Prediction and early biomarkers of cognitive decline in Parkinson disease and atypical parkinsonism: a population-based study

Abstract

Abstract The progression of cognitive decline is heterogeneous in the three most common idiopathic parkinsonian diseases: Parkinson disease, multiple system atrophy and progressive supranuclear palsy. The causes for this heterogeneity are not fully understood, and there are no validated biomarkers that can accurately identify patients who will develop dementia and when. In this population-based, prospective study, comprehensive neuropsychological testing was performed repeatedly in new-onset, idiopathic parkinsonism. Dementia was diagnosed until 10 years and participants (N = 210) were deeply phenotyped by multimodal clinical, biochemical, genetic and brain imaging measures. At baseline, before the start of dopaminergic treatment, mild cognitive impairment was prevalent in 43.4% of the patients with Parkinson disease, 23.1% of the patients with multiple system atrophy and 77.8% of the patients with progressive supranuclear palsy. Longitudinally, all three diseases had a higher incidence of cognitive decline compared with healthy controls, but the types and severity of cognitive dysfunctions differed. In Parkinson disease, psychomotor speed and attention showed signs of improvement after dopaminergic treatment, while no such improvement was seen in other diseases. The 10-year cumulative probability of dementia was 54% in Parkinson disease and 71% in progressive supranuclear palsy, while there were no cases of dementia in multiple system atrophy. An easy-to-use, multivariable model that predicts the risk of dementia in Parkinson disease within 10 years with high accuracy (area under the curve: 0.86, P < 0.001) was developed. The optimized model adds CSF biomarkers to four easily measurable clinical features at baseline (mild cognitive impairment, olfactory function, motor disease severity and age). The model demonstrates a highly variable but predictable risk of dementia in Parkinson disease, e.g. a 9% risk within 10 years in a patient with normal cognition and CSF amyloid-β42 in the highest tertile, compared with an 85% risk in a patient with mild cognitive impairment and CSF amyloid-β42 in the lowest tertile. Only small or no associations with cognitive decline were found for factors that could be easily modifiable (such as thyroid dysfunction). Risk factors for cognitive decline in multiple system atrophy and progressive supranuclear palsy included signs of systemic inflammation and eye movement abnormalities. The predictive model has high accuracy in Parkinson disease and might be used for the selection of patients into clinical trials or as an aid to improve the prevention of dementia.