The mechanisms behind transcranial focused ultrasound (tFUS) for neuromodulation remain elusive. There are several hypotheses, but conflicting results across a range of experiments in both animals and humans have hampered a nice, clean model which permits simple predictions. Some studies report inhibitory effects, others facilitatory effects. Some studies suggest tFUS has effects which persist for minutes to hours following sonication, while others report little to no activity at all. What’s going on? Might a lot of the confusion be coming from multiple sensitivities to ultrasound within the brain?
I’ve been pondering the involvement of the vasculature in tFUS experiments. Many papers seem to assume that tFUS triggers neurons in such a way that the standard neurovascular coupling model should follow, with a causal chain of events leading to functional hyperemia the way we think of task fMRI. But what if the vasculature is also impacted directly by ultrasound? Or, in the limit, what if it’s the vasculature which has the dominant sensitivity to ultrasound and we have vascular-neural coupling, not the reverse? Might tFUS produce transient vasodilation or brief, local ischemia in the tissue? Shouldn’t these possibilities be explored?
While cell culture experiments have demonstrated that neurons are directly sensitive to ultrasound, in a live animal the presence of vasculature complicates the potential mechanisms. Notably, arteries are responsive to changes in global and local blood pressure. They’re compliance vessels responsible for regulation of blood flow within the brain and mechanosensitive ion channels are widely expressed in vascular smooth muscle cells (VSMCs). Given this natural sensitivity to pressure, it is reasonable to speculate that tFUS may have some direct effects.
All this has got me thinking about the sorts of experiments which could separate neurovascular (neuron-first) from vascular-first mechanisms. Cererebrovascular reactivity (CVR), e.g. the response to hypercapnia, might be a good candidate probe of vascular involvement. So for the rest of this post I will develop some potential experiments which might help clarify the actions of ultrasound applied to the human brain.
Experimental components
For what I have in mind we will need the tFUS perturbation of interest, a CVR challenge, and some sort of simple task to assess neural effects.
The tFUS perturbation could be any protocol someone thinks is interesting. Targeting the motor cortex offers several attractive features, including reaction time as a behavioral measure, a robust BOLD response to be exploited in a task fMRI experiment, and the option to use transcranial magnetic stimulation (TMS) as a probe of tFUS activity outside the MRI, if desired.
For CVR we need a vascular challenge and a way to detect it. If we use hypercapnia as the challenge then we would ideally target a specific inspired CO2 over a subject’s end tidal CO2, but a simple breath holding task might suffice. To keep things simple, I’ll ignore the details and stick with the basic concept: that we have a way to produce hypercapnia in a controlled fashion. I’ll also assume that the hypercapnic response time can be measured until it attains a steady state maximal value.
BOLD is a convenient probe for assessing CVR, although ASL could be used. Again, for simplicity I’ll assume a BOLD time series with typical parameters, e.g. a repetition time (TR) of 2 seconds and voxels of around 3-mm on a side. The TR is sufficient to resolve the temporal dynamics of both CVR and task fMRI, and the voxel dimensions should permit evaluation of the brain response to a smallish focal target of ultrasound.
These considerations make a visuomotor task a good, basic fMRI experiment for evaluating both the neural and vascular consequences of CVR and tFUS. For example, we might employ a visual task in which the subject must press a button when a certain target appears, allowing a reaction time measurement. The BOLD response in visual cortex would presumably be unaffected by tFUS applied to the motor cortex.
Thought experiments
Now that we’ve got the basic tools, let’s develop some experimental designs and think through the implications of each part.
The first step is to determine the CVR to the hypercapnic challenge alone. In this initial test we are most interested in the responses in visual and motor areas, for comparison with later steps, but many other regions could also be evaluated as controls. We would want to evaluate both the CVR response timing as well as the magnitude. Then, once we know the baseline CVR we can apply our tFUS protocol before and then during periods of hypercapnia. How long before hypercapnia should tFUS be applied? The exact timing would presumably be related to the expected response to the tFUS protocol of interest. Certain tFUS paradigms might be expected to have persistent effects lasting minutes or tens of minutes.
What might these initial CVR experiments reveal? Perhaps tFUS prior to hypercapnia will reduce the subsequent cerebrovascular reactivity in the motor cortex. Alternatively, it might facilitate a greater subsequent CVR if VSMCs or pericytes are relaxed by tFUS. And of course it might make no difference at all.
When tFUS is applied during steady state hypercapnia we might see nothing at all if the primary mechanism of tFUS is also to produce vasodilation, or we might see a focal reduction of regional blood flow if the ultrasound acts primarily as a vasoconstrictor. We might even see a slight enhancement of vasodilation in the motor cortex if the ultrasound triggers neurovascular coupling there.
Next, we need to evaluate the effects of hypercapnia on the visuomotor task. The task response must first be assessed during regular breathing, and then in the presence of steady state hypercapnia. Our primary ROIs are the responses in the motor and visual cortices. We need to assess the task-induced functional hyperemia in the presence of vasodilation - is it additive? - and assess whether task performance, e.g. the reaction times in the motor task, is affected by hypercapnia.
Now we can combine all the parts. First, we should apply tFUS to motor cortex during task blocks and assess visual and motor cortical responses, as well as task performance, with normal breathing. We should expect the visual cortex to have the same response with or without tFUS, while the motor cortex should be perturbed according to the tFUS protocol being evaluated. And finally, we can repeat the experiment during steady state hypercapnia. Responses in the visual cortex should again be the same with or without tFUS. In the motor cortex we will see the combined effects of tFUS and hypercapnia on the task performance and on the functional response to the task. Does the tFUS paradigm work in the same way in the presence of vasodilation from hypercapnia, or have we managed to alter things somehow?
There are other approaches we might consider. If hypercapnia produces interesting effects, perhaps it would be useful to test hyperoxia and see if opposite effects can be produced, for example. Or, it might prove to be more interesting to try titrating the effects of varying degrees of hypercapnia if it turns out the vasculature is involved but not in an all-or-nothing manner.
Complicating issues
What about confounds? It’s well known that tFUS can produce auditory effects, so we would need appropriate masking and a good sham for control blocks. A more interesting confound might arise from the magnetic field. Lorentz effects inside the MRI may alter the dominant tFUS mechanism, suggesting that the entire experiment might need to be repeated outside the MRI. In that case we would need to replace BOLD as the measure of CVR if we want to measure directly the effects of tFUS in the vasculature. Alternatively, the effects of tFUS on task performance could be evaluated with and without hypercapnia while assuming vasodilation, and the task design modified or extended to include other brain regions.
Something else I’ve been pondering is the involvement of large versus small vessels. Large feeding arteries are highly reactive compliance vessels, and I’m wondering if tFUS intentionally targeted on an artery might produce interesting results across an entire vascular territory. I would like to see someone try sonicating an artery and measuring an angiogram and a CBF map, e.g. with ASL. Few studies map the vascular system prior to applying tFUS, so it would be all too easy to sonicate a rather large artery inadvertently and, if it reacts, alter the vascular dynamics of a sizable region of the brain. A hypercapnic challenge could also be employed, to test whether tFUS alters the reactivity in that vascular territory.
And then what?
Whatever the results, there are clear implications for tFUS. At a minimum, demonstrating no significant vascular effects would have safety implications, especially for certain populations with compromised blood flow who might be prime candidates for tFUS therapies. But the complete absence of direct vascular effects from tFUS would be a surprise to me. Right now, my suspicion is that there should be some interesting vascular effects. They might even be separable from neuron-first perturbations, offering the possibility of tunable responses. Neuroscientists might be disappointed, but those of us with an interest in neurophysiology would perk right up!
___________________
PS I haven't included references yet. I may add them at a later date. I'm not totally up to speed on the tFUS literature so what references I do eventually include will likely be incomplete. Apologies if that puts anyone's nose out of joint!
No comments:
Post a Comment