Study identifies a continuous plasma proteomic axis in major depression

Major depressive disorder (MDD) is clinically heterogeneous, and case-control comparisons may miss molecular variation within the disorder. Researchers developed BioNet-SPACC, a clustering method that incorporates protein-interaction information, and applied it to plasma proteomic data from 4,123 UK Biobank participants with MDD. The analysis used separate discovery and held-out test samples; matched healthy controls were added after MDD-only discovery to provide a reference. A 15-protein panel was also assessed in 42 participants from a separate UK Biobank wave, and pathway-level findings were compared with an independent dataset of 38 participants with MDD and 40 controls.

The method produced a stable two-group partition, but the authors describe the groups as positions along a predominantly continuous molecular axis. The first principal component accounted for 91.3% of variance in the core 85-protein signature, and distributional analyses did not find evidence against a unimodal pattern. Discovery and held-out assignments showed substantial agreement (κ=0.888). The 15-protein panel recovered assignments in held-out data with an AUC of 0.997; cross-wave concordance was exploratory.

The results support further investigation of molecular variation within MDD, not clinically established subtypes. The small cross-wave sample and exploratory external comparison limit validation, and the study does not show that the proteomic axis improves diagnosis or guides treatment.