Bayesian dynamical system analysis of the effects of methylphenidate in children with attention-deficit/hyperactivity disorder: a randomized trial

Abstract

Methylphenidate is a widely used and effective treatment for attention-deficit/hyperactivity disorder (ADHD), yet the underlying neural mechanisms and their relationship to changes in behavior are not fully understood. Specifically, it remains unclear how methylphenidate affects brain and behavioral dynamics, and the interplay between these dynamics, in individuals with ADHD. To address this gap, we used a novel Bayesian dynamical system model to investigate the effects of methylphenidate on latent brain states in 27 children with ADHD and 49 typically developing children using a double-blind, placebo-controlled crossover design. Methylphenidate remediated greater behavioral variability on a continuous performance task in children with ADHD. Children with ADHD exhibited aberrant latent brain state dynamics compared to typically developing children, with a single latent state showing particularly abnormal dynamics, which was remediated by methylphenidate. Additionally, children with ADHD showed brain state-dependent hyper-connectivity in the default mode network, which was also remediated by methylphenidate. Finally, we found that methylphenidate-induced changes in latent brain state dynamics, as well as brain state-related functional connectivity between salience and default mode networks, were correlated with improvements in behavioral variability. Taken together, our findings reveal a novel latent brain state dynamical process and circuit mechanism underlying the therapeutic effects of methylphenidate in childhood ADHD. We suggest that Bayesian dynamical system models may be particularly useful for capturing complex nonlinear changes in neural activity and behavioral variability associated with ADHD. Our approach may be of value to clinicians and researchers investigating the neural mechanisms underlying pharmacological treatment of psychiatric disorders.

Fig. 1. Key steps in Bayesian dynamical systems modeling.

A Schematic illustration of the Bayesian switching linear dynamical systems (BSDS) model. B Regions of interest (ROIS) include key nodes in the salience (SN), frontal-parietal (FPN) and default mode networks (DMN). C Temporal evolution of latent brain states during rest-state fMRI of all participants. Each row represents one subject, each column represents one data point (fMRI volume). lAI left anterior insula, rAI right anterior insula; DMPFC Dorsomedial prefrontal cortex, lMFG left middle frontal gyrus, rMFG right middle frontal gyrus, lFEF left frontal eye field, rFEF right frontal eye field lIPL left inferior parietal lobule, rIPL right inferior parietal lobule, PCC Posterior cingulate cortex, VMPFC Ventromedial prefrontal cortex.

Fig. 2. Methylphenidate normalizes aberrant latent brain states.

Occupancy rate (OR) of the latent brain state S2 was significantly higher in children with ADHD than TD children (p < 0.05, FDR corrected) and methylphenidate significantly reduced hyper-OR of the latent brain state S2 in children with ADHD (p < 0.05).

Fig. 3. Methylphenidate reduces aberrant state-specific hyper-connectivity.

Children with ADHD under placebo condition showed (A) hyper-connectivity between PCC and VMPFC in the state S2 than TD children (p < 0.05, FDR corrected) and B methylphenidate reduced hyper-connectivity between the DMN nodes as well as connectivity between SN, FPN and DMN nodes in the state S2(p < 0.05, FDR corrected). State S2 was the latent brain state with higher occupancy rate in children with ADHD, compared to controls. lAI left anterior insula, rAI right anterior insula, DMPFC Dorsomedial prefrontal cortex, lMFG left middle frontal gyrus, rMFG right middle frontal gyrus, lFEF left frontal eye field, rFEF right frontal eye field, lIPL left intraparietal lobule, rIPL right intraparietal lobule, PCC Posterior cingulate cortex, VMPFC Ventromedial prefrontal cortex.

Fig. 4. State-specific effects of methylphenidate on brain and behavioral dynamics.

Methylphenidate effect on the functional connectivity of lAI and PCC of the latent brain state S2 was significantly associated with the medication effect on the IIRV (r = 0.62, p = 0.0005). State S2 was the latent brain state with higher occupancy rate in children with ADHD, compared to controls. lAI left anterior insula, rAI; PCC Posterior cingulate cortex.