Serial Clinical Studies And Computer-Assisted Clustering Analysis Of Congenital Scoliosis

Part ?:Serial Clinical Studies of Congenital ScoliosisChapter 1 Morphological study of hemivertebra in congenital scoliosis:vertebral body,posterior structure,and intraspinal anomaliesBackground:Hemivertebra(HV)is one of the most prevalent vertebral malformations in congenital scoliosis(CS).The natural history of CS caused by HV varies greatly.The controversies about the indication and timing of surgical intervention remain to be solved.Little is known about the influence of geometric parameters of isolated(single fully segmented)HV on the progression of CS.Existing studies about HV mainly focused on the segmentation pattern of the anterior component and hardly analyze the morphology of the posterior structure.Previous studies have reported considerable rates of intraspinal anomalies(ISA)in patients with CS caused by single hemivertebra(SHV).However,whether the presence of ISA alters the clinical spectrum of SHV related CS is unknown.Objectives:To explore the correlation between the morphology of HV itself and the magnitude of deformity to identify risk factors of progression;to analyze the posterior structure and anteroposterior(AP)relationship of HV from a three-dimensional(3D)perspective and propose detailed classification;to review CS patients with SHV at the apex of the major curve and compare those who had ISA(ISA+)with those who had normal intraspinal structure(ISA-).Methods:Patients treated at our center with a definite diagnosis of CS,an HV induced major curve and complete clinical and radiographic data were retrospectively enrolled.For isolated HV,three morphological parameters were measured and compared.Firstly,the HVs were categorized into two types according to whether the width extends across the"midline",meaning the central vertical line of lower adjacent vertebra.Secondly,lateral height ratio(LHR)from HV-1 to HV+1 was measured and calculated by 2XHV/(HV-1+HV+1),with the cut-point being 0.9.Finally,HV's sagittal position was divided into lateral and posterolateral groups according to whether the HV extended ventrally to the anterior half of the column.Osseous deformities of vertebral body and lamina were reviewed on 3D-CT.According to the segmentation,number,and AP discordance of HV,a detailed classification was developed.The type,number,and location of ISA were identified on MRI.Comparisons of demographic and radiographic characteristics were performed between ISA+and ISA-groups.The analysis of covariance was performed to adjust for the influence of age when necessary.Results:A total of 215 patients with CS caused by HV were enrolled.According to the inclusion and exclusion criteria,eligible cases were allocated to each subpart of the present study as appropriate.After being screened by radiography,132 cases of isolated HV were identified(mean age 10.0±6.4 years,70 males).After adjusting by age,HVs across the midline had significantly larger scoliosis,kyphosis and apical vertebral translation(AVT)compared to those within midline(59.1 ± 20.7° vs.44.7 ± 16.6°,p<0.001;46.4 ± 32.6°vs.31.2± 27.0°,p=0.023;40.1 ± 17.3 mm vs.32.5 ± 14.2 mm,p=0.033,respectively).HV with LHR?0.9 was associated with larger scoliosis(56.8±20.6° vs.43.0 ± 15.4°,p<0.001).Compared to lateral HV,significantly larger kyphosis and AVT was observed in posterolateral HV after adjusting by age(52.9±31.8° vs.20.8±16.9°,p<0.001;40.0 ±16.7 mm vs.31.4 ± 14.3 mm,p=0.039,respectively).A total of 153 HVs from 116 patients(mean age,10.9±7.0;61 males)who had high-quality CT images were reviewed.There were 85 cases of single HV and 31 cases of multiple HVs.The AP relationships of the enrolled HVs were classified into 4 types.Type 0(n=57):AP unison group,the morphology and segmentation pattern of anterior and posterior structures were consistent.Type 1(n=58):mild AP discordance:inconsistency of the morphology or segmentation pattern between anterior and posterior structures without mismatch phenomenon.Type 2(n=28):moderate AP discordance:bi-segmental mismatch between anterior and posterior structures.Type 3(n=10),severe AP discordance:long segmental(?3)mismatch with extensive malformations of posterior structures.No significant difference in radiographic parameters was identified between the 4 groups.In patients with CS caused by single HV,129 patients with complete MRI images were reviewed for ISA(mean age 12.2 ± 6.4 years,61 males).ISA was observed in 37(28.7%)patients,including split cord malformation in 21 cases(16.3%),syringomyelia in 19(14.7%),and low-lying conus in 19(14.7%).ISA+group had a significantly higher percentage of female,larger cobb angle of scoliosis and more malformed vertebrae than ISA-group(67.6%vs.46.7%,p=0.032;63.0 ± 26.8° vs.50.6 ± 20.9°,p=0.006;6.3 ± 3.3 vs.3.2 ± 2.8,p<0.001,respectively).The co-occurrence of rib deformity and contralateral bar were more prevalent in ISA+group(67.6%vs.21.7%,p<0.001;45.9%vs.13.0%,p<0.001,respectively).Conclusions:HVs with a width across the midline,an LHR? 0.9,and a posterolateral position were associated with a more severe deformity in CS caused by isolated HV.HVs could be classified into 4 types according to AP relationship:1)type 0,AP unison;2)type 1,mild AP discordance;3)type 2,moderate AP discordance;4)type 3,severe AP discordance.Type 2 and 3 are characterized by the mismatch between the vertebral body and the lamina.In CS caused by single HV,the incidence of ISA was 28.7%.The ISA+group had more complex deformity and a higher proportion of females,rib deformity,and contralateral bar.Chapter 2 Surgical intervention in patients with early-onset congenital scoliosis:influences of traditional growing rod technique on the growth of normal vertebrae and unsegmented levelsBackground:Previous studies indicated that the dual growing rods(DGR)were likely to preserve or even stimulate spinal growth.However,reports about the effect of the growing rod technique on the growth of an individual vertebral body were rare.Existing studies indicated that the concave side of unsegmented levels(USLs)in patients with congenital scoliosis(CS)had limited growth potential.Whether the traditional single growing rod(TSGR)can alter the natural history of early-onset mixed-type congenital scoliosis(EOMTCS)remains unknown.Objectives:To investigate the growth of the vertebrae around distal instrumented vertebrae(DIV)in patients with early-onset congenital scoliosis(EOCS)who underwent DGR,and to evaluate the growth of unsegmented levels(USLs)in EOMTCS patients who underwent TSGR.Methods:The EOCS patients treated with GR who received at least 4 lengthenings were enrolled.The demographic data of enrolled patients were retrospectively collected.Spine radiographs before index surgery and at the latest follow-up were evaluated.For patients treated with DGR,the height,width,and depth of the vertebral body from DIV-2 to DIV+2 were measured.For patients treated with TSGR,the length of the concave and convex side of USLs and thoracic parameters were measured.The absolute growth and percent growth(absolute growth/baseline value*100)were calculated and compared.Results:Twenty-one patients treated with DGR(mean age,5.8 ± 2.5 years old)met the inclusion criteria,and 66.7%(14/21)of them were female.The mean follow-up duration was 6.9 ± 1.6 years.An average of 7.7 ± 2.5 lengthenings were performed for each patient.The measured vertebrae were divided into DIV group(n=40),DIV-group(DIV-1 and DIV-2,n=44)and DIV+group(DIV+1 and DIV+2,n=29).The total percent growth of vertebral height was significantly higher in DIV-group than that in DIV and DIV+groups(59.5 ± 16.4%vs.47.5 ± 15.0%and 46.1 ± 15.8%,p=0.003 respectively).The vertebrae in DIV-group also had the highest annual height growth rate(8.9±2.7%vs.7.1±2.5%and 6.9 ± 2.2%,p=0.003 and 0.005 respectively).The vertebrae around DIV became"slimmer"(an increase of height/width ratio)after DGR treatment,and this phenomenon was more obvious in the DIV-group(p<0.001).Fourteen patients(mean age 7.3±2.8 years,8 females)treated with TSGR were enrolled.The mean follow-up duration was 4.9±1.2 years.An average of 6.0 ± 2.0 lengthenings were performed for each patient.The number of USLs ranged from 4 to 10 segments.At the latest follow-up,the total and annual percent growth at concave side of USLs was significantly higher than at convex side(40.0± 13.5%vs.20.3 ± 12.8%,p<0.001;8.5 ± 3.1%vs.4.3± 2.5%,p<0.001,respectively);the average correction rate of the major curve was 27.3 ± 13.4%;the annual growth of T1-S1I height was 16.5± 6.3 mm;the Campbell's space available for lung ratio showed a significant increase(from 74.9± 11.1%to 89.6±7.0%,p<0.001).Conclusions:Traditional DGR with periodical distraction stimulated the longitudinal growth of the two segments immediately above DIV in patients with EOCS.Unilateral repetitive lengthening with TSGR can accelerate the growth of the concave side of USLs and improve the symmetry of the thorax.Part ?:Computer-assisted morphological study and clustering analysis of congenital scoliosisBackground:In the research field of scoliosis,classification studies about adolescent idiopathic scoliosis and adult spinal deformity are featured with a larger quantity and faster update.Concerning CS,the development and application of classification are relatively falling behind.In clinical practice,the classification proposed by Winter et al.fifty years ago as formation failure,segmentation failure,and mixed type is still widely used.This classification was based on radiograph findings of vertebral malformation and failed to describe the general morphology of the whole spine.Besides,it did not contain information from CT and MRI.Therefore,it is essential to develop a new clinical classification system that integrates radiograph,CT,MRI,and clinical data.Duo to the complexity of morphological changes in CS,traditional manpower-based analysis of imaging data has limited efficiency.Using computer technology as an enhancement of human performance,to conduct radiographic analysis and to explore the morphological pattern of CS by clustering algorithm is of great importance in developing a classification system for CS.Objective:To explore the morphological pattern of CS by performing radiographic data analysis with the aid of computer science,building a standardized image-omics database of CS,and developing a clustering analysis algorithm.To synthesize the results of clustering analysis,literature research,and group decision-making,and to establish an innovative clinical classification system for CS with pilot validation.Methods:Retrospectively enrolled patients with a definite diagnosis of CS,meaning the existence of congenital bony malformation of the vertebra and/or rib that can be identified on images.Besides,the bony malformation should be the primary cause of the scoliosis.Baseline data and imaging data were collected and a standardized and structuralized database was built.A systematic literature review was performed to identify the widely accepted classification systems of scoliosis,of which the core concept and parameters were extracted.After panel discussions and expert consultation,candidate parameters were determined.For parameters based on the radiograph,measurement and subsequent clustering analysis were performed to develop the core classification module.For parameters related to clinical data,CT and MRI,the corresponding classification modules were developed according to previous reports,current consensus,and the manifestations of the enrolled patients.A multi-rater and multi-measurement approach was applied to evaluate the reliability of the new clinical classification system.Results:A total of 352 patients with CS were enrolled(mean age 13.2±6.1 years,149 males).Thirteen classification systems were identified from previous studies,from which six groups of core parameters were extracted,being summarized as follows:1)age and growth potential;2)the number,location,and shape of the curvature;3)sagittal alignment of the spine;4)global balance in both coronal and sagittal plane;5)the type and extent of vertebral malformation;6)intraspinal anomalies and the function of the spinal cord.For the first group of parameters,a comprehensive classification module was developed,which comprising 5 groups:1)EO group,age<10 years;2)AA group,10?age<19,Risser=0,open triradiate cartilage;3)AB group,10 ? age<19,Risser=0 to 1,closing or closed triradiate cartilage;4)AC group,10?age<19,Risser? 2;5)AD group,age? 19.Concerning the second to fourth groups of parameters,a radiograph-based computer-assisted clustering module was developed,which comprised 3 groups:type I,angular scoliosis/scoliokyphosis(failure-of-formation dominance);type II,arcuate scoliosis(failure-of-segmentation dominance);type III,severe kyphoscoliosis(both the failure of formation and failure of segmentation being major teratogenic factors).As for the fifth group of parameters,a CT-based scoring system of the asymmetry of spinal growth in the major curve was established,which comprised 4 groups:1)S group,score ?2;2)M group,2<score?6;3)L group,6<score?10;4)X group,score>10.Finally,for the sixth group of parameters,a grading system of spinal cord function was developed according to MRI and physical examination,which comprised 3 groups:1)N group,no ISA,no symptoms or signs;2)+group,ISA identified on MRI,no symptoms or signs;3)++group,positive neurological symptoms and/or signs can be observed with or without ISA.The four above modules constitute the new clinical classification system of CS.Ths system showed moderate interobserver reliability(Fleiss Kappa,0.526-0.709)and substantial intraobserver reliability(Cohen Kappa,0.524-0.813).Conclusions:A new clinical classification system of CS was developed,which consists of four modules:1)prefix module,the comprehensive classification of age and growth potential;2)primary core module,the radiograph-based computer-assisted clustering;3)secondary core module,the CT-based scoring system of the asymmetry of spinal growth in the major curve;4)add-on module,the grading system of spinal function.This classification integrates information from baseline data,radiograph,CT,and MRI and covers all related core parameters.With moderate to substantial reliability,it can be applied in clinical practice.This classification provides a meaningful reference and template for future studies.