Effective in treatment-resistant depression (TRD), repetitive transcranial magnetic stimulation (rTMS) presents a remission rate between 30% and 40%. Although genetics could be one potential source of inter-individual variability in rTMS responsiveness, few studies have sought to identify possible genetic basis of rTMS response [1,2]. Therefore, we used an extreme-phenotype design in which we compared genome-wide allelic variation between rigorously defined rTMS responders and non-responders. A total of 99 TRD patients provided informed consent and this study was approved by the Human Research and Ethics Committee of the Alfred Hospital. All were submitted to a rTMS protocol and completed at least 18 sessions of 10 Hz at left dorsolateral prefrontal cortex (DLPFC). Clinical outcomes (responder or non-responder) were determined based on scores on the Montgomery Asberg Depression Rating Scale (MADRS). We used the extreme-phenotype design. Patients were separated into two groups, responders patients with at least 60% reduction on the MADRS scale (n 1⁄4 29) and non-responders patients with 10% or less reduction on the MADRS scale (n 1⁄4 19). We used this criteria considering that extreme scoring patients (<10 >60) may present more representative genetic results, improving power to detect phenotype-genotype associations [3,4]. Genotyping was performed using the Infinium PsychArray-24 BeadChip (Illumina, Inc., San Diego, CA, USA) and analyzed with PLINK 1.9. We performed quality control to remove individuals or markers with high error rates. Data quality control parameters were: call rate (GENO) > 90%, maximum individual missingness rate (MIND) < 10%, minor allele frequency (MAF) > 5% and Hardy-Weinberg equilibrium (HWE) p-value > 10e6. We performed standard association analysis to compare allele frequency in both groups with a 95% CI. After identifying the significant SNPs and genes a functional enrichment analysis using STRING and Cytoscape databases was performed. Of the initial 593,260 SNPs in 48 individuals, 958 variants were removed due to missing genotype data, 310,522 SNPs were removed due to minor allele threshold and 4 people were removed due to missing genotype data. This left 281,780 SNPs and 44 subjects for the association study. In order to estimate the effective number of significant SNPs, we applied the False Discovery Rate (FDR) correction considering sample and SNPs per chromosome. A new p value was then determined for each chromosome (Supplementary Table 01). Analysis revealed 53 significantly SNPs associations mapped to coding genomic regions: 2 SNPs associated

中文翻译：

---

重度抑郁症难治性个体的遗传特征及其对经颅磁刺激的反应

对难治性抑郁症 (TRD) 有效，重复经颅磁刺激 (rTMS) 的缓解率在 30% 到 40% 之间。尽管遗传学可能是 rTMS 反应性个体间变异的潜在来源之一，但很少有研究试图确定 rTMS 反应的可能遗传基础 [1,2]。因此，我们使用了一种极端表型设计，其中我们比较了严格定义的 rTMS 响应者和非响应者之间的全基因组等位基因变异。共有 99 名 TRD 患者提供了知情同意，该研究得到了阿尔弗雷德医院人类研究和伦理委员会的批准。所有人都接受了 rTMS 协议，并在左背外侧前额叶皮层 (DLPFC) 完成了至少 18 次 10 Hz 的会话。基于蒙哥马利阿斯伯格抑郁评定量表 (MADRS) 的分数确定临床结果（有反应者或无反应者）。我们使用了极端表型设计。患者被分为两组，MADRS 量表降低至少 60% 的反应患者 (n 1⁄4 29) 和 MADRS 量表降低 10% 或更少的无反应患者 (n 1⁄4 19)。我们使用此标准是考虑到极端评分患者 (<10 >60) 可能会呈现更具代表性的遗传结果，从而提高检测表型-基因型关联的能力 [3,4]。使用 Infinium PsychArray-24 BeadChip（Illumina, Inc., San Diego, CA, USA）进行基因分型，并使用 PLINK 1.9 进行分析。我们执行质量控制以去除具有高错误率的个体或标记。数据质量控制参数为：调用率（GENO）> 90%，最大个体缺失率 (MIND) < 10%，次要等位基因频率 (MAF) > 5% 和 Hardy-Weinberg 平衡 (HWE) p 值 > 10e6。我们进行了标准关联分析，以 95% CI 比较两组的等位基因频率。在确定重要的 SNP 和基因后，使用 STRING 和 Cytoscape 数据库进行功能富集分析。在 48 个个体的最初 593,260 个 SNP 中，958 个变异因缺失基因型数据而被去除，310,522 个 SNP 由于次要等位基因阈值而被去除，4 个人因基因型数据缺失而被去除。这为关联研究留下了 281,780 个 SNP 和 44 个受试者。为了估计重要 SNP 的有效数量，我们考虑了每条染色体的样本和 SNP，应用了错误发现率 (FDR) 校正。然后为每个染色体确定一个新的 p 值（补充表 01）。分析显示 53 个显着的 SNP 关联映射到编码基因组区域：2 个 SNP 关联