Current State of the Art of Transcranial Magnetic Stimulation in Psychiatry: Innovations and Challenges for the Future

The field of noninvasive brain stimulation, and specifically transcranial magnetic stimulation (TMS), is a fast-developing area of research, fueled by neuroscientific, technical, and clinical advances. Various forms of TMS are used in neuroscience and psychiatry research labs to study perceptual, cognitive, and affective processes underlying health and disease. Therapeutically, repetitive TMS (rTMS) is used to modulate these processes to reduce symptoms or to increase compensatory strategies in a myriad of psychiatric and neurological conditions.

This special issue of Biological Psychiatry focuses on current innovations aimed at enhancing the neuromodulatory effects of TMS and discusses the challenges in bringing these methods and insights to the clinic. In general, the progress made in the field of TMS is encouraging, but there is still considerable work to be done to improve the reliability, efficacy, and accessibility of this technology. The commentary by Lisanby (1) provides historical context, discusses some of the challenges that TMS researchers have faced, and describes how the National Institutes of Health (where TMS was first developed) continues to support research in the field. The commentary by Cabrera and van den Heuvel (2) focuses on ethical considerations, such as the unknown long-term risks, especially in special populations, and accessibility to underserved populations. They also call for responsible use of TMS in the context of the ongoing proliferation of private clinics. The invited reviews in this special issue highlight the factors that influence the clinical efficacy of TMS and the neuroimaging tools and techniques that have been developed to measure and manipulate these factors for optimal brain-target engagement.

TMS coils are designed to induce an electric field within a defined targeted cortical region. It is that induced electric field that determines the amount of effective stimulation of the cortex and the resulting neuromodulatory effect of the TMS on the brain network. Dannhauer et al. (3) describe the development of modeling tools that enable researchers to determine the optimal coil position, device intensity, and other stimulation parameters to ensure that a sufficient electric field is delivered to the targeted region of interest.

Target selection is one domain of research aimed at improving the effects of rTMS by incorporating knowledge of interindividual variability in clinical symptom presentation, structural and functional neuroanatomy, and underlying pathophysiology. Siddiqi and Fox (4) propose a framework of symptom-specific target selection, recognizing that symptoms differ between patients with the same diagnosis (e.g., anxious vs. agitated depression) and that optimal targets may overlap between patients with different diagnoses (e.g., executive dysfunctions in individuals with depression vs. those with schizophrenia). A transdiagnostic symptom-based targeting approach necessitates the identification of targets based on retrospective data. These approaches then need to be tested prospectively to confirm whether targeting different circuits does indeed modulate different symptoms. To facilitate the development and implementation of this framework, a validated open-source library of potentials targets is warranted. Cash and Zalesky (5) propose another approach to individualize targeting using information on the patient’s brain network structure and/or function. Specifically, they describe the development of network connectivity–guided neuronavigation using resting-state functional magnetic resonance imaging (fMRI)–based anticorrelation between the subgenual anterior cingulate cortex and the dorsolateral prefrontal cortex. Such an approach is also one of the elements of the Stanford neuromodulation therapy protocol, which has been recently approved by the U.S. Food and Drug Administration for treatment-resistant depression. Cole et al. (6) describe how Stanford neuromodulation therapy combines individualized fMRI network targeting with the accelerated application of multiple daily sessions of intermittent theta burst stimulation (iTBS) with effects that exceeded previous response and remission rates of standard rTMS protocols in a fraction of the time.

This brings us to the second factor contributing to variation in TMS effects: intersession interval and dosing. Accelerated iTBS protocols enable the delivery of a higher number of pulses and sessions in a shorter period, reducing the standard treatment from 4 to 6 weeks to 1 week. Accelerated protocols enable clinicians to provide faster and potentially more clinically effective treatments for patients who are severely ill and suicidal. Although such accelerated iTBS protocols seem to be safe in adults with treatment-resistant depression, the safety and tolerability in more vulnerable populations (e.g., children/adolescents and the elderly) need further evaluation. In addition, randomized controlled trials are needed to directly compare the clinical efficacy, durability, and cost-effectiveness of maintenance protocols associated with accelerated iTBS compared with nonaccelerated (once per day) iTBS when all other stimulation parameters (such as resting-state fMRI–guided targeting) are kept constant.

A third factor of variance in TMS response is the cognitive and oscillatory state of the brain at the time of stimulation. Sack et al. (7) describe ways to incorporate information on the cognitive and/or emotional state of the individual during or just prior to stimulation for optimal target engagement. The state of the brain can be visualized by concurrent TMS-fMRI, TMS-electroencephalography (EEG), or even TMS-fMRI-EEG. Symptom provocation prior to rTMS, application of concurrent TMS with psychotherapy, or circuit activation using a cognitive task during stimulation are all strategies to manipulate the cognitive or emotional state of the patient within the context of an rTMS session. Another aspect of state dependence concerns the oscillatory state of the brain during stimulation. Zrenner and Ziemann (8) describe developments in EEG-based closed-loop TMS whereby TMS is applied when the endogenous EEG oscillations are at a certain predetermined optimal state (e.g., a certain phase or minimum power of a specific frequency band). EEG-synchronized TMS is thought to induce larger effects and engage the human cortex more effectively than open-loop TMS, which is uninformed of brain state. Although technically possible, the application of an EEG-based closed-loop TMS system in the clinic will depend on more advanced, next-generation EEG-TMS systems with a limited number of EEG sensors and automatic signal processing.

Variation in TMS effects also depends on the properties of the stimulated brain networks. This idea inspired researchers to identify biomarkers of TMS response, using diffusion MRI and fMRI, EEG, and even concurrent TMS-EEG. In their article, Klooster et al. (9) summarize the current state of MRI- and EEG-based biomarkers for rTMS response in depression. Their meta-analyses show that the most robust and replicable biomarkers are fMRI-based functional connectivity between the dorsolateral prefrontal cortex and the subgenual anterior cingulate cortex, fMRI-based lesion network mapping, task-induced EEG frontal-midline theta, and EEG individual alpha frequency. Farzan (10) describes the use of concurrent TMS-EEG to predict response to rTMS. Concurrent TMS-EEG can also be used to explore safety biomarkers aimed at predicting side effects to other treatments in psychiatry, such as clozapine for psychotic disorders and seizure therapy in depression.

Oberman and Benussi (11) address an additional potential source of variability, i.e., typical neurodevelopmental and neurodegenerative processes that occur across the lifespan. TMS has been used to probe neurotransmitter and functional network maturation and to identify pathophysiological mechanisms underlying neurodevelopmental/pediatric-onset disorders and neurodegenerative/late adult–onset disorders. They also discuss the growing body of evidence supporting the therapeutic use of rTMS for symptom management/remediation and prevention (in the context of neurodegenerative disorders) for a variety of disorders of development and aging.

Variation in the rTMS-induced effects on the clinical level also reflects variation in TMS-induced plasticity effects at the neural circuit level. Understanding the neuroplastic effects of rTMS at different biological scales (from synapse to connectome) and different timescales (directly after stimulation and after multiple sessions) across disorders and neurodevelopmental stages is crucial. Fitzsimmons et al. (12) summarize the potential clinical relevance of neuroimaging and blood biomarkers of rTMS-induced neuroplasticity but also show that a more harmonized and longitudinal multimodal approach (i.e., combining methods that visualize effects on multiple neurobiological scales) is needed to understand how the different aspects and scales of plasticity interact and relate to optimal clinical response.

The invited review papers included in this special issue represent the current state of the art of TMS. Given the relative nascency of TMS as a research tool and rTMS as a therapeutic intervention for psychiatric disorders, and the mechanism of action of TMS necessarily interacting with a brain with unique and ever-changing neurophysiology, it is not surprising that there is variability in response. Technological advances have allowed for precise measurement of the electrical field and the neurophysiological state both at baseline for protocol and patient stratification and after stimulation as a metric of immediate or long-term effects, but these tools may not always be accessible outside of major research institutions.

Many unanswered questions remain. Much of the literature includes small-scale, underpowered, and/or noncontrolled studies. Additional large-scale, ideally randomized prospective trials comparing multiple conditions, retrospective data projects, and meta-analyses are needed to evaluate the impact of each of these sources of variability on clinical outcomes. In addition, studies assessing the accessibility and feasibility of some of these protocols in general clinical practice still need to be conducted. Through “bench to bedside” collaborations and clinical translational approaches such as experimental therapeutics platforms, future studies may be able to bridge this gap and bring the field closer to elucidating the most optimal TMS parameters for a given individual with a psychiatric condition and to bringing this novel treatment to the masses.