Development and Validation of the Asia-Pacific Proximal Colon Neoplasia Risk Score
Martin C.S. Wong,Rungsun Rerknimitr,Khean Lee Goh,Takahisa Matsuda,Hyun-Soo Kim,Deng-Chyang Wu,Kai Chun Wu,Khay Guan Yeoh,Vui Heng Chong,Furqaan Ahmed,Jose D. Sollano,Jayaram Menon,Han-Mo Chiu,Jingnan Li,Jessica Y.L. Ching,Joseph J.Y. Sung,Martin C S Wong,Rungsun Rerknimitr,Khean Lee Goh,Takahisa Matsuda,Hyun-Soo Kim,Deng-Chyang Wu,Kai Chun Wu,Khay Guan Yeoh,Vui Heng Chong,Furqaan Ahmed,Jose D Sollano,Jayaram Menon,Han-Mo Chiu,Jingnan Li,Jessica Y L Ching,Joseph J Y Sung
DOI: https://doi.org/10.1016/j.cgh.2019.12.031
IF: 13.576
2021-01-01
Clinical Gastroenterology and Hepatology
Abstract:<h3 class="u-h4 u-margin-m-top u-margin-xs-bottom">Background & Aims</h3><p>Patients found to be at high risk of advanced proximal neoplasia (APN) after flexible sigmoidoscopy screening should be considered for colonoscopy examination. We developed and validated a scoring system to identify persons at risk for APN.</p><h3 class="u-h4 u-margin-m-top u-margin-xs-bottom">Methods</h3><p>We collected data from 7954 asymptomatic subjects (50–75 years old) who received screening colonoscopy examinations at 14 sites in Asia. We randomly assigned 5303 subjects to the derivation cohort and the remaining 2651 to the validation cohort. We collected data from the derivation cohort on age, sex, family history of colorectal cancer, smoking, drinking, body mass index, medical conditions, and use of non-steroidal anti-inflammatory drugs or aspirin. Associations between the colonoscopic findings of APN and each risk factor were examined using the Pearson χ<sup>2</sup> test, and we assigned each participant a risk score (0-15), with scores of 0-3 as average risk and scores of ≥4 as high risk. The scoring system was tested in the validation cohort. We used the Cochran-Armitage test of trend to compare the prevalence of APN among subjects in each group.</p><h3 class="u-h4 u-margin-m-top u-margin-xs-bottom">Results</h3><p>In the validation cohort, 79.5% of patients were classified as average risk and 20.5% were classified as high risk. The prevalence of APN in the average-risk group was 1.9% and in the high-risk group was 9.4% (adjusted relative risk 5.08; 95% CI, 3.38–7.62; P<0.001). The score included age (61-70 years: 3; ≥70 years: 4), smoking habits (current/past: 2), family history of colorectal cancer (present in a first-degree relative: 2), and the presence of neoplasia in the distal colorectum (non-advanced adenoma 5-9mm: 2; advanced neoplasia: 7). The c-statistics of the score was 0.74 (95% CI, 0.68–0.79), and that for distal findings alone was 0.67 (95% CI=0.60-0.74). The Hosemer-Lemeshow goodness-of-fit test statistic was greater than 0.05, indicating the reliability of the validation set. The number needed to refer was 11 (95% CI, 10–13) and the number needed to screen was 15 (95% CI, 12–17).</p><h3 class="u-h4 u-margin-m-top u-margin-xs-bottom">Conclusions</h3><p>We developed and validated a scoring system to identify persons at risk for APN. Screening participants who undergo flexible sigmoidoscopy screening with a scores of 4 points or more should undergo colonoscopy evaluation.</p>
gastroenterology & hepatology