基于最优子集回归和临床病理/血液学参数的早期宫颈癌预后预测研究

Prognostic Prediction Study of Early Cervical Cancer Based on Best Subset Regression and Clinicopathological/Hematological Parameters

目的:探讨早期宫颈癌预后相关因素并建立预测模型,为术后辅助治疗提供临床依据。方法:收集172例接受根治性子宫切除术的宫颈癌(术前分期为IB1、IB2、IIA1)患者的临床病理资料和术前血液学参数。通过最优子集回归分析,构建一个早期宫颈癌预后模型,并对其进行了预测效能评估。结果:全集患者中位发病年龄50岁;中位无病生存期(disease free survival,DFS)33.5个月,1年和3年的DFS分别是94.8%、89.7%。病理分期、肿瘤大小、浸润深度、阴道切缘状态、手术方式(腹腔镜或开腹)放射治疗情况与宫颈癌预后密切相关。此外,中性粒细胞比例、单核细胞比例和红细胞压积也与预后有关。基于以上因素建立的预后模型的一致性指数为0.822(95%CI:0.781~0.863),经过1年和3年的校正曲线和决策曲线分析验证,该模型显示出较高的预测准确度并具有明显的净收益。同时Bootstrap内部验证得到C-index平均值为0.681(95%CI:0.637~0.729)。结论:本研究建立的基于临床病理、血液学参数的宫颈癌预后模型,能够直观地预测早期宫颈癌患者在术后1年和3年的DFS。

Objective:Toexplore prognostic factorsof early-stage cervical cancer and establish a predictive model,in order to provide clinical evidence for postoperative adjuvant therapy.Methods:Clinicalpathological factors as well as preoperative hematological parameters were collected from 172 cervical cancer patients(IB1,IB2 and IA1)who underwent radical hysterectomy.A prognostic model based on these data was established through best subset regression,and its prediction efficacy was evaluated.Results:The median age of the entire cohort was 50 years old;the median disease-free survival(DFS)was 33.5 months,with 1-and 3-year DFS rates of 94.8%and 89.7%,respectively.We found that pathological stage,tumor size,depth of invasion,vaginal margin status,surgical approach(laparoscopy or laparotomy),and radiation therapy were closely associated with the prognosis of cervical cancer patients.Additionally,the proportion of the neutrophils,the proportion of monocytes,and hematocrit were also related to the prognosis of cervical cancer patients.The consistency index of the prognostic model based on these factors was 0.822(95%Cl:0.781~0.863).Through 1-and 3-year calibration curve analysis and decision curve analysis,the model demonstrated higher predictive accuracy and significant net benefits.The average C-index obtained from internal validation using Bootstrap was 0.681(95%Cl:0.637~0.729).Conclusion:The prognostic model for cervical cancer developed in this study based on clinicalpathological factors and hematological parameters is capable of providing intuitive predictions for the I-and 3-year DFS of early-stage cervical cancerpatients.