# Integrated transcriptomic and machine learning analysis reveals novel diagnostic biomarkers for adolescent major depressive disorder

## TL;DR

This study identifies three genes that could help diagnose adolescent depression by combining blood RNA data and machine learning.

## Contribution

A three-gene biosignature (SLC4A1, IGF1, MMP9) is proposed as a novel diagnostic tool for adolescent MDD with high accuracy.

## Key findings

- Transcriptomic analysis revealed 367 differentially expressed genes linked to immune activation and hypoxia.
- Eight core hub genes showed a distinct expression pattern, validated experimentally.
- A three-gene panel achieved 86.7% diagnostic accuracy for adolescent MDD.

## Abstract

The lack of objective biomarkers and mechanistic understanding of adolescent Major Depressive Disorder (MDD) impedes early diagnosis and targeted intervention.

To elucidate peripheral molecular biomarkers for adolescent MDD, we performed RNA sequencing on peripheral blood mononuclear cells (PBMCs) from 15 adolescents with MDD and 15 age- and sex-matched healthy controls. Differential expression analysis and protein-protein interaction (PPI) network construction were utilized to identify key regulatory genes. The expression of core targets was validated using RT-qPCR and ELISA. To establish a robust diagnostic model, an integrated feature selection strategy combining Least Absolute Shrinkage and Selection Operator (LASSO), Support Vector Machine-Recursive Feature Elimination (SVM-RFE), and Random Forest algorithms was applied to screen candidate biomarkers.

Transcriptomic profiling identified 367 differentially expressed genes characterized by a dual signature of innate immune activation and compensatory hypoxic responses. Eight core hub genes were identified and experimentally validated, revealing a dichotomous expression pattern: upregulation of erythroid-related and inflammatory factors (SLC4A1, HBB, GYPA, IL6) and downregulation of neurotrophic and remodeling factors (IGF1, CSF2, MMP9, CXCR1). Notably, lower expression levels of MMP9 and CXCR1 were significantly correlated with higher Hamilton Depression Rating Scale (HAMD) scores, indicating greater symptom severity. The multi-algorithm machine learning approach identified a consensus three-gene diagnostic panel comprising SLC4A1, IGF1, and MMP9, which achieved a high classification accuracy with an Area Under the Curve (AUC) of 0.867.

This study delineates a systemic molecular landscape of adolescent MDD defined by the coexistence of hypoxic compensation and neurotrophic/remodeling failure. The identified three-gene biosignature (SLC4A1, IGF1, MMP9) offers a promising, objective tool for the early diagnosis of adolescent depression, highlighting the immune-metabolic interface as a critical avenue for future precision medicine.