Optimization Of The Schedules Of Prophylactic Irradiation To Neck In Supraglottic Cancer

Part Ⅰ:Patterns and risk factors of lymph node metastasis in locally advanced supraglottic cancerObjective:To investigate the prevalence and distribution of cervical lymph node metastasis(LNM)in locally advanced supraglottic cancer(SGC)and guide the delineation of clinical lymph node target volumes.Methods:We reviewed patients defined as locally advanced SGC from January 2000 to December 2017 in our hospital.The primary tumor was operated on using partial or total laryngectomy,and all patients underwent bilateral neck dissection(levels Ⅱ-Ⅳ at least).Univariate and multivariate logistic regressions were used to find risk factors associated with LNM.Results:A total of 206 patients were enrolled.In the whole group,the rate of ipsilateral metastasis(IM)was 60.9%(67 patients),whereas contralateral metastasis was 25.5%(28 patients).Only positive ipsilateral lymph nodes contributed to contralateral metastasis(CM)(p=0.001).Seventy-six cases were diagnosed with clinically positive lymph nodes(cN+).IM of primary lesions mainly located within the unilateral sites(n=49 patients)was detected in levels Ⅱ,Ⅲ,and Ⅳ with lymph node metastasis ratios of 73.5%(36 patients),63.3%(31 patients),and 20.4%(10 patients),respectively,and CM of 36.7%(18 patients),16.3%(8 patients),and 6.1%(3 patients),respectively.Involvement of level Ⅱ or Ⅲ was associated with metastasis of level Ⅳ.No one developed contralateral level Ⅳ involvement without metastasis of contralateral levelsⅡ and Ⅲ.A total of 130 cases had clinically negative neck lymph nodes(cN0).The prevalence of occult metastasis(OM)was 35.4%.Among 62 patients with unilateral lesions,the rates of OM to ipsilateral neck levels Ⅱ,Ⅲ,and Ⅳ were 21,11.1,and 1.6%,respectively,whereas contralateral neck levels were 6.3,4.8,and 0%,respectively.In terms of the risk factors,histopathological differentiation was related to OM(p=0.003).Two of 25 people were with level Ⅵb metastasis,and both of them were with subglottic involvement.Conclusion:Neck levels Ⅱ to Ⅳ are most frequently involved and should be included in clinical target volume(CTV)in cN+patients.Contralateral Ⅳ may be omitted when contralateral levels Ⅱ and Ⅲ are negative.In cN0 patients,ipsilateral levels Ⅱ and Ⅲare suggested to be included in the CTV,whereas whether contralateral levels Ⅱ and Ⅲshould be included needs further research.Part Ⅱ:A radiomics prediction of occult lymph node metastasis in cN0 stage patients with supraglottic cancerObjective:We aimed to develop a radiomics Nomogram for predicting occult lymph node metastasis in patients with supraglottic cancer(SGC)and determine its validation.Methods:The training set included 131 patients with SGC gathered from 1/2010 to 12/2019 in our hospital,and the validation set included 108 patients in other hospital.All patients’ lymph node metastasis(LNM)status were firstly diagnosed as negative by computed tomography(CT)and ultrasound or magnetic resonance imaging of the neck,and later confirmed by postoperative pathology exam.The primary tumor regions were outlined in each patient’s CT image layer by two radiotherapists.Using the least absolute shrinkage and selection operator(LASSO)method to select radiomics features.The clinical risk factors associated with LNM were selected by correlation analysis.The multivariable logistic regression analysis was adopted to develop a radiomics model and a radiomics Nomogram combining radiomics score and clinical variables.Decision curve analysis was used to evaluate the clinical utility.The predictive performance of the two models was evaluated by the area under the curve(AUC).Results:Six radiomics features were selected to constitute radiomics signature,which was associated with occult metastasis.Clinical T stage(p=0.01)and pathology differentiation(p=0.02)were also significant associated with occult metastasis.The multiparametric radiomics Nomogram incorporate these factors and radiomics signature,which showing better prediction efficacy(AUC 0.808,95%CI:0.720～0.890)than radiomics model.(AUC 0.795,95%CI:0.710～0.872)in training set.Decision curve analysis showed our model achieved favorable discrimination and calibration.However,the area under curve yielded was 0.639(95%CI:0.528-0.750)in the validation set,which was less than satisfied.Conclusion:The newly developed radiomics Nomogram including radiomics signature,clinical T stage,and pathology differentiation had favorable individualized prediction of occult LNM in SGC patients of training set.However,the forecast performance of models applying to validation set was less satisfied,which need further optimization.In order to improve the accuracy of model prediction,we plan to enlarge the sample size and train a 3-dimensional(3D)residual convolutional neural network(rCNN)model by inputting 3D CT images,clinical risk factors and outputting occult metastasis.Part Ⅲ:Delineation of neck node levels for patients with locally advanced supraglottic cancer receiving radical IMRT:A CrossSectional Study in Mainland ChinaObjective:We aim to survey the diversity of lymph node clinical target volumes(CTV)for locally advanced supraglottic cancer(SGC)with radical radiotherapy in mainland China and provide bases for improving delineation consistency.Methods:Radiation oncologists from one provincial cancer hospital,one randomly chosen provincial general hospital and one randomly chosen municipal general hospital from each of the 30 provinces of mainland China participated.The study included four representative cases(T2N1,T3N2b,T4N0,T4N2c)of locally advanced SGC chosen from the following four different groups:non-T4,low nodal burden(T2-3N0-1);non-T4,high nodal burden(T2-3N2-3);T4,low nodal burden(T4N0-1);and T4,high nodal burden(T4N2-3).Respondents were asked which lymph node levels should be included in high-risk(HR)or low-risk(LR)CTV for nodal prophylactic irradiation.The impact of risk factors was also assessed.Results:Altogether,164 chief or attending physicians completed valid questionnaires from all 82 hospitals in China.The criteria that HR-CTV included the node levels with positive lymph nodes and the next lower adjacent level(83.8%-90%agreement)were followed by most physicians(n=160,97.6%).In the NO-1 stage(cases 1 and 3),ipsilateral levels Ⅱ and Ⅲ selected as HR-CTV and level Ⅳ as LR-CTV reached good agreement.Whether contralateral levels Ⅱ and Ⅲ should be included in HR-or LR-CTV remained controversial;more respondents were inclined to choose them as HR-CTV in case 3(61.3%).Some respondents supported including contralateral level Ⅳa in LRCTV(61.9%-68.1%).In the N2 stage(cases 2 and 4),bilateral levels Ⅱ-Ⅳb other than HR-CTV regions were all included in LR-CTV was indicated in most respondents(75%92.5%).Levels Ⅰb and Ⅴ were more likely included in CTV when there were multiple positive lymph nodes in the ipsilateral neck,and more respondents selected level Ⅴ as HR-CTV in case 4.Nearly half of respondents selected ipsilateral level Ⅵb as CTV when the subglottic region was involved(50.6%,46.2%and 56.2%in cases 2 to 4,respectively).Tumors crossing the midline(141,86%),extracapsular spread(132,80.5%),T stage(142,86.5%)and N stage(154,93.9%)as risk factors influencing nodal level selection were shown to have good agreement(≥80%).Conclusion:Most physicians selected involved nodal levels and lower adjacent levels as HR-CTVs in mainland China.Whether bilateral levels Ⅱ-Ⅳ are included in CTV reached relative consensus but poor agreement for HR-or LR-CTV.The selection conditions of levels Ⅰb,Ⅴa/b and Ⅵb as CTVs is no consensus.Part Ⅳ:Comparing different postoperative radiotherapy dose to lymph node levels for locally advanced supraglottic cancerObjective:To evaluate whether reduce postoperative radiotherapy(PORT)dose to lymph node levels is possible in supraglottic cancer(SGC).Methods:Patients with SGC were derived from two cancer centers,treated with primary site surgery,neck dissection,and PORT.In arm 1,the involved nodal levels(high-risk levels,HRL)and the next level received a dose of 60.06Gy/1.82Gy per fraction,while the other uninvolved levels(low-risk levels,LRL)received 50.96Gy/1.82Gy per fraction.In arm 2,all nodal levels received 50Gy/2Gy per fraction.The rates of high-risk levels control(HRC),regional control(RC),overall survival(OS),progression-free survival(PFS)and distant metastasis-free survival(DMFS)were estimated calculated by KaplanMeier method.The statistically significant difference between the two arms was evaluated using the log-rank test.The patterns of regional failure were evaluated based on the relevance of recurrent sites to clinical target volume and radiation dose.Results:In all,124 patients were included(62 in arm 1,62 in arm 2).Most patients(106,85.5%)had a stage T3/N+tumor,while more patients had stage T2N0 in arm 2(27.4%vs.1.6%).Nearly half of patients in arm 1 were N2 stage,while in arm 2 were NO stage.All patients were treated with selective neck dissection,of which 91.9%underwent bilateral neck dissection and a few patients received ipsilateral neck dissection in arm 1(10,16.1%).The rates of extracapsular spread(ECS)were 21%and 9.7%,and receiving concurrent chemotherapy were 43.5%,14.9%in arm 1 and arm 2,respectively.The median follow-up was 45 months(range 1-163 months).There were no significant differences in terms of OS(p=0.126),RC(p=0.514),PFS(p=0.195)and DMFS(p=0.834)between the two arms.Most regional recurrences(4,80%)occurred within three years of treatment,and all occurred within the target volumes.No regional failure occurred in HRL in arm 1,while three(3/4)failures occurred in arm 2.Dose reduction prescription to HRL led to a lower HRC rate(100%vs.90.6%,p=0.009).While the rates of LRL control(98.4%)were equal between the two arms.Conclusion:Compared with a standard dose(60Gy),the reduced dose(50Gy)to involved nodal levels showed inferior regional control for PORT,while uninvolved nodal levels showed equal outcomes.A dose of 50Gy for HRL may be an unfavorable treatment option for SGC.