Literature Review

Online Algorithm Identifies People at Risk of Parkinson’s Disease

The web-based approach can be scaled up to enable the identification of larger numbers of cases that reflect the spectrum of disease.


 

The online PREDICT-PD algorithm effectively identifies people at increased risk of Parkinson’s disease, according to research published online ahead of print January 16 in Movement Disorders. The risk scores that the algorithm assigns are significantly associated with intermediate markers of Parkinson’s disease. Higher PREDICT-PD risk score at baseline also is associated with an increased rate of incident Parkinson’s disease.

Previous research with the goal of identifying patients at risk of Parkinson’s disease mainly has focused on people with a family history of the disease, asymptomatic carriers of genes associated with the disease, or people with imaging abnormalities associated with the disease (eg, hyperechogenicity on transcranial sonography). The rarity, the representativeness, and the feasibility of identifying these factors have limited these studies, however.

Internet-Based Testing

To overcome these limitations, Anette-Eleonore Schrag, PhD, Professor of Clinical Neurosciences at University College London, and colleagues developed an online, evidence-based algorithm to identify risk indicators of Parkinson’s disease in the UK population. The investigators used a study website to recruit participants between ages 60 and 80 who did not have Parkinson’s disease, any other movement disorder, or stroke. Participants were prompted to return to the website yearly and complete tests.

Anette-Eleonore Schrag, PhD

Among the tests was a survey that included demographic questions and validated questionnaires about early nonmotor features of and risk factors for Parkinson’s disease (eg, the Hospital Anxiety Depression Scale and the REM sleep behavior disorder Screening Questionnaire [RBDSQ]). Participants also took the Bradykinesia Akinesia Incoordination (BRAIN) test, an online keyboard-tapping task. At baseline and at year three, participants took the University of Pennsylvania Smell Identification Test (UPSIT). The annual survey included a question about whether the participants had received any new diagnoses, including Parkinson’s disease or movement disorder. At year three, participants also gave saliva samples that were genotyped for mutations in exons 8 to 11 of glucocerebrosidase (GBA) and exon 41 in the LRRK2 gene.

The investigators used participants’ early features and risk factors to calculate their risk scores and rank participants according to their risk. Dr. Schrag and colleagues also tested support for enrichment of the population at risk of Parkinson’s disease by examining associations between risk scores and the following intermediate markers of Parkinson’s disease: reduced sense of smell, presence of subjective RBD, and slowing of finger tapping speed.

Risk Scores Were Associated With Incident Disease

The researchers recruited 1,323 eligible volunteers. At baseline, their mean age was 66, and approximately 61% of the population was female. In all, 1,040 of the participants completed follow-up testing at year 1, 939 completed year 2 follow-up, and 846 completed year 3 follow-up. A total of 223 participants completed the baseline assessment only.

Baseline risk scores were associated with significantly higher rates of all intermediate markers of Parkinson’s disease during each year of follow-up. The group of patients at higher risk (ie, those above the 15th percentile) had significantly worse UPSIT score, RBDSQ score, and finger tapping in all years, compared with patients at lower risk (ie, those below the 85th percentile).

Risk scores in the entire sample were strongly associated with intermediate markers of Parkinson’s disease each year. Higher- and lower-risk groups differed significantly in median UPSIT score, RBDSQ score, and mean finger tapping speed in all years of follow-up.

At year 1, three patients had been newly diagnosed with Parkinson’s disease. An additional participant received the diagnosis in year 2, and three other participants received the diagnosis in year 3. All of the participants with newly diagnosed Parkinson’s disease at year 1 were in the higher risk group, and two had also been in the higher risk group at baseline. The participant diagnosed at year 2 was in the higher risk group at baseline, year 1, and year 2. All of the three participants diagnosed by year 3 were in the middle risk group at baseline. “There was substantial heterogeneity in the occurrence of intermediate markers in these individuals,” said the researchers.

The incidence of independently diagnosed Parkinson’s disease in participants who had been in the higher risk group during three years of follow-up was 1.6% per year and 0.2% across the whole population. Exploratory Cox regression analysis using incident Parkinson’s disease over three years as the outcome indicated an association with baseline risk estimate. In participants for whom GBA and LRRK2 status was known, the association between baseline risk and incident Parkinson’s disease was weaker. But the addition of GBA and LRRK2 variants in the algorithm improved the strength of association between baseline risk and incident Parkinson’s disease. Among the intermediate markers of disease, baseline finger tapping was associated with incident Parkinson’s disease at three years.

Results Could Reflect Selection Bias

“Although only a small number of individuals have been independently diagnosed with Parkinson’s disease during follow-up so far, the overall incidence of 0.2% is consistent with the expected incidence rate in the age group of 60 to 80 from the general population (one to three per 1,000 per year) and supports the representativeness of our sample,” said Dr. Schrag.

The recruitment of participants may have introduced selection bias into the study, she acknowledged. “However, we did not use family history as an entry criterion, and the proportion with a positive family history was lower than in other landmark studies.” The fact that 17% of participants only completed the baseline assessment also may have introduced selection bias, “although there were no differences between those with or without follow-up, meaning that bias as a result of loss to follow-up was less likely.

“This Internet-based approach may be useful for population screening because it can easily be scaled upward,” added Dr. Schrag. “It will allow larger numbers of Parkinson’s disease cases that represent the spectrum of the disease to be identified, rather than what would be possible from cohorts of carriers of specific risk factors.”

Erik Greb

Suggested Readings

Noyce AJ, R’Bibo L, Peress L, et al. PREDICT-PD: An online approach to prospectively identify risk indicators of Parkinson’s disease. Mov Disord. 2017 Jan 16 [Epub ahead of print].

Recommended Reading

Treatment Advances in Parkinson’s Disease and Other Movement Disorders
MDedge Neurology
Gene Mutation Linked to Early Onset of Parkinson’s Disease
MDedge Neurology
Patient Expectations May Be Potent Factor in DBS for Parkinson’s Disease
MDedge Neurology
What Are the Best Treatments for Nonmotor Symptoms in Parkinson’s Disease?
MDedge Neurology
Light therapy eases Parkinson’s-related sleep disturbances
MDedge Neurology
Is Closed-Loop DBS Ready for Clinical Application?
MDedge Neurology
Blood NfL Accurately Differentiates Parkinson’s Disease From Atypical Parkinsonian Disorders
MDedge Neurology
MCI May Predict Dementia in Patients With Parkinson’s Disease
MDedge Neurology
Sublingual Apomorphine Film May Induce On State in Patients With Parkinson’s Disease
MDedge Neurology
Which Cognitive Domains Predict Progression From MCI to Dementia in Parkinson’s Disease?
MDedge Neurology

Related Articles