孤立性肺结节恶性概率临床预测模型的建立

查小久; 黄礼年; 贡会源; 王安生

doi:10.13898/j.cnki.issn.1000-2200.2023.11.007

孤立性肺结节恶性概率临床预测模型的建立

Establishment of clinical predictive model of malignant probability of solitary pulmonary nodules

摘要

摘要:
目的筛选出恶性孤立性肺结节独立危险因素，构建判断孤立性肺结节良恶性的列线图模型。
方法收集胸外科行手术治疗并有明确病理的孤立性肺结节病人的临床资料及胸部影像学特征。通过单因素及多因素logistic回归分析，筛选出与孤立性肺结节良恶性相关的独立影响因素，并利用列线图构建预测孤立性肺结节恶性概率的数学模型。最后，利用受试者工作特征（ROC）曲线下面积（AUC）及校准曲线验证风险预测模型。
结果605例病人中，恶性病变398例（65.8%），良性病变207例。多因素分析中，病人年龄、油烟吸入、含磨玻璃成分、长毛刺、短毛刺、胸膜凹陷征、血管集束征、空泡征、钙化、深分叶是恶性孤立性肺结节的独立影响因素（P < 0.05~P < 0.01）。将上述独立影响因素纳入列线图中，构建预测恶性孤立性肺结节概率的数学模型。本列线图模型显示出较好的区分度及一致性，ROC的AUC为0.913（95%CI：0.888~0.938），当截取T=0.55时，约登指数最大，此时模型的敏感性为89.2%，特异性为80.2%，校正曲线显示预测孤立性肺结节的恶性肿瘤概率与实际恶性肿瘤的概率基本平行，为斜率大约45°的曲线。校准图显示该模型充分拟合，可以预测孤立性肺结节的恶性肿瘤概率。
结论年龄、油烟吸入史、含磨玻璃成分、短毛刺、深分叶、血管集束征、胸膜凹陷征、空泡征是恶性孤立性肺结节的独立危险因素，而钙化、长毛刺更常见于良性孤立性肺结节。列线图模型可使临床医生对孤立性肺结节的恶性概率进行个体化、可视化和精确预测。

Abstract:
ObjectiveTo screen the independent risk factors for malignant solitary pulmonary nodules, and build a nomogram model to estimate the probability of malignancy.
MethodsThe clinical data and chest imaging characteristics of patients with solitary pulmonary nodules who underwent surgical treatment and had definite pathology in the Department of Thoracic Surgery were collected.The independent risk factors were screened by univariate and multivariate logistic regression analysis, and a nomogram mathematical model was established to predict the malignant probability of solitary pulmonary nodules.Finally, the risk prediction model was validated using the area under the curve (AUC) of receiver operating characteristic (ROC) and the calibration curve.
ResultsAmong 605 patients, there had 398 (65.8%) cases of malignant lesions, and 207 benign lesions.In multivariate analysis, patient's age, lampblack, ground-glass opacity, long-spiculation, short- spiculation, pleural indentation, vessel convergence, vacuole sign, calcification and deep-lobulation were independent influencing factors of malignant solitary pulmonary nodules (P < 0.05 to P < 0.01).The above independent influencing factors were included in the nomogram to construct a mathematical model to predict the malignant probability of solitary pulmonary nodules.The nomogram model showed better discrimination and consistency.The AUC was 0.913 (95%CI: 0.888-0.938).When T=0.55 was intercepted, the Youden index was the largest, with a sensitivity of 89.2% and a specificity of 80.2%.The calibration curve showed that the predicted malignancy probability of solitary pulmonary nodules was substantially parallel to the probability of actual malignant tumors, with a slope of approximately 45°.Calibration graphic revealed adequate fit of the model predicting the risk of malignancy probability of solitary pulmonary nodules.
ConclusionsAge, lampblack, ground-glass opacity, short-spiculation, pleural indentation, vessel convergence, vacuole sign, and deep-lobulation are independent risk factors for malignant solitary pulmonary nodules, while calcification and long-spiculation are more common in benign solitary pulmonary nodules.A new nomogram model, allowing clinicians to individualize, visualize and accurately predict the malignant probability of solitary pulmonary nodules.

HTML全文

参考文献(31)

施引文献

资源附件(0)