The study covered in this summary was published on medRxiv.org as a preprint and has not yet been peer reviewed.
Key Takeaways
-
A machine-learning algorithm demonstrated efficacy in accurately distinguishing patients with Parkinson’s disease (PD) who had depression from patients with PD without depression and controls with depression.
-
This information lends insight into cortical mechanisms of depression and could prompt the identification of innovative neurophysiologically-based biomarkers for nonmotor symptoms of PD.
-
EEG, which records activity from the cortex via an array of scalp electrodes, is especially well-suited to capture cortical neurophysiology.
Why This Matters
-
Depression is a prominent nonmotor symptom of PD. It is difficult to diagnose and often missed by physicians, and the neurophysiological basis is poorly understood.
-
PD-related depression afflicts approximately 20%–40% of patients with PD, several times the anticipated prevalence within this population.
-
Despite the magnitude of PD-related depression, it is uncertain which brain circuits play a role. Establishing which brain circuits contribute to PD-related depression could spark the development of new diagnostic tools and targeted treatments such as neuromodulation.
-
Depression can affect cortical function considerably, which implies that EEG may be able to identify PD-related depression.
-
Improved understanding of depression in PD may help shed light on basic mechanisms of both diseases.
Study Design
-
The researchers enrolled 18 patients with PD, 18 patients with PD and with depression, and 12 demographically-similar non-PD patients with clinical depression.
-
All patients stayed on their usual medications.
-
Resting-state EEG was obtained in all patients while they remained seated in a quiet room with their eyes open for 2 minutes.
-
Cortical brain signal features were compared between patients with and without depression.
-
A machine-learning algorithm was utilized to leverage the entire power spectrum (linear predictive coding of EEG Algorithm for PD: LEAPD), to differentiate between groups.
Key Results
-
Depressed patients with PD had distinct spectral EEG features.
-
The researchers found contrasts between patients with PD with and without depression in the alpha band (8-13 Hz) globally and in the beta (13-30 Hz) and gamma (30-80 Hz) bands in the central electrodes.
-
From 2 minutes of resting-state EEG, they discovered that LEAPD-based machine learning could strongly differentiate between PD patients with and without depression with 97% accuracy, and between PD patients with depression and non-PD patients with depression with 100% accuracy.
-
The researchers confirmed the strength of their finding by ascertaining that the classification accuracy diminishes gracefully as data are truncated.
Limitations
-
The sample size was small, although similar to previous EEG studies in patients with PD and with depression.
-
All patients were medicated, and it could be that medications may affect EEG signals.
-
The researchers’ approach to diagnosing depression and determining the level of symptom burden in patients with PD was distinct from the method employed with patients without PD, which limited comparisons between these groups.
-
An out-of-sample prospective test was not part of the LEAPD technique, although the truncation analysis eliminates concerns of overfitting.
Study Disclosures
-
The authors have declared no competing interest.
This is a summary of a preprint research study, “Resting-state EEG distinguishes depression in Parkinson’s disease,” written by Arturo I. Espinoza from the University of Iowa, on medRxiv, provided to you by Medscape. This study has not yet been peer reviewed. The full text of the study can be found on medRxiv.org.
For more news, follow Medscape on Facebook, Twitter, Instagram, YouTube, and LinkedIn