Clinical And Basic Research On The Effect Of Vertebral Augmentation On Untreated Vertebra At Different Location Based On Texture Analysis And QCT/FEA

Part Ⅰ Clinical study of the association between vertebral augmentation and new symptomatic fractureObjective:Vertebral augmentation has been widely used in the treatment of osteoporotic vertebral compression fractures(OVCF)for many years,but whether it increases the risk of new fractures has been controversial.This study aims to explore the association between the vertebral augmentation and the new symptomatic fracture(NSF)with different distances from the augmented vertebra and different types at the vertebra-specific level,and analyze the risk factors for the NSF,so as to provide reference for the prevention of new fractures.Materials and Methods:1.The data of patients with NSF after vertebral augmentation in our hospital from May 2014 to April 2019 was investigated retrospectively.The untreated vertebrae located in T4-L5 segments of the patients meeting the inclusion and exclusion criteria were stratified according to the different distance from the nearest augmented vertebrae,and the proportion of NSF among the groups was compared.The life table was used to calculate and compare the 1-year and 5-year fracture-free probabilities of the untreated vertebrae at different distances from the nearest augmented vertebra.Kaplan-Meier method was used to draw and compare the risk curves of NSF of untreated vertebrae with different distances and Cox proportional risk model was used to analyze the risk factors of NSF.2.The data of patients with sandwich vertebra after vertebral augmentation in our hospital from May 2014 to April 2019 was analyzed retrospectively.According to the relative position to the nearest augmented vertebra,the untreated vertebrae were divided into sandwich vertebrae,ordinary-adjacent and non-adjacent vertebrae,and the ratio of NSF among the three groups was compared.The life table was used to calculate and compare the 1-year and 5-year fracture-free probabilities of the three groups.Kaplan-Meier method was used to draw and compare the risk curves of NSF of the three groups,and Cox proportional risk model was used to analyze the risk factors of NSF.Results:1.A total of 228 NSF events were reported in 162 patients,of which 56.6%(129/228)occurred within 1 year after procedures of augmentation.In 228 NSF events,a total of 432 augmented vertebrae and 2760 untreated vertebrae at T4-L5 levels were accumulated,and 273(9.89%)unenhanced vertebrae had NSFs.The proportions of cumulative NSF of the untreated vertebrae were 22.01%(125/568),10.62%(46/433),10.55%(40/379),6.69%(22/329)and 3.81%(40/1051)at 1,2,3,4,5 or more vertebrae away from the neatest augmented vertebrae,respectively,with significant differences(χ2=141.452,P<0.001).The fracture-free probabilities for the untreated vertebrae located 1,2,3,4,5 or more segments from the nearest augmented vertebra within 1 year after were 0.86,0.96,0.95,0.97,and 0.99,respectively,and the fracture-free probabilities within 5 years were 0.75,0.87,0.86,0.92,and 0.95,respectively.Kaplan-Meier method showed significant differences in NSF risk curves among untreated vertebrae located 1,2,3,4,5 or more segments from the nearest augmented vertebrae(χ2=150.525,P<0.001).Cox analysis showed that distance from the nearest augmented vertebra,thoracolumbar location,augmentation method and number of augmented vertebrae were risk factors for NSF.The relative risk(RR)of NSF of the untreated vertebrae located at 1,2,and 3 segments away was 3.989,1.967,and 2.258 times of that located at 5 or more segments away,respectively(P<0.05),while there was no significant difference in the risk of NSF between the untreated vertebrae located at 4 segments and that located at 5 or more segments away(P>0.05).In addition,the RR of NSF at the thoracolumbar region(T11-L2)was 2.322 times higher than that at the non-thoracolumbar region.The RR of NSF after percutaneous vertebroplasty(PVP)was 1.499 times than that after percutaneous kyphoplasty(PKP).The risk of NSF increased by 1.115-fold for each increase in the number of augmented vertebrae.2.A total of 125 patients with sandwich vertebrae were included in this study,including 1750 vertebrae at T4-L5 segments.Except for 342 augmented vertebrae,there were 147 sandwich vertebrae,307 ordinary-adjacent vertebrae,and 954 non-adjacent vertebrae.The NSF rates of sandwich vertebrae,ordinary-adjacent and non-adjacent vertebrae were 12.93%(19/147),6.19%(19/307)and 1.68%(16/954),respectively(χ2=49.591,P<0.001).The fracture-free probabilities of sandwich,ordinary-adjacent,and non-adjacent vertebrae within 1 year after augmentation were 0.89,0.95,and 0.99,respectively,and 0.85,0.92,and 0.98 within 5 years,respectively.Kaplan-Meier method showed significant difference in NSF risk curves among sandwich,ordinary-adjacent and non-adjacent vertebrae(χ2=52.976,P<0.001).Cox analysis showed that the type and location of the untreated vertebrae,the number of the augmented vertebrae and the puncture method were the important risk factors for NSF of different types vertebrae.The RR of NSF of sandwich vertebrae was 4.040 times than that of non-adjacent vertebrae and 2.325 times of ordinary adjacent than that of non-adjacent vertebrae.The RR of untreated vertebrae located at the thoracolumbar region(T11-L2)was 3.005 times higher than at the non-thoracolumbar region.The RR of NSF after augmentation with unilateral puncture was 1.925 times that with bilateral puncture.The RR of NSF increased 1.507 times for each increase in the number of augmented vertebrae.Conclusion:1.Vertebral augmentation increases the risk of NSF in untreated vertebrae at the vertebra-specific level.There was a significant increase in the risk of NSF in the untreated vertebrae within 3 segments on both sides of the augmented vertebra in dependence on the distance from the treated vertebra.In terms of the type of untreated vertebrae,the risk of NSF in sandwich vertebrae was higher than that of ordinary-adjacent vertebrae,risk of which in turn was higher than that of non-adjacent vertebrae.2.The types of untreated vertebrae,the distance between the untreated vertebrae and augmented vertebrae,thoracolumbar location,the number of augmented vertebrae,the augmentation method and the puncture method are the important risk factors for NSF.3.Since the fracture-free probability of sandwich vertebrae within 5 years after augmentation is still as high as 85%,prophylactic augmentation of all sandwich vertebrae is not recommended.However,for the sandwich vertebrae located in the thoracolumbar segments with a high risk of NSF,selective prophylactic augmentation can be used.Part Ⅱ The correlation between vertebral MRI texture features and vertebral bone microstructureObjective:To analyze the correlation between MRI texture features of vertebral body and bone microstructure,and investigate the feasibility of using vertebral MRI texture features to evaluate the bone microstructure of vertebral body Materials and Methods:The biopsy specimens of vertebral body that met the inclusion and exclusion criteria since 2020 were collected for micro-CT scanning and bone microstructure measurement,and the original vertebral MRI image data of patients were collected to extract the texture features of Tlweighted image(T1WI),T2weighted image(T2WI)and T2 weighted image and fat suppression(T2WI-FS).Bone microstructure indicators include:trabecular thickness(Tb.Th),trabecular spacing(Tb.Sp),bone surface(BS),total volume(TV),bone volume(BV),Volume fraction(BV/TV),Connectivity density(Conn.D),and degree of anisotropy(DA).The correlation between the texture features of different MR sequences and the bone microstructure indexes was analyzed.Results:A total of 35 qualified specimens were collected in this study,including 20 osteoporosis specimens and 15 non-osteoporosis specimens.Tb.Th were 241.83μm,Tb.Sp were 744.3μm,BV/TV were 0.2 in osteoporosis samples,and 280.9μm,611.8μm,0.3 in non-osteoporosis samples,respectively.There were significant differences between the two groups(p<0.05).However,there was no significant difference in Tb.Th Max,Tb.Sp Max,BS,BV,TV,Conn.D and DA between the two groups(P>0.05).Among the 1746 texture features in each sequence of T1WI,T2WI and T2WI-FS,there were 1013,899 and 812 texture features significantly correlated with bone microstructure,respectively(P<0.001).The texture features of T1WI had the highest correlation with TV,BV/TV,and the correlation coefficients were 0.57(P<0.001);The texture features of T2WI had the highest correlation with BV/TV,and the correlation coefficient was-0.57(P<0.001).The texture features of T2WI-FS have the highest correlation with Tb.Th Mean,and the correlation coefficient is-0.64(P<0.001).Conclusion:The vertebral texture features of T1WI,T2WI and T2WI-FS are correlated with the vertebral bone microstructure,which can indirectly reflect the bone microstructure.Among them,the correlation between texture features of T1WI and vertebral bone microstructure is more extensive and stable,which is more suitable for OVCF related studies.Part Ⅲ Impact of vertebral augmentation on MRI texture features of untreated vertebraeObjective:To investigate the changes in the microstructure of untreated vertebrae after augmentation by comparing the differences in MRI texture features of vertebral body before and after augmentation and those from patients receiving conservative treatment,and to seek evidence for the increasing risk of subsequent new fracture after augmentation.Materials and Methods:Data of patients diagnosed with OVCF who received vertebral augmentation or conservative treatment in our hospital from May 2014 to April 2019 were collected.Vertebral texture features of T1WI before and after treatment were extracted from the two groups respectively,and the texture features with differences before and after augmentation but no difference before and after conservative treatment were screened.The least absolute shrinkage and selection operator(LASSO)method was used to reduce the dimensionality of the omics data,and Adaptive Boosting(AdaBoost),Logistic Regression(LR)and support vector machine(SVM)models were constructed to distinguish the vertebrae of the augmentation group from the control group,and the specificity of texture feature alterations was determined.Results:A total of 178 patients meeting the criteria were analyzed,including 143 patients in the augmentation group and 35 patients in the control group.The age of the augmentation group was 74.8 years old and the interval between two MR examinations was 12.3 months,which was significantly different from 70.9 years old and 19.1 months interval of the control group(P<0.05).However,there were no significant differences between the two groups in gender,anti-osteoporosis treatment and vertebral proportion of each segment(P<0.05).A total of 1020 vertebrae meeting the criteria were included in the augmentation group and 249 vertebrae in the control group,and 1746 texture features were extracted from T1WI.Among them,1104 texture features altered significantly between the two MRI examinations in the augmentation group and 311 in the control group,and 873 texture features that altered in the augmentation group but did not change in the control group.In the augmentation group,there were 687,551,300,385 and 633 features significantly different from the two MRI examinations of the vertebrae with a distance of 1,2,3,4 and more than 5 segments from the treated vertebra,respectively.A total of 14 important features were obtained by LASSO dimensionality reduction,of which 10 features were changed between two MRI examinations in the augmentation group but not in the control group.The AUC of AdaBoost,LR and SVM models were 0.880,0.818 and 0.931 in the training set,and 0.816,0.832 and 0.842 in the validation set,respectively,indicating some characteristic alterations in the texture of the untreated vertebrae after augmentation.Conclusion:The study confirmed that the vertebral augmentation resulted in the changes of texture features associated with augmentation in the untreated vertebrae,suggesting that the augmentation resulted in the changes of the microstructure of the untreated vertebrae.These changes may be the basis for the increased risk of subsequent new fractures of untreated vertebrae.Part Ⅵ Impact of vertebral augmentation on biomechanicalparameters of unaugmented vertebrae at different locations basedon QCT/FEAObjective:To analyze the influence of bone mineral density(BMD)on the finite element results,compare the differences of biomechanical parameters of the untreated vertebrae with different distances from the augmented vertebrae before and after augmentation and explore the biomechanical basis of NSF after augmentation.Materials and Methods:Using data of quantitative computed tomography(QCT),vertebral compression fracture models under osteoporosis(OVCF model)and non-osteoporosis(NOVCF model)were constructed respectively,and the difference of bone strength between the two models was compared.On the basis of 150N vertical load applied to the model,1.5N·m torsional torque was applied in different directions to simulate forward flexion,back extension,left and right sides bend,left and right rotation.The overall displacement and stress changes of the two models in different states were compared,and the stress changes of the vertebral body,intervertebral disc and facet joints before and after bone cement implantation were compared between the two models.Results:1.The cortical bone strength of OVCF model decreased by 23.04%and cancellous bone strength decreased by 83.79%compared with that of NOVCF model.The overall displacement amplitude of OVCF model in each motion state was higher than that of NOVCF model,and the overall displacement amplitude of both models decreased after augmentation procedures.The preoperative stress of the fractured vertebra and its caudal adj acent vertebral body in the OVCF model was about 23%higher than that in the NOVCF model,and the stress of the proximal disc was about 8%higher on average.Except for L3-5 segments in flexion state,the stress of the left facet joint was about 4%and 5%in right higher than that of the NOVCF model.After vertebral augmentation,the stress of the five proximal vertebrae in the OVCF model was about 30%and 40%higher than that in the NOVCF model,the stress of the intervertebral disc was about 7%higher than that in the NOVCF model,the stress of the left facet joint was about 5%and the right side was about 6%higher than that in the NOVCF model,respectively.2.In the OVCF model,the stress of the five proximal vertebrae of the augmented vertebrae increased by about 20%on average compared with that before augmentation.In the cephalic direction of augmented vertebra,the stress of the first disc decreased by about 20%,while the stress of the second and third disc increased by about 4%and 2%,respectively.The stress of the facet joints of the augmented vertebra and its caudal adjacent vertebrae decreased by about 6%compared with that before augmentation.The stress of the facet joints of the other caudal facet joints and 4 pair of cephalic facet joints increased in varying degrees compared with that before augmentation.In both models,the stress changes decreased as the distance away from the treated vertebra increased.Conclusion:BMD will significantly affect the results of finite element analysis related to vertebral augmentation.Vertebral augmentation will lead to stress changes within 3-5 segments of vertebral augmentation,intervertebral discs and facet joints at both the cephalic and caudal sides of the augmented vertebra,and the stress changes decreased as the distance away from the treated vertebra increased.This stress change may account for the increasing risk of new fracture.