2026 ECNP Congress

PS03-3163

The association of cholesterol with rich-club organisation and global efficiency in the structural connectome

Delia Schepers — M. Gruber, F. Timm, H. Staub, P. Rehm, S. Fischer, L. Krätzig, A. Reif, U. Dannlowski, T. Kircher, S. Edwin Thanarajah, J. Repple · University Hospital Frankfurt, Department of Psychiatry- Psychosomatic Medicine and Psychotherapy, Frankfurt, Germany

2026 ECNP Congress

Background

Major depressive disorder (MDD), bipolar disorder (BD), and schizophrenia (SZ) are increasingly seen as disorders of large-scale brain network organization. The rich-club architecture of the structural connectome has been proposed as a transdiagnostic vulnerability backbone, partly because of its high metabolic demand. Lipid dysregulation has been observed across various mental disorders and may be considered as a shared metabolic pathway leading to structural dysconnectivity. However, the relationship between rich-club organization and cholesterol levels, including high-density lipoprotein (HDL) and low-density-lipoprotein (LDL), remains unclear. Moreover, no study has investigated whether healthy controls and patients with mental disorders differ in this relationship.

Methods

Diffusion MRI-based structural connectomes were analyzed in N=533 participants (healthy controls (HC) n=207; patients with MDD, BD, or SZ n=326). Weighted rich-club coefficients were estimated to define significant rich-club regimes and peak degree. Rich-club node membership and the total streamline weights of rich-club, feeder, and local connection were quantified. Logistic regression tested classification of patients versus controls based on rich-club features. The association between cholesterol markers and rich-club organization was investigated using generalized least squares (GLS) models. HDL and LDL were tested separately, including interaction terms for lipid x edge type (rich-club vs. feeder vs. local edges) and lipid x group (HCs vs. patients). Likelihood ratio tests were utilized for model selection.

Results

Rich-club organization was robustly present in both groups. Controls showed higher k peak, whereas patients showed a greater number of rich-club nodes (p < 0.027). Patients exhibited higher total streamline weight, driven by significantly increased rich-club and feeder connection weights (p < 0.034), with no significant differences in local weights. Classification performance based on rich-club features was near chance (BACC = 0.52-0.54). GLS analyses revealed a significant HDL × edge type interaction (p=0.003). Higher HDL was positively associated with rich-club connection weights but negatively associated with feeder and local connections, with the rich-club vs. feeder slope contrast reaching significance after Bonferroni correction (p=0.008). No significant HDL × group interaction was found (p=0.168). For LDL, a significant LDL × group interaction was identified (p=0.003): LDL was positively associated with connectivity in HC (slope=21.25, 95% CI [7.26, 35.24]) but showed no significant association in patients (slope=−3.79, 95% CI [−14.59, 7.01]). LDL × edge type interactions were not significant. Effect sizes for all lipid interactions were small (Cohen’s f =0.003- 0.009).

Conclusions

Our analyses reveal a redistribution of connectivity at the hub-periphery interface in mental disorders, rather than a loss of rich-club structure. HDL was found to be selectively associated with rich-club connectivity, whereas LDL showed different relationships with structural connectivity in healthy controls compared to patients. This suggests that lipid metabolism may interact with connectome organization in a diagnostic manner. Importantly, effect sizes and explanatory power were small across most analyses, highlighting the complexity and multifaceted nature of the relationships between lipids and the connectome. Therefore, our findings emphasize the need for multimodal and longitudinal studies to elucidate the metabolic basis of structural dysconnectivity.

References

No references