Transcranial Magnetic Stimulation for Non-Invasive Brain Modulation
Postdoctoral Research Fellow, Dartmouth College
Dr Aman Aberra is a postdoctoral fellow at Dartmouth College with a biomedical engineering and physics background. During his undergraduate studies, he developed an interest in electromagnetic fields and how they interact with biology, particularly the nervous system. This interest led him to complete his PhD in biomedical engineering at Duke University in the labs of Drs. Warren Grill and Angel Peterchev, where he started to concentrate on cellular-level understanding of mechanisms of non-invasive brain scanning. He then focused on developing mathematical models that integrated the biological components that underlie the direct response to brain stimulation in Transcranial Magnetic Stimulation (TMS). This inspired him to pursue postdoctoral research in which he performs experiments to validate predictions and offer an experimental perspective on how electric fields influence neurons at the cellular and subcellular levels.
What is the motivation behind focusing on Non-Invasive Neuromodulation instead of invasive forms such as Deep Brain Stimulation (DBS), and what are each approach's potential benefits and drawbacks?
Aman: The motivation behind focusing on non-invasive neuromodulation techniques, such as Transcranial Magnetic Stimulation (TMS), is the potential for rapid translation from novel ideas and protocol designs to treatments in patients. Non-invasive techniques allow for the exploration and characterisation of a wide range of parameters, potentially producing desired outcomes without invasive surgery. While there are some limitations to non-invasive procedures, such as the inability to target deep structures, they have shown promise in treating various neurological and psychiatric disorders. In comparison, invasive techniques like deep brain stimulation (DBS) can offer more precise targeting of specific brain regions but require invasive surgery and carry potential risks such as infection and hardware failure. I am particularly interested in the fundamental effects of electric fields, which are common to all forms of electrical stimulation. They aim to develop general principles that can be used to understand and improve all types of neuromodulation techniques, focusing on understanding how neural activity changes at both the acute and circuit levels. Ultimately, the goal is to modulate the neural code precisely and specifically to treat neurological and psychiatric disorders.
What are the potential limitations of Deep Brain Stimulation (DBS) regarding scalability, and why is non-invasive neuromodulation considered to have a higher potential for delivering treatments at scale to address health challenges such as Depression, OCD, and Obesity?
Jonathan: Absolutely, and I agree that the lower barrier for non-invasive approaches, such as TMS, allows for easier translation from pre-clinical to clinical settings. However, established techniques like Deep Brain Stimulation (DBS) face a significant bottleneck due to the surgical requirements. It is challenging to envision DBS being delivered at scale considering the large number of people worldwide who could potentially benefit from it. For health challenges like depression, OCD, and obesity that affect populations globally, scalable treatments capable of reaching tens or even hundreds of millions of individuals are needed.
Aman: In terms of the use of Deep Brain Stimulation (DBS) for depression, there are specific patient populations where it may be difficult to justify the economic costs and invasiveness of the procedure. However, as new indications for DBS outside of movement disorders and Parkinson’s disease are being explored, there is potential for an expansion in the use of this technique. However, it is unlikely that DBS will be able to scale up to meet the needs of many people suffering from depression.
Given the lifetime prevalence of depression in the United States, which is estimated to be around one in five, it is clear that we need multiple approaches to address this complex disorder. One possibility is to stratify the patient population and use invasive methods targeted to specific brain structures for certain types of depression. In other patients, non-invasive approaches such as Transcranial Magnetic Stimulation (TMS) may be more appropriate, especially for those with treatment-resistant depression or at a particular stage of symptom severity.
It is worth noting that the efficacy of TMS and other non-invasive approaches could improve with the development of new protocols and technology, leading to broader adoption of these methods. Additionally, there is still a role for Electro Convulsive Therapy (ECT) in treating severe depression, despite the challenges posed by its side effect profile.
What is the primary sequence of events during a Transcranial Magnetic Stimulation (TMS), including the induction of a magnetic field and the method of neuron excitation?
Jonathan:Could you walk us through a typical sequence of events during a figure-eight TMS, including how a magnetic field is induced, how long it lasts, and how neurons are excited based on Hodgkin-Huxley-style cable theory models? Additionally, what do we currently understand about the subsequent steps, and where does our understanding become less clear? While the initial steps involving basic physics are relatively straightforward, does the process quickly become uncharted or, at the very least, not well-understood?
Aman: Yes, you’re right that it is still pretty uncharted territory, even at the first stage. However, we have good biophysical theories that have worked in many cases, such as Hodgkin-Huxley or cable theory. Cable theory, which has been around for about a century, gives us a theory of how currents flow in mineral processes, similar to how they flow through wires or leaky wires, mainly where current can flow both within the membrane inside the axon dendrites, the “wire” of the brain, but also flow outside through the membrane. This cable theory, combined with ideas and models of the dynamics of neurons, which we get from things like Hodgkin-Huxley, gives us some predictions.
The idea is that first, you induce an electric field through magnetic induction. The magnetic field generated by the coil is time-varying. One of the fundamental laws of electromagnetics gives us the idea that a time-varying brain magnetic field induces an electric field. So we think it’s the electric field, even though obviously, it’s called transcranial magnetic stimulation, that’s really interacting with the neurons because that’s able to change their membrane potential, which is the critical parameter or property of neurons, their state, as their membrane potential is a function of the currents across the membrane, which is through these ion channels.
The electric field, which for most TMS pulses is less than 100 microseconds, so around 100 microseconds or less, can polarise neurons, which means it can shift their membrane potentials up or down. Some populations of them will be shifted enough to cause them to fire an action potential. This is happening with the TMS electric field, which is produced in a tangential way to the brain, so it has this particular direction where it flows in these lines of current that are tangential to the scalp. This implicates a specific set of neural structures that can be activated, which are neural structures that are parallel to those currents, and this falls out of cable theory.
This gives us a set of possible elements that TMS might activate, and when we plug this into our model, where we combine a whole population of neurons. We’ve modelled their morphologies or their geometry, their shape, and we’ve modelled using this kind of Hodgkin-Huxley ideas; we’ve modelled their dynamics. When we plug that into these models, we find that this field ends up activating predominantly axons and the models predominantly axons at their terminals that are aligned to the electric field. As you change the electric field direction, we expect you to get activation of a different set of branches of axons.
When we did this in a model that combined models of cells from different layers in the cortex, which is composed of up to six layers where you have other cellular properties of cells in each of those layers and additional connectivity, we found that neurons in the middle layers 2, 3, 4, and 5 seem to be activated. These are particularly excitatory cells in a class of inhibitory cells called large basket cells. This gives us a reasonably mixed activation, where we seem to activate both excitatory and inhibitory neurons. The balance of excitation inhibition can vary based on obviously the intensity you apply, the pulse shape you apply and the direction you apply it with.
We think that the TMS electric field, which is decaying with depth, so it decreases in intensity as you go deeper into the brain, is predominantly activating neurons in the most superficial parts of the cortex, which might be limiting in some ways as this means we’re only simulating the superficial parts of the cortex, and we can’t simulate.
Do backside H3 and inhibitory neurons have similar activation dynamics during TMS, or is one population slower? Is there a specific pattern of activation and inhibition during TMS?
Jonathan:During TMS, we stimulate a mixed blend of excitatory (H3) and inhibitory neurons. Do these two populations of neurons exhibit similar dynamics in terms of activation, or is one population slower than the other? Is the excitation followed by inhibition happening in a staggered manner?
Aman: That’s an excellent question. From the direct response, we expect that whichever neurons are activated by the pulse will be activated at the same time. But as soon as the pulse evolves over a few milliseconds, you start to see effects. This is being observed in monkey research, where a single pulse of TMS seems to activate neurons in a way that causes an increase in firing rate, followed by a long pause where the firing rate drops below baseline, and then a rebound effect where the firing rate goes above baseline. This is known as the single approach response, and it’s similar to what’s seen with invasive cortical microstimulation using a technique called ICMS or interpretive microstimulation. When this is done in mice and rats, a similar burst pause rebound response is seen, which is a conserved cortical response to this type of stimulation that potentially activates a mix of solitary inhibitory neurons. However, we still need a better sense of whether doing this multiple times and at different cell types would result in uniform facilitation or suppression in the early phase of excitatory-inhibitory cells. If you stimulate where you get that pause or rebound, it might underlie these frequency-dependent effects because you can get a buildup of either inhibition or excitation as you stimulate different parts of the temporal profile observed after a single pulse. It’s a great question, and we don’t have the full answer yet, but we should be thinking about these asymmetries and non-uniformities in the response when combining effects across pulses. Although TMS is not as focal as optogenetic stimulation, where molecularly identified populations can be stimulated, we can still take advantage of these asymmetries and use them to understand what’s happening with what we’re already applying.
Do you think that the relatively crude and nonspecific nature of TMS, compared to optogenetic approaches, limits its ability to target specific neuron types and anatomical locations, or is it still a valuable tool despite these limitations?
Jonathan: It is indeed true that compared to optogenetic approaches, TMS lacks the specificity of targeting specific cell types and precise anatomical locations. With TMS, the magnetic field induces an electrical field, and its effects tend to decay, at least in part, with distance following a rapid decay pattern. Consequently, TMS can be considered a relatively blunt and crude instrument in terms of its precision. The question arises whether this limitation is problematic or acceptable when it comes to targeting different neuron types and anatomical structures.
Aman: The spatial resolution is around a square centimeter. Still, if you look at the number of neurons – hundreds of thousands, even millions of neurons of different cell types – you need to get resolution within the different layers, at least as far as we know. The field doesn’t change so much within layers that we can selectively activate only superficial or deep layers. The first thought of an engineer is to strive for more precision and resolution, and I lean in that direction. However, the caveat is always worth remembering: what would we do with that precision? Do we know enough to say, “If I could stimulate exactly in these places, I could achieve this specific effect”? We’re setting these goals, but in the meantime, we have relatively crude but still effective methods for mixed brain activation. We have some spatial targeting at the centimetre scale – we can target macroscopic cortical regions separately or even simulate different hand muscles. But we don’t have the level of resolution that would allow us to stimulate specific cell populations that are molecularly identified, for example. I mention this because there are ideas around stimulating more broad regions, such as Transcranial Direct Current Stimulation (tDCS) or other non-focal methods that affect a broader swath of neurons across a much broader region. There might be applications where we do want to stimulate a lot of the brain in different ways. These are different parts of the design domain for non-invasive brain stimulation, which includes ranges of focal and non-focal stimulation, as well as ranges in the temporal domain. So, while it’s a limitation in a direct sense, we still have a lot of resolution to work with for now. We have to figure out how to use it to its full potential and be open to new approaches as they are developed, as there are constantly new ideas around how to modulate brain activity.