Objective: To examine a predictive model that incorporating high risk pathological factors for the prognosis of stage Ⅰ to Ⅲ colon cancer. Methods: This study retrospectively collected clinicopathological information and survival outcomes of stage Ⅰ~Ⅲ colon cancer patients who underwent curative surgery in 7 tertiary hospitals in China from January 1, 2016 to December 31, 2017. A total of 1 650 patients were enrolled, aged (M(IQR)) 62 (18) years (range: 14 to 100). There were 963 males and 687 females. The median follow-up period was 51 months. The Cox proportional hazardous regression model was utilized to select high-risk pathological factors, establish the nomogram and scoring system. The Bootstrap resampling method was utilized for internal validation of the model, the concordance index (C-index) was used to assess discrimination and calibration curves were presented to assess model calibration. The Kaplan-Meier method was used to plot survival curves after risk grouping, and Cox regression was used to compare disease-free survival between subgroups. Results: Age (HR=1.020, 95%CI: 1.008 to 1.033, P=0.001), T stage (T3:HR=1.995,95%CI:1.062 to 3.750,P=0.032;T4:HR=4.196, 95%CI: 2.188 to 8.045, P<0.01), N stage (N1: HR=1.834, 95%CI: 1.307 to 2.574, P<0.01; N2: HR=3.970, 95%CI: 2.724 to 5.787, P<0.01) and number of lymph nodes examined (≥36: HR=0.438, 95%CI: 0.242 to 0.790, P=0.006) were independently associated with disease-free survival. The C-index of the scoring model (model 1) based on age, T stage, N stage, and dichotomous variables of the lymph nodes examined (<12 and ≥12) was 0.723, and the C-index of the scoring model (model 2) based on age, T stage, N stage, and multi-categorical variables of the lymph nodes examined (<12, 12 to <24, 24 to <36, and ≥36) was 0.726. A scoring system was established based on age, T stage, N stage, and multi-categorical variables of lymph nodes examined, the 3-year DFS of the low-risk (≤1), middle-risk (2 to 4) and high-risk (≥5) group were 96.3% (n=711), 89.0% (n=626) and 71.4% (n=313), respectively. Statistically significant difference was observed among groups (P<0.01). Conclusions: The number of lymph nodes examined was an independent prognostic factor for disease-free survival after curative surgery in patients with stage Ⅰ to Ⅲ colon cancer. Incorporating the number of lymph nodes examined as a multi-categorical variable into the T and N staging system could improve prognostic predictive validity.
目的： 探讨基于术后不良病理学特征建立Ⅰ~Ⅲ期结肠癌预后预测模型的效力。 方法： 回顾性收集2016年1月1日至2017年12月31日就诊于国内7家三甲医院的Ⅰ~Ⅲ期结肠癌患者的病理学及预后信息，共纳入1 650例患者，男性963例，女性687例，年龄［M（IQR）］62（18）岁（范围：14~100岁）。中位随访时间51个月。采用Cox比例风险回归模型筛选不良病理学特征、制作列线图和评分模型，选用Bootstrap重抽样法进行模型的内部验证，使用一致性指数评价模型的区分度，利用校准曲线评估模型的校准度，使用Kaplan-Meier法绘制风险分组后的生存曲线，使用Cox单因素回归比较亚组间生存率的差异。 结果： 无病生存的独立预后因素包括年龄（HR=1.020，95%CI：1.008~1.033，P=0.001）、T分期（T3：HR=1.995，95%CI：1.062~3.750，P=0.032；T4：HR=4.196，95%CI：2.188~8.045，P<0.01）、N分期（N1：HR=1.834，95%CI：1.307~2.574，P<0.01；N2：HR=3.970，95%CI：2.724~5.787，P<0.01）和淋巴结获取数目（≥36枚：HR=0.438，95%CI：0.242~0.790，P=0.006）。基于年龄、T分期、N分期、淋巴结分类二分类变量（<12枚和≥12枚）的评分模型（模型1）的一致性指数为0.723，基于年龄、T分期、N分期、淋巴结分类多分类变量（<12枚、12~<24枚、24~<36枚、≥36枚）的评分模型（模型2）的一致性指数为0.726。建立基于淋巴结多分类变量的评分系统，将风险评分≤1 划分为低危组（n=711），2~4划分为中危组（n=626），≥5 划分为高危组（n=313），3 年无病生存率分别为96.3%、89.0%、71.4%，差异有统计学意义（P<0.01）。 结论： 淋巴结获取数量是Ⅰ~Ⅲ期结肠癌患者术后无病生存的独立预后因素，将其作为多分类变量与T、N分期联合具有良好的预后预测效力。.