整合多组学和众包评估胎盘-脑轴生物标志物预测神经发育障碍

Integration of multi-omics and crowdsourcing assessment of placenta-brain axis biomarkers for predicting neurodevelopmental disorders

整合多组学和众包评估胎盘-脑轴生物标志物预测神经发育障碍

Authors：Yimin Zhang, Heyue Jin, Juan Tong, Hong Gan, Fangbiao Tao, Yumin Zhu

Source：J Affect Disord

PMID: 40544887

DOI: 10.1016/j.jad.2025.119732

Abstract

Neurodevelopmental disorders (NDDs) are heterogeneous and multifactorial psychiatric disorders with abnormalities in multiple biological domains. It is increasingly recognized that the placenta profoundly influences fetal neurodevelopment due to the finding of the placenta-brain axis. However, few studies have investigated the interplay between placenta dysfunction and NDDs, especially autism spectrum disorder (ASD) symptoms, attention-deficit/hyperactivity disorder (ADHD) symptoms, and intellectual disability (ID) symptoms, by using integrative multi-omics data. Here, we performed an analysis of transcriptomic and non-targeted metabolomic individually and integratively to characterize the placental multi-omics profiles of children with NDDs in Ma'anshan Birth Cohort, and to identify biomarkers associated with the placenta-brain axis. Integrating transcriptome and metabolome perspectives, we further conducted a multi-omics machine learning workflow to discover reliable placental biomarkers for early diagnosis of these NDDs in the participants. Integrative analysis of differentially expressed genes and metabolites revealed a common intrauterine regulation mechanism for ASD symptoms and ADHD symptoms. Combined with machine learning, prediction models were constructed and 99.7 % of ASD symptoms, 99.0 % of ADHD symptoms, and 95.7 % of ID symptoms were correctly classified. This is the first study combining transcriptomics and metabolomics from the perspective of the placental-brain axis in humans, which contributes to a deeper understanding about the pathogenesis of the NDDs and may potentially pave the way toward molecular diagnosis of different disorders.

Keywords: Biomarker; Machine learning; Metabolome; Neurodevelopmental disorder; Placenta; Transcriptome.

摘要

神经发育障碍是异质性和多因素的精神疾病，在多个生物学领域存在异常。由于胎盘-大脑轴的发现，人们越来越认识到胎盘对胎儿神经发育有着深远的影响。然而，很少有研究通过整合多组学数据来探究胎盘功能障碍与神经发育障碍（尤其是自闭症谱系障碍症状、注意缺陷/多动障碍症状和智力障碍症状）之间的相互作用。在此，我们对马鞍山优生优育队列中患有神经发育障碍的儿童进行了转录组学和非靶向代谢组学的单独及整合分析，以表征其胎盘多组学特征，并识别与胎盘-大脑轴相关的生物标志物。结合转录组和代谢组的视角，我们进一步开展了一项多组学机器学习工作流程，以发现可靠的胎盘生物标志物，用于这些神经发育障碍的早期诊断。对差异表达基因和代谢物的综合分析揭示了自闭症谱系障碍症状和注意缺陷多动障碍症状在子宫内存在共同的调节机制。结合机器学习，构建了预测模型，正确分类了 99.7% 的自闭症谱系障碍症状、99.0% 的注意缺陷多动障碍症状和 95.7% 的智力障碍症状。这是首次从人类胎盘-大脑轴的角度结合转录组学和代谢组学的研究，有助于更深入地了解神经发育障碍的发病机制，并可能为不同障碍的分子诊断铺平道路。

关键词：生物标志物；机器学习；代谢组；神经发育障碍；胎盘；转录组。