Danny Pine, M.D.: Can clinical fMRI research transfer to the clinic? (NIMH Podcast)

Summary

For 25 years, functional magnetic resonance imaging or functional MRI (fMRI) has generated excitement in the research community and public, shedding insight into systems level organization of the brain in healthy and clinical population. But can fMRI provide actionable information or have a meaningful impact on clinical practice? NIMH’s Dr. Peter Bandettini talks with NIIMH IRP psychiatrist and fMRI researcher, Dr. Danny Pine, about how and when fMRI might ‘go clinical.’

Transcript

Time: 00:30:43 | Size: 42 MB

Title: Danny Pine, M.D.: Can clinical fMRI research transfer to the clinic?

Description: For 25 years, functional magnetic resonance imaging or functional MRI (fMRI) has generated excitement in the research community and public, shedding insight into systems level organization of the brain in healthy and clinical population. But can fMRI provide actionable information or have a meaningful impact on clinical practice? NIMH’s Dr. Peter Bandettini talks with NIIMH IRP psychiatrist and fMRI researcher, Dr. Danny Pine, about how and when fMRI might ‘go clinical.’

>> PETER BANDETTINI: Welcome to the Brain Experts podcast, where we meet neuroscience experts and talk about their work, the field in general, and where it’s going. We hope to provide both education and inspiration. I’m Peter Bandettini with the National Institute of Mental Health, which is part of the National Institutes of Health, Department of Health and Human Services. Views expressed by the guests do not reflect NIMH policy. This is episode one. Danny Pine, M.D.: Can Clinical fMRI research transfer to the Clinic? Let’s chat.

>> PETER BANDETTINI: In today's podcast, I interview Dr. Danny Pine who leads a Section on Development and Affective Neuroscience called SDAN at the National Institute of Mental Health. He received his MD from the University of Chicago and trained at Columbia University. His work here since about 2000 has examined the neuroscience of normal and abnormal fear, including related symptoms of anxiety. His emphasis is on development in children and adolescence. His particular focus is on using insights from neuroscience to inform understandings of risk, diagnosis, and treatment. His use of neuroscience applies brain-focused tools through three kinds of studies, which we'll get into the podcast. These include understanding the heterogeneity in clinical aspects of anxiety disorders, brain-based predictors of response to treatment, and implementation of trials that use novel therapies discovered through basic research. He has been prolific publishing about 500 papers in the past 10 years with an h-index around 120, which means that he has 120 papers with at least 120 citations.

>> PETER BANDETTINI: Okay. All right. So today, we're here with Danny Pine, and Dr. Pine is leading the Section on Development Affective Neuroscience, SDAN, in IRP. He's at the NIH. And you've been here about the same amount of time that I've been here. We both started in the early 2000s or 1999. Well, I started in '99, but I think you started maybe a year later or something like that. And you've been working on more the clinical end of neuroimaging. I've known you for all these years, and you've been doing a lot of functional MRI, and you've been doing other things as well, but you've really been honing your functional MRI research. And so maybe backing up a little bit and telling us how you got started and what made you interested in brain imaging and what made you finally decide to sort of shift more towards research as well?

>> DANNY PINE: Yeah, so probably the first thing I would say is, that for as long as I can remember, I wanted to be a research physician. That was always my goal. And I always wanted to work with children. So all my life I had always worked with children. I began exploring all the different aspects of helping children that were interesting. In medical school, I pretty much tried every single one.  I went to New York at the medical school really to start pediatric training and to get trained as a child psychiatrist. When I got to Columbia, it was this totally different world that I had no idea even existed. It was this amazingly rich combination of kind of mental health advocacy, on the one hand; psychiatry and psychology in this very deep way, on the other hand; and neuroscience on the third hand. That all the most brilliant people, this is what they were doing. And I really took off in terms of thinking about what I wanted to do, and that's where I really decided how I would be an academic.

>> PETER BANDETTINI: So you decided to get into more the methods. So what was your thought? I mean, was your thought that we'll develop better ways of more precisely predicting outcome or giving treatments? And somehow if you have a brain-based phenotype or brain-based marker - not just behavior, not just a questionnaire - that will give you insight?

>> DANNY PINE: And at the time - this is 1990 - functional MRI did not even exist. So I really was trained as primarily a psychophysiologist, an epidemiologist, and a clinical trialist. So I spent 10 years doing that work at Columbia. So we kind of learned from epidemiology that you could define a group of at-risk children, and if you followed those children over time you would capture virtually all the adults who would have problems, but most of the kids that you would identify they would do fine. So very few of the kids who were doing really well would develop problems. A lot of the kids who were having trouble would have problems over time, but a lot of them would go away. And our ability to predict was just terrible, and no matter what we did, no matter what we thought of, no matter how many measures we collected, the amount of variance you could predict was very little. So that was one observation. And then the other was the same thing. With treatment, you saw that overall treatments generally worked. I did some of the early SSRI trials in kids and they were much stronger than placebo, but again your ability to predict which kids would respond to an SSRI, which kid could would respond to placebo, which kid would respond to cognitive behavioral therapy, it was really, really not very good. So this is mid-1990s, and functional MRI had just started. And you and I, actually, our paths almost crossed in Charleston. I want to say it was maybe 1996 that I went and I took that fMRI course with Robert Savoy.

>> PETER BANDETTINI: Yeah. I was helping to teach it.

>> DANNY PINE: That's exactly right. So this is 1996, 1997. And I started thinking, "Okay. This kind of--" because I was doing a lot of experimental work in psychophysiology, so I really started to think that fMRI was really the way to go, but that there was going to be this huge kind of methods development stage. It's interesting. I had no idea at the time - I don't think anybody had an idea at the time - how long it was going to take, and--

>> PETER BANDETTINI: We're still at a methods development stage.

>> DANNY PINE: That's exactly right. We're still in a developmental stage. That's exactly right. So I kind of went back to Columbia, and I said, "This is really what I wanted to do." I came here in 2000 … it was a major change. The last thing I'll say about that, the way that I thought about the change when I was at Columbia it was very much heavily focused on neuroscience on the one hand. People in the psychiatry department looked at me as a basic scientist. They thought I was a basic scientist because I wanted to do all this methods development in imaging. But then when I came here, and it's very much like this right now, I'm one of the few clinicians. I really am a clinician. That's what I am at heart.

>> PETER BANDETTINI: Yeah, yeah. I mean, do you still practice as well?

>> DANNY PINE: I practice a lot. A lot. I would say probably about 25 hours a week I'm actually delivering care, and it's about 20-ish hours here at the NIH. So every single kid that comes into my study, I assess every single one. I know them in a lot of detail.

>> PETER BANDETTINI: So pretty much there's this interplay between your subjects and your treatment as well?

>> DANNY PINE: Exactly. Exactly.

>> PETER BANDETTINI: So you decided to get into more the methods. So what was your thought? I mean, was your thought that we'll develop better ways of more precisely predicting outcome or giving treatments? And somehow if you have a brain-based phenotype or brain-based marker - not just behavior, not just a questionnaire - that will give you insight?

>> DANNY PINE: Yeah. I guess there were two general thoughts. One, and probably the biggest, dominant one that remains, is that when you look at the history of medicine, and you look at the kind of quantile changes in survival and ability to predict outcome, they all really came when people got at mechanism. So when you looked at infectious disease once you developed germ theory, antibiotics made a huge change. If you looked at cardiovascular illness once you understood pathophysiology of ischemic vascular disease. When you looked at cancer, it was when you understood the process. And so I had the thought that, "Well, the brain is obviously the organ that we need to study. We need to study it directly. We need to get a direct measure of it, and least get going on that." At the time, and I think it's still the case now, fMRI was the best tool of the time. So that was one big thought. The other big thought and this hasn't happened yet, again, when you looked at other areas of clinical medicine what happened with progress was when the basic scientists and the clinical scientists were talking about the same phenomenon, I kind of knew and still know that we cannot just talk about behavior. We have to talk about brain-behavior connections. And the problem is even now, and I don't know if it's the scale of fMRI isn't close enough or if neuroscience keeps advancing so far ahead, or the incentives aren't right in terms of focusing on clinical things I kind of knew that we needed to talk about brain-behavior relationships with a direct measure of brain function that at least came close to what basic scientists were working on. The other thing that again was kind of an inkling when I started doing this that now we know much better is that there's this tremendous heterogeneity in clinical presentation. So you can see a clinical presentation in all kinds of vivid detail, and then you can see another one in another kid that looks very, very similar, but the response to treatment in these two identical kids will be radically different. And we think that the behavior is kind of a superficial manifestation of what's going on.

>> PETER BANDETTINI: Yeah. So if you image them, you'll see maybe a radically different activation pattern. That seems like still a challenge, in some sense, because there's a lot of heterogeneity across subjects in terms of their activation patterns.

>> DANNY PINE: Exactly.

>> PETER BANDETTINI: I was just talking with Francisco Pereira trying to understand, how can you apply fMRI to individual subjects? You can actually draw inferences about groups really well, and that takes maybe only 50 or 100 subjects or something like that, but individual subjects-- Do you envision ultimately being able to put someone in the scanner and say, "With this set of tasks, we're going to give you this drug, or you'll have this likelihood of being treated properly?"

>> DANNY PINE: I think we're still fairly far from having clinical applications in fMRI … So then the question becomes, well, if we're really far away from doing this, why even bother? And I think that that's a legitimate question. So something that I spend a lot of time on is that I try to develop ideas about novel treatments based on what we know about the underlying circuity … that we learn a lot about the neural circuitry from basic research, and it gave us a lot of ideas about using principles of learning to come up with novel treatments that then we tested. And in the work that I do, I'm very interested in implicit learning because it turns out that the underlying circuitry, particularly of attention dysfunction in anxiety, a lot of the problems are in implicit systems. So that led to all this work kind of using training techniques. So there, the one approach is using neuroscience to understand the brain--

>> PETER BANDETTINI: Just to back up really quickly, can you define implicit learning, as opposed to explicit--?

>> DANNY PINE: Yeah. When we think about learning, we kind of divide it into two general ways that we learn. One way is that we call explicit or declarative. And what that means is that you can describe what you're learning. So it's basically how we usually think about learning in school. You read something. It's described to you with language. The individual can tell you exactly what they're doing. And when we think about psychotherapy, it's usually what we think about in psychotherapy. You tell somebody when you're feeling nervous, you should look over there, or you should think of this, and you remember that and you can describe it. So that's kind of declarative learning. Implicit learning refers to things that you cannot describe.

>> PETER BANDETTINI: Like a skill--

>> DANNY PINE: Like a skill. That's exactly right. So probably the best common analogy is when you learn to drive a car, and when people have been driving for many years they kind of realize that they can do five or six other things while they're driving, but if they tried to do that the first time that they drove, they would wreck. And that is an example where they've implicitly learned how to drive a car. You could spend 25 hours talking to somebody about how to drive and they wouldn't improve their skills. It turns out that attention and anxiety is just like that. It works just like that. And so I think that's probably the lowest hanging fruit for imaging that it's going to generate these new ideas. And then related to that, I think if we can get basic scientists to try to better understand those clinically relevant phenomenon I think those ideas are going to be even better. They're trying to prioritize behaviors that clinicians have shown are both clinically relevant, but also understandable from a brain function standpoint. I think that's probably the place where it's already taking off. It's not usually what people think of because you generate the ideas for the treatments from neuroscience but when you examine the therapies, you're not using the neuroscience tools.

>> PETER BANDETTINI: Yeah. Yeah. Can you give an example of a therapy that involves implicit learning?

>> DANNY PINE: Yeah. Yeah. So probably the best example in neurology is when patients have brain lesions, and they develop problems using their motor skill. You don't just tell people what to do. You do a systematic series of exercises that are tailored towards the functions that that person still has, and they practice them over and over and over again to build up the skill. So that's a form of implicit learning. The therapy that we use relates to attention orienting, and we found that children with anxiety disorders have this problem in their implicit brain function and that we use video games where they have to practice an attention reaction over and over and over again. And when you tell them what to do, it really doesn't help. It doesn't make a difference.

>> PETER BANDETTINI: Yeah. They just learn it.

>> DANNY PINE: Well, it turns out their symptoms are better. Their symptoms get better.

>> PETER BANDETTINI: Wow. Cool.

>> DANNY PINE: Yeah. So we've got a bunch of those things that we're trying to do. I don't want to over-hype the benefits. They're not to the point where I would widely disseminate them to everybody, but the treatments are getting better with each subsequent one that we do. So one way is that we develop ideas from imaging. The other way, which I think is probably the next in line, would be that you actually stratified people at baseline, and that you say that based on the pattern of brain imaging perturbation they have you assign them to this treatment or that treatment, or maybe you monitor their treatment with that brain imaging parameter, and you adjust the treatment based on what their brain profile looks like. I mean, my feeling about that - and this is something that you and I were talking about a little while ago - the methods are still developing, and I'm not sure that the magnitude of the association between the brain profiles and the symptomatic outcomes are nearly strong enough that we're going to be ready to do anything like that at least for 5 years but more like 10.

>> PETER BANDETTINI: Yeah. So try to develop a litmus test for the brain?

>> DANNY PINE: Exactly. The other thing I do for that when thinking about those ideas - again, a little bit like we were talking about before - I think of other areas of medicine. And so I look at Alzheimer's disease, and I look at where brain imaging is in Alzheimer's disease, and it's clearly far ahead of where we are in psychiatry. The third one is treatments where you're actually going to stimulate the brain, and you're going to use imaging to guide where you stimulate. And there are a couple of examples already. TMS. Deep brain stimulation. There's something in kids called tDCS. Transcranial Direct-Cortical Stimulation. They're more invasive. They're, I think, less widely applicable.

>> PETER BANDETTINI: Yes. Yeah, yeah. Right. So looking at interventions, either there's drugs or like you said neuromodulation. What I've heard more and more is just simple exercise.

>> DANNY PINE: Yeah. I do think exercise is a very good area to pursue, but the research hasn't dealt with those two huge problems. Finding a credible control, number one. And then number two, dealing with the fact that the sickest people don't do the intervention.

>> PETER BANDETTINI: If there was some way of exercising a person

>> DANNY PINE: or maybe giving incentives, or maybe developing a coach.

>> PETER BANDETTINI: Yeah. Yeah, something like that.

>> DANNY PINE: Something like that.

>> PETER BANDETTINI: So as far as imaging is concerned, you're producing a lot of research related to imaging. You're pushing the field. What do you think the final-- maybe, eventually, fMRI will just flatten out and reach its peak in terms of what it can do, what it can offer. Where do you see that? Where do you see that end game?

>> DANNY PINE: You're probably better than me in predicting where it's going to go in terms of how much more advanced we can get in the method. I would say two things. One thing I would say is that fundamentally it comes down to, how much of a behavior trajectory can we predict with measures of brain function. I think we need more data, not less. I think we need to kind of use what we already know about how to predict people's behavior. So we need to get richer fMRI data with more tasks and get every ounce of information out of the data that we can. I think that's the best hope with the current technique that we have.  I think one of two things is going to happen. In the way that I was just mentioning, we'll develop better ways with the current technique that we have to maximize the amount of variance that we can predict. It will be meaningful and it will be linked to a treatment, but it will only be in people who are treatment refractory. And then what'll happen is there'll be kind of stages that once you get to a certain point in terms of your treatment failures, your chronicity, then imaging will be used for that. Either that'll happen, or a little bit like what happened in Alzheimer's disease, where people used to talk about FDG. This was when I was in medical school, people were talking about using FDG clinically, but they really don't--

>> PETER BANDETTINI: A measure of metabolism?

>> DANNY PINE: Yeah, FDG's a measure of overall brain metabolism. It was only when chemical ligands for amyloid or for tau, the two proteins that go awry in Alzheimer's disease, it was only when those could be identified that it really became clinically relevant. So that's the other thing that could happen in imaging. There could be some--

>> PETER BANDETTINI: Some measure that's sort of under our noses right now that we don't realize?

>> DANNY PINE: Yeah. Exactly. Some major technical advance. There isn't a clue really about it now. Bcause when you look at the improvements, they're kind of graded. They're not--

>> PETER BANDETTINI: Right. There's no breakthroughs yet. I mean, there might be something we might be able to figure out how to see the lymphatic system or see inflammation in the brain with MRI.

>> DANNY PINE: Things like that. Things like that. Yeah.

>> PETER BANDETTINI: But as far as a biomarker, not even with Alzheimer's, but as far as a biomarker where you can put someone in and say, "Oh, yeah. You have this type of pattern of activation. It corresponds to this drug." Maybe if you could do that--

>> DANNY PINE: I do not think with the current techniques we've had over the last 10 years, and the ones that you and I can probably both see in the next 5, I do not think that's going to happen with these techniques. They just cannot predict any behaviors well enough, and I think that's a prerequisite. I mean, it's hard enough to predict normal variations in behavior. All these different varieties of pathological behaviors, I just can't see it happening.

>> PETER BANDETTINI: Yep. And I tend to agree. It's funny, I was just seeing this movie that just came out called Free Solo. This climber, Alex Honnold. And part of the movie, [crosstalk]--

>> DANNY PINE: Yeah. He lives, though. He lives, though. I read about it.

>> PETER BANDETTINI: Oh, yeah. Yeah, yeah. He did it. He climbed 3,000 feet. It's unbelievable. The point is, is that they actually had a part of the documentary where they put him in an fMRI scanner. They just put him in and they showed him intense scenes that would typically activate the amygdala of a normal person. They'd pull him out, and they were like, "Oh, your amygdala has flatlined." Immediately, the scientist in me is like, "You can't make that [inference?]."

>> DANNY PINE: Exactly. Of course you can't. Of course you can't.

>> PETER BANDETTINI: It's cool, [crosstalk], it's an observation, but you can't do that.

>> DANNY PINE: Oh, it's awful. I'll tell you the one thing that I spend most of my time on now. Reliability. I would never use an fMRI paradigm now kind of broadly until I was convinced it was reliable in terms of what I got today reasonably predicted what I get tomorrow on a scale that's similar to my clinical measures. It's pretty amazing how bad most of the fMRI measures are. They're just not very reliable. And that's a really good example. Amygdala activation to just about anything it's just not very reliable on an individual difference basis. That guy, he was, "Flatlined," as they say, if they brought him back in a week he might be white hot for all you know.

>> PETER BANDETTINI: Right. Right. Exactly. Exactly. So that week to week you have to a wide range of subjects.

>> DANNY PINE: Exactly. Exactly.

>> PETER BANDETTINI: Yeah. Maybe that specific thing. So as I mentioned before, your publications have been really prolific, and not only that, you've been pushing the field, but is there an example of either your work or is there anything that's sort of blown you away? Like, "Oh, this is really cool." Is there any recent work that you've done that you've been excited about or that you've seen?

>> DANNY PINE: I tend not to be a blown away kind of guy to tell you the truth.

>> PETER BANDETTINI: Okay. That's good.

>> DANNY PINE: I tend not to be. I tend not to be. Probably the thing that I am most excited about is that I do feel like the science of mental health is growing some healthy skepticism, I don't know if you read it. Did you see those papers in Nature last week about the UK Biobank? They scanned 10,000 people--

>> PETER BANDETTINI: Oh, yes. Yes, yes.

>> DANNY PINE: Yeah. And you look at the amount of variance that you could explain, and it was unbelievably small in terms in they did a whole genome--

>> PETER BANDETTINI: Right. And a behavioral profile and everything.

>> DANNY PINE: --and they mapped it to brain imaging data, and the amount of relationship was just really, really small. Yeah. So the fact that that's the kind of thing that's in Nature - published in Nature - I think that's really good, and that people are kind of skeptical about that. So the thing that I get most excited about are positive results in clinical trials. And you do see them, you don't see as many as you used to see, but people kind of getting on the same page and doing the same treatments, I wish that there was more of that.

>> PETER BANDETTINI: Yeah. It still seems right. It's still this early stage where each clinician is sort of like their own scientist. They try something, they kind of think about it, they try something else.

>> DANNY PINE: Yeah. The other thing I'm really excited about. I do really like the fact that it does feel-- compared to psychophysiology where I was studying heart rate variability, which was very interesting if you were trying to predict a heart attack, but not very related to the brain. Yeah. It's pretty interesting. I spent a lot of time doing that. It does feel a lot better that at least people are talking about the same things. We're talking about amygdala function, the basic scientists and the clinicians, so least we're on the same page.

>> PETER BANDETTINI: That's interesting. So it's at least getting somewhere, starting with something?

>> DANNY PINE: That's what I think. At least the foundations feel a little more linear.

>> PETER BANDETTINI: Yeah. Yeah. Okay. Okay. Right. So at least you have to have a firm grounding and say, "Okay. These are the important structures, or this is the important sort of thing?"

>> DANNY PINE: Exactly. This is the relevant behavior we want to understand. The associated brain structures that relate to this behavior. This behavior is relevant because it does relate to things out in the real world. At least coming to some agreements. And we're not there yet. I often compare us to the Human Genome Project. That was kind of a starting point. We're not even at that starting point.

>> PETER BANDETTINI: Right. Right. Yeah. Exactly. I mean, it seems like even things like ADHD, or--

>> DANNY PINE: Absolutely. Absolutely.

>> PETER BANDETTINI: And then people are trying treatments that are sort of black box treatments

>> DANNY PINE: I'm okay with that as long as you call it what it is.

>> PETER BANDETTINI: The cognitive training, bringing it all the way back to the beginning, isn't it a form of implicit behavioral therapy?

>> DANNY PINE: It is. It is. It is. The problem is that when you evaluate that cognitive training, that form of implicit learning, the criteria you have to use to judge it, it has to be clinical. It has to have a clinical impact. So it's not enough to say that people with ADHD have this problem with implicit learning. We know they do. We're going to improve it. And look, we've improved their implicit learning. That's not enough. You have to show that when you improve that implicit learning, you improve them clinically. That's not happening. People are improving their implicit learning, and they're saying, "This is a great treatment. Go use it." And people are not being sufficiently critical about that, about saying that, and about saying, "We're not ready to sell this."

>> PETER BANDETTINI: Interesting. Okay. Well, I think that we're getting close to the end.

>> DANNY PINE: There you go.

>> PETER BANDETTINI: Well, thanks.

[crosstalk]