王冬青,张建,李宝生,孙洪福.剂量体积直方图参数对放射性肺炎的工作特征曲线预测分析[J].中华放射医学与防护杂志,2012,32(5):505-508
剂量体积直方图参数对放射性肺炎的工作特征曲线预测分析
Dose-volume histogram parameters for predicting radiation pneumonitis using receiver operating characteristic curve
投稿时间:2012-02-09  
DOI:10.3760/cma.j.issn.0254-5098.2012.05.014
中文关键词:  非小细胞肺癌  剂量体积直方图  放射性肺炎  受试者工作特征曲线
英文关键词:Non-small cell lung cancer  Dose-volume histogram  Radiation pneumonitis  Receiver operating characteristic curve
基金项目:
作者单位E-mail
王冬青 250117 济南, 山东省肿瘤医院  
张建 250117 济南, 山东省肿瘤医院  
李宝生 250117 济南, 山东省肿瘤医院 baoshli@yahoo.com 
孙洪福 250117 济南, 山东省肿瘤医院  
摘要点击次数: 3123
全文下载次数: 2226
中文摘要:
      目的 应用受试者工作特征曲线(ROC)对剂量体积直方图参数(DVH)预测放射性肺炎(RP)进行分析,探究DVH参数预测RP的准确性(ACC)、敏感性(SEN)和特异性(SPE)。方法 收集118例接受三维适形调强放疗和化疗的非小细胞肺癌患者资料,回顾分析三维放疗计划系统中双肺V5V10V13V20V30 (Vx为接受≥x Gy的相对肺体积) 和平均肺剂量 (MLD) 与治疗后出现≥2级RP (CTCAE3.0) 的相关性。对上述DVH参数应用ROC曲线进行回顾性分析,确定预测RP的ACC、SEN和SPE。结果 单因素分析显示,双肺V5V10V13V20和MLD均与RP发生显著相关(χ2=4.786、5.771、6.366、7.367、6.945,P<0.05);双肺V30、患者因素(年龄、性别、KPS评分、肿瘤位置、病理类型)和治疗因素(放疗总剂量、照射技术、化疗方案、化疗时机)与RP的发生风险无显著相关性。多因素分析显示双肺V20与RP发生风险相关(χ2=10.96,OR=4.16, 95%CI 1.40~12.36,P<0.05),与其他DVH参数具有显著共线性(r=0.767~0.902,P<0.05)。ROC曲线证实双肺V20能够预测RP的发生(Z=2.038,P<0.05),其预测的ACC、SEN和SPE分别为0.645(95%CI 0.498~0.793),0.650(95%CI 0.408~0.864) 和0.674(95%CI 0.571~0.765),其阳性预测值仅为28.9%。结论 双肺V20与RP的发生风险相关,能够预测RP的发生,但是预测能力有限。
英文摘要:
      Objective To assess the accuracy (ACC), sensitivity (SEN), and specificity (SPE) of dose-volume histogram (DVH) parameters in predicting the radiation pneumonitis (RP) using receiver operating characteristic (ROC) curve.Methods Complete clinical data of 118 non-small cell lung cancer patients treated with three-dimensional conformal and intensity-modulated radiotherapy plus chemotherapy were included. Chi-square and logistic regression were retrospectively applied to analyze the correlations between DVH parameters [relative lung volume received ≥ 5 Gy (V5), 10 Gy (V10), 13 Gy (V13), 20 Gy (V20) and 30 Gy (V30) and mean lung dose (MLD)] and grade 2 (and above) RP defined by the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0. ROC curve was adopted to investigate the predictive ACC, SEN and SPE of potential DVH parameters associated with RP.Results Total lungs V5,V10,V13,V20 and MLD were all correlated to the development of RP (χ2=4.786,5.771,6.366,7.367 and 6.945,P<0.05) according to univariate analysis. However, total lungs V30, patient characteristics (age, sex, KPS, tumor location, pathology) and treatment factors (prescription dose, radiotherapy technique, chemotherapy method and timing) were not contributors to RP. Logistic regression showed that V20 of both lungs remains tight by associated with RP (χ2=10.96,OR=4.16, 95%CI 1.40~12.36,P<0.05), although significant colinearity was found between V20 and other DVH parameters (r=0.767- 0.902,P<0.05). ROC curve confirmed that V20 of both lungs could act as a predictor for RP (Z=2.038,P<0.05). The predictive ACC, SEN, and SPE were 0.645 (95% CI 0.498-0.793), 0.650 (95% CI 0.408 - 0.864), and 0.674 (95% CI 0.571 - 0.765), respectively. However, the positive predictive value was only 28.9%.Conclusions V20 of both lungs was correlated to the development of RP. It could act as a predictor for RP though the predictability is limited.
HTML  查看全文  查看/发表评论  下载PDF阅读器
关闭