The Experimental And Clinical Research For Evaluation And Optimization Of Breast Cancer Radiotherapy Treatment Plans With Radiobiological Models

Adjuvant radiotherapy given after operation for patients with breast cancer can improve local control, and has become one of the most important local control treatment modalities. Designing of the radiotherapy treatment has a profound effect on the delivery and clinical efficacy of the radiotherapy. So evaluation and optimization radiotherapy plans are always the important parts of radiotherapy. Conventional evaluation of treatment plans which judges a "beast treatment plan" using traditional and practical knowledge alone is no longer adequate to answer the issues continually arising in modern practice. At present, the most common tool of plan evaluation is based on physical endpoint, but it has many pitfalls include that it dose not only consider the radiobiological factors such as the fraction dose size effects, treatment time and the heterogeneity of tissues/organs or tumor radiation sensitivity, but also lack quantitative analysis. With the already availability of modern three dimensional (3-D) treatment planning systems (TPS), physician could produce various treatment plans. It will be very helpful to supplement physical methods by evaluation and optimization radiotherapy plans with radiobiological models. Many radiotherapy techniques for breast-conserving treatment such as: conventional tangents radiotherapy (CONv), 3-dimensional conformal radiotherapy (3-DCRT), forward-planning Intensity-modulated radiotherapy (FMRT) and inverse-planning Intensity-modulated radiotherapy (IMRT) an so on. How to evaluate and chose the best radiotherapy plan for patients is emergent issue. Nowadays, there was no system analysis of breast-conserving radiotherapy techniques with radiobiological models, which is worthy to studying. Moreover, radiation-induced heart disease (RIHD) may occur after radiotherapy for breast cancer, which is the most important complication and compromise the overall survival rate. The common use of heart toxicity drugs in adjuvant breast cancer therapy, such as Anthracyclines, Herceptin and so on, may aggravate the seriousness of RIHD. Thus it is necessary to study the risk of RIHD after modern radiotherapy and chemotherapy. Among the numerous species of RIHD spectrum, the radiation induced cardiac mortality (RICM ) was a common endpoint for NTCP model. But to effectively evaluate the RICM it takes at least 10 years of follow-up was impossible at some degree. So we hope to know the relationship between the results predicted by NTCP models and the variances of clinical early phase examinations. It will be useful for NTCP models to find a good surrogate endpoint which can indirectly and sensitively reflect RIHD early.ObjectivesTo explore the merits and pitfalls of radiobiological models, compare the four common breast-conserving radiotherapy techniques, find the reasonable process of evaluation breast-conserving radiotherapy treatment with radiobiological models. And focusing on the most important RIHD complication of breast cancer radiotherapy, we studied the relationship between the results of RIHD predicted by NTCP models and the variances of clinical examinations, and sought for rational endpoints and parameters for radiobiological models for optimizing breast cancer radiotherapy treatment plans.Methods and materials1 Results obtained from various types of radiological models (NTCP/TCP) was analyzed and compared , to investigate the characteristics of each model and effects of bins sizes of inputted DVH data and the value of parameters on results and provide information to help selection of appropriate radiological models for comparison of radiotherapy treatment planning.2 22 cases of early breast cancer (12 cases with left breast cancer and 10 with right breast cancer) were obtained, for each case four plans were designed for Conventional tangents radiotherapy (CONv), 3-dimensional conformal radiotherapy (3-DCRT), forward-planning Intensity-modulated radiotherapy ( FMRT ) and inverse-planning Intensity-modulated radiotherapy (IMRT) respectively. We apply NTCP-Lyman model compute the probability of radiation pneumonitis and skin ulceration, NTCP-RSM model caculate cardiac mortality, Schneider model count the probability of radiation-induced cancer, LQ-Poisson-TCP model predict the probability of tumor control. Radiobiological indexes such as Normal Tissue Complication Probability (NTCP) and Tumor Control Probability (TCP) were calculated for each of the four plans and Uncomplicated Tumor Control Probability(UTCP) were determined according to NTCP/TCP, and the most favorable plan was determined.3 24 white rabbits irradiated with 3-DCRT, single dose from 0 to 18 Gy made up the clinical related dose group; 8 irradiated dose from 22 to 80Gy made up the high dose group. Detect the serum cTnI/CKMB, ECG, and heart SPECT (using the agents of ⁹Tc^m -MIBI, ⁹⁹Tc^m -HL91 and ⁹⁹Tc^m -Annexin V) before and after irradiation, follow up 5 months, then perform the biopsy of rabbit heart and pathologic test by HE. Stain. To observe the changes of detection results of radiation-induced heart disease (RIHD) in rabbits irradiated in clinical related dose, and to evaluate the role of different detective methods and the apoptosis and hypoxia of the irradiated heart by the new scintigrahic agents of ⁹⁹Tc^m -HL91 and ⁹⁹Tc^m -Annexin V of heart SPECT.4 35 patients with breast cancer after the operation were eligible for this study, who accepted anthracyclines based adjuvant chemotherapy, no history of heart disease. All patients received either 3DCRT or FMRT, and the targets included chest wall and internal mammary node region. There are 22 patients only irradiated the chest wall, 13 patients with irradiated the chest wall and IMN. according the V₂₅ level of heart, that is V₂₅≤1 %, 1 %-5% and >5%, we divided the patients intio three groups. Among follow-up 1 year after treatment, the inspection item included LVEF, ECG, cTnI et al. We defined the event which is above grade 2 as a CDE. RICM was estimated by NTCP-RSM model. We conduct an analysis to see if there is correlation between the results predicted by NTCP model and the results of the cardiac examination. We also made a multiplicity analysis of the CDE.Result1 Comparison among various models NTCP-Lyman model might be better for radiation pneumonitis computation and NTCP-RSM model might be better for radiation-induced cardiac mortality prediction, and LQ-Poisson-TCP might be the better of TCP models in clinical application. Setting various differential DVH (dDVH) bin (e.g. dDVH₁₀ and dDVH₅₀) when data input has minimal effect on the NTCP/TCP results for a single model. Adopting physical dose or LQED₂ affected results with greater results for physical dose. Variation inαorβvalue had significant affect upon TCP results. For example, whenα=0.3Gy^-1 andβ=0.03Gy^-2 was changed toα=0.52 Gy^-1 andβ=0.091^-2 Gy ,it resulted in average differences in TCPs of up to 52.09%( t=19.137,P=0.0001) for LQ-Poisson-TCP. Tumor clonogen cell density had great affect on LQ-Poisson-TCP results as well, There was an average difference of 20.49% (t=8.627, P=0.0001) between TCP results with tumor clonogen cell density of 1×10^2.1 and 7.8×10⁶.2 The Mean PTV doses of all the cases for four plans were 5055.55cGy, 5164.87cGy, 4915.19cGy and 5070.08cGy respectively, and the physical dose difference was not statistically significant (F=1.94, P=0.132). For the 12 cases of left breast cancer, RICM differences among the four planning were statistically significant (F=25.1000,P=0.0001). With the exception for radiation pneumonia., Differences of radiation-induced cancer, skin ulceration and Tumor Control Probability were also of statistical significance, The mean UTCP value of four plans were 0.4397,0.9635,0.9700 and 0.9485 respectively, with statistically significant difference among the four planning(F=25.1000, P=0.0001). Taking UTCP as the Optimization index, FMRT was the best plan in 11 out of the 12 cases of left breast cancer and 8 out of the 10 cases of right breast cancer.3 In the 16 rabbits of clinical related dose group, none died of RIHD. But there were 2 rabbits died of RIHD in the high dose group. The serum value of the cTnI, increased at the 12^th hour after irradiation, and reached the peak maintaining for 4 months, however it decreased at the 5^th month. It suggested that the process of early and delaying myocardial cell radiation damage. The value of cTnI could predict the heart early radiation damage at some extent. The SPECT images displayed that 99Tcm -HL91 and 99mTc-Annexin V did not accumulate in the irradiated heart.4 For 35 cases, the D_mean and the V25 of the hearts is 442.56cGy, 5% respectively; The D_mean of the LAD, LADP and V₂₅ of LAD in the patients with 7 left breast cancer patients only irradiated with 3-DCRT is 1935.73 cGy, 4235.53 cGy, 32% respectively; in the 15 patients who received the FMRT is 1114.01 cGy, 1282.66 cGy, 18% respectively(P<0.05), implying the lower dose of LAD in the FMRT group. After one year's follow-up, 3 patients had complains of palpitations. There is no pericardial effusion, thickening pericardium, myocardial infarction and chronic heart failure reported. The value of cTnI began increasing after the radiotherapy and became normal 6 months later after the RT. It also suggested that the process of early and delaying myocardial cell radiation damage as in experiment research. There was no abnormity variance detected of LVEF and ECG. RICM predicted by the NTCP-RSM model in the three groups (heart V₂₅≤1%, 1%-5% and >5%) is 0.2%,0.5% and 2.4% respectively (P=0.01) ; The correlation among RICM and ECG, LVEF were not significant (P>0.05) But at the end of radiotherapy, the correlation among RICM and cTnI was significant (t=9.834, P=0.001) in heart V₂₅>5% patients, that is RICM= -0.004+3.071×(end of RT cTnI-before RT cTnI). Multivariate analysis showed V₂₅ as an independent prognostic factor of CDE.Conclusion Comparison among various models suggested that NTCP-Lyman model might be better for radiation pneumonitis computation and NTCP-RSM model might be better for radiation-induced cardiac mortality prediction, and LQ-Poisson-TCP might be the better of TCP models in clinical application. Analysis of data for a single model suggested that it was important to understand the distinct characteristics of various models, given that factors such as data input, dosage for calculation and particularly selection of parameters could have substantial effects upon results.The NTCP/TCP results suggest the whole breast radiotherapy after breast conservation techniques such as IMRT, 3-DCRT and FMRT, the TCP was satisfying possibly, but the NTCP results are different. The UTCP results, which incorporated both NTCP and TCP, suggested that FMRT obviously outweigh over CONv and might be better than 3-DCRT and IMRT in general. According experimental and clinical research, the results implied the risk of RIHD is low in modern breast cancer radiotherapy. There were no CDE above grade 3 reported during the one-year follow-up, without irreversible heart damage but the research for the long-term sequelae was warranted. At the end of radiotherapy, the correlation among RICM and cTnI was significant in heart V₂₅>5% patients. Which suggested the value of cTnI at the end of treatment maybe predict the probability of RIHD if V₂₅ of the heart over 5%. Since the incidence of CDE was higher in group 3 than other two groups, and V₂₅ was suggested to be an independent prognostic factor of CDE, it was crucial to judiciously control the dose of the heart and it would be beneficial to control V₂₅ of heart less than 5%.In a word, it is feasible and reasonable to evaluate and optimize the radiotherapy treatment plans for breast cancer with radiobiological models. To accumulate the clinical data for making radiobiological models perfectly and optimize the radiotherapy treatment plans rationally.