Background
Craving is a core factor driving drug-seeking and -taking, representing a significant risk factor for relapse. This study aims to identify neuroanatomical biomarkers for quantifying and predicting craving.
Methods
The study enrolled 94 ketamine-dependent users and 103 healthy controls (HC). Utilizing support vector regression (SVR) with 10-fold cross-validated framework, we developed a neuroanatomical craving model based on measures of regional cortical thickness (CT), surface area (SA), and subcortical volume (SV) derived from T1 images. The generalizability of neuroanatomical craving model was examined in an independent set. Spatial correlation analysis was employed to assess the relationship between the regional contribution to craving and density maps of receptors/transporters from previous molecular imaging studies.
Results
The neuroanatomical craving model identified neuroanatomical biomarkers that predicted self-report craving (r = 0.635). The most importance of predictors of craving included the SA of the left medial orbitofrontal cortex and the left supramarginal gyrus, CT in the left caudal anterior cingulate, the left cuneus, the right lateral occipital cortex and the right lingual gyrus, as well as the left amygdala GMV. Importantly, these predictors were generalized to an independent sample. Moreover, nodal contribution to predicted craving scores were associated with DA2, 5-HTa, 5-HTb receptor and serotonin reuptake transporter densities.
Conclusion
The results offer a key perspective on craving prediction among ketamine-dependent users, and identify neuroanatomical areas associated with craving in the frontal and parietal regions. Additionally, the underlying neuroanatomical structures involved in the craving process may be linked to the dopaminergic and serotonergic systems.