TMS in Neurologic Diseases: TMS - a tool and a therapeutic strategy
May 4, 2014
Highlights from the address by Alvaro Pascual-Leone Summarized by Ted Williams, MD and John Fleming, MD DLFAPA
There are many potential TMS applications out there. The numbers of publications on brain stimulation continue to rise annually. Only about one third relate to psychiatry. The others are all over the place in part because TMS is a tool that one can apply to any disorder of brain functioning and by extension many problems in overall health. However understanding and interpreting the literature is complex process.
How should we interpret studies if they include only small numbers of patients, some with positive and others with negative results and the studies lack replication? What if studies on the same illness have not have targeted the same circuits because the methodology was disparate? What if the illness studied is itself diverse, such as in epilepsy, Alzheimer’s disease or Parkinson’s disease? In such diseases, differing problems with differing circuits in different patients can complicate analysis if patient specific factors are not taken into account in the study design. Further, some disease states may be helped most by using TMS in combination with other interventions so that appropriate modifications in brain plasticity can lead to therapeutic outcomes. Finally, which potential applications should become part of clinical practice and which should remain in the field of research? And is there a way to make progress in both clinical practice and research simultaneously? These questions were addressed by Dr. Pascual-Leone in his presentation.
TMS is two things: a tool and a therapeutic strategy, often at the same time.
As we look at TMS, two principles are operative: TMS is both a tool to learn about circuits as well as a strategy to modulate diseased circuits in a therapeutic way and induce durable neuroplastic changes in the brain. It is important to recognize that TMS does not only produce local changes. TMS stimulation spreads throughout a circuit network. When used as a mapping tool, TMS used in a pre-surgical, noninvasive way produces data with a resolution that matches within a couple of millimeters that determined by invasive techniques. When used therapeutically, TMS allows us to learn about the circuits that underlie the substrate of symptoms of neuropsychiatric diseases and give us the opportunity for targeted interventions.
The specific application of TMS as a therapeutic strategy for migraine is interesting. Only 10% of treatment resistant migraineurs have aura. But at least so far, TMS treatment during aura is essential. Thus any TMS device for this purpose would need to be readily at hand and therefore the device has to be portable. In other words, you need to carry it with you. “This has huge implications for the field because it illustrates the possibility of a commercially viable device that is self-delivered and home usable.” And the availability of such devices could help with other conditions, at least theoretically.
The data for epilepsy looks good under one condition: the cortical malformation associated with the focus of the seizure must be able to be identified and reachable in a targeted way by TMS. In that case, the results are very positive. The results of studies for multifocal, primary, generalized seizures have been disappointing suggesting that unless one can indeed localize and target the patient specific lesion(s) generic application of TMS will be ineffective. The work on epilepsy underscores the important need for larger, well-designed follow up trials with larger samples sizes to verify the hypotheses generated by smaller studies. “Twenty small proof-of-principle, hypothesis-generating studies with small sample sizes are just that, twenty interesting studies. We cannot really generalize to the population and then go back to the individual with this type of data.”
High frequency stimulation to the motor cortex is effective in some patients, but not others. Why? The likely reason is individual differences in networks.
A number of studies have been done in patients with Parkinson’s disease. Unfortunately, the wide variety of techniques and populations in the studies of TMS in Parkinson’s has made it difficult to generalize results. It appears that higher frequency stimulation of the motor cortex has produced beneficial effects, more so than targeting the cerebellum. In larger populations, however, only about 30% have a significant response. So the question is, “Why do some patients with Parkinson’s respond, and others do not?” The answer likely has to do with individual differences in networks. Our goal in treating Parkinson’s is to alter functioning in the Sub Thalamic Nuclei (STN) or alternately the Globus Pallidus Interna (GPI). Different individuals however have different networks requiring different interventions to lead to the same outcome. For example, we might be able to say, “Well, if we want to decrease STN or decrease GDI functioning, then we need to increase this person’s motor cortex, or we may need to suppress this person’s SMA.” We need to individualize treatment based on the individual’s own network configuration. Resting-state fMRI can non-invasively evaluate connectivity and could identify correlation areas that would then point to the area that needs to be targeted to alter STN or GDI functioning.
But even if we know where to stimulate, lots of questions remain. Should we suppress activity or increase activity? It remains difficult for us to know for sure at times what we are doing.
In the work to learn how to individualize treatment, it is worth thinking about the question, “What do we really know? Do we need to drive connected area in the same direction? When DBS is used in Parkinson’s does it really suppress activity, or does it in fact increase the activity?’” There is evidence that DBS in Parkinson’s does not suppress activity after all, but may actually be enhancing it. Does high or low frequency TMS increase or decrease activity in the target region, at the end of the circuit that we stimulate? We have known for a long time that individuals vary with regard to their motor systems. If we want to target yet some other area, what biological principles guide our thinking? We need additional biological insights to guide our strategies. Otherwise, we’re limited. Another consideration is that a big reason why we are seeing such differences in Parkinson’s could be the placebo effect that is enhanced in Parkinson’s disease. So it may well be that TMS as a standard treatment for Parkinson’s may be the wrong way to think about it.
The Paired-pulse TMS technique
What is ‘paired-pulse’ TMS? Paired-pulse TMS is using a single TMS pulse to condition the motor cortex to respond differently to a follow up pulse. A single TMS pulse of sufficient strength to the motor cortex will cause a measurable motor response. If a preceding pulse is given prior to the test pulse, the degree of response of the test pulse changes. The preceding pulse, or ‘conditioning’ pulse, may either facilitate or suppress the test pulse that follows, depending on when the preceding pulse is given.
Using Paired-pulse TMS technique to study TMS in neuropathic pain
Jean-Pascal Lefaucheur, in 2007, examined measures of cortical inhibition in patients with neuropathic pain. He hypothesized that neuropathic pain may be due to cortical hyper-sensitivity, mediated by a deficit in GABA-A, which mediates cortical inhibition. Lefaucheur went on to show that with repetitive stimulation at high frequency, but not at low frequency and not with sham, cortical inhibition can be enhanced, and is correlated with the subjective report of pain improvement. This is true for neuropathic pain, but not acute pain. Dr. Pascual-Leone has shown that about 60% of the subjective improvement is correlated with this measure of inhibition. Measurements of cortical inhibition could give us a valuable marker for treating neuropathic pain, but work needs to be done establish that both are true: that TMS inhibits neuropathic pain, and that paired-pulse responses are indeed a marker predicting response to treatment.
The problem of reliably demonstrating improvement
Using TMS in Alzheimer’s disorder is complex. Dementia is characterized by multiple functional deficits in multiple domains by the dysfunction of multiple circuits. Tests of working memory in patients with Alzheimer’s disorder show improvement when repeatedly given, a training effect. Such improvement in testing measures doesn’t demonstrate improvement in the dementia itself however. Therefore, demonstrating an actual positive treatment effect can be tricky. Having a biomarker would be helpful.
Short-latency afferent inhibition with TMS may be a useful biomarker
There is evidence in Alzheimer’s disorder that some of the dysfunction of central cholinergic neurons may result from excess in inhibition by short-latency afferent neurons. Stimulation of peripheral nerves inhibits the motor cortex, and the process is called afferent inhibition. When a TMS pulse of the primary motor cortex is preceded by electrical stimulation of a peripheral nerve, the motor evoked potential evoked by the TMS stimulus substantially decreases, a phenomenon known as short-latency afferent inhibition(SAI). Short-latency afferent inhibition (SAI) is likely correlated with central cholinergic transmission and has been found to be altered in Alzheimer's disease. Using paired-pulse stimulation, the SAI can be measured and may be a useful biomarker to determine if any TMS intervention is helping with the basic pathophysiologic cholinergic processes.
Activating a targeted circuit by a cognitive task concurrently with TMS may enhance that network
Since Alzheimer’s disease affects multiple networks, mapping onto multiple cognitive functions, one approach may be to target multiple networks and activate them by giving a cognitive task, concurrently enhancing them with TMS. We know that the cognitive training alone does not make a substantial difference in Alzheimer’s disease, and that stimulation alone does not, but the combination appears to have a significant effect. There are now two groups of sham-controlled trials utilizing such an approach. To summarize them: when patients receive the combination of real TMS with real cognitive training versus sham TMS with real cognitive training, a significant improvement occurs that is about twice the benefit of any medication currently available. “How durable is this? We do not know. Does it come with any costs? We do not know. But the combination is potentially transformative for the patient and illustrates the value of the combination and the possibility of targeting multiple networks at the same time.”
TMS is a ‘very cool tool’, and a promising treatment, but studies are needed
There is a long list of disease states that people have tried and will think of to try, with various numbers of subjects, “but they only really tell us that TMS is a tool, a fantastically cool tool. We like this tool, but we need to think about the details of what we are really doing, the details of the subjects we are doing it to, and the details of what we are delivering, because it obviously makes a difference.”
Controlled studies need to be done, but they can be done while helping patients
Appropriate studies are needed to establish efficacy. These should not be small, additive, meta-analyzed trials, but appropriately powered controlled trials. As to the question of whether the traditional trials are the best way to go, or the only way to go, Dr. Pascual-Leone would argue that the answer is “No.” “There is a real opportunity to do what oncology has done, and that is to treat patients while, at the same time, gaining new insights into this field.“
Studies need to be driven by hypotheses, not, “Let’s stimulate and see what happens”
Lots of small studies or even multiple meta-analyses can generate hypotheses. They can sometimes be right and they can sometimes be wrong. They are worth pursuing, but they do not establish the evidence of efficacy. In order to design a trial, one better have fundamental logic that guides the parameters because otherwise one is going to be faced with the same challenges that (we see) all of the time, which is, “Why do we do the things we do? What parameters should we use and on the basis of what?” There is such an infinite parameter space that trials need to be driven by some hypothesis, and hopefully the hypothesis proven to be right, so that at least one is testing something with some sense. It should be based on biology ideally, but (at least) it should have some internal logic to be followed.
Final thoughts – plasticity takes time, and combining treatments, or combining them with imaging, makes sense.
It seems pretty clear from all applications, both psychiatric and neurologic, that multiple treatments are necessary, likely because plastic changes take time. We need to be aware about the placebo effects. A combination of interventions is going to be proven the way to go in therapeutics, the combination with behavioral and pharmacological interventions and the combination of some neurological imaging measures and other biomarkers will most likely be needed.
Note: Unless in quotations, italics do not necessarily indicate a direct quote, but may be an approximation.