Use of ColonFlag score for prioritisation of endoscopy in colorectal cancer

Abstract

ObjectiveColorectal cancer (CRC) is the fourth most common cancer in UK. Symptomatic patients are referred via an urgent pathway and although most are investigated with colonoscopy <4% are diagnosed with cancer. There is therefore a need for a suitable triage tool to prioritise investigations. This study retrospectively examined performance of various triage tools in patients awaiting investigation on the urgent lower gastrointestinal cancer pathwayDesignAll patients over 40 years of age on the urgent pathway awaiting investigation for suspected CRC on 1 May were included. After 6 months, outcomes were evaluated and the performance of the faecal immunochemical test (FIT), faecal haemoglobin concentration, age and sex test (FAST) and the artificial intelligence algorithm ColonFlag were examined.Results532 completed investigations and received a diagnosis; 15 had CRC. 388 had a valid FIT result, of whom 11 had CRC; FAST Score ≥4.5 had sensitivity of 72.7%, specificity of 80.6% and would have failed to detect three tumours. Faecal haemoglobin (f-Hb) at cut-off of 10 µg/g and ColonFlag had equal sensitivity of 81.82%, ColonFlag had greater specificity 73.47%, compared with 64.99%. Both tests would have failed to detect two tumours but not in the same patients; when used in combination, sensitivity and specificity were 100% and 49.4%. When ColonFlag was applied to the cohort of 532, an additional four tumours would have been detected in patients without a valid FIT.ConclusionThis study showed ColonFlag to have equal sensitivity and greater specificity than f-Hb at a cut-off of 10 µg/g as a triage tool for CRC

Faecal immunochemical testing (FIT) in patients with signs or symptoms of suspected colorectal cancer (CRC): a joint guideline from the Association of Coloproctology of Great Britain and Ireland (ACPGBI) and the British Society of Gastroenterology (BSG)

“…Specificity was poor at 50% with ColonFlag but 81% with FAST score and NPVs of 99% and 100% for FAST score and ColonFlag, respectively. 52 Using a lower of two proposed cut-offs for ColonFlag, CRC accuracy resulted in a sensitivity of 80%, specificity of 48% and NPV of 99% (sample size was limited to 21 cases). 106 ColonPredict 107 which used a combination of symptoms, fHb, serum haemoglobin and mean cell volume was deemed superior to symptoms alone while ColonoFIT, 60 which uses three serial fHb measurements in a week, patient questionnaire, medication consumption (eg, NSAIDs) had a ninefold higher OR of detecting CRC than serial fHb on its own.…”

“…Modelling studies to date have not demonstrated the benefit of combining FIT with other clinical features and blood test results to enhance sensitivity by reducing false negative FITs. A comparison of FIT at ≥10 µg Hb/g faeces alone, with the FAST score (combining FIT age and sex), and ColonFlag (a machine learning algorithm using age, sex and FBC indices to derive a risk score), showed that FIT and ColonFlag missed a different 18% of CRCs, respectively, and FAST score missed 27.3% 52. Combining simple blood tests with FIT at best matches the sensitivity of FIT alone in patients tested in primary care, whether as pairs of results or within multivariable model 20…”

“…CRC accuracy was reported to provide sensitivity of 100% compared with 73% with FAST score. Specificity was poor at 50% with ColonFlag but 81% with FAST score and NPVs of 99% and 100% for FAST score and ColonFlag, respectively 52. Using a lower of two proposed cut-offs for ColonFlag, CRC accuracy resulted in a sensitivity of 80%, specificity of 48% and NPV of 99% (sample size was limited to 21 cases) 106…”

“… 28 It is not clear whether the use of different platforms affected these findings, as well as other intrinsic differences in the cohorts evaluated. The ColonFlag score 52 takes this approach a step further, combining FBC results with fHb, age and sex, and showed improved specificity compared with FIT—taking an AND/OR approach sensitivity improved to 100%. However, this study is relatively small and needs further validation.…”

Artificial intelligence and machine learning for early detection and diagnosis of colorectal cancer in sub-Saharan Africa

“…Presymptomatic longitudinal CBC patterns may be imperceptible to clinicians but would be readily detectable by statistical algorithms or ‘prediction models,’ often referred to as Multianalyte Assays with Algorithmic Analysis (MAAA) 32. Currently, proprietary MAAA exist that were built and validated in high-income countries; these MAAA use CBC and demographic data to identify patients at high risk of CRC 33–36. Similarly, we have developed a MAAA prediction model in a US cohort using longitudinal and single timepoint laboratory studies and patient characteristics (accepted to Digestive Disease Week 2022).…”