预测小细胞肺癌一线化疗导致严重中性粒细胞减少的Nomogram的构建

Establishment of a nomogram to predict severe neutropenia induced by first-line chemotherapy in patients with small cell lung cancer

目的建立一个Nomogram来预测小细胞肺癌一线化疗引起严重中性粒细胞减少(3级或4级)的概率。方法选择经组织学证实并接受一线化疗的小细胞肺癌患者263例纳入预测队列。在多变量Logistic回归分析中,采用前进法、后退法和逐步法的似然比检验,以最小赤池信息准则值作为停止准则,在此基础上绘制Nomogram;评估其鉴别、校准能力以及临床实用性。结果将年龄、目前吸烟状态、卡氏评分、肿瘤分期、血小板、化疗方案等因素提取到预测模型中建立了Nomogram。评价结果表明该Nomogram具有较好的区分度(接收器工作特性曲线下面积为0.73),以及良好的校准曲线。决策性曲线分析显示,当阈概率的范围在15%~90%之间时,使用该Nomogram预测严重的中性粒细胞减少症将比全干预和无干预方案获得更多的净收益。建议医生将干预临界或偏好设定在15%~30%。结论建立的Nomogram能准确预测小细胞肺癌每次化疗发生严重中性粒细胞减少的可能性,具有一定的临床实用价值。

Objective To establish a Nomogram to predictive the probability of severe neutropenia(grade 3 or 4)caused by first-line chemotherapy in patients with small cell lung cancer(SCLC).Methods In total,263 cases who had histologically confirmed SCLC and received first-line chemotherapy were selected and enrolled in the predictive cohort.Three procedures(forward,backward and stepwise selections)were applied by using the likelihood ratio test with smallest Akaike′s information criterion as the stopping rule in the multivariable logistic regression analysis to establish a Nomogram.The assessment of this Nomogram was performed by evaluating its discrimination and calibration abilities,as well as clinical utility.Results Age,current smoking,karnofsky performance status,tumor stage,platelet and chemotherapy regimen were extracted into the predictive model to establish the Nomogram.The results of assessment showed this Nomogram model had a good discrimination with area under the receiver operating characteristics(ROC)curve(AUC)of 0.73,and an acceptable calibration curve.In addition,the decision curve analysis showed when the range of threshold probability was approximately between 15%and 90%,using the nomogram to predict severe neutropenia would get more net benefits than the intervention-all scheme and intervention-none scheme.We suggested doctors should set the threshold or preference at 15%-30%.Conclusion The established Nomogram can accurately predict the probability of severe neutropenia before each chemotherapy in SCLC and has certain application value in clinical practice.