Differentiation between Parkinson’s and Parkinson’s-like patients in MDS-UPDRS-based diagnosis
Sven C. van Dijkman (1), Gopichand Gottipati (1), Elodie L. Plan (1), Mats O. Karlsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: It has been shown in the PPMI study [1] that approximately 15% of patients with MDS-UPDRS scores suggesting Parkinson’s disease (PD) are later found to not show dopaminergic deficits on a SPECT scan (SWEDD cohort). For confirmed PD patients (DeNoPD cohort), an Item Response Theory (IRT) model (DeNoPD model) was previously built on the MDS-UPDRS individual item scores. Here, it was our aim to determine if patient type differentiation could be performed by expanding the DeNoPD model to SWEDD patients.
Methods: Data used in this work were obtained from the Parkinson’s Progression Markers Initiative (PPMI) database [1] A SWEDD model was developed based on the DeNoPD model and a shift parameter for each UPDRS item, estimated on the SWEDD data alone. Based on ratios of individual likelihoods from evaluating both models on screening and baseline data of both cohorts, the probability of belonging to either cohort could be estimated for each patient. Differentiation power residing in individual items was assessed using subsets of individuals´ data, by an iterative item addition process.
Results: IRT model based differentiation between PD and PD-like patients was found to be 86.3% sensitive and 62.7% specific to detecting PD. A subset of 14 items was found to be 94.9% sensitive and 57.8% specific compared to the full 68 item MDS-UPDRS rating scale for differentiation of these patients.
Conclusions: The proposed model allows a distinction between PD and PD-like patients to be made to a higher degree of sensitivity and similar specificity compared to visual clinical examination [2], while avoiding the need for expensive and burdensome radiological testing such as SPECT or MRI scans.
References:
[1] The Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/data). For up-to-date information on the study, visit www.ppmi-info.org
[2] Bajaj NPS, Gontu V, Birchall J, Patterson J, Grosset DG, Lees AJ. Accuracy of clinical diagnosis in tremulous parkinsonian patients: a blinded video study. J Neurol Neurosurg Psychiatry. 2010;81:1223–8
