Predicting the Source of Low Back Pain to Inform Personalized Medicine

Introduction: Low back pain (LBP) is now considered to be the leading cause of disability worldwide, accounting for billions of dollars in medical and indirect costs annually. Most patients with acute LBP suffer from “non-specific” LBP since the source cannot be identified clinically, but biomechanical causes are often implicated. Understanding the link between whole-body biomechanics (where abnormal movement strategies occur) and tissue-level mechanics (where the pain originates) is critical to treat this type of LBP. However, it is a challenge to obtain subject-specific tissue mechanics without invasive measurements. Computational modeling allows for estimation of biomechanical quantities that are difficult or impossible to measure in vivo. Musculoskeletal (MS) simulations of whole-body motion can be used to obtain estimates of muscle force distribution and finite element (FE) modeling in Abaqus can be used to obtain tissue-level mechanics. The objective of our study was to create a multiscale model of the human lumbar spine using MS and FE modeling techniques to predict lumbar spine mechanics from subject-specific whole-body dynamic motion. Modeling: A validated FE model of the lumbar spine created in Abaqus/Standard was used to replace the lumbar spine of a whole-body MS model. Bones and endplates of the lumbar vertebrae were modeled as rigid and 7 spinal ligaments were represented with sets of nonlinear tension-only connector elements. The annulus fibrosis of each intervertebral disc was modeled using the Holzapfel-Gasser-Ogden material formulation in Abaqus with fiber stiffness and angle properties taken from the literature. The nucleus pulposus of each disc was modeled as a fluid cavity with SFM3D4R elements and cartilage of the facet joints was modeled as rigid with rigid contact surfaces for computational efficiency. Lumbar spine muscle attachments for 132 muscle fibers were defined as node sets on the vertebral bodies and fixed centers-of-rotation were defined for the intervertebral joints using multi-point constraints (*MPC, BEAM). Subject-specific muscle force distribution was estimated using the MS model with experimentally-collected kinematics of individuals with and without a transtibial amputation performing a sit-to-stand motion (*CLOAD). Lower limb amputees were included in the study because they suffer from LBP at a much higher rate than their able-bodied peers. Regression comparisons were made between FE tissue mechanics (S, CFNM, PCAV) of the discs and facet joints vs. whole-body kinematics. Predictive power was stronger for participants with a lower-limb amputation, suggesting that they may have more consistent movement strategies that can help to inform specialized treatment. Significance: The use of computational modeling allows for an efficient, cost-effective, and non-invasive method for understanding how subject-specific whole-body dynamic motion affects lumbar spine tissue mechanics and leads to pain activation. This insight can help shed light on how biomechanical low back pain develops for certain populations and better inform therapeutic intervention (e.g. movement retraining) to alleviate pain.