摘要目的 比较分析不同放射生物模型的特性,以寻求评价乳腺癌放疗计划合理的放射生物模型.方法 比较预测放射性肺炎发生率和放射性心脏病死亡率的NTCP两种模型和TCP四种模型,计算相同DVH数据所得结果的差异;并分析同一模型中,输入DVH数据的形式、参数的选择等对结果的影响.结果 假设全肺平均照射30 Gy剂量时,NTCP-RSM模型预测的放射性肺炎发生率为32%,NTCP-Lyman模型预测的为54%.以发生放射性心脏病死亡率1%为例,NTCP-RSM模型对应的心脏平均照射剂量为28 Gy,而NTCP-Lyman模型对应的为40 Cy.应用LQ-Poisson-TCP模型、Poisson-TCP模型、Logit-TCP模型和Zaider-TCP模型,计算相同DVH数据库的平均TCP分别为21.1%、38.4%、41.0%和80.8%(P=0.000).采用不同栅格大小计算的NTCP/TCP结果差别较小.计算时采用物理剂量或LQED2剂量对NTCP/TCP结果有一定影响,采用物理剂量时的结果稍大.ft.和p值、肿瘤细胞密度、D50值和DVH简化方法对TCP的影响显著(P=0.000).结论 评价和优化乳腺癌放疗计划选择放射生物模型时,以NTCP-Lyman模型计算放射性肺炎和以NTCP-RSM模型计算放射性心脏病死亡率比较合理.TCP模型以LQ-Poisson-TCP模型比较符合临床实际.影响预测结果最大的是模型参数值的选取,选择时需要加以注意.这些模型目前有助于对不同治疗模式进行研究和比较,而不是给出对临床实际结果的精确预测.

abstractsObjective To find an appropriate r3diobiological model for analyzing the biological effect of the radiotherapy for breast cancer by comparing different results computed by various types of radiobiological models. Methods DVHs database simulating breast conserving radiotherapy was set up,based on clinical DVHs data of the heart.the lung and PTV of 22 patients with early breast cancer given conventional tangential radiotherapy.Two models assessing NTCP of radiation pneumonitis and cardiac mortality and four models assessing TCP were compared by analyzing the effects of the parameters and DVH database input methods on the results. Results When mean irradiation dose of the whole lung was 30 Gy.the incidence of radiation pneumonitis was 32%and 54%predicted by NTCP-RSM model and NTCP-Lyman model,respectively.When 1%cardiac mortality of radiation was assumed,28 Gy and 40 Gy isodose should cover the heart assessed by the two models.The mean TCP were 21.1%.80.8%.38.4%and 41.0%when assessed by LQ-Poisson-TCP,Zaider-TCP,Poisson-TCP and Logit-TCP models,respectively.Setting various differential DVH(dDVH)bins had very few effect on the NTCP/TCP results in a certain model.Adopting physical dose or LQED2 affected the results with greater resu]ts for physical dose.Variation in α or β value,tumor cell density and D50 had significant effect upon TCP results in LQ-Poisson-TCP(P:0.000). Conclusions NTCP-Lyman model is better for predicting the incidence of radiation pneumonitis while NTCP-RSM model is better for predicting radiation-induced cardiac mortality.LQ-Poisson-TCP is the best TCP model for clinical application.Parameters selected for model can significantly affect the results.It is imporrant to understand the distinct characteristics of different models.