Epidemiology and Prevention Division, Research Center for Cancer Prevention and Screening, National Cancer Center, 5-1-1 Tsukiji, Chuo-ku, Tokyo 104-0045, Japan.
Cancer Epidemiol. 2010 Oct;34(5):534-41. doi: 10.1016/j.canep.2010.04.021. Epub 2010 May 31.
To estimate an individual's probability of developing colorectal cancer (CRC) may aid health professionals and individuals in improving lifestyle behaviors or deciding the screening regimens. As fewer studies on cancer risk prediction were seen so far, we initially developed an assessment tool with synthesizing key information from a variety of CRC risk factors through a large population-based cohort study.
The prediction model was derived from 28,115 men in the Japan Public Health Center-based (JPHC) Prospective Study Cohort II (follow-up: 1993-2005), with risk factors selected by Cox proportion hazard regression. 18,256 men in the JPHC Study Cohort I (follow-up: 1995-2005) were used to evaluate the model's performance.
543 and 398 CRCs were diagnosed during the follow-up period in Cohorts II and I, respectively. The prediction model, including age, BMI, alcohol consumption, smoking status, and the daily physical activity level, showed modest discrimination ability for CRC (C=0.70; 95% confidential interval, 0.68-0.72) in Cohort II and well calibrated in Cohort I (Hosmer-Lemeshow χ(2)=14.2, P=0.08).
The 10-year CRC risk prediction model may be used to estimate CRC risk in Japanese men. It may also play a role in the promotion of CRC prevention strategies.
估计个体罹患结直肠癌（CRC）的概率可能有助于医疗专业人员和个体改善生活方式行为或决定筛查方案。由于到目前为止，关于癌症风险预测的研究较少，我们最初通过一项基于大型人群队列研究的综合分析，从各种 CRC 风险因素中提取关键信息，开发了一种评估工具。
该预测模型来源于日本公共卫生中心（JPHC）前瞻性研究队列 II 中的 28115 名男性（随访时间：1993-2005 年），通过 Cox 比例风险回归选择风险因素。JPHC 研究队列 I 中的 18256 名男性（随访时间：1995-2005 年）用于评估模型的性能。
在队列 II 和队列 I 的随访期间，分别诊断出 543 例和 398 例 CRC。该预测模型包括年龄、BMI、饮酒、吸烟状况和日常体力活动水平，在队列 II 中对 CRC 具有适度的区分能力（C=0.70；95%置信区间，0.68-0.72），在队列 I 中具有良好的校准度（Hosmer-Lemeshow χ²=14.2，P=0.08）。
该 10 年 CRC 风险预测模型可用于估计日本男性的 CRC 风险。它也可能在促进 CRC 预防策略方面发挥作用。
