Parkinson’s disease is notoriously varied. Some people have tremors, while others have rigid muscles—or slow movements, or difficulty smelling things, or even a loss of automatic movements like blinking. Some experience episodes of psychosis. Currently, doctors have no way of predicting what kinds of symptoms any given person will develop.
But now, researchers funded in part by the Knight Initiative for Brain Resilience have taken a step toward solving that problem. In the process, the team reported recently in npj Parkinson’s Disease, they uncovered new links between Parkinson’s, inflammation, and metabolism that could change how doctors diagnose and treat the disorder in all its varied forms.
Ted Wilson, the new study’s first author and an instructor in the Department of Neurology and Neurological Sciences at Stanford Medicine, said the team was inspired in part by recent advances in Alzheimer’s disease research.
“We have in the last ten years really come to understand the biological underpinnings of Alzheimer’s, and that biological definition has allowed us to develop the first approved disease-modifying drugs for the disease,” Wilson said. Now, to move towards more effective treatments, he said, “we want to define Parkinson’s disease based on its biological underpinnings as well.”
A pathway to disease
In the new paper, Wilson; Katrin Andreasson, the Edward F. and Irene Thiele Pimley Professor in the Department of Neurology and Neurological Sciences; and colleagues focused on a metabolic process called the kynurenine pathway (KP), which has been on the radar of Parkinsons’ researchers for a while. Ordinarily, this metabolic pathway turns tryptophan—the molecule apocryphally linked to post-turkey dinner sleepiness—into food for cells, but inflammation can throw a wrench in the works. In particular, inflammation can nudge the kynurenine pathway to make more of a neurotoxin called quinolinic acid (QA) and less of a neuroprotective molecule called kynurenic acid (KA). That may seem like a lot of intimidating alphabet soup, but ultimately it suggests the kynurenine pathway is linked to Parkinson’s, but researchers haven’t figured out exactly how.
To learn more, Wilson and his team gathered biological samples from 177 people with Parkinson’s and 158 healthy but otherwise similar adults from Stanford’s Alzheimer’s Disease Research Center and Movement Disorders Clinic, the Stanford Aging and Memory Study, and similar programs at the University of Washington, Oregon Health & Science University, and the University of Pennsylvania. The researchers gathered samples of blood plasma and cerebrospinal fluid (CSF), which flows in and around the brain, from each participant. In addition, each of the Parkinson’s patients symptoms were assessed with the Movement Disorder Society Unified Parkinson’s Disease Rating Scale.
The team next analyzed the blood and CSF samples to figure out how molecules related to the kynurenine pathway varied in people with and without Parkinson’s.
As the team had suspected, more quinolinic and less kynurenic acid were both correlated with the disease. Compared with healthy adults, people with Parkinson’s had lower overall levels of neuroprotective kynurenic acid and higher relative levels of neurotoxic quinolinic acid (a higher QA-to-QP ratio) in both blood plasma and CSF. Of particular interest, the researchers found that more severe Parkinson’s symptoms were associated with increased levels of quinolinic acid in blood plasma and CSF.
Discovering new patterns
However, there was a great deal of variability in the data, Wilson said, perhaps mirroring the variability in Parkinson’s disease symptoms themselves. That got Wilson wondering whether there were distinct patterns in their biochemical data that might reveal something new about Parkinson’s.
To take a closer look, the team connected with Nima Aghaeepour, an associate professor of pediatrics and of anesthesiology, perioperative and pain medicine, who specializes in biomedical applications of machine learning.
Focusing solely on the blood and CSF data, Aghaeepour’s lab used machine learning to show that many patients fell into distinct subgroups, each comprising a few dozen patients—and each with its own distinct molecular profile.
One group, for example, was marked by a particularly high ratio of neurotoxic quinolinic acid vs. protective kynurenic acid. Another had a normal ratio of those compounds but very low levels of kynurenine itself—the molecular pathway’s namesake, which acts as an intermediate between tryptophan and the pathway’s end products.
Remarkably, those molecular patterns corresponded to distinct patterns of Parkinson’s symptoms. People in the first group (high quinolinic-to-kyurenic-acid ratio) had more uncontrollable muscle contractions, more severe motor symptoms overall, and more non-motor symptoms such as insomnia, anxiety, and pain. In contrast, people in the second group (low overall kynurenine levels) experienced more tremors and slowed movement than others.
“I was surprised,” said Kathleen Poston, Edward F. and Irene Thiele Pimley Professor in the Department of Neurology and Neurological Sciences and a coauthor on the study. “On the clinical side, we had noticed relationships between Parkinson’s and blood chemistry,” but doctors generally assumed medications—or other factors not directly connected to the underlying disease—were behind those links. “Ted’s data shows this is part of Parkinson’s itself.”
The new results could help researchers develop a new understanding of Parkinson’s variability, new tests to diagnose different forms of the disorder, and more targeted treatments for those variants.
“We can start thinking about what type of therapy each person might respond to,” depending on that person’s pattern of KP malfunction, Wilson said. “All of this needs to be tested in the lab with careful studies, but this precision medicine approach could be transformational.”
Publication Details
Research Team
Study authors were Edward N. Wilson, Jacob Umans, Michelle S. Swarovski, Paras S. Minhas, Justin H. Mendiola, Marian Shahid-Besanti, Patricia Linortner, Siddhita D. Mhatre, Qian Wang, Divya Channappa, Nicole K. Corso, Brenna Cholerton, Sharon J. Sha, Frank M. Longo, Victor W. Henderson, Tony Wyss-Coray, Elizabeth C. Mormino, Kathleen L. Poston, Katrin I. Andreasson, Edward D. Plowey, Thomas J. Montine, Anthony D. Wagner & Nima Aghaeepour from Stanford University; Øivind Midttun, Arve Ulvik & Per M. Ueland from Bevital, Bergen, Norway; Carolyn A. Fredericks from Yale University; Geoffrey A. Kerchner from Pharma Research and Early Development, F. Hoffmann-La Roche, Ltd., Basel, Switzerland; Cyrus P. Zabetian from the VA Puget Sound Health Care System and the University of Washington; Nora E. Gray from Oregon Health & Sciences University; Joseph F. Quinn from OHSU and the Portland VA Medical Center; and David A. Wolk & Alice Chen-Plotkin from the University of Pennsylvania.
Research Support
The research was supported by the National Institutes of Health, National Institutes on Aging, and National Institute of Neurological Disease and Stroke (RF1AG058047, R01AG048232, NS115114, K23 NS075097, R35GM138353, R01AG048076, R01AG048076, R01AG074339, P30 AG072979, RO1NS115139, P30 AG066518, P30 AG066515, P30 AG072979, P50 NS062684), the Michael J. Fox Foundation (6440.0), the American Heart Association (19PABHI34580007), the Jean Perkins Foundation, and the Scully Research Initiative. Wilson, Wyss-Coray, Poston, and Andreasson are supported by the Phil & Penny Knight Initiative for Brain Resilience.
Competing Interests
The authors declare no competing interests.