RESUMEN
OBJECTIVES: Pressured healthcare resources make risk stratification and patient prioritisation fundamental issues for the investigation of colorectal cancer (CRC) in symptomatic patients. The present study uses machine learning algorithms and decision strategies to improve the appropriate use of colonoscopy. DESIGN: All symptomatic patients in a single health board (2018-2021) proceeding to colonoscopy to investigate for CRC were included. Machine learning algorithms (NeuralNetwork, randomForest, Logistic regression, Naïve-Bayes and Adaboost) were used to risk-stratify patients for CRC using demographics, symptoms, quantitative faecal immunochemical test (qFIT) and haematological tests. Decision curve analyses were performed to determine the optimal decision strategies. RESULTS: 3776 patients were included (median age, 65; M:F,0.9:1.0) and CRC was identified in 217 patients (5.7%). qFIT > 400 µg Hb/g was the most important variable (%IncMSE = 78.5). RandomForrest had the highest area under curve (0.91) and accuracy (0.80) for CRC. When utilising decision curve analysis (DCA), 30%, 46% and 54% of colonoscopies were saved at accepted CRC probabilities of 1%, 2% and 3%, respectively. RandomForrest modelling had superior net clinical benefit compared to default colonoscopy strategies. CONCLUSIONS: MLA-derived decision strategies that account for patient and referrer risk preference reduce colonoscopy demand and carry net clinical benefit compared to default colonoscopy strategies.
RESUMEN
BACKGROUND: Barrett's oesophagus surveillance places significant burden on endoscopy services yet is vital to detect early cancerous change. Oesophageal cell collection device (OCCD) testing was introduced across Scotland for Barrett's surveillance in response to the COVID-19 pandemic. This national pragmatic retrospective study presents the CytoSCOT programme results and evaluates whether OCCD testing is successfully identifying high-risk Barrett's patients requiring urgent endoscopy. METHODS: All patients undergoing OCCD testing for Barrett's surveillance across 11 Scottish health boards over a 32-month period were identified. Patients who underwent endoscopy within 12 months of OCCD test were included. Individual patient records were interrogated to record clinical information and OCCD test result to categorize patients into risk groups. Endoscopic histopathology results were analysed according to risk group and segment length. Patients were deemed high risk if the OCCD test demonstrated atypia and/or p53 positivity. RESULTS: 4204 OCCD tests were performed in 3745 patients: 608 patients underwent endoscopy within 12 months and were included in this analysis. Patients with longer Barrett's segments were significantly more likely to have an abnormal OCCD test. 50/608 patients (8.2%) had high-grade dysplasia or cancer on endoscopic biopsies: this equates to 1.3% of the total group (50/3745). 46/50 patients (92.0%) were deemed high risk, triggering urgent endoscopy: this rose to 100% with insufficient tests removed. There were no cancers diagnosed within 12 months post-OCCD in the low-risk group. CONCLUSION: OCCD testing is an effective triage tool to identify high-risk patients with Barrett's oesophagus requiring further investigation with endoscopy within the real-world setting.
Asunto(s)
Esófago de Barrett , Neoplasias Esofágicas , Esofagoscopía , Humanos , Esófago de Barrett/patología , Esófago de Barrett/diagnóstico , Masculino , Femenino , Estudios Retrospectivos , Persona de Mediana Edad , Anciano , Esofagoscopía/métodos , Neoplasias Esofágicas/diagnóstico , Neoplasias Esofágicas/patología , COVID-19/diagnóstico , Escocia/epidemiología , Biomarcadores/metabolismo , Medición de Riesgo , Esófago/patología , Detección Precoz del Cáncer/métodos , AdultoRESUMEN
High quality Barrett's esophagus surveillance is crucial to detect early neoplastic changes. An esophageal cell collection device (OCCD) was introduced as a triage tool for Barrett's surveillance. This study aims to evaluate whether the Scottish OCCD program (CytoSCOT) has reduced delays to Barrett's surveillance, and whether delayed surveillance negatively impacts endoscopic pathology. All patients undergoing OCCD testing for Barrett's surveillance across 11 Scottish health boards between 14/9/2020 and 13/9/2022 were identified. Patients were dichotomised into two groups (Year 1 vs. Year 2), with individual records interrogated to record demographics, recommended surveillance interval, time from last endoscopy to OCCD test, and OCCD result. Patients were deemed high-risk if the OCCD demonstrated atypia and/or p53 positivity. Further analysis was performed on patients who underwent endoscopy within 12 months of OCCD testing. A total of 3223 OCCD tests were included in the analysis (1478 in Year 1; 1745 in Year 2). In Year 1 versus Year 2, there was a longer median delay to surveillance (9 vs. 5 months; P < 0.001), increased proportion of patients with delayed surveillance (72.6% vs. 57.0%; P < 0.001), and more high-risk patients (12.0% vs. 5.3%; P < 0.001). 425/3223 patients (13.2%) were further investigated with upper gastrointestinal endoscopy, 57.9% of which were high-risk. As surveillance delay increased beyond 24 months, high-risk patients were significantly more likely to develop dysplasia or malignancy (P = 0.004). Delayed Barrett's esophagus surveillance beyond 24 months is associated with increased risk of pre-cancerous pathology. The CytoSCOT program has reduced delays in surveillance, promoting earlier detection of dysplasia and reducing burden on endoscopy services.
Asunto(s)
Esófago de Barrett , Neoplasias Esofágicas , Esofagoscopía , Esófago de Barrett/patología , Humanos , Masculino , Femenino , Persona de Mediana Edad , Anciano , Neoplasias Esofágicas/patología , Esofagoscopía/métodos , Esofagoscopía/estadística & datos numéricos , Escocia/epidemiología , Factores de Tiempo , Detección Precoz del Cáncer/métodos , Esófago/patología , Diagnóstico Tardío/estadística & datos numéricos , Lesiones Precancerosas/patología , Adenocarcinoma/patologíaRESUMEN
AIM: Lower gastrointestinal (GI) symptoms are poor predictors of colorectal cancer (CRC). The aim of this study was to examine the diagnostic yield of colonoscopy by faecal haemoglobin (f-Hb) concentration in symptomatic patients assessed in primary care by faecal immunochemical testing (FIT). METHOD: In three Scottish NHS Boards, FIT kits (HM-JACKarc, Hitachi Chemical Diagnostics Systems Co., Ltd, Tokyo, Japan) were used by general practitioners to guide referrals for patients with lower GI symptoms (laboratory data studied for 12 months from December 2015 onwards in Tayside, 18 months from June 2018 onwards in Fife and 5 months from September 2018 onwards in Greater Glasgow and Clyde). Cases of CRC diagnosed at colonoscopy were ascertained from colonoscopy and pathology records. RESULTS: Four thousand eight hundred and forty one symptomatic patients who underwent colonoscopy after FIT submission were included. Of the 2166 patients (44.7%) with f-Hb <10 µg Hb/g faeces (µg/g), 14 (0.6%) were diagnosed with CRC, with a number needed to scope (NNS) of 155. Of the 2675 patients (55.3%) with f-Hb ≥10 µg/g, 252 were diagnosed with CRC (9.4%) with a NNS of 11. Of the 705 patients with f-Hb ≥400 µg/g, 158 (22.4%) were diagnosed with CRC with a NNS of 5. Over half of those diagnosed with CRC with f-Hb <10 µg/g had coexisting anaemia. CONCLUSION: Symptomatic patients with f-Hb ≥10 µg/g should undergo further investigation for CRC, while higher f-Hb concentrations could be used to triage for urgency during the COVID-19 recovery phase. Patients with f-Hb <10 µg/g and without anaemia are very unlikely to be diagnosed with CRC and the majority need no further investigation.