非小细胞肺癌同期放化疗后重度急性放射性肺炎的预测模型研究

Study on prediction models for severe acute radiation pneumonitis in patients with non-small cell lung cancer after concurrent chemoradiotherapy

目的 利用剂量体积直方图（DVH）参数建立Logistic剂量反应及Lyman-Kutcher-Burman正常组织并发症概率（LKB-NTCP）模型,并评估其对非小细胞肺癌同期放化疗后重度急性放射性肺炎（SARP）的预测价值.方法 搜集2006-2010年间行三维适形放疗同期化疗的147例非小细胞肺癌患者资料.按美国RTOG毒性评价标准定义超过3级的ARP为SARP.根据DVH剂量学信息建立Logistic剂量反应模型和LKB-NTCP模型.结果 SARP发生率为9.5％（14/147）.Logistic剂量反应模型参数：常数b0=-6.66、b1=0.252,TD50=26.43 Gy,γ50=1.67；模型曲线在17 Gy以下相对平坦,17～18 Gy处变为陡峭,SARP风险增大.LKB-NTCP模型参数：体积效应因子n=0.87±0.40,曲线斜率倒数m=0.27 ±0.10,TD50（1）=（29.5±8.0）Gy;Logistic回归及ROC分析均发现此参数下计算出NTCP值对SARP有良好预测价值（P=0.013、0.019）.结论 NTCP值对SARP的预测价值优于简单剂量参数,2个模型曲线均提示最大限制剂量在约17 Gy.

Objective To establish the Logistic dose response model and Lyman-Kutcher-Burman （LKB）-normal tissue complication probability （NTCP） model using dose-volume histogram （DVH） parameters and to evaluate their predictive values for severe acute radiation pneumonitis （SARP） in patients with non-small cell lung cancer （NSCLC） after concurrent chemotherapy and three-dimensional conformal radiotherapy （3DCRT）.Methods The clinical data of 147 NSCLC patients who were treated with concurrent chemotherapy and 3DCRT from 2006 to 2010 were collected.According to RTOG criteria,grade 3 or even severer acute radiation pneumonitis was defined as SARP.The Logistic dose response model and LKB-NTCP model were established according to DVH dosimetric information.Results The incidence of SARP was 9.5% （14/147）.The best-fit parameter values for Logistic dose response model were shown as follows：constant b0 =-6.66;constant b1 =0.252;TD50 =26.43 Gy;γ50 =1.67.The fit curve was relatively flat when the maximum limit dose （MLD） was 〈 17 Gy,and it became sharper when the MLD was 17-18 Gy,which implied that the risk of SARP increased.The best-fit parameter values for LKB-NTCP model were shown as follows：volume factor n=0.87 ± 0.40;slope factor m =0.27 ±0.10;TD50（1） =（29.5 ±8.0）Gy.The Logistic regression analysis and receiver operating characteristic （ROC） analysis showed that the NTCP value calculated using the parameter values had a good predictive value for SARP （Logistic regression：P =0.013 ;area under the ROC curve：0.707,P =0.019）.Conclusions The predictive value of NTCP for SARP is better than simple dose parameters.The two model curves suggest that MLD is above 17 Gy.