An objective biomarker of pain – it’s what many consider the Holy Grail of pain research. In recent years, brain activity signatures from neuroimaging studies have received a great deal of attention from pain researchers as possible candidate biomarkers. Now, a new imaging biomarker contender has emerged.
In a study led by Choong-Wan Woo, Sungkyunkwan University, Suwon, South Korea, and Tor Wager, Dartmouth College, Hanover, US, researchers present a functional magnetic resonance (fMRI) signature for sustained pain based on whole-brain functional connectivity that pushes the field closer to an objective pain biomarker. This pain signature, dubbed the Tonic Pain Signature (ToPS), reliably predicted sustained pain in healthy volunteers, and clinical pain in people suffering from subacute or chronic low back pain.
By reliably predicting sustained experimental and clinical pain, ToPS goes a big step further than previous pain signatures identified by Wager’s group, which were based on responses to phasic heat pain (Wager et al., 2013Woo et al., 2017 PRF related news story).
“This paper is significant in that it provides a novel brain signature for the sustained pain experience, and by doing so brings us incrementally closer to a brain signature of chronic pain, which is inherently elusive,” said Katherine Martucci, a pain neuroimaging researcher at Duke University, Durham, US, who was not part of the new study.
The study was published in Nature Medicine on January 4, 2021.
The Tonic Pain Signature (ToPS)
Pain brain imagers are seeking a reliable pattern of activity in the brain that would characterize the clinical pain that patients experience, in order to improve the assessment of pain and guide treatment. A key feature of clinical pain is its sustained nature, but sustained pain has been difficult to objectively assess since it is affected by many different factors such as mood, emotion, and attention.
To overcome this challenge, Woo, Wager, first author Jae-Joong Lee, also at Sungkyunkwan University, and colleagues decided to focus on sustained (tonic) experimental pain. They reasoned that this would more closely approximate chronic pain than a study of acute experimental pain would.
The group’s first task was to identify an fMRI-based signature for tonic experimental pain (fMRI measures blood oxygen level-dependent [BOLD] signals in the brain, a correlate of neural activity). Nineteen healthy participants received capsaicin-rich hot sauce on the tongue to induce tonic pain and then underwent fMRI neuroimaging, over the course of five minutes. The researchers then generated candidate models that could predict pain based on functional connectivity patterns during tonic pain. Ultimately, Lee landed on the ToPS model, which reliably predicted capsaicin-induced sustained pain.
The group then sought to validate ToPS in their next experiment, which included 42 healthy volunteers who received the hot sauce on the tongue. Here, the researchers asked the study subjects how much they wanted to avoid this painful experience. ToPS predicted within-individual variation in subjects’ avoidance ratings during the course of the tonic pain. ToPS also discriminated capsaicin-induced pain from a bitter (quinine) taste, with 76% accuracy. This was important as it showed the specificity of ToPS to painful stimuli.
Similar findings emerged from another independent dataset of 48 volunteers at a different site, in South Korea. In this case, once again, ToPS performed well when tracking variation in avoidance ratings within individuals. And, it discriminated capsaicin pain from bitter taste, and from aversive odor (caused by fermented skate, a type of fish), with 85% accuracy.
“ToPS is a predictive model based on whole-brain functional connectivity patterns, aimed to capture the sustained nature of pain. It’s important to understand that our ToPS model is based on experimentally-induced sustained pain because this is different from acute pain that is usually used in fMRI experiments, and more similar to clinically important chronic pain diseases,” said Lee.
What about clinical pain?
The next question was whether ToPS could predict clinical pain. To answer this question, the authors tested ToPS on open fMRI datasets from patients with subacute back pain or chronic back pain, each of which is characterized by sustained pain.
The first dataset included 70 people with subacute back pain and 25 with chronic back pain. ToPS reliably predicted individual differences in pain severity in these two patient groups, but this depended on the type of task. That is, in the case of subacute back pain, ToPS reliably predicted pain when patients were asked to report their spontaneous pain, while in patients with chronic back pain, the model predicted pain even when participants rested without performing any other task.
The investigators then tested ToPS on two more chronic back pain datasets from Japan (24 patients, 39 healthy controls) and the United Kingdom (17 patients and 17 controls). They found that ToPS discriminated patients with chronic back pain from healthy controls with 71%-73% accuracy.
Together, the findings suggested that brain connectivity patterns were similar under conditions of experimental tonic pain and clinical pain.
Sustained pain activates diverse brain networks
The researchers next asked what specific brain activity patterns underlie ToPS. For their experimental tonic pain model, the team found reliable positive predictive weights (that is, higher pain with increasing connectivity) in the somatomotor network, frontoparietal network, visual network, and dorsal attention network. The authors say this reflects a role for multi-sensory integration and top-down attention processes.
Negative weights (lower pain with increasing connectivity) were found in the limbic and paralimbic cortical regions, and in brainstem regions. These brain areas are important in context processing and descending pain modulation.
Looking in more detail at specific regions of the brain commonly examined in pain brain imaging studies, including pain processing and modulatory regions, they saw that connectivity between the dorsolateral prefrontal cortex (dlPFC) and classical pain processing regions like the somatosensory cortices and the dorsal posterior insula predicted higher sustained pain. Meanwhile, connectivity between the dlPFC and the brainstem predicted lower pain.
“This new signature aligns with and is able to frame other individual neuroimaging findings that have shown us brain regions and connections relevant for chronic pain. It’s both reassuring and exciting that their advanced and intricate brain signature of tonic pain is still complementary to what we’ve learned about the brain related to chronic pain,” Martucci told PRF.
ToPS is top among models
The authors next compared their ToPS model to the subacute back pain model and to an experimental phasic pain model trained on data from healthy subjects who received a painful heat stimulus. These comparisons showed that patterns of brain activity in ToPS were more similar to the back pain model than to the phasic pain model. Meanwhile, ToPS performed the best in predicting tonic pain, and the back pain model performed better than the phasic pain model.
Further, patterns of brain activity in regions including the sensorimotor network, frontoparietal network, and dorsal attention network were similar between ToPS and the back pain model, whereas the phasic pain model showed a different pattern of activity. Together, the results indicated that brain patterns look different in sustained pain versus phasic pain, with a number of different brain networks becoming involved.
“These findings suggest that when pain becomes sustained, it recruits multiple processes like salience, attention, and affective systems in the brain. Important sensory regions like the insula, ACC, and thalamus were also activated, but with strong connections with these non-sensory areas,” Woo told PRF.
“It is an easy assumption to make that painful stimuli that last a few seconds and those that last a few minutes are similar, with many of the same pathways involved,” Wager said. “We found that this is not the case – they are represented differently.”
Ultimately, the findings “offer us a glimpse of a future with neuroimaging-based characterization of clinical pain and sets us one more step of the way towards that future,” Martucci said. “We need these types of models to be continually presented and refined in order to advance brain signatures of pain.”
Fred Schwaller, PhD, is a freelance science writer based in Germany.
Image credit: Lee et al. A neuroimaging biomarker for sustained experimental and clinical pain. Nat Med. 2021 Jan;27(1):174-182.