Deep brain stimulation for neurodegenerative disease: a computational blueprint using dynamic causal modeling.

Abstract

Advances in deep brain stimulation (DBS) therapeutics for neurological and psychiatric disorders represent a new clinical avenue that may potentially augment or adjunct traditional pharmacological approaches to disease treatment. Using modern molecular biology and genomics, pharmacological development proceeds through an albeit lengthy and expensive pipeline from candidate compound to preclinical and clinical trials. Such a pathway, however, is lacking in the field of neurostimulation, with developments arising from a selection of early sources and motivated by diverse fields including surgery and neuroscience. In this chapter, I propose that biophysical models of connected brain networks optimized using empirical neuroimaging data from patients and healthy controls can provide a principled computational pipeline for testing and developing neurostimulation interventions. Dynamic causal modeling (DCM) provides such a computational framework, serving as a method to test effective connectivity between and within regions of an active brain network. Importantly, the methodology links brain dynamics with behavior by directly assessing experimental task effects under different behavioral or cognitive sets. Therefore, healthy brain dynamics in circuits of interest can be defined mathematically with stimulation interventions in pathological counterparts simulated with the goal of restoring normal functionality. In this chapter, I outline the dynamic characterization of brain circuits using DCM and propose a blueprint for testing in silico, the effects of stimulation in neurodegenerative disorders affecting cognition. In particular, the models can be simulated to test whether neuroimaging correlates of nondiseased brain dynamics can be reinstantiated in a pathological setting using DBS. Thus, the key advantage of this framework is that distributed effects of DBS on neural circuitry and network connectivity can be predicted in silico. The chapter also includes a review of how DCM has been used to characterize the effects of DBS in Parkinson's disease.