"LIFE-PPD: Uncovering Prodromal Markers of Parkinson's Disease in the Dutch Population" A collaboration with the LifeLines biobank

From 1990 to 2016, the prevalence of Parkinson’s disease (PD) increased from 2.5 million to more than 6 million cases worldwide, making PD the second most common neurodegenerative disorder (1). In the light of an aging population, this number is expected to rise in the coming years. PD poses a substantial burden to society and has a great impact on the quality of life of patients and their relatives/caregivers. So far, only symptomatic treatments are available for PD, and attempts to halt or slow down the neurodegenerative process have failed. 
Parkinson’s disease is a clinical diagnosis, which can be made when cardinal motor features are present: bradykinesia, rigidity and/or rest tremor (2). These symptoms are caused by a loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc), but only occur after more than 50% of the dopaminergic neurons are lost (3). This means that the disease process has been present long before the clinical diagnosis can be made. Indeed, in PD, three phases can be recognized: [1] a preclinical phase, in which risk factors are present but the individual does not have any symptoms or signs; [2] a prodromal phase, in which an individual has symptoms but the typical motor features are not yet present or not sufficiently pronounced to be able to make a clinical diagnosis of PD; and [3] the clinical phase, in which the diagnosis can be made based on the typical motor features (4). This implies that in the clinical phase, neurodegeneration has already been present for several years and may be too advanced for any potential disease-modifying therapies to have effect. The preclinical and prodromal stages provide opportunities for meaningful interventions (5). To this end, it is essential that preclinical and prodromal individuals can be detected accurately and reliably for potential inclusion in clinical trials. 
In recent years, there have been new insights into the early stages of PD and its pathophysiology. PD is associated with abnormal aggregations of the synaptic protein alpha-synuclein in neurons. These abnormal aggregations can spread from neuron to neuron in a prion-like manner. The symptoms that PD patients experience depend on which neuronal systems are affected, and in which sequence. The common denominator is that once pathology substantially impairs the mesencephalic substantia nigra, motor symptoms start to occur. However, PD can affect many other systems, giving rise to a plethora of non-motor symptoms. This includes symptoms of the gut and autonomic system (constipation, urinary dysfunction, orthostatic hypotension). A very specific feature of prodromal PD is REM sleep behavior disorder (RBD). This is a parasomnia in which normal muscle atonia during the REM sleep phase is lost, causing dream-enacting behaviors during sleep. 
The movement disorders society (MDS) has developed research criteria for prodromal PD (10, 11). These criteria provide a strategy to calculate the probability that an individual has prodromal PD. This calculation combines the likelihood ratios (LR) of prodromal symptoms that the individual has (e.g. constipation, idiopathic RBD, hyposmia) and several risk markers (e.g. gender and age). The probability is expressed as a percentage, and a percentage of ≥80% represents probable prodromal PD. It includes prodromal factors that can be easily assessed with questionnaires, in addition to factors that have to be proven with ancillary investigations. For instance, RBD can be suspected based on a questionnaire of specific symptoms during sleep, but has to be confirmed by a video-polysomnography. 
The validity of the MDS research criteria has been investigated in several studies. These studies were performed either in the general population, or in enriched cohorts (i.e. subjects were selected based on the presence of criteria that are associated with an increased risk of PD, such as RBD or genetic mutations). In a cohort of 121 subjects with iRBD, the MDS research criteria at baseline were found to have a sensitivity of 81.3% and a specificity of 67.9% for conversion to PD in the next 4 years (12). Sensitivity and specificity increased with longer follow-up duration. Specificity and positive predictive value for the criteria were 100% after 10-year follow-up. This means that all of the converters met the MDS criteria before conversion. Indeed, the authors found a dose-response association between the calculated prodromal LR and the time to conversion. It is important to note that all of the participants in those studies already had RBD at baseline, which translates to a likelihood ratio of 130 for prodromal PD at the time of recruitment. 
Similar results were found in a cohort of 171 iRBD patients with a mean follow-up of approximately 2 years (13). 74% of iRBD patients reached the 80%-threshold set by the MDS criteria, while 92.4% reached the 50% threshold. The MDS criteria were also applied to 119 de novo PD patients, based only on simple clinical measures. Without any invasive investigations (such as polysomnography or dopaminergic imaging), 22% reached the 80% threshold and 51% reached the 50% threshold. Obviously, these patients have clinically established PD and should all reach the 80% threshold; this simply indicates the reliance of these criteria on specialist investigations in order to obtain high sensitivity. Of 296 controls, only 0.3% reached the threshold of 80%, and 1,4% reached the threshold of 50%. 
Sensitivity of the MDS criteria is lower when measured in the general population. The criteria were tested in the Bruneck study, which is a cohort of 539 participants aged 55-94, representative of the elderly population. During a follow-up of 10 years, 12 individuals developed PD. The MDS criteria for prodromal PD had a specificity of 99%, a sensitivity of 67%, and a positive predictive value of 40%, considering a follow-up of 3 years. With increasing follow-up time, specificity remained stable, sensitivity decreased and the positive predictive value rose (60% at 5 years, 78% at 10 years) (14, 15). 
Similarly, in a study that investigated data of the TREND (n=650, follow-up of 6 years, PD n=10) and PRIPS (n=715, follow-up of 5 years; PD n=7) community-based cohorts, the specificity and negative predictive value was high (>98%), and sensitivity was low (TREND, 30%, PRIPS, 14%) (16). This was lower than for the Bruneck study, but fewer prodromal markers were assessed in these cohorts. Only a minority of PD patients reached the threshold of 80% prior to diagnosis. However, the risk of prodromal PD was strongly negatively associated with the time to conversion.  
Finally, the Hellenic Longitudinal Investigation of Aging and Diet cohort (n=961, follow-up of approximately 3 years, 22 incident PD or DLB cases) found a prevalence of prodromal PD of 0.8% using the updated criteria (17). None of the converters sufficed the 80% threshold for prodromal PD before diagnosis. This was probably because the markers with large LR could not be evaluated (substantia nigra hyperechogenicity, hyposmia, polysomnography-confirmed RBD). 
In addition to the MDS criteria, other methods of screening for prodromal PD may also be considered. Noyce et al. developed and improved PREDICT-PD, which is an online screening tool for the detection of increased risk of developing PD (18, 19). The tool is based on online surveys and is easy to employ. It was found to be a useful tool for identification of increased risk for PD (20). The PREDICT-PD algorithm was validated in another cohort, although a different study that applied the algorithm found only a slight added value for detection of an increased risk for PD (21,22). 
Population-based studies that have been performed so far (see above) have been relatively small, with only 12-22 incident PD cases to investigate. Lifelines provides a unique opportunity: it includes over 167,000 participants from the Northern Netherlands with clinical data and biosamples collected since 2006. Between 2011 and 2017, 110 Lifelines participants received a PD diagnosis. To the best of our knowledge, this is the largest cohort of incident PD cases from a population-based study so far.  Our objectives for the proposed retrospective study are to test whether we can reliably identify the individuals with an increased risk of having prodromal PD based on risk markers and prodromal features that were assessed using Lifelines questionnaires. To do this, we will retrospectively compare a group who have developed PD since the start of Lifelines with a group of healthy controls.
                This first project will pave the way to understanding the specific features of prodromal PD in the Northern Netherlands. The next step will be to test serum biomarkers for PD in the Lifelines population, and investigate genetics of the northern population. Our goal is to identify a risk profile for PD in the Northern Netherlands. Ultimately, we wish to identify individuals with a high risk of PD from the Lifelines cohort prospectively (‘profile-based’). The ultimate goal is to include high-risk individuals in disease modification trials. These projects fall outside of the scope of the current application.