual content. NO, SW, LB, BL, JC created the sample collection kits, recruited patients and completed documentation. MGT performed the statistical analysis. CT and CLT performed biochemical analysis and were involved in the study design. SS supervised the biochemical analysis. All authors approved the final version of the manuscript. ACKNOWLEDGEMENTS Declaration of personal interests: All the patients attending UHCW NHS Trust for their altruistic contributions. Matthew Davies at Alpha Laboratories Ltd. for provision of the reagents and FIT devices. The NHS Bowel Cancer Screening Hub at Rugby for allowing us to use their HM-JACKarc analyser and Angela Ryder for analysing the FIT devices. The Department of Biochemistry and the Tissue Bank at UHCW NHS Trust for their help with collection and storage of the stool samples. The Surgical Department, the Outpatient Department and all the Colorectal Surgeons at UHCW NHS Trust for their help in facilitating the recruitment of patients for this study. RPA and CT have provided educational lectures on behalf of Thermo Fisher Scientific Ltd. which provide FCP analysis. Declaration of funding interests: Bowel Disease Research Foundation, UK; Thermo Fisher Scientific Ltd.; Alpha Laboratories Ltd. REFERENCES 1. Cancer Research UK. Bowel Cancer Statistics. Available at: (accessed September 2015). 2. Hewitson P, Woodrow C, Austoker J (NHS Bowel Cancer Screening Programme). Evidence Summary: Patient Information. November Available at: publications/nhsbcsp04.pdf (accessed September 2015). 3. Pande R, Leung E, McCullough P, Smith S, Harmston C. Impact of the United kingdom national bowel cancer awareness campaign on colorectal services. Dis Colon Rectum 2014; 57(1): Jellema P, van der Windt DA, Bruinvels DJ, et al. Value of symptoms and additional diagnostic tests for colorectal cancer in primary care: systematic review and meta-analysis. BMJ 2010; 340: c Thorne K, Hutchings HA, Elwyn G. The effects of the Two-Week Rule on NHS colorectal cancer diagnostic services: a systematic literature review. BMC Health Serv Res. 2006; 6: National Institute for Health and Care Excellence (NICE). Suspected Cancer: Recognition and Referral (NG12). June 2015, London. Available at: (accessed October 2015). 7. Fraser C, Matthew C, Mowat N, Wilson J, Carey F, Steel R. Immunochemical testing in individuals positive for guaiac faecl occult blood test in a screening programme for colorecatal cancer: an observational study. Lancet Oncol 2006; 7: Hol L, van Leerdam ME, van Ballegooijen M, et al. Screening for colorectal cancer: randomised trial comparing guaiac-based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut 2010; 59: Dhaliwal A, Zeino Z, Tomkins C, et al. Utility of faecal calprotectin in inflammatory bowel disease (IBD): what cut-offs should we apply? Frontline Gastroenterol 2015; 6: McFarlane M, Chambers S, Malik A, et al. Clinical outcomes at 12 months and risk of inflammatory bowel disease in patients with an intermediate raised fecal calprotectin: a real-world view. BMJ Open 2016; 6: e Tibble J, Sigthorsson G, Foster R, Sherwood R, Fagerhol M, Bjarnason I. Faecal calprotectin and faecal occult blood tests in the diagnosis of colorectal carcinoma and adenoma. Gut 2001; 49: Mowat C, Digby J, Strachan J, et al. Faecal haemoglobin and faecal calprotectin as indicators of bowel disease in patients presenting to primary care with bowel symptoms. Gut 2016; 65: Cubiella J, Salve M, Diaz-Ondina M, et al. Diagnostic accuracy of the faecal immunochemical test for colorectal cancer in symptomatic patients: comparison with NICE and SIGN referral criteria. Colorectal Dis 2014; 16: O Rozen P, Levi Z, Hazazi R, et al. Quantitative colonoscopic evaluation of relative efficiencies of an immunochemical faecal occult blood test and a sensitive guaiac test for detecting significant colorectal neoplasms. Aliment Pharmacol Ther 2009; 29: Rozen P, Levi Z, Hazazi R, et al. Identification of colorectal adenomas by a quantitative immunochemical faecal occult blood screening test depends on 362 Aliment Pharmacol Ther 2017; 45:

10 Faecal biomarkers for detecting colorectal cancer and adenoma adenoma characteristics, development threshold used and number of tests performed. Aliment Pharmacol Ther 2009; 29: Digby J, Fraser CG, Carey FA, et al. Faecal haemoglobin concentration is related to severity of colorectal neoplasia. J Clin Pathol 2013; 66: McDonald PJ, Digby J, Innes C, et al. Low faecal haemoglobin concentration potentially rules out significant colorectal disease. Colorectal Dis 2013; 15: e Hazazi R, Rozen O, Leshno M, et al. Can patients at high risk for significant colorectal neoplasms and having normal quantitative faecal occult blood test postpone elective colonoscopy? Aliment Phramacol Ther 2010; 31: Whitehead S, French J, Brookes M, Ford C, Gama R. Between assay variability of faecal calprotectin enzyme-linked immunosorbent assay kits. AnnClinBiochem2013; 50: Aliment Pharmacol Ther 2017; 45: