<![CDATA[Background. Baseline DNA methylation has been linked to antidepressant outcomes. Candidate signals need cross-treatment testing before translational interpretation. Objective. To test whether five antidepressant-response CpGs in CHN2, JAK2, SCN7A, and IQANK1 transfer to outcomes after repetitive transcranial magnetic stimulation (rTMS) in treatment-resistant depression. Methods. This secondary analysis used 46 participants of the public GSE300009 cohort with baseline blood methylation and paired Patient Health Questionnaire-9 (PHQ-9) scores. The primary outcome was PHQ-9 improvement. Response (≥50% improvement) and remission (end-treatment PHQ-9 <5) were secondary. Effects per 1-SD higher CpG M-value were estimated with covariate-adjusted linear models (HC3 standard errors) and bias-reduced logistic models. Holm correction, rank correlations, leave-one-out diagnostics, sensitivity analyses, and repeated nested five-fold cross-validation tested robustness and incremental prediction. Results. Mean PHQ-9 fell from 22.3 to 8.9. Calculated response occurred in 31 participants (67.4%) and remission in 18 (39.1%). After multiplicity correction, no CpG was associated with continuous improvement (Holm-adjusted p = 1.000). Two CHN2 CpGs showed nominal inverse associations with remission, not response: cg06926818, odds ratio 0.36 (95% confidence interval 0.15–0.84), p = 0.018, Holm-adjusted p = 0.089; cg23687322, odds ratio 0.41 (0.19–0.89), p = 0.024, Holm-adjusted p = 0.095. Under one-standard-error regularization, adding the CpGs left root-mean-square error unchanged. The area under the receiver-operating-characteristic curve rose by 0.012 for response and 0.059 for remission, with intervals compatible with no benefit. Conclusion. The panel did not transfer as a dependable marker of continuous improvement or response. CHN2-remission signals remain hypothesis-generating. Larger, prospectively harmonized external validation must precede clinical use.]]>
Copyrights © 2026
