Risk Prediction Of Motor Fluctuations In Patients With Parkinson Disease

ObjectiveMotor fluctuations are one of the important markers of disease progression in patients with Parkinson’s disease.Motor fluctuations refer to the decreased therapeutic response and pharmacodynamic stability of patients to Parkinson’s disease drugs,which is one of the important causes of disability in patients with advanced Parkinson’s disease and often leads to a series of psycho-psychological problems in patients.Therefore,early identification and timely management are particularly important.The aim of this study was to analyze the risk factors of motor fluctuations in patients with Parkinson’s disease and to construct a predictive model for survival without motor fluctuations.This result can be applied clinically as survival prediction tools without motion fluctuations.MethodsThis study was a retrospective study based on the PPMI database.The raw data was obtained from access and download of the PPMI database on 15 November 2021.The distribution of each variable in the two groups of patients with or without motor fluctuations within 8 years was analyzed and KM curves were plotted.The data were randomized to generate training and validation sets.Univariate COX regression was applied to screen for variables followed by backward stepwise multivariate regression to select 9 variables as Model 1.8 variables were selected as Model 2 by applying LASSO regression followed by backward stepwise multivariate regression.By performing receiver operating characteristic curve analysis,calibration curve analysis,and decision curve analysis of the two models,Model 2 was finally selected to draw the Nomogram.Results401 patients were finally included after screening.The data were randomized in a 7:3 ratio into a training set with 280 patients and a validation set with 141 patients.Model 2 had a higher AUC than Model 1 for the prediction of no movement fluctuations at 2 years(0.69,0.73),4 years(0.72,0.72),and 7years(0.76,0.70)in both the training and validation sets.In the calibration curve analysis,model 2 was better than model 1 in predicting the probability of survival without motor fluctuations at 4 and 7 years in the training set.The survival probability prediction without motor fluctuations at 2 and 4 years in the training set was better than that in model 1.In the decision curve analysis,model 2 was superior to model1 in terms of clinical benefit over a larger range of risk thresholds predicted at 2,4,7 years in both the training and validation sets.Model 2 was selected after comparative analysis of the 2 models.Model 2included eight variables,which were age < = 57.23,MDS-UPDRS I < = 3,MDS-UPDRS II < = 9,MDS-UPDRS III < = 32,Aβ1-42 < = 1333,serum uric acid < = 269,p-tau/t-tau < = 0.078,and striatal binding < = 2.05.The C index of model 2 in validation set was 0.688,indicating that model 2 had some predictive ability for the occurrence time of motor fluctuations in Parkinson’s disease patients.Finally,Model 2 was visualized as a Nomogram map to facilitate clinical application.ConclusionIn this study,we explored the risk factors of motor fluctuations in Parkinson’s disease through data analysis of a public database,thus providing a clinical basis for further studies on the pathogenesis of Parkinson’s disease.The prediction model constructed in this study can predict the time when motor fluctuations occur to a certain extent,and then analyze the progression of Parkinson’s disease through clinical practice application,providing reference for early and timely intervention to delay the progression of Parkinson’s disease.