孤立性肺小结节及微小结节恶性概率的预测模型建立

The Establishment of Prediction Model for Malignant Probability of Solitary Pulmonary Nodules and Tiny Nodules

目的探讨孤立性肺小结节及肺微小结节的独立危险因素,并建立其恶性概率的临床预测模型。方法收集苏州大学附属第一医院2018年01月—2018年11月肺结节直径≤10 mm,且经手术取得病理结果的患者的胸部CT和临床危险因素资料,建立恶性概率预测模型。结果301例患者中,术后病理为恶性结节209例,良性结节92例。吸烟(OR=4.522)、环境或高危职业暴露史(OR=5.818)、家族恶性肿瘤史(OR=51.836)、结节密度为部分实性(OR=46.694)、结节密度为纯磨玻璃(OR=7.486),体积越大(100~300 mm^(3),OR=12.090,>300 mm^(3),OR=16.025)在肺小结节及肺微小结节良恶性两组之间的差异具有统计学意义(P<0.05);根据结果建立二元Logistic回归预测模型,与传统梅奥模型绘制的ROC曲线比较,预测概率模型AUC曲线下面积大于传统梅奥模型AUC曲线下面积。结论吸烟、有环境或高危职业暴露史、有家族恶性肿瘤史、结节密度为部分实性或纯磨玻璃、体积较大是孤立性肺小结节与肺微小结节的独立危险因素。构建的二元Logistic回归预测模型可以辅助预测CT初筛结果为直径≤10 mm的肺结节患者的恶性概率,相对于传统梅奥模型具有更高的诊断价值。

Objective To investigate the independent risk factors of solitary pulmonary nodules and tiny nodules,and to establish a clinical prediction model for their malignant probability.Methods From January 2018 to November 2018 in the First Affiliated Hospital of Soochow University,the data of chest CT and clinical risk factors of patients with lung nodules≤10 mm in diameter and pathological results obtained by surgery were collected to establish a prediction model of malignancy probability.Results Among 301 patients,there were 209 cases of malignant nodules and 92 cases of benign nodules.Smoking(OR=4.522),environmental or high-risk occupational exposure history(OR=5.818),family history of malignant tumors(OR=51.836),nodule density is partially solid(OR=46.694),nodule density is pure ground glass(OR=7.486),larger volumes(100-300 mm^(3),OR=12.090,>300 mm^(3),OR=16.025)were statistically significant between the two groups of benign and malignant solitary pulmonary nodules and tiny nodules(P<0.05).A logistic binary regression prediction model was established based on the results,and compared with the ROC curve drawn by the traditional Mayo model.The area under the AUC curve of the predictive probability model is larger than the traditional Mayo clinic model.Conclusion Smoking,history of environmental or high-risk occupational exposure,family history of malignant tumors,nodule density of partially solid or pure ground glass and large size are independent risk factors for solitary pulmonary nodules and tiny nodules.The constructed binary logistic regression prediction model can assist in predicting the malignant probability of patients with pulmonary nodules with a diameter≤10 mm as a result of CT preliminary screening,which has a higher diagnostic value than the traditional Mayo model.