Novel Brain Biomarker Accurately Predicts Pain Sensitivity

A novel biomarker signature may offer a more objective way of predicting pain sensitivity, new research suggested.

Investigators utilized the biomarker, which consisted of two measures — sensorimotor peak alpha frequency (PAF) and corticomotor excitability (CME) — to predict pain response in patients who received an injection that induced prolonged temporomandibular pain.

They used machine learning models to see if PAF and CME could predict pain sensitivity in a training set and then compared the predicted and the actual pain reported by patients.

The PAF/CME biomarker successfully categorized participants with “high” and “low” pain sensitivity, with an area under the curve (AUC) of 1.00. Results were reproduced across a variety of methodological parameters.

“Using the PAF/CME biomarker signature, we were able to accurately and reliability predict who would develop high vs low pain with excellent accuracy,” study investigator Siobhan Schabrun, PhD, professor and research chair in mobility and activity, University of Western Ontario, London, Ontario, Canada, told Medscape Medical News.

“This is important because individuals who experience higher pain soon after injury are more likely to develop chronic pain.” The findings “pave the way for new treatments and preventatives for chronic pain,” she stated.

This study was published online on January 27 in JAMA Neurology.

Transition From Acute to Chronic Pain

“Musculoskeletal pain is a leading cause of disability worldwide, and once pain becomes chronic — defined as pain lasting more than 6 months — there are few effective treatments,” Schabrun said.

“Understanding who is at risk of developing chronic pain at the time of injury would help in the diagnosis, prevention, and treatment of chronic pain, reducing the prevalence of a major global health problem,” she added.

Two features of cortical activity are important in shaping the subjective experience of pain. These are neural oscillatory rhythms involved in processing nociceptive input and corticospinal signaling involved in subsequent motor response. Studying these features of cortical activity has led to the identification of a “promising” sensorimotor cortical biomarker signature, consisting of two metrics: PAF and CME.

“Given that individuals who experience higher pain in the early stages of a prolonged pain episode (eg, post-surgery) are more likely to develop chronic pain in the future, slow PAF before an anticipated prolonged pain episode and/or CME depression during the acute stages of pain could be predictors for the transition to chronic pain,” the researchers wrote.

Previous research has suggested that slower PAF prior to pain onset and CME depression during prolonged pain were linked with more pain. So in the current study, the researchers “undertook analytical validation” of this new brain-based biomarker signature for musculoskeletal pain, Schabrun said.

“The biomarker consisted of measures of corticomotor excitability — the excitability of the pathway between the brain and muscle — made using transcranial magnetic stimulation and peak alpha frequency made using electroencephalography,” she added.

To find whether individuals can accurately be classified as having “high” or “low” pain sensitivity, based on these two metrics, the investigators used a model of prolonged myofascial temporomandibular pain — masseter intramuscular injection of nerve growth factor (NGF) — to produce progressively developing pain lasting up to 4 weeks.

They collected outcomes over a over a 30-day period. Participants (n = 150; mean [SD] age, 25.1 [6.2] years; 66 women) visited the laboratory on days 0, 2, and 5 when pressure pain thresholds, PAF, and CME were measured. Of the total group, 100 participants were included in the training set and 50 in the test set. All were healthy adults with no history of chronic pain and no neurologic or psychiatric conditions.

On day 0, following two laboratory sessions, NGF was injected into the right masseter muscle. The participants submitted electronic pain diaries twice daily, where they rated their pain (0-10) during various activities, with chewing and yawning used as primary pain outcomes.

The researchers employed five machine learning models on the label training set, with pain sensitivity (high/low) as the dependent variable and sensorimotor PAF and change in CME as independent variables.

Fivefold cross-validation was used to identify “optimized parameters” in the training set, and these were then used to predict labels in the validation set to validate model selection. The model with the best performance (AUC) on the validation set was then locked in.

‘And the Winner Is…’

Logistic regression emerged as the “winning classifier.” The authors described its AUC as “outstanding,” at 1.00 (95% CI, 1.00-1.00) when applied to the validation set. Slower PAC and CME depression predicted higher pain.

The locked model assessed on the test set also had “excellent performance” (AUC, 0.88; 95% CI, 0.78-0.99). Results were reproducible across a wide range of methodological parameters.

Inclusion of sex and pain catastrophizing as covariates didn’t improve the model, “suggesting the model including biomarkers only was more robust.” PAF and CME biomarkers showed good to “excellent” test-retest reliability.

The researchers noted that the use of a cohort of exclusively healthy participants and an experimental pain model limits the generalizability to clinical populations.

“We are currently undertaking studies to show the accuracy and reliability of using the biomarker in clinical populations (jaw and low back pain) to predict who will transition from acute to chronic pain,” Schabrun reported.

“If accuracy and reliability remain high, the biomarker could be used clinically to identify those at risk of developing chronic pain and to target early interventions toward this group, interfering with the development of chronic pain. In the future, these findings could also be used to develop novel brain-based treatments and preventatives that reduce the prevalence of chronic pain,” she added.

Seeking the ‘Holy Grail’

In an accompanying editorial, Prasad Shirvalkar, MD, PhD, of the University of California San Francisco, and Christopher Rozell, PhD, of the School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, described the quest to identify objective biomarkers that track individual pain severity as the “holy grail” of pain neuroscience.

They describe the current study as “technically rigorous and innovative.”

However, they warned, “While progress in quantifying subjective percepts is exciting…we must take care to ensure that quantitative measures do not supplant lived experience reports, introduce distrust in the physician-patient relationship, set unrealistic patient expectations, or exacerbate existing inequalities in pain treatment across this vulnerable population.”

This project was funded by the National Institutes of Health. Andrew J. Furman, PhD, and David A. Seminowicz, PhD, reported having a patent issued for PAF through the University of Maryland, Baltimore. The other study authors disclosed no relevant financial relationships. Shirvalkar reported receiving grants from the National Institutes of Health, nonfinancial support from Medtronic and QuantalX and having had a patent pending for closed-loop deep brain stimulation for chronic pain. Rozell reported receiving personal fees from Motif Neurotech and having had a patent pending for a system to identify transitions in brain states from electrophysiological markers in DBS for depression.

Batya Swift Yasgur, MA, LSW, is a freelance writer with a counseling practice in Teaneck, New Jersey. She is a regular contributor to numerous medical publications, including Medscape Medical News and WebMD, and is the author of several consumer-oriented health books as well as Behind the Burqa: Our Lives in Afghanistan and How We Escaped to Freedom(the memoir of two brave Afghan sisters who told her their story).