Editor's note: David Yarnitsky, MD, is chair of Neurology at Rambam Health Care Campus, and professor and head of the Laboratory of Clinical Neurophysiology at Technion Faculty of Medicine, Haifa, Israel. He is also the founding and current editor-in-chief of PAIN Reports. Dr. Yarnitsky's research seeks to understand how the nervous system processes pain, mainly using psychophysical and neurophysiological tools. His lab explores the modulation of pain via ascending and descending pathways, in healthy subjects and pain patients, in a variety of clinical syndromes, and how it can be used to improve pain alleviation.
In this interview, Yarnitsky chats with PRF Correspondent Caroline Phelps, PhD, a postdoctoral research associate, University of Arizona, Tucson, US, to discuss his path to pain research, his work on conditioned pain modulation, and more. Below is an edited transcript of their conversation.
What inspired you to do pain research?
That’s an interesting question. I finished med school and had to do my military service as a physician, and then I had to make a decision about becoming a specialist in something. It was between pediatrics and neurology because I liked kids, but the brain was very interesting to me.
I did like pediatrics, and I was on-duty in the hospital, staying up most nights. But then I got married and had my first baby, and suddenly sick babies became too disturbing. I decided to leave pediatrics and go into neurology.
In neurology, pain is interesting to me because it deviates from the solid lines of the profession. Neurology is very rigid in a way; neurologists look at “hard” things like weakness and reflexes, but this leaves little room for your imagination, for some free thinking, for a little more elaboration on what’s going on with your patient.
The sensory system seemed much more attractive to me than the motor system. There was a professor of pain in my hospital, Jesmond Birkhan, who was the head of anesthesiology and pain – a really dear person. He was very welcoming to me, and I started visiting his pain clinic. He led my first steps in the field of pain research, and it became so interesting that I stayed there.
What was the first pain research that you did?
The very first was a small clinical observation when I was a resident in neurology. There was a patient with a stroke, and I did the usual medical work with him, taking his history. He said that he used to have phantom pain and just noticed a few hours ago, when he had the stroke, that there was no phantom pain anymore. This really ignited my interest. I was intrigued.
So, I started looking around in the literature. In those days, it was not as easy as going to the computer; I had to go to the library. I found very little support in the literature for this patient’s experience, but it makes perfect sense. If you damage the region of the brain where the perception of pain arises, then the patient can lose the phantom sensation, which is a good thing. So I wrote a little paper; it was my first pain paper. It was published in PAIN as a case report.
My first real research was during my fellowship with Jose Ochoa in Portland, Oregon. This research was about heat pain thresholds, a cornerstone in pain psychophysics.
You’re well known for your work on conditioned pain modulation. When did you start working on that?
When I came to work with Jose Ochoa in Oregon, I got acquainted with quantitative sensory testing; it was really the very beginning of this field. Everybody was busy at that time with very simple, what we now call static, pain parameters. I kept working on these static parameters for quite a few years after coming back home to Israel. Then, I realized that pain thresholds are important specifically when there is neurological damage, like in neuropathic pain cases. In these cases, sensory loss or gain can be documented. However, it seemed to me that pain thresholds are less relevant for the description of the pain experience the patient is having.
We started looking for something that we could do in the lab that would mimic the clinical pain experience a little more. So instead of just using a single stimulus like a threshold – which is an observation at one point in time – let’s evoke a process. Let’s do something in the lab that will work similarly to how clinical pain works.
There was already data from two groups. One of the groups was in Paris, including Daniel Le Bars, Tony Dickenson, and Didier Bouhassira, and they had data on DNIC [diffuse noxious inhibitory controls – the idea that “pain inhibits pain”]. Many people worked on DNIC and developed the concept in animals, and the idea that you could test the ability of a person to inhibit pain by providing a set of stimuli became very attractive.
And then, on the other side, there was Clifford Woolf and his work on temporal summation [a phenomenon where people report increased pain in response to repetitive painful stimuli] and sensitization. So there are two processes: One is inhibiting pain, one is enhancing pain, and they are dynamic processes that we can evoke in the lab, and they seemed relevant to the clinical situation. This is wonderful. Now psychophysics can go forward instead of dealing with the simple static parameters of threshold and suprathreshold responses. We can now be a little more lively in the lab, evoke the pain, and see how the patient responds.
So we started working on the techniques. How do you do it? How do you get a sound pain inhibition or pain enhancement? How do you develop a protocol that would be fitting for most patients? These questions are not resolved yet; many labs use different parameters.
Then we had difficulty getting this work published because I got responses from referees that DNIC is an animal-related phenomenon – and we were dealing with people, not animals. So, some modifications of the term came up like “DNIC-like” – all sorts of weird names! So I thought maybe we needed a new name. At the European EFIC meeting in 2009 in Lisbon, I decided I would call a meeting the day before. Some 20 people came to this meeting, and we discussed how we should do human DNIC. I came with the idea for the name and it was accepted by the group: conditioned pain modulation [CPM]. Then we published this recommendation in the European Journal of Pain, and it was the most-cited paper in that journal both for that year and the year after.
Your work on CPM shows differences in this “pain inhibits pain” ability within the human population. What underlies these differences?
It’s very difficult to tell. The most relevant thing for people to know is that not everybody is going to get the pain inhibits pain response under standard test conditions. You could get a universal response if you use very intense conditioning stimuli, but for the standard stimuli that most people use, there’s a good one-fourth, one-fifth, or even one-third of people who are not going to inhibit but rather summate the two pains.
If you want to look at specific populations, we thought at certain points that women will have less efficient CPM, but although some studies claim so, this is not a universal finding. In my country we have many different populations: Jews of Western and of Eastern origin, Arabs, those emigrating from Africa, and so on. But we couldn’t find a very specific pattern for each one of these populations. So I think it’s mostly kind of an innate ability: Either you were born with a good innate ability to inhibit pain or not.
We saw that CPM can predict your forthcoming pain experiences, although it’s not the only factor. You can use CPM and it will give you a certain percentage of the variance of the pain to come in your patients. For instance, if a patient is going to have surgery, pre-surgery CPM is going to correlate with how much pain there is after surgery.
But CPM is just one parameter, and to become better at prediction, we need to do a few more things. We don’t really understand which factors are involved in CPM. We could examine a patient or a healthy person today, two weeks later, and then another few weeks after that, and we will not get the same result. This is psychophysics, so there’s built-in variation. I could imagine that if somebody had a very bad morning arguing with their spouse, for example, or not finding a place to park or having gotten a parking ticket, they will perform differently than if everything went smoothly and they came in smiling.
Pain involves so many systems in the brain and in the mind that you don’t expect just one parameter, like how much you can inhibit pain in the lab, to characterize all your pain experiences. So there must be a combination of other parameters. It’s timely that many people study CPM in large cohorts of patients so as to include many parameters, using modern, big data analysis techniques. This will enable us to find out how pain modulation, hand-in-hand with other parameters, can predict how much pain a patient is going to experience after pain-generating procedures, or after an accident, and so on, and also the response to treatment.
Where do you see the work on conditioned pain modulation and temporal summation going in the future, both in the clinic and in research?
The most important thing will be to find which treatment is going to be most beneficial for the specific patient. When I’m treating patients, I toss a coin between drugs like Cymbalta, Lyrica, Neurontin, Elavil – there’s no way to know which will work best. These drugs do not work immediately; each one can take a few weeks, sometimes two to three months before you reach the optimal dose, and after this you may discover the drug is not going to work for the patient. Then you have to reduce the dose and start another drug, and sometimes you find yourself working a few months with one patient who is still in pain all the time.
Now, if you had a test that you could give the patient in your clinic before the first prescription, and the test tells you which patient should go with drug A or B or C, that would be a big step forward. So if I see any real clinical use for CPM, temporal summation, and the other parameters, it is for telling us which drug should go to which patient.
Could you tell me a bit about your recent work on remote electrical neuromodulation?
This is another application of CPM. Let me take a step back. When we were working on building the CPM protocol, I noticed that it was possible to evoke the inhibition with a conditioning stimulus that is not painful, but nearly painful. So if you do the conditioning at a stimulation level that is a little below the pain threshold – the stimulation is felt but not painful – that was sufficient to inhibit the test pain for many healthy people who went through the experiment.
That gave me the opening to think that we could use a nonpainful conditioning stimulus to inhibit pain. If you have to use painful conditioning, it doesn’t really make sense to give the patient one pain to inhibit the other pain. But if you could give a stimulus that is not painful and still evoke pain inhibition, this starts sounding okay.
We tested remote electrical modulation in migraine because it is a relatively easy situation to conquer using a pain inhibitor tool that is not very strong. If you take a patient right after surgery, when everything is painful, and you give them conditioning pain in the leg to inhibit pain that they had in the hand because of surgery, it’s not going to do much because the clinical pain is so high. But migraine is a periodic pain; it’s easy to identify the very beginning of the migraine attack when the pain is still low.
When I did my previous sabbatical with Rami Burstein at Beth Israel in Boston, we explored patients with migraine and went through the development of the migraine attack. Once the attack starts, there is a two-hour process where everything is flaring up, the thresholds are going down, the suprathresholds are going up, everything is becoming more and more sensitive. Things are at their worst at about two hours.
If you ask every migraineur, they’ll tell you that if they take a medication immediately at the beginning of an attack, it’s going to be much more effective than if they waited two or three hours into the attack. So it made sense to me that it might be useful to exert the pain inhibition at the beginning of the migraine attack. The idea was to come up with a stimulus applied within 20 or 30 minutes into the migraine attack, before everything gets worse. We found that giving a migraineur a remote stimulus that is near to being painful, as early as possible after the attack has started, will abort the migraine attack.
Is there anything else you’d like to add?
I want to tell you two stories that people really like. The first one is about how DNIC was discovered; Tony Dickenson told me this story. He said they were exploring spinal second-order sensory neurons – this was in the rat or the cat – and they had an electrode installed in the spinal cord. Pain was given at the tail, and they recorded the responses. At some point they wanted to check on the level of anesthesia, so they pulled the whiskers. They noted that instead of showing an increase in activity, the second-order neurons in the spinal cord, which were activated by the pain in the tail, started reducing their activity simultaneously with the whisker being pulled.
It takes a clever researcher to stop, think, and say, “Wow, there is something here,” rather than typically saying that something went wrong and we should do it again or differently. The people in that lab were clever enough to look at it and understand it.
The second story comes from Clifford Woolf in London. He was working on second-order neurons in the spinal cord in the morning and then went for lunch – fish and chips – and then came back to the lab. At a certain point, he noted that the second-order neurons in the spinal cord in the afternoon were much more responsive to stimuli than in the morning. I guess he realized that since the animal did not have fish and chips, this was not a result of the food. It took some more exploration to understand that the neurons had changed; because of the poking in the morning and the activity in the neurons that was produced by the peripheral stimuli, the neurons started becoming sensitized. And then the group realized there’s a windup, and the concept of temporal summation came along.
These are two stories that people remember because they’re so nice about how serendipity plays a role in scientific developments.
Caroline Phelps, PhD, postdoctoral research associate, University of Arizona, US.