Episode 33: Combating the “Multi-Dimensional Beast” of Chronic Pain

Chronic pain, according to a 2023 study, affects more Americans than diabetes, depression, and hypertension. Yet the disease is poorly understood, often undiagnosed or misdiagnosed, and effective treatments are in short supply.

A recent study in Nature Neuroscience provides new insights into how the disease affects the nervous system. For the first time, researchers recorded data from inside the brains of individuals who were suffering chronic pain and found distinct biomarkers for the disease. These insights are an important first step toward better diagnosing and treating chronic pain.

In this episode, the lead author of that study, Prasad Shirvalkar, a neurologist and interventional pain medicine specialist at the University of California, San Francisco, talks with managing editor Jason Lloyd about his research and how it could transform physicians’ understanding and treatment of what Shirvalkar calls a “multi-dimensional beast.”

SpotifyApple PodcastsStitcherGoogle PodcastsOvercast



(This transcript has not yet been corrected and may contain errors.)

Jason Lloyd: Welcome to The Ongoing Transformation, a podcast from Issues in Science and Technology. Issues is a quarterly journal published by the National Academies of Sciences, Engineering, and Medicine and Arizona State University.

Although chronic pain affects nearly 20% of Americans, the disease is poorly understood and effective treatments are in short supply. But a recent study in Nature Neuroscience provides new insights into chronic pain. For the first time, researchers recorded data from inside the brains of individuals experiencing chronic pain. This is an initial step toward developing better clinical treatments. I’m Jason Lloyd, managing editor of Issues. On this episode, I’m joined by the lead author of that study, Dr. Prasad Shirvalkar, a neurologist and interventional pain medicine specialist at the University of California, San Francisco. Prasad, welcome. Thank you so much for joining us today.

Prasad Shirvalkar: My pleasure. It’s so great to be here. Thanks for the invite.

Lloyd: Before we get to your research, just to define the terms, could you talk about what chronic pain is and how it’s different from normal or acute pain? I guess that might be the better term.

Shirvalkar: Chronic pain, which is a huge problem, not only in the US, but internationally, is defined as pain that persists for most days or every day for longer than three months. And so, it’s, by definition, persistent and enduring. And in fact, it’s so prevalent that one in five Americans, 20% of the US population, at some point in their lives, lives with chronic pain. And the real challenge is that we currently don’t have great therapies for many chronic pain syndromes. And in fact, most patients who live with chronic pain go misdiagnosed or undiagnosed for almost one year, which is really challenging. And so, chronic pain, according to a new study in JAMA, has a higher incidence per year, in the US, than diabetes, than high blood pressure and a higher incidence than depression, so it’s very common. And acute pain or pain when you stub your toe or bang your funny bone on the table, is really a different phenomenon. And so, I think of chronic pain as actually an individual disease.

It’s a disease state that develops through rewiring of the nervous system, of your brain and spine, over the course of three months, that starts to involve different dimensions of experience than just a somatic or what we call nociception. So acute pain, you can think of as, primarily, kind of an unpleasant experience that somebody has. And really, the point of acute pain, it’s like an alarm bell. So it’s like an alarm signal that your body uses to alert you to potential tissue damage or injury like a broken bone or a cut. And chronic pain, it’s when the fire alarm is going off, but there’s no obvious fire. And so, it’s, somehow, a maladaptive response of the nervous system over time that tends to amplify pain signals and then it starts to involve more emotional processing and interferes with relationships, interferes with mood regulation and control. It also affects cognitive faculties like people’s ability to pay attention or shift attention, it interferes with memory. So chronic pain really is this kind of multidimensional beast where something has hijacked the nervous system to amplify pain when it no longer serves a useful purpose for health.

Lloyd: You mentioned that treating chronic pain is really difficult. I’m wondering just what the current treatment looks like for chronic pain. I mean, often we think of treating pain with medication, taking Advil when you have a headache or something like that, but I think, I’ve seen elsewhere that you’ve talked about this, that there is, potentially, a relationship with behavioral or psychological therapies as well.

Shirvalkar: Yeah, absolutely. So I’m a neurologist and a pain physician, and so, I will say, across medicine, people approach treatment of pain quite differently, but, at least in pain management, we tend to have different buckets of therapies, let’s say, that we use for treatment. The first bucket is kind of conservative management. So if somebody’s had an injury in their leg or has had surgery like a hernia repair, but they continue to have pain six months later, what we typically first do is, we make sure that they’ve engaged in some kind of physical therapy, training your body to understand what signals are coming from certain locations, strengthening muscles, improving posture is all an important part of healing and even reducing impact of pain on life. Dietary changes are also really important. Another conservative therapy, which often can really help, psychological therapy. There’s a specific branch of pain psychology which actually teaches patients, one, how to recognize signals in their own body, and moreover, how to recognize how they’re reacting to that pain.

Pain psychology approaches can help a lot with coping, which can help with function. And so, the conservative therapies are all aimed at maximizing function as well as reducing suffering. But of course, if those don’t work, there’s a host of medications. We classify chronic pain in different ways. We call some types of chronic pain related to arthritis or inflammation, often we refer to it as nociceptive or inflammatory. If a nerve injury is involved or if there’s burning, stinging, stabbing, it falls in the category of neuropathic pain. Of course, there’s pain that’s deeper, from the gut or organs, yet a third type of pain called visceral pain. So different medications have better efficacy for different types of pain. And the third bucket, often we’ll try nerve blocks or even injections such as an epidural. Or, for example, if someone’s having headaches, we’ll often do an occipital nerve block.

These are transient. They typically last a few weeks and at best, they may last six months to a year. And last but not least, and this is really emerging as a new therapy, is kind of the idea of neuromodulation. That is nerve stimulation. We can stimulate peripheral nerves, we can stimulate the spine and we can stimulate even the brain. And I kind of think, in human experience and in the function of the nervous system, the buck stops at the brain. The question is, how can we identify where pain perception is being generated and then target those areas using various modalities? It’s been the case, so far, that a multimodal approach using more than one bucket is usually more successful than relying on any one therapy.

Lloyd: That really provides a good sense of the context in which you’re conducting your research. So let’s talk about your experiment. So what did you do and then what did you discover?

Shirvalkar: Right. So I run a lab at UCSF, and the main goal of the experiments that we did is to develop a new brain stimulation therapy for chronic pain. So we use something called DBS device and DBS, Deep Brain Stimulation. Deep Brain Stimulation is a therapy that’s FDA approved for Parkinson’s disease, for epilepsy, for movement disorders like tremor. However, it’s completely experimental for pain, but it was one of the first, at least in the research world, it was one of the first therapies that was actually tested for pain starting from the 1950s. There’s some pain syndromes that don’t respond to any treatments whatsoever, and they’re really refractory. So for example, there’s something called post-stroke pain. And so, when patients have a stroke, often, their brain will fill in half of their body with burning, stabbing, stinging sensation that may fluctuate throughout the day, but it doesn’t go away. Often, phantom limb pain is really hard to treat.

And so, our study was couched in a clinical trial that aims to try to find, number one, what are the best brain regions to stimulate to try to treat pain in folks where nothing else is working for phantom limb pain and post stroke pain. And number two, can we identify where in the brain there might be biomarkers or there might be signals that actually track pain in the effort to try to map out how these circuits communicate to actually generate pain. And so, our study that we recently published was a part of that overall clinical trial where we actually show, for the first time, that we can use electrical fingerprints or electrical signatures, recorded directly from the brain activity of four patients, that provide kind of objective biomarkers of their subjective pain state over many, many months. And we were able to do this with really novel technology using kind of a brain stimulation device that not only allows us to stimulate, but allows us to record electrical signals directly from electrodes that are implanted in certain brain regions that are connected to this device.

Lloyd: So how did you connect the signals that you were getting from the electrodes to their experience of pain?

Shirvalkar: It’s so hard to quantify a subjective experience. So what we did is, we implanted four patients with standard deep brain stimulator, DBS, electrodes in the anterior cingulate cortex, which is a region of the brain that’s thought to play a role in emotion or affect in general, but also related to pain, and in another brain region called the orbital frontal cortex, the OFC. It’s literally above your orbits, behind your eyes, behind your forehead, and that’s a region of the brain that’s been implicated in flexible cognition and decision making and planning. And so, we implanted electrodes in these regions and connected them to recording device slash stimulator that was implanted in the chest, and that’s pretty standard for DBS for other diseases as well, at least the construct, the setup.

These patients, after we implanted them, we didn’t turn the stimulator on because we wanted to identify these brain signatures. And the goal was to try to use their brain signatures, ultimately, in a kind of personalized or a on demand brain stimulation that responds to the signals in their brain. And so, those studies are ongoing. Like I said, after we implanted the patients, for up to six months, we ask them to report their pain symptoms multiple times a day, at least three times a day, and they did this through either an online reporting system or text message system. And what they do is hit a button on a remote control and their brain recording device would actually trigger a 30 second recording. So it would take a snapshot of the brain activity from these two brain regions on both sides of the brain, and they would save that snapshot. And then, of course, they would report all of their symptoms. So in each of four patients, we got hundreds of snapshots of brain recordings and we got hundreds of simultaneously reported pain scores.

And so, what we did is, we built machine learning models. We kind of trained these models to try to identify, okay, one, are there features in their brain signals that help us to predict their zero to 10 pain score? Zero being no pain, 10 is the worst pain you’ve ever had. It’s not a great metric, but it’s the most commonly used one. So we use that, it’s called the numerical rating scale, and then we also use a bunch of other metrics to see if they also correlated or they tracked it. It was in the wild, so to speak. They were at home, they were going about their regular lives, walking their dog, cooking dinner, et cetera, and they would give these reports in a naturalistic setting, which has never been done before. And so, what was really amazing is, what we could do very accurately is predict when individual patients would be in a very high pain state or a low pain state.

So if we dichotomized or broke up the reports into binary high or low pain, our models helped us identify which brain signals really helped to discriminate high versus low pain in each patient.

Lloyd: That’s really interesting. Okay, you said there were four participants in this experiments. Do you think if you had more participants, or have a larger study in the future, that you could get more granular with that predictive power of what you’re seeing on the brain scan? Like, rather than just sort of a binary, could you get closer to that 10 point scale on their subjective experience of pain?

Shirvalkar: That’s a really good question. I think instead of more participants, I think what would really help us to get more granular and to get more nuanced in predicting individual symptoms would be sampling from more brain regions. And I think our study was a proof of concept, it was a proof of principle. You can track signals from even one brain region and actually do pretty well on a forced choice kind of test or paradigm in low pain or high pain. You could still track that using, even, only a single brain region. Most of the studies in the past that have used brain scans like FMRI or Magnetic Resonance Imaging looking at blood flow, some famous studies have tracked pain fluctuations in healthy participants in response to a stimulus or response to experimental pain stimulus. And so, one, we know that that’s quite different from chronic pain, but two, what’s really remarkable is these studies often were able to predict the exact number, but they showed that in order to do so, you had to use up to 13 or 14 different brain regions in your model.

And so, pain is so widely distributed, the representation of pain is so spread out over your brain that I think there’s features across different brain regions that contain a lot of information that would probably help us to make more accurate models.

Lloyd: So from what you’re seeing from this particular study is, the experience of chronic pain in the brain, is it different than the acute pain that has been measured in the past in these other, like FMRI, study?

Shirvalkar: Yeah, so that’s a second part that we actually demonstrated in our study. So we included, in our study, an acute pain task. So we actually took those same participants and we asked them to come into the laboratory. And what we did is, we performed the same tasks that are very popular in the literature, specifically, one acute pain task that people use is, they have folks sit down and we basically are recording their brain activity while they’re doing this, but you give somebody a heat stimulus or thermal stimulus, so basically, it’s like a hot plate. You have them put their hand or some body part on a little probe, and that probe will heat up to, let’s say, like 45, 46 degrees Celsius or 48 degrees Celsius. Different people rate the zero to 10 intensity of pain in response to something painfully hot. People might have different thresholds.

So we calibrated like, okay, your six out of 10 for this acute heat pain, what temperature does that correspond to? And we do that for each patient. And so, we tried to, basically, do a bunch of trials where we exposed, in a kind of pseudo random order, we presented patients and kind of gave them this heat pain and asked them to report what number or pain they were feeling. When we generated models to try to predict the acute pain from their brain signatures, one, surprisingly, we could only accurately predict the acute pain in two patients. We couldn’t predict in all the patients. I think that’s probably a function that… We were recording from too few brain regions or maybe not the right brain regions, but nonetheless, even in those patients in whom we could predict acute pain, their electrical biomarkers or their fingerprints in their brain of what was representing high acute pain versus low acute pain, it was very different. It didn’t overlap very much. Like, you couldn’t generalize the biomarkers from the chronic pain to their acute pain state.

So this is something that’s kind of intuitive. I think doctors are like, “Oh yeah, we recognize that chronic pain is not just an extension or prolonged version of acute pain, but…” I think we demonstrated this, we provided direct evidence for this intuition for the first time based on the brain recordings.

Lloyd: I think, if I understood correctly, that what you found, the signatures that we’ve been talking about, were actually different for each of the participants. What does that mean for your study and what does that mean for potential treatments? Does everything have to be customized for patients if all of their signatures are going to be all different?

Shirvalkar: It’s absolutely right, that everyone’s signature was kind of unique, and that’s why I would refer to it as kind of a fingerprint. The really notable thing, and I think it’s important for science because in science, we hope that we do kind of a bunch of subjects or animal studies, you’ll do many animals and you hope that there’s some finding on the group level that generalizes so you can make some kind of inference about, “Oh, okay, this is something that is a mechanism in common.” So what we were really surprised to find was that there was one signal, one biomarker, that did generalize across all the patients. Even though most of their features in the brain were different, the low frequency oscillations or low frequency vibrations in the orbital frontal cortex actually seemed to be a common feature that discriminated when their pain was going to go up versus down across folks.

And so I think that’s really helpful, the fact that there’s some generalizable signal. The question is, will it generalize to many more people, people with other pain syndromes or other pain types? That remains to be seen, but you’re right. On the flip side though, I think, in a way, it’s very validating that everyone’s signature was different. Everyone’s experience of pain is informed by their upbringing, past traumas, perspectives, interpretations about what causes what, even expectations for improvement influence how they might perceive pain. So all these variables are so hard to control, and I think they probably account for why, or at least partially, for why these brain signatures are so different. For treatment or diagnosis, it’s a problem if there’s no generalizable signal, but my hunch is, though, that if we start to look at more brain regions, this is just the tip of the iceberg, this discovery, and I think it’s kind of opening up the possibility that if we start to expand across brain regions or even different syndromes, we will be able to learn something that’s generalizable.

And the hope is that even if people’s individual brain signatures are unique, that there’ll be some common property or some common kind of picture that emerges that will allow us to say, “Oh, based on your brain activity, your diagnosis is more likely this disease versus that disease,” Or, “Based on your brain activity, this treatment might work better for you than that treatment.” And so, what we’re actually trying to do is take these individual signatures for these patients, these biomarkers, and actually feed them into an algorithm on their device. So we’re using these research grade devices, so we’re trying to feed their brain activity to an algorithm that can be used to then control their stimulator. So if their pain is climbing or they’re entering a high pain state, the stimulator can turn on, for example, or if they don’t need it, the stimulator could turn off. And so, it’s almost akin to a thermostat for chronic pain. So basically trying to give on demand or adaptive brain stimulation in response to some signal that tracks individual symptoms.

Lloyd: Wow, that’s really cool. Is what you just talked about in terms of being able to measure someone’s brain activity in such a way that you can anticipate or know when they are experiencing chronic pain in a way that can then start the treatment if they have an implant like this, is that kind of the next step in treatment based on this research, or are there other things that this kind of signifies?

Shirvalkar: In the scope of the clinical trials that are going on, that’s our next step. The hope is that you wouldn’t need to have a hole drilled in your skull and have electrodes put in to actually get similar signals in the future. People are doing similar studies with FMRI or EEG. They kind of lack the resolution. I mean, you can’t really send someone home with EEG for three months. It’s hard to have reliable data, and FMRI, of course, you got to come in the lab. So the hope is that we’ll be able to use some kind of non-invasive method, in the real world, that would track symptoms or even right now, the Apple Watch or the Fitbit can get you EKG, and I really think that there may be something similar, whether it’s a headband or some kind of sensor you can put on your head that might be able to pick up a similar signal, one that we’ve established exists in the brain. So I guess the question is, how can we detect it outside the brain?

So I think the next step is understanding or trying to identify how can we do this less invasively and scale it, one, for diagnosis or two… I mean, biomarkers can be used for so many things. So the thing I mentioned was, in chronic pain treatment or in pain management, unfortunately, we’re still kind of in the era of trial and error. I know there’s a bunch of drugs, I don’t know which one’s going to help you, so we’ll kind of go through them one at a time. But the era of precision medicine, it’s lifted up so many fields already, but it has yet to make, I think, a big impact on pain medicine. And so my hope is that by identifying biomarkers like these, we can actually start developing precision medicine approaches where we’re using kind of data-driven treatment strategies for patients.

Lloyd: One of the things that people are so, I think, excited about, in terms of this research, is that managing pain is such a difficult thing to do. And as the opioid epidemic illustrates, it can be risky, really risky at both the individual and at the societal level. And then as you talked about before, there’s also the subjective aspect of asking people to report how they feel rather than having something that a doctor can look at and say, “Okay, here’s where you’re at,” Through this kind of objective measure. And so, I guess if I could ask you to speculate, where do you see this research kind of taking treatment in, I don’t know, maybe the next five years, maybe next 10 years? And maybe even beyond some of the diseases that you’re looking at, particularly in the study like phantom limb pain or post-stroke kind of pain. If I go get my wisdom teeth out, am I going to get a short term implant or a headband to wear rather than a prescription for oxycodone?

Shirvalkar: No, that would be wild, right? It’d be amazing if we could, one day, treat, even acute pain, with noninvasive stimulation of some kind. And it might be possible, but yeah, no, I could definitely try to speculate a little bit. It’s amazing how, in neuroscience, in the last decade, brain stimulation as an approach to treating multiple neuropsychiatric disorders has really exploded. And I’d say most of the research on the new indications for brain stimulation is probably being done on depression right now. There’s a lot of studies that are involved, including at UCSF and others that are involved in trying to help folks with refractory depression. There’s a lot of research on OCD, obsessive compulsive disorder. Also, folks are trying to treat addiction or substance use disorder. That’s still somewhat nascent, but I think the technology, as we learn more about different disorders, learn more about even how stimulation works.

I think that we fundamentally don’t even know, often, why stimulation works when it does. Even for diseases like Parkinson’s disease or epilepsy. We have models but they’re definitely incomplete. What’s going to happen in five years or 10 years? I think, hopefully you don’t get an implant to treat acute pain when you have your wisdom teeth pulled, but you’re right, we desperately, desperately need opioid sparing therapies. We need treatment for pain that are not addictive and that aren’t going to cause horrible personal problems for individuals, that aren’t going to cause huge social concerns like accidents and deaths from misuse or opioid overdose. One of the things, I think, that has to happen in the next five or 10 years, one… Opioids work. We know they work and that’s why we use them, but they’re obviously very dangerous.

So, one, we have to figure out how to treat addiction in a humane way, and part of that is reducing the stigma. I think in the next five to 10 years, I hope the concept of addiction or the concept of substance use disorder is no longer seen as a moral failing, or it’s no longer seen as a personal vice, but rather actually acknowledged, it’s a brain disease. If addiction is a verifiable biological disease that we need to treat and people shouldn’t be kind of discounted for having this disease. If we’re going to effectively treat addiction at all, I think that definitely needs to happen.

But two, I think with the advent of the new technology I was kind of talking about, we heard recently that, famously, in Neuralink, Elon Musk’s company, has IDE approval for doing human trials. I think brain simulation’s going to become more mainstream in the sense that people are going to be trying to achieve things on a larger scale in terms of disease modification, improving symptoms, and even, this worries me and it worries a lot of people, people are probably even going to try, if they’re not already, human augmentation. Try and make healthy people smarter, fitter, faster instead of just curing disease. And so I think we really need to have some, I’d say at a national level, policy discussions and guidelines in place to at least decide what’s acceptable, what’s worth pursuing, and to identify and clarify risks of this technology. But I think brain simulations can get more popular. I think the non-invasive approaches… Scientists and clinicians are going to start discovering more about how we might be able to use brain activity and influence brain activity without entering the skull surgically.

Lloyd: Yeah. So you had talked about, at the top a little bit, the prevalence of chronic pain. I think the Centers for Disease Control and Prevention, they did a survey and found that in, at least in 2019, one in five Americans has experienced chronic pain in the past three months. And one of the things I find interesting about that research is that it’s not distributed evenly across the board, and the experience of chronic pain is disproportionately focused in people who are non-Hispanic white, older, and more rural. And one of the things that wasn’t a metric for that study, or maybe they didn’t report it, but education level, I’m sure, probably, also correlates a little bit with the experience of chronic pain. That population kind of significantly overlaps with the one that Anne Case and Angus Deaton write about in their study of deaths of despair, kind of massive number of deaths in America from overdose, suicide and alcoholism.

So I guess my question, as it connects to your research, is that, it seems to be relatively easy to get kind of cutting edge treatments of the ones that you might envision in the next five years or to treat patients who are relatively well off, sophisticated and insured, and this population that suffers from chronic pain is probably not those things or does not have those characteristics. So I’m wondering how you think about that in terms of the clinical trial that you’re a part of, and maybe your research, more generally.

Shirvalkar: Yeah. Thanks for asking that. That’s so important. Brain simulation cannot and should not be therapies just for those who can afford it. But the sad reality is that, often, patients who have better access to healthcare, there’s studies supporting that they tend to be socioeconomically better off. I mean, it’s a fundamental disparity in access to healthcare. And so, no, that’s really unfortunate. One of the things that we tried to do for our clinical trials is make real efforts to recruit from underrepresented populations in scientific research to make sure that folks are well-informed, to make sure that money has nothing to do with it. So the NIH has been really generous in, at least, paying for everything for these patients, so it doesn’t matter if they have insurance or not. And so, yeah, the truth is, though, within the realms of a clinical trial, you can have this ideal setting, but you’re right, what happens once it gets approved? You’re right, this data from 2019 where they looked at variables that predict chronic pain, it makes sense that older adults have a higher incidence.

I mean, as you age, you have degeneration, your body’s break it down. So, yeah.

Lloyd: I’m experiencing it now.

Shirvalkar: Yeah. Yeah, me too, as I’m shifting in my chair. So I think that that’s explained. What’s interesting is that when you look at folks who are more likely to have chronic pain, their demographics kind of overlap with those of deaths of despair. There’s an interesting paradox there, in a sense, or maybe it’s not a paradox, but I think of chronic pain, almost, as a disease of despair. When you have acute pain, you expect, and you’re usually right, that the pain’s going to resolve and you’ll get on with your life, but because chronic pain pervades into people’s personal life, personal relationships, their goals, their expectations for the future, even their ability to work and basically pursue what’s fulfilling to them, chronic pain, almost by definition, is often associated with despair. And so, there’s an element of suffering that’s just not present in acute pain. If the incidence of chronic pain is higher in the same population that experiences deaths of despair, the question is, is chronic pain a consequence of preexisting despairing situation, or is it potentially the cause? I think that that important arrow, the directional arrow, needs to be worked out.

Lloyd: The causality. Yeah. So what’s on the horizon for you for this research? What are you looking at next?

Shirvalkar: Yeah. We’re just so excited and grateful for the attention and the exposure that this study has gotten, and I really hope that other real world studies like this will be done where folks actually try to measure pain in an ecologically valid way out in the world. And I think for us, we’re actually actively testing the hypothesis that these biomarkers are causal. So in the sense that, if we move the needle, if we could stimulate the brain, move the needle on the biomarker, can we move the needle on a patient’s pain experience? So we’re trying to, one, establish or test that hypothesis, and two, we’re really trying to integrate these and trying to fine tune the pain thermostat, trying to develop closed loop control algorithms that can help each patient based on their individual needs. And so, with success, always comes with more work. The hope is, we’re applying for more grants and planning experiments to honestly try to help many more people if we can.

Lloyd: Thank you for joining us, Prasad. This has been a really fascinating conversation about groundbreaking research and understanding brain signals for chronic pain.

To learn more about Prasad Shirvalkar’s research, we have a link to his lab’s website, his recent paper in Nature Neuroscience, and much more in our show notes. Have you experienced chronic pain? Do you have ideas about how this research could be used? Email us at [email protected] with your thoughts, and please subscribe to The Ongoing Transformation wherever you get your podcasts. Thanks to our podcast producer, Kimberly Quach and audio engineer, Shannon Lynch. I’m Jason Lloyd, managing editor of Issues in Science and Technology. Thank you for joining us.