Disrupted brain network topology and structure-function coupling in non-specific low back pain

Central nervous system mechanisms contribute to non-specific low back pain (LBP), with widespread alterations in brain function and structure reported. It remains unclear how these alterations propagate to brain network topology and structure-function coupling. In this cross-sectional study, 282 participants with non-specific LBP (mean ± standard deviation, pain intensity 4.3 ± 1.7 [0-10 numerical rating scale]; current pain duration 54 ± 105 weeks) and 71 pain-free controls underwent resting-state functional and diffusion-weighted MRI to derive functional and structural connectomes. Subgroups were analyzed by pain status during scanning and pain duration. The main graph-theory analysis revealed that participants with non-specific LBP had lower segregation (d = 0.39-0.45, P = 0.009-0.019) in the whole-brain functional network and higher integration (d = 0.30-0.38, P = 0.005-0.027) in the functional ventral attention network. Subgroup analysis showed that participants with chronic non-specific LBP had lower segregation (d = 0.48, P = 0.002) and higher integration (d = 0.45, P = 0.004) within the structural default mode network. By modeling functional connectivity using structural communication predictors in a multivariable linear regression, we identified structure-function decoupling in participants with non-specific LBP across the whole-brain (d = 0.25-0.28, P = 0.026-0.041), ventral attention (d = 0.28-0.32, P = 0.005-0.048), and default mode network (d = 0.22-0.24, P = 0.031-0.046), alongside complex regional patterns of both decoupling and hypercoupling. Explorative partial correlation analyses associated disability, rather than pain intensity, with differences in brain network topology and structure-function decoupling. These findings advance the mechanistic understanding of the severity and persistence of non-specific LBP, showing disrupted bottom-up attentional processing and self-referential circuitry, resulting in large-scale structure-function decoupling.