转录组网络分析揭示创伤后应激障碍中创伤类型与祖先背景相关的异质性
该研究对南非466名孕妇(215例PTSD和/或MDD,251名创伤暴露对照)进行血液转录组网络分析,识别出与童年逆境、亲密伴侣暴力、近期生活应激、祖先背景及抑郁症状相关的基因表达模块。随机森林模型预测创伤相关表型达到中等显著性能(平均AUROC约0.61-0.69),但预测PTSD或MDD诊断接近随机水平(平均AUROC=0.52)。结论提示外周血转录组特征与创伤暴露特征和人群背景的关联强于诊断状态。
要点
- 外周血转录组特征与创伤暴露特征及祖先背景关联更强,与PTSD/MDD诊断关联弱。
- 预测创伤相关表型的模型平均AUROC约0.61-0.69,预测诊断仅0.52。
- 研究支持基于暴露特征的分层可作为传统诊断分类的补充框架。
核心结论
- 外周血转录组特征与创伤暴露特征和人群背景的关联强于与PTSD或MDD诊断状态的关联。
- 结果支持暴露信息分层可作为传统诊断分类的补充框架,以捕捉有生物学意义的异质性。
研究设计与样本
- 研究对南非466名孕妇的血液转录组进行网络分析,其中215例PTSD和/或MDD(平均年龄26.34岁,SD=5.77),251名创伤暴露对照(平均年龄27.14岁,SD=5.82)。
- 采用随机森林模型,嵌套交叉验证框架及置换检验评估转录组特征的预测相关性。
主要发现
- 识别出与童年逆境、亲密伴侣暴力、近期生活应激、祖先背景及抑郁症状相关的基因表达模块,提示多源分子变异。
- 模型预测创伤相关表型的平均AUROC约0.61-0.69(SD=0.07-0.10),具有中等且统计学显著的预测性能。
- 预测PTSD或MDD诊断的平均AUROC=0.52(SD=0.03),接近随机水平。
临床怎么用
- 临床可考虑将创伤暴露特征(如童年逆境、亲密伴侣暴力、近期生活应激)纳入评估,以辅助理解PTSD和MDD的生物学异质性。
- 外周血转录组特征不宜单独用于PTSD或MDD的诊断预测,但可能有助于创伤相关表型的风险分层。
局限性
- 样本仅限南非孕妇,结果外推至其他人群需谨慎。
- 预测创伤相关表型的性能仅为中等(AUROC约0.61-0.69),预测诊断接近随机。
- 横断面设计无法推断因果关系。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVE: Posttraumatic stress disorder (PTSD) is a debilitating psychiatric disorder that frequently co-occurs with major depressive disorder (MDD), a comorbidity that is partly attributable to shared genetic heritability and common environmental risk factors. Despite extensive efforts to identify peripheral transcriptomic biomarkers, findings have been inconsistent, potentially reflecting the substantial biological heterogeneity underlying these disorders. The primary goal of this study was to identify molecular signatures that reflect the clinical and biological heterogeneity of PTSD and MDD. METHODS: The authors present a network analysis of blood-based transcriptomics from 466 pregnant women in South Africa, including 215 individuals with PTSD and/or MDD (mean age, 26.34 years [SD=5.77]) and 251 trauma-exposed control participants (mean age, 27.14 years [SD=5.82]). Gene expression modules associated with trauma-related phenotypes were identified, including childhood adversity, intimate partner violence, and recent life stress, as well as ancestry and depressive symptoms, highlighting multiple sources of molecular variation. To evaluate the predictive relevance of these transcriptomic features, random forest models were applied using a nested cross-validation framework with permutation testing. RESULTS: Models achieved moderate and statistically significant performance in predicting trauma-related phenotypes (mean area under the receiver operating characteristic curve [AUROC], ~0.61-0.69 across models [SD=0.07-0.10]), whereas prediction of PTSD or MDD diagnosis showed near-chance performance (mean AUROC=0.52, SD=0.03). CONCLUSIONS: These findings suggest that peripheral blood transcriptomic signatures are more strongly associated with trauma exposure profiles and population background than with diagnostic status. Collectively, the results support exposure-informed stratification as a complementary framework to traditional diagnostic categories for capturing biologically meaningful heterogeneity.
引用
- Jiang, Bronk, Lake, et al. Transcriptomic Network Analysis Reveals Trauma Type- and Ancestry-Related Heterogeneity in Posttraumatic Stress Disorder. Am J Psychiatry. 2026. PMID: 42946686. doi:10.1176/appi.ajp.20251098
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。