Clinical Subtypes of Parkinson’s Disease Stephanie M. van Rooden, MSc, 1 * Fabrice Colas, PhD, 2 Pablo Martı ´nez-Martı´n, MD, 3 Martine Visser, PhD, 1 Dagmar Verbaan, PhD, 1 Johan Marinus, PhD, 1 Ray K. Chaudhuri, MD, 4 Joost N. Kok, PhD, 2,5 and Jacobus J. van Hilten, MD 1 1 Department of Neurology, Leiden University Medical Centre, Leiden, The Netherlands 2 Leiden Institute for Advanced Computer Science, Leiden University, Leiden, The Netherlands 3 Alzheimer Disease Research Unit, National Centre for Epidemiology and CIBERNED, Carlos III Institute of Health, Madrid, Spain 4 National Parkinson Foundation Centre of Excellence, King’s College Hospital and University, Hospital Lewisham, London, United Kingdom/King’s College and Institute of Psychiatry, London, United Kingdom 5 Department of Medical Statistics and Bioinformatics/Molecular Epidemiology, Leiden University Medical Center, Leiden, The Netherlands ABSTRACT: The clinical heterogeneity of Parkin- son’s disease (PD) may point at the existence of sub- types. Because subtypes likely reflect distinct underlying etiologies, their identification may facilitate future genetic and pharmacotherapeutic studies. Aim of this study was to identify subtypes by a data-driven approach applied to a broad spectrum of motor and nonmotor features of PD. Data of motor and nonmotor PD symptoms were col- lected in 802 patients in two different European prevalent cohorts. A model-based cluster analysis was conducted on baseline data of 344 patients of a Dutch cohort (PROPARK). Reproducibility of these results was tested in data of the second annual assessment of the same cohort and validated in an independent Spanish cohort (ELEP) of 357 patients. The subtypes were subsequently characterized on clinical and demographic variables. Four similar PD subtypes were identified in two different populations and are largely characterized by differences in the severity of nondopaminergic features and motor complications: Subtype 1 was mildly affected in all domains, Subtype 2 was predominantly characterized by severe motor complications, Subtype 3 was affected mainly on nondopaminergic domains without prominent motor complications, while Subtype 4 was severely affected on all domains. The subtypes had largely similar mean disease durations (nonsignificant differences between three clusters) but showed considerable differ- ences with respect to their association with demographic and clinical variables. In prevalent disease, PD subtypes are largely characterized by the severity of nondopami- nergic features and motor complications and likely reflect complex interactions between disease mechanisms, treatment, aging, and gender. V C 2010 Movement Disorder Society Key Words: Parkinson’s disease; subtypes; cluster analysis; nondopaminergic; motor complications Parkinson’s disease (PD) is generally known as a movement disorder, but there is an increasing aware- ness that the clinical spectrum of PD encompasses also many nonmotor domains like cognition and auto- nomic function. 1 Patients with PD exhibit conspicuous differences in the disease profile and progression rate. 2 This clinical heterogeneity may indicate the existence of subtypes. Identification of subtypes is important as homogeneous groups likely reflect stronger clinical, pathological, and genetic coherence, which, in turn, may facilitate our understanding of involved biological pathways. This may ultimately lead to tailored treat- ment strategies. In studies on PD subtypes, patients have often been classified according to predefined criteria [e.g., young versus old age at onset (AO) or dominance of tremor versus bradykinesia/rigidity], after which other clinical variables were compared between the resulting groups. 2 Alternatively, subtypes may be identified through a data-driven approach like cluster analysis (CA), in which the profile of the subtypes arise from the data without a priori clinical assumptions. ------------------------------------------------------------ Additional Supporting Information may be found in the online version of this article. *Correspondence to: Ms. Stephanie M. van Rooden, Department of Neurology, K5Q-92 Leiden University Medical Center, P.O. Box 9600, NL 2300 RC Leiden, The Netherlands; s.m.van_rooden@lumc.nl Relevant conflicts of interest: Nothing to report. Full financial disclosures and author roles may be found in the online version of this article. Received: 19 April 2010; Revised: 2 June 2010; Accepted: 18 June 2010 Published online 16 November 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/mds.23346 RESEARCH ARTICLE Movement Disorders, Vol. 26, No. 1, 2011 51