Cortical Plasticity & Phantom Limb Pain Linked

artificial intelligence
artificial intelligence
"Phantom limb pain is causally related to cortical activity."

Though phantom limb pain is a well-established phenomenon, its causes have yet to be fully elucidated. In a study reported in Nature Communications, researchers used a brain-machine interface (BMI) to investigate the role of cortical plasticity and pain in 10 patients with loss of movement in one hand due to brachial plexus root avulsion or amputation.1 

The BMI used real-time magnetoencephalography (MEG) sensorimotor cortex signals to decode the neural activity of the mental action involved in moving the hand, and then channeled the movement into a neuroprosthetic (robotic) hand. 

Participants underwent 2 separate training sessions to learn to control the robotic hand, and each session was conducted with a different decoder: A phantom decoder used MEG signals from the phantom hand to control the robotic hand, and a random decoder used MEG signals with randomly relabeled movement types. In each session, participants were directed to grasp and release a ball with the robotic hand by attempting to move the phantom hand while viewing the robotic hand.

Following training sessions with the phantom decoder, the average F-values of the z-scored cortical currents increased in the contralateral sensorimotor cortex. 

Contrary to the authors’ hypothesis, there was also a significant increase in pain with the phantom decoder: Visual analogue scale (VAS) scores increased from 38.2±18.5 [mean±95% confidence interval (CI)] to 45.8±18.4 (1/100; n=10, P=.0066, uncorrected, t(9)=3.51, paired Student’s t-test). After training with the random decoder, the F-values did not increase, and there was no significant change in pain scores.

The results further reveal that the increase in VAS positively correlated with an increase in F-values. “That is, pain increased as the discriminability in cortical currents representing phantom movements increased,” wrote the authors. 

To reduce discriminability, a subsequent experiment used a real hand decoder that was constructed with MEG signals from each participant’s intact hand so they would “unknowingly associate the phantom movements with the cortical representation of the intact hand’s movements, which were different from the cortical representation of the phantom movements in pre-BMI training,” according to the paper.

As expected, training with the real hand decoder led to reduced F-values of the phantom hand movements for the sensorimotor cortex contralateral to the phantom hand. 

There was also a significant reduction in VAS pain scores, from 38.3±15.5 (mean±95% CI) to 34.6±14.8 (1/100; n=10, P=.029, uncorrected, t(9)=2.60, paired Student’s t-test), as well as a decrease in scores on the short-form McGill Pain Questionnaire 2 [SF-MPQ2], from 26.0±21.0 to 20.7±16.3 (n=10, P=.016, uncorrected, Wilcoxon signed-rank test).

Study co-author Takufumi Yanagisawa, MD, PhD, of Osaka University in Japan, believes the increased pain with the phantom limb decoder is caused by the discrepancy between expected and actual sensory feedback. 

This is in line with the incongruence hypothesis of pain that has been proposed by other researchers.2,3 “Of course, there is no sensory feedback in the phantom limb. However, when patients intend to move their phantom hand, their brain expects the sensory feedback–therefore there is a discrepancy between them,” he told Clinical Pain Advisor

The real limb decoder, which was intended to dissociate the phantom hand from the robotic hand, may have resulted in less of a sensory discrepancy and therefore less pain.

“Our study suggests that phantom limb pain is causally related to cortical activity–especially to the activity corresponding with motor intention,” said Dr Yanagisawa. He and colleagues are developing a device that would train patients to decrease pain by modulating cortical activity. Future studies are needed to understand the causal relationship between pain and cortical plasticity as it applies to phantom limb pain and other types of chronic pain. 

Summary and Clinical Applicability

These findings support a causal relationship between pain and cortical plasticity and show promise for the development of novel treatments based on brain-machine interface neurofeedback.

Follow @ClinicalPainAdv

References

  1. Yanagisawa T, Fukuma R, Seymour B, et al. Induced sensorimotor brain plasticity controls pain in phantom limb patients. Nature Communications. 2016; doi:10.1038/ncomms13209
  2. Harris AJ. Cortical origin of pathological pain. Lancet. 1999; 354(9188):1464-6.
  3. Ploghaus ATracey IClare SGati JSRawlins JN, Matthews PM. Learning about pain: the neural substrate of the prediction error for aversive events. Proc Natl Acad Sci USA. 2000; 97(16):9281-6.