Background & Aims

Chronic pain’s diverse etiology poses a challenge to finding a universal biomarker1. This study aims to identify and test biomarkers for pain-associated medical conditions, with the hypothesis that distinct pathophysiologies will express through different, unrelated biomarkers. The study also postulates that a biological signature predicting medical conditions would partly reflect chronic pain’s clinical classifications (i.e., nociceptive, nociplastic, and neuropathic).

Methods

he study used data from 493,211 UK Biobank participants. We trained machine learning models to predict 35 pain-associated medical conditions (i.e., derive biomarkers) based on brain imaging, blood immunoassays, and genome-wide associations.

Results

Our study findings underline the fact that different biological factors predict various medical conditions. For instance, blood immunoassays proved most effective in predicting gout and Crohns Disease, while brain imaging (fMRI) was superior in forecasting conditions such as fibromyalgia and chronic fatigue snydrome. Further, we found that medical conditions characterized by similar clinical classifications of pain (nociplastic, nociceptive, neuropathic) tended to be predicted by similar biological indicators. For example, resting-state functional magnetic resonance imaging (rsfMRI) proved most effective in predicting nociplastic pain conditions such as widespread pain, fibromyalgia and chronic fatigue syndrome, while blood assays were effective as predicting nociceptive/inflammatory conditions such as gout, rheumatoid arthritis, and Crohn’s disease.

Conclusions

Our findings indicate that various biological markers could potentially predict diverse types of chronic pain conditions, furthering our understanding of their etiology and potentially guiding targeted treatment strategies. Additionally, it emphasizes the importance of characterizing pain conditions based on etiology and biological factors to identify clinically useful biomarkers.

References

1.Suzanne Galloway, Maryann Chimhanda, Jayme Sloan, Charles Anderson, James Sinacore, Linda Brubaker, “Pain Scores Are Not Predictive of Pain Medication Utilization”, Pain Research and Treatment, vol. 2011, Article ID 987468, 5 pages, 2011.

Presenting Author

Matt Fillingim

Poster Authors

Matt Fillingim

BSc

Mcgill

Lead Author

E-mail

Christophe Tanguay-Sabourin

University of Montreal

Lead Author

E-mail

Gianluca Guglietti

McGill University

Lead Author

E-mail

Azin Zare

McGill University

Lead Author

E-mail

Jax Norman

McGill University

Lead Author

E-mail

Topics

  • Models: Chronic Pain - Inflammatory