A team of researchers has developed a simple and non-invasive approach that uses genetic data in combination with other demographic and biomarker information to accurately distinguish those with and without Parkinson’s disease, an illness that is notoriously difficult to detect.
In a study published this week in the journal Lancet Neurology, the group detailed their findings showing how they used a small set of key data points to discern individuals with Parkinson’s from those who do not have the disease. The predictive model offers hope for earlier detection and diagnosis for the disease, according to the study.
“Accurate diagnosis and early detection of complex diseases, such as Parkinson’s, has the potential to be of great benefit for researchers and clinical practice,” the authors said.
Simply making a correct diagnosis of a progressive neurodegenerative disease like Parkinson’s can be very difficult, and take time. It’s often misdiagnosed and the result is that many patients do not get early treatment. Although there is not currently a cure for Parkinson’s, existing treatments have been able to significantly mitigate symptoms, and new treatments being tested now may help slow the progression of the disease. The earlier those treatments can begin, the better the prognosis. That makes early diagnosis that much more important. There currently is no cure for Parkinson’s so treatment has focused on slowing the progression of the disease and mitigating the symptoms.
“The model could have immediate application in helping refine cohorts used in clinical research,” said Paul Cannon, 23andMe’s Parkinson’s Research community manager.
“With more data and further refinement, it may be able to contribute to better diagnosis of individuals with Parkinson-like symptoms and the identification of prodromal (pre-symptomatic) and “at risk” populations, which will be critical for the development of disease modifying therapies.”
This paper shows that a model, which combines genetic risk score, demographic and biomarker data, can more accurately discriminate individuals with Parkinson’s from healthy controls as compared to genetics alone. That is partly what makes this study so interesting.
Lead by scientists with the National Institutes of Health, the researchers tested their model against a group of more than 1000 individuals, some with Parkinson’s and some without the disease. Researchers from 23andMe, the Michael J. Fox Foundation, the University of Pennsylvania and the University of Rochester Medical Center also participated in the work. The researchers also tested their model against another set of patients who had had image scans of their brain, and are referred to as Scans Without Evidence of Dopaminergic Deficit or SWEDD. The model classified 17 patients as having Parkinson’s and within the first year, four of them developed the disease. Only one of the 38 patients the model classified as not having Parkinson’s developed the disease in the same time frame.
The model includes genetic risks, family history, age, gender and biomarkers such as the ability to detect certain smells into account in diagnosing Parkinson’s. The information used in the model can be gathered remotely — without visiting a medical center, patients can send in saliva for genetic testing and answer surveys regarding family history, the study authors said.
After they created an algorithm using this information, the scientists then tested it by using data on individuals from five different study groups, including 23andMe’s Parkinson’s Research Community. What they found was that in each case their method had a very high accuracy rate for predicting those with Parkinson’s from those without the disease. The model accurately classified Parkinson’s in patients more than 90 percent of the time. The model also did well classifying different Parkinson’s subtypes.
The researcher used many methods of validating the results, including using data from 23andMe’s LRRK2 biomarker cohort, and the data from the Parkinson’s Progression Marker Initiative.
What the researchers found was that not only was the new method highly accurate at diagnosing Parkinson’s, it also significantly outperformed any single-classifier method of diagnosing the disease.
The study authors believe that with further refinement the tool may be useful for screening and diagnosing individuals in high risk groups. It may also help in differentiating Parkinson’s disease from other neurological conditions.