Electrical Stimulation modulates Multidien Biomarkers of Human Treatment-Resistant Depression - an Emerging Neuromodulation Target
DOI: Pending DOI
Synopsis
Closed-loop neuromodulation is an emerging therapy for psychiatric disorders such as treatment-resistant depression (TRD). Here, we show temporal variability in closed-loop biomarkers arises from multidien rhythmicity in the local field potential and behavior. Intracranial EEG (iEEG) & self-reported VAS-D were sampled daily for 2-4 yrs in 5 TRD patients in the PRESIDIO clinical trial. Multiday rhythms (MDR) were quantified using Morlet wavelet analysis, phase coherence and Ridge regression. MDR of 10-18 & 25-32 day periods were observed in iEEG and VAS-D, and were synchronized. Ridge regression using neural MDR achieved VAS-D forecasting. Closed-loop stimulation increased the occurrence of iEEG and mood MDRs, and concurrently disrupted their synchronization. Forecasting rhythms, and MDR biomarker-based neuromodulation may enable novel chronotherapeutic strategies.
Background
Closed-loop neuromodulation is an emerging therapy for psychiatric disorders such as treatment-resistant depression (TRD). This approach delivers direct electrical stimulation in response to an acute neural biomarker of symptoms. However, the stability of these biomarkers over long-term therapy is poorly understood. Here, we show that temporal variability underlying closed-loop biomarkers arises, in part, from multidien rhythmicity in both the local field potential and behavior. Using chronic intracranial recordings in individuals with TRD, we show the encoding of depressive symptoms in multiday rhythms in intracranial physiology, and their modulation by electrical stimulation. Uncovering and modulating these rhythms may enable forecasting depressive episodes and improving closed-loop therapies.
Methods
Five TRD patients in the PRESIDIO trial (NCT04004169) were implanted with the Responsive Neurostimulation System (NeuroPace, Inc), each with intracranial leads sampling two of these regions: orbitofrontal cortex, amygdala, subgenual cingulate and ventral capsule/ventral striatum. Intracranial EEG (iEEG) and self-reported Visual Analog Scale of Depression (VAS-D) were sampled multiple times daily over 2-4 years. Clinical depression severity was evaluated through Montgomery-Asberg Depression Rating Scale (MADRS). Multiday rhythms in iEEG and VAS-D were quantified using Morlet wavelet analysis and their relationship dynamics modeled using phase coherence. Neural rhythms were used to forecast VAS-D using Ridge regression, and the effect of stimulation on the neural and VAS-D multidien rhythms was quantified.
Results
Multiday rhythms of 10-18 and 25-32 day periods were observed in iEEG across brain regions and in mood (VAS-D) (p<0.05) (Fig 1.a, b). Neural and mood rhythms are synchronized (coherence: 0.1-0.7, p<0.05) (Fig 1.c), and phase lag analysis indicates neural rhythms lead mood rhythms by 4-35 days (p<10-6) (Fig 1.d). Ridge regression model using neural and behavioral multidien features achieved VAS-D forecasting (Pearson correlation: 0.43-0.62, p<10-8). Closed-loop stimulation increased multi-day rhythmicity of VAS-D scores and of the local field potential (p<0.05) (Fig 1.e), and disrupted the relationship between the neural and VAS-D multidien rhythms (Fig 1.f).
Discussion
We demonstrate multidien biomarkers of depression and their modulation by neuromodulation. Forecasting models that track rhythms and predict depressive episodes, and multidien biomarker-based neuromodulation may enable novel chronotherapeutic strategies.
Disclosures
N/A