Integrated transcriptomic and network analysis reveals candidate immune–metabolic biomarkers in children with the inattentive type of ADHD

Qiaoyan ShaoYanhui Chen^*

• Department of Pediatrics, Fujian Medical University Union Hospital, Fuzhou, China

Abstract Background: Attention-Deficit/Hyperactivity Disorder (ADHD) is a clinically heterogeneous neurodevelopmental disorder. Its inattentive presentation (ADHD-I) is a common subtype characterized predominantly by difficulties in sustaining attention, organization skills, and task completion. The biological foundations of ADHD-I remain unclear, hampering the development of effective treatments. This study aimed to identify potential ADHD-I biomarker candidates to guide the therapeutic strategies. Methods: We analyzed transcriptome sequencing data from a cohort of 32 children (15 control, 17 ADHD-I; aged 6–12 years;81.2% male). All ADHD-I participants were medication-naïve and without comorbid neurodevelopmental or major psychiatric conditions) to systematically identify potential biomarkers for ADHD-I. Candidate genes were identified by integrating differential expression analysis with weighted gene co-expression network analysis (WGCNA) modules. High-confidence biomarkers were selected via a multi-step pipeline combining protein-protein interaction (PPI) network analysis and machine learning feature selection (LASSO regression, Boruta algorithm). Biomarker performance was evaluated using ROC and gene expression analyses, and a predictive nomogram was developed. The ADHD-I molecular landscape was explored through functional enrichment, immune cell profiling, pharmacological screening, and ligand-receptor interaction modeling. Results: Cluster of Differentiation 180(CD180) and Cytochrome c Oxidase Assembly Factor 3(COA3) were identified as potential ADHD-I biomarker candidates. Both showed high preliminary diagnostic accuracy (AUC > 0.8) and significantly elevated expression in ADHD – I cohorts. The nomogram incorporating these biomarkers showed preliminary predictive accuracy for ADHD-I risk stratification (AUC = 0.878) in this cohort. Pathway enrichment analysis further localized CD180 and COA3 to the dorsoventral axis formation pathway, suggesting their role in developmental patterning. Five significant differential immune cell types were identified between ADHD-I and control samples. Both biomarkers demonstrated the significant positive correlation with gamma delta T cells and the strongest negative correlation with eosinophils. Compound prediction showed that 20 compounds such as benzo(a)pyrene targeted CD180, and benzo(a)pyrene had a strong binding ability to CD18 (ΔG = –8.1 kcal/mol). Conclusion: The study identified CD180 and COA3 as candidate biomarkers for ADHD-I, which may provided new clues into the mechanism of ADHD-I and potential therapeutic targets.