Network-level convergence of functional and structural alterations in first-episode major depressive disorder

Understanding network-level neuropathological alterations in first-episode major depressive disorder (FED) is essential for clarifying early-stage depression mechanisms. Although numerous studies have investigated FED, reported findings remain highly heterogeneous. We systematically reviewed functional and structural neuroimaging studies comparing patients with FED and healthy individuals, extracting coordinates of reported group differences. Coordinate-level coordinate network mapping (CNM), combined with a normative connectome dataset ( n = 1,000), was applied to derive FED-associated functional and structural networks. These networks were mapped onto canonical large-scale brain networks to provide a functional interpretation within established systems. Cross-modal overlap and spatial correspondence between functional and structural networks were quantified to evaluate their cross-modal alignment. Separately, spatial associations with neurotransmitter distribution maps were examined to contextualize potential neurochemical alterations in FED. Robustness was evaluated using alternative seed sizes, test–retest data, and study-level CNM. Specificity was examined using parallel CNM analyses of first-episode schizophrenia (FES) and drug-naïve obsessive–compulsive disorder (OCD), and sensitivity was assessed using drug-naïve FED samples and leave-one-study-out analyses. Eighty-two studies (4,480 FED and 4,211 controls) revealed convergent FED-associated networks involving the hippocampus, insula, thalamus, precuneus, temporal cortex, middle frontal gyrus, and cingulate cortex. The identified FED-associated networks showed a systematic network organization: the functional network predominantly aligned with the cingulo-opercular, somatomotor, and posterior multimodal networks, whereas the structural network primarily aligned with the default mode, auditory, and language networks. FED-associated probability maps across two modalities showed strong whole-brain correspondence ( r = 0.61, p < 0.001). Furthermore, significant spatial associations were observed between FED-associated probability maps and several neurotransmitter systems, including the dopamine transporter, N-Methyl-D-aspartate receptor, and vesicular acetylcholine transporter. The identified networks were highly consistent across seed sizes, test–retest data, drug-naïve samples, and leave-one-study-out analyses, and showed relatively limited spatial overlap with FES- and OCD-associated networks. Comparison with study-level CNM demonstrated high whole-brain spatial correspondence but differences in the detailed topology of the functional network. By integrating heterogeneous neuroimaging evidence, CNM delineates a coherent large-scale network architecture underlying FED. These findings refine the system-level characterization of early-stage depression and may help inform future neuromodulation strategies.