An Algorithm to Predict Advanced Proximal Colorectal Neoplasia in Chinese Asymptomatic Population

Abstract

This study aims to develop and validate a new algorithm that incorporates distal colonoscopic findings to predict advanced proximal neoplasia (APN) in a Chinese asymptomatic population. We collected age, gender, and colonoscopic findings from a prospectively performed colonoscopy study between 2013 and 2015 in a large hospital-based endoscopy unit in Shanghai, China. Eligible subjects were allocated to a derivation group (n = 3,889) and validation group (n = 1,944) by random sampling. A new index for APN and its cut-off level were evaluated from the derivation cohort by binary logistic regression. The model performance was tested in the validation cohort using area under the curve (AUC). Age, gender, and distal finding were found to be independent predictors of APN in the derivation cohort (p < 0.001). Worldwide, colorectal cancer (CRC) is the third most common cancer in males and ranks the second in females, with an estimated 1.4 million cases and 693,900 deaths reported in 2012 1 . The increasing incidence and mortality of CRC in many low-income and middle-income countries, including China, highlights the pressing need for CRC control 2 . Screening has been proven an effective and cost-efficient strategy to reduce CRC mortality 3 . CRC prevention should be the primary goal of screening. Tests that are employed to detect early cancer and adenomatous polyps have been recommended for average-risk subjects aged 50 years or above 4 , and these include colonoscopy and flexible sigmoidoscopy (FS). In recent decades, proximal shift of CRC and advanced proximal neoplasia (APN) with no distal lesions detected 5-9 are increasingly observed, and this has an implication on the use of FS in clinical practice. Nevertheless, colonoscopy might not be suited as a primary screening test in resource-limited countries with limited colonoscopy capacity 10 . It may be more cost-effective and cost-saving to reserve colonoscopy for subjects with high risk for APN in population-based screening 11,12 . To address this need, Imperiale et al.12 proposed a 7-point scoring system to predict APN by using age, gender, and distal findings at FS as predictors and a high discrimination value was reported in its internal validation group. Its generalizability remain unknown in Chinese subjects, since the incidence and distribution of colorectal neoplasia are different among different ethnic groups 13 . Levitzky et al. has performed an external validation of the model devised by Imperiale et al., and the discriminatory capability was found to be lower in a different population, including black and Hispanic 11 . to the primary objective of this study is to develop and validate a new model specifically tailored to predict APN in a Chinese asymptomatic population. We conducted this cross-sectional study in a large hospital-based endoscopy unit.

From Algorithms to Clinical Utility: A Systematic Review of Individualized Risk Prediction Models for Colorectal Cancer

“…The newly developed risk prediction models between 2011 and 2021 predominantly originated from Asian countries (n = 19), including China [37,46,68,74], Korea [47,48,50,51,61,62,66,70,72], Japan [59,65], Thailand [57], Hong Kong [45,49], and Taiwan [56]. Among these novel risk prediction models based on Asian populations, the most common predictors were age (n = 19), smoking history (n = 17), sex (n = 15), BMI (n = 14), family history of CRC (n = 14), alcohol intake (n = 8), and diabetes mellitus (n = 6) [37,[45][46][47][48][49][50][51]56,57,59,61,62,65,66,68,70,72,74,80].…”

“…All included studies, except for that of Jung et al (2018) and Briggs et al (2023), reported AUC values to assess the discriminatory accuracy of the models (Table 3). Among the 44 included models, 23 conventional risk models [37,40,42,[44][45][46][48][49][50][51]54,[58][59][60][61][63][64][65][67][68][69][70]72,74,75] and 10 composite models [39,41,47,56,57,62,71,75,76,80] reported AUC values with a 95% confidence interval (CI), as presented in Figure 5. Meanwhile, the model calibration was assessed in two different metrics: using the expected-to-observed (E/O) ratios, which were only demonstrated in nine of the studies; or by employing the Hosmer-Lemeshow (HL) goodness-of-fit test, as was the case in 12 out of 41 studies.…”

Challenges to the adoption of risk algorithms for colorectal cancer screening programmes: perspectives for future research

“…Subsequent validation in other populations might not result in similar discriminatory capabilities. 11 Third, there are concerns about the generalisability of these scores in people residing in different regions around the world. It is well recognised that the prevalence and distribution of advanced colorectal neoplasia differ for different ethnicities.…”