| Part Ⅰ:The general clinical features and factors associated with disease severity of Chinese patients with multiple system atrophyObjective:Multiple system atrophy(MSA)is a neurodegenerative disorder characterized by a variable combination of autonomic failure,parkinsonism,cerebellar ataxia,and pyramidal signs.According to foreign epidemiological studies,the annual incidence rate is about 0.11-3/100,000 and the prevalence rate is about 1.9-4.9/100,000.MSA is a rare disease,which has been included in the list of rare diseases of China.The etiology of the disease is unknown,and it develops gradually after the onset.There is no effective treatment.The mean average survival time of MSA patients is only about 6-10 years.The clinical heterogeneity is large,patients onset with autonomic dysfunction,cerebellar symptoms or Parkinsonism.Depending upon the predominant motor symptom,MSA can be classified into the MSA with predominant parkinsonism(MSA-P)subtype and MSA with predominant cerebellar ataxia(MSA-C).The rate of disease progression is inconsistent,which brings challenges to the diagnosis and prognosis of the disease.In addition to the main motor symptoms,non-motor symptoms were common in MSA.Previous studies have reported that some hematological indicators(such as blood lipid and uric acid)and genes(such as SNCA,COQ2,and GBA)were associated with the risk of MSA.In addition,"hot cross bun sign",presence of a putaminal hyperintense rim,pontine atrophy and putamen atrophy were important for the differential diagnosis of MSA.However,the relationship between these non-motor symptoms,hematological indicators,genetic and imaging features and the severity of the disease remain unclear.At present,most of the studies on the clinical characteristics of MSA come from abroad,and there is no complete large-scale clinical study on the clinical characteristics of MSA in China.Therefore,the current study is proposed to explore the clinical characteristics of Chinese MSA patients and their relationship with the severity of the disease from multiple dimensions such as clinical features,hematological indicators,imaging,and genetics.Materials and Methods:The cross-section study was used in this chapter.In this study,patients who were diagnosed with probable MSA in the Department of Neurology,West China Hospital of Sichuan University were included.At the same time,healthy people with age,gender matching and more than 3 years of education were collected as health control group(HC).The general data of patients were collected and Unified multiple system atrophy rating scale(UMSARS)was used to assess the severity of the disease.Rapid eye movement sleep behavior disorder screening questionnaire(RBDSQ),Fatigue severity scale(FSS),Parkinson’s disease sleep Scale-2(PDSS-2),Hamilton Depression Rating Scale-24(HDRS-24),Hamilton Anxiety Rating Scale(HARS),and Montreal Cognitive Assessment(MoCA)was used to evaluate the rapid eye movement sleep behavior disorder(RBD),fatigue symptoms,Parkinson’s disease related sleep disorders,depression symptoms,anxiety symptoms,and cognitive function.MOCA scale was also used to evaluate the cognitive function of HCs.Fasting peripheral blood was collected from patients,and blood routine,biochemical indicators,C-reactive protein(CRP),and immune indicators collected.Genomic DNA was extracted from peripheral blood leukocytes using standard phenol-chloroform procedures and then genotyped on the Illumina Infinium Asian Screening Array-MD v1.0 for a total of～0.66 million single nucleotide polymorphisms(SNPs)via standard protocol.Imputation was performed using Minimac4 with reference panel Genome Asia Pilot after phasing with Eagle v2.4.According to the previous literature,we screened out the following SNPs associated with MSA:GBA rs75822236,rs421016,rs1064651,rs76763715,rs75548401,rs439898;SLC1A4 rs759458;COQ2 rs148156462,rs751185256,rs121918231;SNCA rs11931074,rs3857059;TNF rs1799964;DDC rs921451;LRRK2 rs17466213,rs1427263,rs33995883.In addition,magnetic resonance imaging(MRI)was performed in all MSA patient inorder to exclude other diseases.Among them,255 patients with MSA who had acquired T1 weighted images,T2 weighted images and T2 magnetic resonance imaging fluid attenuated inversion recovery(FLAIR)sequences on the MRI machine of our hospital were studied for structural image severity classification.Seven specific MRI features of MSA were collected,including pontine atrophy,cerebellar atrophy,middle cerebellar peduncle atrophy,putamen atrophy,putamen hypointensity,presence of a putaminal hyperintense rim,and "hot cross bun sign".According to the severity of these features,they were divided into four grades:0=normal,1=mild abnormality,2=moderate abnormality and 3=severe abnormality.The total score of imaging features was the sum of the above 7 MSA special MRI feature grades.According to the diagnosis subtype(MSA-P and MSA-C),gender(male and female),symptom of onset(autonomic symptoms and motor symptoms)were divided into several subgroups,and the general clinical characteristics of subgroups were compared and analyzed.T test,Mann Whitney U rank sum test or chi square test were used to compare the differences among the groups in demographic data,clinical data,hematology data,genetics data and imaging data.Pearson correlation test or Spearman correlation test were used to analyze the correlation.The related factors of disease severity were analyzed by univariate and multiple linear regression model.In order to explore the gene loci related to the disease severity of MSA,we used Plink software to carry out genome wide association study(GWAS)on 458 MSA patients.The general linear regression model was performed to analyze the association between the disease severity of MSA and genotypes.Gender,age,disease duration and five principal components were adjusted.Results:1.General clinical features:A total of 458 patients with MSA were included in this study.The mean age of MSA patients was 59.81±8.54 years,the mean age of onset was 57.17±8.54 years,the median age of onset was 57.68 years,and the mean disease duration was 2.57±1.58 years.Male accounted for 48.9%and female 51.5%.Two hundred and thirteen patients(46.5%)were MSA-P and 245 patients(53.5%)were MSA-C.One hundred and sixty-eight patients(36.75%)onset with autonomic symptoms and 290 patients(63.3%)onset with motor symptoms.34.9%(160/458)of MSA patients had OH and 66.6%(305/458)had urinary incontinence.The mean total score of UMSARS was 36.26±14.36.35.8%of MSA patients had history of smoking,33.4%had history of drinking,19.4%had hypertension,12.7%had hyperlipidemia,and 10.9%had diabetes.72.1%of MS A patients were complicated with symptom of anxiety,68.7%with symptom of depression,65.5%with fatigue,57.9%with RBD,and 19.2%with Parkinson’s disease-related sleep disorder.55.9%patients with cognitive impairment,as abstract thinking(74.2%),delayed recall(61.0%),visuospatial and executive functions(60.5%)were the most frequently in cognitive domains.Subgroup analysis showed that compared with MSA-C patients,MSA-P patients had older age(61.24±8.72 vs 58.56±8.19,P=0.001),older age of onset(58.26±8.87 vs 56.22± 8.15,P=0.010),longer disease duration(2.84±1.76 vs 2.34±1.37,P=0.001),higher proportion of onset of motor symptoms(68.1%vs 59.2%,P=0.049)and urinary incontinence(71.8%vs 62.0%,P=0.027),while the proportion of OH was lower(28.6%vs 40.4%,P=0.008).After adjusting for age and disease duration,there was no significant difference in UMSARS-Ⅱ score and total UMSARS score between MSA-P and MSA-C patients.In addition,the scores of PDSS-2(12.63±6.78 vs 10.82±7.25,P=0.018)and HARS(10.95±7.46 vs 9.01±6.04,P=0.004)in MSA-P patients were higher than those in MSA-C patients,and the proportion of symptom of depression in MSA-P patients(76.5%vs 63.7%,P=0.023)was significantly higher than that in MSA-C patients.Male MSA patients had higher proportion of OH(39.7%vs 30.3%,P=0.035),history of smoking(69.2%vs 3.8%,P<0.001)and history of drinking(65.2%vs 3.0%,P<0.001)than female MSA patients.Compared with MSA patients with onset of autonomic symptoms,the proportion of MSA-P subtypes in MSA patients with onset of motor symptoms was higher(68.1%vs 31.95%,P=0.049),and the proportion of OH was lower(31.0%vs 50.0%,P<0.001).After univariate linear regression analysis,we found that age,diagnosis subtype,disease duration,OH and urinary incontinence were correlated with the total score of UMSARS.Multiple linear regression model further showed that age(P=0.024),disease duration(P<0.001),OH(P<0.001),and urinary incontinence(P<0.001)were related to the total score of UMSARS.In addition,the univariate linear regression analysis showed that total score of FSS,total score of PDSS-2,total score of HDRS-24,total score of HARS,and total score of MoCA were correlated with total score of UMSARS.Multiple linear regression model showed that the total score of FSS(P=0.003),PDSS-2(P=0.020),HDRS-24(P<0.001),HARS(P=0.003)and MoCA(P=0.001)were correlated with the total score of UMSARS,besides the disease duration,OH,urinary incontinence and age.2.Hematological features:A total of 198 MSA patients and 158 HCs who completed the hematological test in our hospital were further analyzed.The mean age of MSA patients was 59.93±8.33 years,and the mean disease duration was 2.52±1.46 years.Compared with the HC group,the MSA group had significantly lower erythrocyte count(P<0.001),hemoglobin(P<0.001),albumin(P<0.001),globulin(P<0.001),cholesterol(P<0.001),high-density lipoprotein(P<0.001)and low-density lipoprotein(P<0.001),but higher platelet count(P=0.004),leukocyte count(P=0.037)and absolute value of neutrophils(P=0.004).In addition,the IgG level of MSA patients(P<0.001)was lower than that of HCs,while the proportion of CD3+(P=0.003)and CD4+(P<0.001)T cells was higher than that of HCs.There was no significant difference in IgA,IgM,CD8+%and CD4+/CD8+between the two groups.The proportion of patients with a CRP level above the normal range was 0.5%.After adjustment of diseae duration,we found that the globulin level(25.12±3.59 vs 26.49±4.09,P=0.008)and triglyceride level(1.13±0.48 vs 1.46±0.80,P=0.001)in MSA-P patients were lower than that in MSA-C patients.There was no significant difference in the remaining hematological indicator between the two subtypes.The results of univariate linear regression analysis showed that the absolute value of neutrophil,albumin,CRP,and CD3+T cells were correlated with the total score of UMSARS.There was no significant correlation between other hematological data and the total score of UMSARS.The leukocyte count,absolute value of neutrophils,albumin,CRP,CD3+T cells,and CD4+T cells were included into the multiple linear regression model,and the disease duration,age,OH and urinary incontinence were adjusted.The results showed that only CRP level(P=0.011)was related to the total score of UMSARS3.Genetic features:A total of 458 MSA patients were included for genetic analysis.The results showed that the proportion of MSA-P(43.1%vs 54.2%,P=0.028)and cognitive impairment(52.4%vs 64.5%,P=0.026)in MSA patients with SNCA rs11931074 "GT+GG" genotype was lower than MSA patients with the "TT" genotype.However,after adjusting for years of education,there was no significant difference in the proportion of MSA patients with cognitive impairment between the two groups(P=0.118).The proportion of MSA-P(42.9%vs 54.1%,P=0.025)in MSA patients with SNCA rs3857059 "AG+AA" genotype was lower than that in MSA patients with "GG"genotype.There were no significant differences in general clinical features,UMSARS total score and non-motor symptoms among the other SLC1A4 rs759458,COQ2 rs148156462,TNF rs1799964,DDC rs921451 and LRRK2 rs1427263 genotypes.In order to explore the gene loci related to the disease severity of MSA,we performed GWAS to analyze 458 MSApatients by linear regression model on Plink software.The genomic inflation factor(GIF)λ was 1.013,indicating that there was no population drift.We found that there was no significant correlation between the gene loci and the severity of the disease(P<5E-8).The results of univariate linear regression analysis showed that the genotypes of SLC1A4 rs759458,COQ2 rs148156462,SNCA rs11931074,SNCA rs3857059,TNF rs1799964,DDC rs921451 and LRRK2 rs1427263 were not associated with the total score of UMSARS.4.Imaging features:A total of 255 MSA patients were included in the imaging analysis(mean age 60.20±8.72 years,mean disease duration 2.57±1.53 years).The proportion of mild,moderate and severe atrophy of pontine was 26.3%,20.4%and 4.3%,respectively;the proportion of mild,moderate and severe atrophy of cerebellum was 42.7%,33.7%and 6.3%,respectively;the proportion of mild,moderate and severe atrophy of middle cerebellar peduncle was 27.5%,45.9%and 14.1%,respectively;The proportion of mild,moderate and severe atrophy of putamen was 49.0%,10.6%and 2.4%,respectively;the proportion of mild,moderate and severe hypointensity of putamen was 29.8%,19.6%and 9.0%respectively;the proportion of mild,moderate and severe putaminal hyperintense rim was 22.4%,3.9%and 1.2%,respectively;the proportion of mild,moderate and severe "hot cross bun sign" was 26.7%,15.7%and 4.3%,respectively.Compared with MSA-P patients,MSA-C patients had a higher proportion of pontine atrophy(P<0.001),cerebellar atrophy(P<0.001),middle cerebellar peduncle atrophy(P<0.001)and moderate to severe abnormality of "hot cross sign"(P<0.001).Patients with MSA-P had a higher proportion of putamen atrophy(P=0.011),hypointensity of putamen(P=0.012),moderate to severe abnormality of presence of a putaminal hyperintense rim(P=0.007)than that of patients with MSA-C.The seven specific MRI features of MSA and the total score of imaging features were correlated with the total score of UMSARS.Multiple linear regression model showed that in addition to the disease duration,OH,urinary incontinence,age and the total score of imaging features(P<0.001)was significantly correlated with the total score of UMSARS.Conclusion:The current study is the first time to explore the clinical characteristics of Chinese MSA patients and their relationship with disease severity from a multi-dimensional perspective based on the clinical cohort of large sample MSA patients,combined with detailed clinical,hematology,genetics,imaging features.The conclusions are as below:1)The median age of Chinese patients with MSA was 57.68 years,MSA-C is slightly more than MSA-P,and patients with motor symptoms onset were more than patients with autonomic symptoms onset.Hypertension was the most common complication of MS A,followed by hyperlipidemia and diabetes mellitus.Patients with longer disease duration,higher proportion of OH and urinary incontinence,and older age had more greater disease severity.2)More than half of the MSA patients were complicated with fatigue,symptom of anxiety,symptom of depression,sleep disorder and cognitive impairment.The severity of fatigue,sleep disorders,depression,anxiety and cognitive impairment were related to the severity of the disease.3)The level of serum lipids,albumin,hemoglobin and IgG in MSA patients were lower than HC,while the leukocyte count,CD3+%and CD4+%were higher than HC,but only CRP was related to the severity of the disease.4)The proportion of MSA-P patients carrying SNCA rs11931074 "TT" genotype was higher than "GT+GG" genotype;the proportion of MSA-P patients carrying SNCA rs3857059 "GG" genotype was higher than "AG+AA" genotype.SLC1A4 rs759458,COQ2 rs148156462,SNCA rs11931074,SNCA rs3857059,TNF rs1799964,DDC rs921451 and LRRK2 rs1427263 were not associated with the disease severity.5)The MRI features of MSA-P patients mainly showed putamen atrophy,hypointensity of putamen and presence of a putaminal hyperintense rim,while those of MSA-C patients were mainly cerebellar atrophy,middle cerebellar peduncle atrophy,pontine atrophy and "hot cross sign".The total score of imaging features of MSA patients was correlated with the disease severity.Part Ⅱ Related factors of disease progression of multiple system atrophyObjective:Due to the rapid progress and no effective drug for MSA at present,it is particularly important to clarify the law of disease development and find the markers of disease progression.However,the clinical heterogeneity of MSA patients is large,although most patients progress quickly,some patients progress slowly.Judging disease progression is very important for clinical management and decision-making of patients.Previous studies were mostly limited to the relationship between clinical features and disease progression.In recent years,blood biomarkers have gradually become a hot spot,such as α-synuclein which is the important component of glial cytoplasmic inclusions(GCIs)as the pathological markers of MSA,neurofilament light chain protein(NFL)that is an important structure for maintaining axonal stability and growth,astrocyte and microglia mediated neuroinflammatory factors,etc.In view of the limitations of cross-sectional studies and the lack of domestic research on the progress of MSA disease.On the basis of the first part,the longitudinal cohort study is used to further explore the relationship between the disease progression and clinical features,hematological data,blood biomarkers,genetics and imaging,so as to find biomarkers that can early monitor the disease progression.Materials and Methods:The longitudinal cohort study was used in this chapter.Patients enrolled in the first part were followed up regularly every year(face-to-face evaluation).A total of 146 MSA with a baseline disease duration of less than 3 years and a one-year follow-up were included in this study.At the same time,healthy people matched with MSA in age and sex were recruited as controls.The levels of α-synuclein,glial fibrillary acidic protein(GFAP),brain-derived neurotrophic factor(BDNF)and triggering receptor expressed on myeloid cell-2(TREM2)were determined by ELISA.NFL was quantified using an ultra-sensitive Simoa technology.The collection of clinical data and evaluation of specific scales,and the detection methods of blood routine,biochemical indicstors,CRP and blood immune indicators,risk gene loci of MSA,and the collection of imaging data are the same as the first part.The annual progression rate of UMSARS total score=(1-year follow-up UMSARS total scorebaseline UMSARS total score)/year,representing the disease progression rate.The annual progression rate of FSS total score=(1-year follow-up FSS total scorebaseline FSS total score)/year,representing the progression rate of fatigue symptoms.T-test,Mann Whitney U-rank sum test or chi square test were used for comparison between groups.The univariate and multiple linear regression model was used to analyze the related factors of MSA disease progression.Results:1.The relationship between general clinical features and disease progression of MSAThe average age of 146 MSA patients was 59.53±7.50 years,and the average age of onset was 57.43±7.34 years,the average disease duration was 1.90±0.84 years.There were 69 patients with MSA-P,77 patients with MSA-C.72 patients were male and 74 patients were female.The average total score of UMSARS was 30.82±10.86.We found that the total score of UMSARS after one-year follow-up in MSA patients was significantly higher than that in baseline(43.32±14.24 vs 30.82±10.86,P<0.001),and the proportion of OH and urinary incontinence was also increased,but there was no significant difference compared with baseline.After one-year follow-up,the severity of symptom of fatigue in MSA patients was significantly worse than that in baseline(44.28±18.53 vs 37.64±19.17,P=0.003).There was no significant difference in the severity of other non-motor symptoms such as RBD,Parkinson’s disease-related sleep disorder,symptom of depression,symptom of anxiety and cognitive impairment between one-year follow-up and baseline.In addition,the proportion of symptom of fatigue(73.3%vs 60.3%,P=0.018)and symptom of depression(71.2%vs 60.3%,P=0.048)in MSA patients after one-year follow-up was significantly higher than that at baseline.The univariate linear regression model showed that the diagnosis subtype,age,and OH were significantly correlated with the annual progression rate of total score of UMSARS;the total score of baseline FSS and the total score of HDRS-24 were correlated with the annual progress rate of total score of FSS.The multiple linear regression model further showed that only baseline OH(P=0.034)was significantly correlated with the annual progression rate of total score of UMSARS and the total score of baseline FSS(P<0.001)was significantly correlated with the annual progression rate of total score of FSS.2.The relationship between baseline hematological indicators and disease progression of MSAAmong the 146 MSA patients included in this longitudinal cohort,146 patients completed CRP detection,only 72 patients completed blood routine and biochemical indicators detection,and 56 patients completed blood immune indicators detection.The univariate linear regression model showed that only CRP was correlated with annual progression rate of UMSARS total score(P<0.001).The other baseline blood routine,biochemical indicators and blood immune indicators were not related to annual progression rate of UMSARS total score.After one-year follow-up,the level of CRP in MSA patients was significantly higher than that in baseline(1.88±2.04 vs 1.06±0.98,P=0.005).Baseline blood routine,biochemical indicators,CRP and blood immune indicators were not associated with annual progression rate of FSS total score.We further included CRP and OH into the multiple linear regression model,and the results further showed that CRP(P=0.023)was significantly correlated with annual progression rate of UMSARS total score.3.Relationship between blood biomarkers and disease progression of MSA Among 146 MSA patients included in this longitudinal cohort,64 MSA patients and 60 age and gender matched HC completed the analysis of NFL,α-synuclein,GFAP,TREM2 and BDNF at baseline and follow-up.Compared with HC,the baseline levels of NFL(31.91±14.16 vs 12.07± 5.88,P<0.001),α-synuclein(2.01±1.04 vs 1.62± 0.89,P=0.026)and TREM2(5721.91 ± 1774.43 vs 4559.19 ± 1187.23,P<0.001)in MSA patients were significantly higher;there was no significant difference in GFAP and BDNF between MSA patients and HC.Due to the significant difference in the disease duration between MSA-P and MSA-C patients,we found that the level ofα-synuclein in MSA-P patients(2.45 ± 1.05 vs 1.66 ± 0.90,P=0.003)was significantly higher than that in MSA-C patients after adjusting for disease duration.There was no significant difference in NFL,GFAP,BDNF and TREM2 between the two subtypes.After one-year follow-up,the levels of NFL(41.60±26.09 vs 31.91±14.16,P=0.010)in MSA patients was significantly higher than those in baseline MSA patients,while the levels of α-synuclein,GFAP,BDNF and TREM2 had no significant difference before and after follow-up(P>0.05).The univariate linear regression model showed that the baseline NFL level was correlated with annual progression rate of the total score of UMSARS but not the annual progression rate of total score of FSS.α-synuclein,GFAP,BDNF and TREM2 were not associated with the annual progression rate of total score of UMSARS and the annual progression rate of total score of FSS.We further included NFL and OH into the multiple linear regression model.The results suggested that NFL(P=0.043)was significantly correlated with annual progression rate of UMSARS total score.4.The relationship between genetics and disease progression of MSA We analyzed the relationship between risk gene SNPs of MS A and disease progression in 146 MSA patients.Correlation analysis showed that SLC1A4 rs759458,COQ2 rs148156462,SNCA rs11931074,SNCA rs3857059,TNF rs1799964,DDC rs921451 and LRRK2 rs1427263 genotypes were not associated with annual progression rate of UMSARS total score and annual progression rate of FSS total score.5.The relationship between imaging features and disease progression of MSA Among the 146 MSA patients included in this longitudinal cohort,81 patients completed MRI scanning in the same machine in our hospital.The univariate linear regression model showed that cerebellar atrophy,hypointensity of putamen,"hot cross sign" and total score of imaging features were correlated with the annual progression rate of the total score of UMSARS.There was no significant correlation between these imaging features and annual progression rate of FSS total score.We further included the total score of imaging features and OH into the multiple linear regression model.The results showed that the total score of imaging features(P=0.010)was significantly correlated with annual progression rate of UMSARS total scoreConclusion:This part of the study first combined with clinical characteristics,hematological indicators,blood biomarkers,genetics and imaging,included MSA patients with baseline disease duration<3 years for follow-up,to explore the factors related to the disease progression of MS A in the early stage of the disease.1)After one-year follow-up,MSA patients had greater disease severity and severer fatigue symptoms.The proportion of MS A patients with fatigue and symptom of depression was significantly higher than that of baseline.Only OH was associated with disease progression2)The level of CRP was not only increased with the progress of the disease,but also related to the progress of the disease.3)The plasma levels of NFL,α-synuclein and TREM2 in MSA patients were significantly higher than those in HCs.However,with the progression of the disease,only plasma NFL increased,but α-synuclein,GFAP,BDNF and TREM2 did not change.NFL was associated with disease progression.4)SNPs of risk genes associated with MSA were not associated with disease progression.5)The total score of imaging features was related to the progression of MSA.Part Ⅲ Construction of prognostic model of multiple system atrophy based on machine learning algorithmsObjective:MSA progresses rapidly with an average survival time of 6-10 years.Wheelchair confinement is an important marker of the progression of MSA.Confined to wheelchair makes MSA patients unable to take care of themselves,and many daily activities such as dressing,getting on and off bed,bathing and going to the toilet can not be completed independently,leading to disability.Previous literature reported that the late age of onset,higher motor symptom score and early onset of autonomic symptoms were related to the wheelchair confinement of MSA patients.Previous studies on prognosis of MSA mainly focused on clinical features,and used traditional statistical analysis methods had limitations.As an artificial intelligence technology,machine learning is flexible and scalable,and has been widely used in disease diagnosis,classification and prognosis prediction.Random forest(RF)and support vector machine(SVM)are two widely used machine learning algorithms.Therefore,based on the longitudinal cohort of MSA with complete clinical,biomarker,genetic,and imaging data.This part is to establish MSA disability prediction model using RF and SVM machine learning algorithms,so as to provide guidance for clinical intervention.Materials and Methods:The longitudinal cohort study was used in this chapter.The patients included in the first part were followed up regularly every year(face-to-face evaluation or telephone follow-up).Patients with MSA with a baseline disease duration less than 3 years were included,and disability(confined to wheelchair)was used as an indicator of clinical outcome.The collection of baseline clinical data,hematological indicators,imaging data and MSA related risk gene loci is the same as the first part,and the detection of blood biomarkers is the same as the second part.At the deadline of December 2020,280 patients with MSA were included in this part of the study.Firstly,the median time of disability in 280 MSA patients was 4.04 years(95%CI:3.81-4.27)using Kaplan Meier curve.Therefore,this study took 4 years as the cut-off time,excluding 86 patients with disease duration less than 4 years without disability,and finally divided 194 MSA patients into two groups:patients with disability group:long-term wheelchair confinement within 4 years;patients without disability group:more than 4 years confine or not confine to wheelchair.T test,Mann Whitney U rank sum test or chi square test were used for subgroup comparison.The following variables were included into machine learning model:gender(female=0,male=1),age of onset,disease duration,diagnosis subtype(MSA-C=0,MSAP=1),symptom of onset(autonomic symptom=0,motor symptom=1),UMSARSⅠ score,UMSARS-Ⅱ score,UMSARS-Ⅳ score,OH,urinary incontinence,NFL,CRP and total score of imaging features according to the results of the first two parts,the results of pairwise comparison between the patients with and without disability group,and the factors related to the disability of MS A reported in the previous literature.RF and SVM were used to establish models to predict whether patients with MSA will be disabled within 4 years.Because only 104 MSA patients in this part of the study completed the baseline imaging data collection,then we intend to establish two groups of prediction models based on the two groups of data(104 MSA with imaging features group and 194 MSA without imaging features group).Machine learning randomly divided the data into two groups,80%of the sample for training data sets,the remaining 20%of the sample for validation data sets.The training data set is used for the construction of MSA disability model,and the validation data set is used for the internal validation of the disability model,and further perform 5-cross validation.In this study,accuracy,sensitivity,specificity and area under curve(AUC)were used to evaluate the performance of the model.Results:A total of 280 MSA patients were included in this section,145 males and 135 females,123 patients with MSA-P,and 157 patients with MSA-C,with an average age of onset of 57.23±7.64 years.The median time of disability was 4.04 years.Taking the median time of disability of 4 years as the cut-off time,excluding the patients with disease duration less than 4 years without disability,194 patients with MSA were finally divided into with disability(96 MSA)and without disability group(98 MSA).The baseline disease duration(1.69±0.67 vs 2.13± 0.73,P<0.001)was shorter in patients with disability than those without.After adjusting for disease duration,the baseline UMSARS-Ⅱ score(17.20±6.90 vs 14.47±5.75,P<0.001),UMSARS-Ⅱscore(20.40 ± 8.20 vs 16.37± 6.17,P<0.001),UMSARS-Ⅳ score(2.28 ± 0.88 vs 1.88± 0.94,P<0.001)and the total score of UMSARS(37.59± 14.36 vs 30.84 ±11.17,P<0.001)were higher in patients with disability than those without.The proportion of OH at baseline in patients with disability was higher than that in patients without disability(43.8%vs 28.6%,P=0.009).Among the 194 patients in the model,100 patients had baseline blood routine and biochemical indexes,and 76 patients had immunological indicators.After adjusting the disease duration and the total score of UMSARS,there was no significant difference in blood routine,biochemical and immunological indicators between the patients with and without disability.There was no significant difference in the genotype frequencies of SLC1A4 rs759458,COQ2 rs148156462,SNCA rs11931074,SNCA rs3857059,TNF rs1799964,DDC rs92145 and LRRK2 rs1427263 between the patients with and without disability.All 194 MSA patients completed the detection of blood biomarkers and CRP,after adjusting the disease duration and the total score of UMSARS,the NFL level(46.35 ± 26.73 vs 37.42 ± 25.37,P=0.018)and CRP level(2.26±2.36 vs 1.72±2.02,but there was no statistical significance)in patients with disability was higher than that in patients without disability.There was no significant difference in the levels of α-synuclein,GFAP,BDNF and TREM2 between the two groups.One hundred and four of 194 patients in the model completed the baseline MRI scan in our hospital.The results showed that the proportion of mild and moderate atrophy of the baseline putamen in patients with disability was higher than patients without disability,but there was no significant difference after adjusting the disease duration and the total score of UMSARS;there was no significant difference in the other MSA imaging features and the total score of imaging features between the two groups.The accuracy of the binary logistic regression model predicting the disability of 104 MSA patient... |
PDF Full Text Request