Hou Can, Liu Haowen, Zeng Yu, Gong Yike, Yang Huazhen, Ye Weimin, Fang Fang, Valdimarsdóttir Unnur A, Song Huan
Mental Health Center and West China Biomedical Big Data Center, West China Hospital, Sichuan University, Chengdu, China.
Med-X Center for Informatics, Sichuan University, Chengdu, China.
Mol Psychiatry. 2025 Jul 18. doi: 10.1038/s41380-025-03120-y.
Depression is strongly associated with a range of subsequent diseases. To elucidate key mechanistic pathways for targeted interventions, this study aimed to determine the main disease networks associated with depression as well as their underlying genetic determinants. We developed a novel three-dimensional network approach which refines disease association verification by incorporating regularized partial correlations, and facilitates robust identification and visualization of disease clusters (i.e., groups of depression-associated diseases with high within-group connectivity) through both non-temporal (illustrating by x-axis and y-axis) and temporal (by z-axis) dimensions. We applied this approach to a matched cohort of 54,284 middle aged patients diagnosed with depression and their 496,005 age- and sex-matched unexposed individuals from the Swedish national registers and validated our findings in a cohort from the UK Biobank. Additionally, we conducted genetic analyses, including polygenic risk score (PRS) and genome-wide association studies (GWAS), using genetic data from 10,754 depression patients in the UK Biobank. Our analysis of the Swedish cohort identified nine reliable disease clusters consisting of 85 component diseases associated with depression, of which six clusters with 30 diseases were successfully validated using the UK Biobank cohort. These were clusters characterized by central nervous system (CNS) diseases, respiratory system diseases, cardiovascular and metabolic diseases, gastrointestinal diseases, musculoskeletal diseases, and mental disorders. PRS analysis revealed a dose-response relationship between genetic liability to depression and the susceptibility for subsequent disease clusters, while GWAS identified eight genome-wide significant loci in four of the clusters. Overall, our novel three-dimensional disease network approach identified six robust disease clusters after depression across two large cohorts, each with shared and cluster-specific genetic underpinnings. These findings warrant further research on genetic-based risk prediction and the development of therapeutic interventions aimed at health improvement for patients with depression.
抑郁症与一系列后续疾病密切相关。为了阐明靶向干预的关键机制途径，本研究旨在确定与抑郁症相关的主要疾病网络及其潜在的遗传决定因素。我们开发了一种新颖的三维网络方法，该方法通过纳入正则化偏相关来完善疾病关联验证，并通过非时间维度（以x轴和y轴表示）和时间维度（以z轴表示）促进对疾病集群（即组内连通性高的抑郁症相关疾病组）的稳健识别和可视化。我们将这种方法应用于来自瑞典国家登记册的54284名被诊断为抑郁症的中年患者及其496005名年龄和性别匹配的未患病个体的匹配队列，并在英国生物银行的队列中验证了我们的发现。此外，我们使用英国生物银行中10754名抑郁症患者的遗传数据进行了遗传分析，包括多基因风险评分（PRS）和全基因组关联研究（GWAS）。我们对瑞典队列的分析确定了九个可靠的疾病集群，由85种与抑郁症相关的组成疾病组成，其中六个包含30种疾病的集群在英国生物银行队列中得到了成功验证。这些集群的特征分别为中枢神经系统（CNS）疾病、呼吸系统疾病、心血管和代谢疾病、胃肠道疾病、肌肉骨骼疾病和精神障碍。PRS分析揭示了抑郁症的遗传易感性与后续疾病集群易感性之间的剂量反应关系，而GWAS在其中四个集群中确定了八个全基因组显著位点。总体而言，我们新颖的三维疾病网络方法在两个大型队列中确定了抑郁症后的六个稳健疾病集群，每个集群都有共同的和特定于集群的遗传基础。这些发现值得进一步研究基于遗传的风险预测以及旨在改善抑郁症患者健康的治疗干预措施的开发。
