The Use Of Four-Dimentional CT And Cone Beam CT In Target Determination For Lung Stereotactic Body Radiation Therapy

According to statistics from the World Health Organization,lung cancer is the leading cause of morbidity and mortality in all malignant tumors.High precision radiotherapy represented by stereotactic body radiation therapy(SBRT)has played a significant role in multiple stages of lung cancer treatment.High precision radiotherapy requires high conformal radiation dose delivery.Respiratory-induced tumor motion is a major cause of geometric uncertainty of targets during radiation delivery.The uniform margin may result in a geographical miss or high risk of radiation injuries.A four-dimensional CT(4DCT)scan including respiratory motion information could display a reliable tumor motion for treatment.On-board free-breathing cone-beam computed tomography(CBCT)can provide online internal target volumes(ITVs).The use of 4DCT combined with CBCT ensures the success of lung SBRT.With the advent of the era of systemic targeted and immunotherapy,many new requirements are raised,including fraction dose,segmentation mumble,coverage,side effects control,and so on.Therefore,we should comprehensively consider the influence of targetrelated and clinical factors on the therapeutic strategies in order to obtain maximum patient benefit.Most of previous studies on 4DCT and CBCT biased toward radiation physics technology.There is a lack of studies on the influence of clinical factors on the targets derived for 4DCT or CBCT.In this study,The tumor motion analysis,comparison of ITVs,and optimization of planning target volumes(PTVs)were conducted in a large sample.We focused on the influence of the target-related and clinical factors on tumor motion and construction of different targets,and put forward an optimized scheme for individualized determining targets of lung tumors based on 4DCT and CBCT images.ObjectiveFirstly,this study aimed to systematically analyze the influence of the target-related and clinical factors on tumor motion and to definite the tumor motion characteristics for different subgroups of patients.The information obtained may be valuable in reviewing the reliability of the ITV for individual patients,and providing motion data for generating individual IT Vs to patients with limited access to 4DCT scanning.The second aim was to evaluate the influence of the target-related and clinical features on differences of ITVs derived from 4DCT and CBCT.Furthermore,we tried to establish a predictive model in relation to the similarity of the targets based on the significant features.The availability of such information may contribute to a reasonable application of the ITVs derived from 4DCT MIP and CBCT images in clinical practice.Lastly,we comprehensively evaluated the inclusion relationship between the PTVs derived from 4DCT and the fused ITV derived all CBCT images,and analyzed the impact of target-related and clinical features on the inclusion degree.The ultimate aim was to establish an optimized scheme for detennining patient-specific targets of lung tumors based on 4DCT and CBCT images.MethodsWe firstly analyzed the data of 438 patients who underwent lung SBRT between May 2015 and December 2019 at Shandong Cancer Hospital and Institute.Different inclusion and exclusion criteria were set up based on the requirement of each part of the study.267 tumors of 246 patients were enrolled in the study of tumor motion analysis.When comparing the ITVs,we enrolled 210 tumors of 195 patients.Lastly,175 tumors of 167 patients were enrolled in the study of optimization of PTVs.For each patient,a conventional 3DCT scan of the thoracic region was performed,followed by a 4DCT scan during free breathing on a Brilliance Bores CT simulator.The first CBCT scan data was acquired for comparing the ITVs,and all CBCT scan data was obtained for optimization of PTVs.Gross tumor volumes(GTVs)were delineated based on 3DCT images,maximum intensity projection(MIP)and each of the 10 4DCT phases,and each CBCT images.All delineations were carried out by an experienced radiation oncologist using the same contouring protocol,and another oncologist reviewed all contours and rectified the contour if necessary.Target-related factors included the size,location and motion.Clinical factors included patient sex,age,body mass index(BMI),body surface area(BSA),Karnofsky Performance Scale(KPS),pulmonary surgery history,tumor origin,and so on.The influence of target-related and clinical factors on tumor motion and target variation were analyzed.Multiple linear regression models were used to explore the tumor motion-related risk factors.Akaike information criterion-based backward stepwise regression was used to identify important variables.While metastatic tumor and surgery were identified as protective factors,stratification analyses were used to explore the different effects of selected variables in patients with metastatic tumors and in surgery recipients.When comparing ITVs,we accessed the variation among IGTV-10,IGTV-MIP,and IGTV-CBCT derived from 4DCT ten phases,MIP,and the first CBCT images respectively.Multiple logistic regression models were used to explore the risk factors for dice similarity coefficient(DSC)of IGTV-MIP and IGTV-10 and for DSC of IGTV-CBCT and IGTV-10.Backward stepwise regression based on the Akaike information criterion was used to select important variables.Once the model was established,we used it to predict risk,and the effect of the prediction was presented using a receiver operating characteristic(ROC)curve and the area under the curve(AUC).In the part of optimized analysis of PTVs,we used a uniform 5 mm and a uniform 3mm margin to generate PTV-5mm and PTV-3mm based on 4DCT respectively.IGTV-CBn were obtained by contouring visible tumors on each CBCT images.ITV-CB were constructed by fusing all the CBCT-based IGTV.IGTV-10-CB1 was generated by fusing IGTV-10 and IGTV-CB1,which added a uniform 3mm margin to generate PTV-CB1-3mm.We compared the variations of different targets.Multiple logistic regression models were used to explore the risk factors for inclusion degree(ID),and used it to predict risk,and the effect of the prediction based the above method.ResultsFirstly,the mean(median)tumor motion amplitudes were 1.5±1.2(1.2),2.2±1.5(1.9),5.3±5.3(3.2),and 6.4±5.2(4.4)mm in the left-right(LR),anterior-posterior(AP),and cranial-caudal(CC),and three dimentional(3D)directions,respectively.We discovered that 95%of the tumors moved less than 3.7,5.1,16.4,and 16.7 mm in the LR,AP,CC,and 3D directions,respectively.The tumor segment location correlated to the tumor motion in each direction.Tumor size was predictive of tumor motion in the 3D(P=0.023)and AP directions(P=0.049).The tumor motion for metastatic tumors was smaller than that for primary tumors in the LR(P=0.019)and AP directions and(P=0.008).The CC motion for pulmonary surgery recipients was less than that for patients who had not undergone surgery,and no significant clinical factor was observed.BSA and BMI positively correlated with the motion in the CC(P=0.02)and LR direction(P=0.002).Secondly,the mean size ratios of the GTV-3D,IGTV-CBCT,and IGTV-MIP compared to the IGTV-10 for all patients were 0.63[0.52,0.76],0.71[0.58,0.82]and 0.82[0.70,0.89],respectively.The mean size ratio of the IGTV-CBCT to the IGTV-MIP was 0.91± 0.26.The size of the IGTV-10 was greater than the other target volumes(all P<0.001).The DSC of IGTV-MIP and IGTV-10 was 0.87[0.76,1.00],and the AUC value of the DSC of the IGTVMIP and IGTV-10 prediction models was 0.756.Female sex,greater BSA,and larger target size were protective factors,while the KPS,BMI,and motion were risk factors for the similarity between IGTV-MIP and IGTV-10.The DSC of IGTV-CBCT and IGTV-10 was 0.65[0.52,0.75],and the AUC value of the DSC of the IGTV-CBCT and IGTV-10 prediction models was 0.834,Older age and larger target size were protective factors,while adhesion to the heart,coexistence with cardiopathy,and tumor motion were risk factors for the similarity between IGTV-CBCT and IGTV-10.The median[IQR]ID of PTV-5mm,PTV-3mm,PTV-CB1-3mm to ITV-CB were 0.96[0.92,0.98],0.91[0.83,0.96]and 0.95[0.90,0.97],respectively.The median ID ITV-CB to PTV-5mm,PTV-3mm,PTV-CBl-3mm were 0.33[0.25,0.42],0.46[0.37,0.57],0.45[0.36,0.54].The performance of PTV-CB1-3mm was close to PTV-5mm,and it can reduce the side effect.The female,BSA,and≥6 fractions were risk factors for the DI of PTV-5mm to ITVCB,and the AUC value of the DI was 0.683.Tumor size and fractions were protective factors and lung surgery was a risk factor of the DI of ITV-CB to PTV-5mm,and the AUC value was 0.911.BSA,female,and≥6 fractions were risk factors of the DI of PTV-3mm to ITV-CB,and the the AUC value was 0.724.ConclusionIn this study,we found that the tumor segment location was a good predictive factor for the tumor motion.A larger tumor tended to have a smaller motion.Patients with metastatic tumors or those who have undergone pulmonary surgery showed a smaller and more unpredictable tumor motion.It is especially important to individually account for the tumor motion for these patients.BMI and BSA had a significant correlation to the tumor motion.Clinical factors combined with targeted-related factors can be used to predict the tumor motion for an individual.Secondly,the clinical factors should be considered when using MIP images for defining the ITV and when using CBCT images for verifying the treatment targets.The prediction models of the DSC of IGTVs derived from 4DCT and CBCT showed good predictive value,which may be useful in clinical setting.Lastly,PTV-5mm was able to include the real tumor range during treatment well,but the use of PTV-5mm would result in more normal tissue irradiated.PTV-CB1-3mm had a good inclusion of the real tumor range as well,meanwhile,the use of PTV-3mm would reduce normal tissue irradiated.PTV-CB1-3mm showed a good potential value in clinical setting.In addition,several clinical factors significantly impact on the inclusion effect of PTVs.The models of predicting the inclusion effect of PTVs established on targeted-related and clinical factors could be used to individually generate the optimal PTV.In brief,clinical factors should be considered when establishing the treatment targets based on 4DCT and CBCT images.The use of a combination of clinical factors and targetrelated factors contributes to generating individual treatment targets preferably.