Last month, TMS Clinicians and Scientists gathered in Vancouver, BC, Canada for the Clinical TMS Society Meeting followed by the International Brain Stimulation Conference. Though the attendance at the Clinical TMS Society was mostly clinicians and the Brain Stimulation Conference mostly researchers, I chose to attend both meetings. As a clinical researcher, there was much to learn from both. Here are some of my musings…
To personalize or not to personalize? Though there was much discussion about personalized approaches much of the discussion revolved around what aspect of TMS to personalize (the frequency or location) and the best strategy for personalizing (e.g. EEG oscillations, cardiac rhythms, functional MRI, structural MRI etc.). For example, TMS applied at personalized frequencies (e.g. individual alpha frequency) appears to have just as good efficacy as a standard 10 Hz stimulation protocol and (though there is less literature on this) 1 Hz stimulation also seems to be effective. So is personalization necessary or are the standard protocols sufficient? I would argue that personalization will almost certainly be better. We just have to know what we are personalizing to. Perhaps the “secret sauce” so to speak is to figure out what strategy is best for whom. Some individuals may require a larger stimulation region at a specific frequency while others may not be sensitive to frequency, but experience a greater response based on treatment location. This brings me to my second musing…
Biotypes. Both Clinicians and Researchers accept that depression and other mental disorders are not one thing (at a brain level). So why are we expecting that one treatment (at a brain level) will treat everyone with this heterogeneous behaviorally defined condition? At the Brain Stimulation conference, there were several proposed strategies to subgroup or “biotype” depressed patients with the hypothesis that different TMS protocols may be more or less effective for different individuals. Again, how to define biotypes (based on clinical symptoms or brain imaging or both), and how many unique biotypes there are, remain open for debate. Thus far, for depression, only three targeted regions have been adequately trialed. These include Dorsomedial Prefrontal Cortex, Orbitofrontal Cortex, and the most popular target, Dorsolateral Prefrontal Cortex. Some speculate that if you have clinical data with domains and subdomains assessing anxiety, anhedonia, psychomotor retardation, obsessive and/or compulsive behavior, anergia and fatigue and functional connectivity data between frontal cortices and deeper limbic and midbrain structures, you could get as high as an 80-90% response rate to the correct stimulation target and protocol (see Drysdale et al., 2017, Nature Medicine). A prospective test of this hypothesis, however, where biotypes are prospectively assigned to the protocol based on this or similar algorithms has not been conducted. But this type of study is on the horizon…stay tuned! Also on the horizon are large-scale clinical trials for other indications including OCD, PTSD, Suicidality, and Substance Use Disorders as well as special populations including children, veterans and military personnel, and concussed patients.
Another take-home from the conferences was a call to learn from our own and other’s mistakes and failed attempts. There is no doubt in my mind that TMS works and can work for a number of disorders. That being said, many trials including two recent large-scale ones (Adolescent Depression and VA trial) have failed. Not from lack of efficacy, but from unusually high placebo response rates. These trials cost millions of dollars (and if funded by NIH these are your tax-payer dollars) and years to conduct. Therefore, so as to optimize effect size and reduce placebo effects, some key factors for research scientists to consider:
- Do the necessary background research. TMS is a brain modulation tool. So we need to ask ourselves why are we stimulating this region with a specific protocol? Is this region implicated in the symptom we are targeting or in the desired therapeutic effect? Does this protocol shift the brain in a positive direction? There is a large body of literature to review that can inform these decisions. Don’t simply apply high-frequency stimulation to DLPFC and expect it to work for everything.
- Pick patients who have the best chance to show the desired effect (see biotypes above). If the efficacy of a brain stimulation protocol presumes a specific brain-based baseline dysfunction, make sure the patient has that dysfunction. There is not always a one-to-one correspondence between symptom and brain dysfunction.
- State dependency is critical to both the safety and efficacy of TMS. These states include drug use and drug withdrawal, sleep, attention, symptom provocation, and arousal state which impact both risks for side effects as well as the engagement of certain brain regions. A still outstanding question is which state is best for synergistic effects? Do you want to provoke symptoms (as many OCD and PTSD trials have tried) or focus on positive or productive cognitive states as is sometimes the case in depression?.
- Control (as best as you can) placebo effects. Many trials have failed, not because the treatment didn’t work, but because the placebo worked just as well. It is true that coming into a clinic and receiving what you think may help you from a researcher or clinician who genuinely cares about you and asks you about your symptoms is therapeutic. But so is taking medication (if you were previously noncompliant), being less treatment-resistant oriented in the first place (for some people these nonspecific therapeutic benefits are all they need), or intentionally or unintentionally giving the placebo group active therapy (via talk therapy or otherwise). Effect sizes increase not only by improving the treatment effect but also by reducing the placebo effects.
And last, but certainly not least take-home from the Brain Stimulation conferences was the safety of TMS and the importance of accurate characterization and documentation of side effects. As far as we know the only serious adverse event associated with TMS is the exceedingly rare induction of a seizure. The exact number of seizures and exact risk is difficult to quantify as many seizures may not be reported or mischaracterized (leading to uncertainty in the numerator) and the number of total sessions or total number of patients is often estimated as not all are documented in a way that can be queried (leading to uncertainty in the denominator). That being said, two recent efforts (with papers forthcoming) have attempted to survey TMS clinicians and have come to similar conclusions. Risk of seizure is estimated to be between 1/1000-1/10,000 patients. Factors known to contribute to seizure risk (such as drug/drug withdrawal, family history of epilepsy, brain pathology, and parameters which exceeded safety guidelines in terms of intensity, frequency, and intertrain interval) often contribute to the risk. That being said, when compared to antidepressant medication, TMS is as safe or safer than many pharmacological agents commonly used to treat depression. Thus for those who do not benefit from medications or psychotherapy TMS represents a safe and often effective option.
In conclusion, the future of TMS remains promising. There is a growing number of smart, dedicated researchers and clinicians interested in developing and testing novel protocols across a range of neuropsychiatric conditions. There are device companies and engineers pushing the technology in new directions including improved focality and depth, “closed loop” systems that trigger stimulation only when the brain is in a certain state, integrated neuronavigation systems, and additional software and hardware designed to improve user experience and patient comfort. I continue to be proud and honored to be working in this field amongst esteemed colleagues and friends.