检验医学 ›› 2025, Vol. 40 ›› Issue (7): 666-672.

• 论著 • 上一篇    下一篇

基于血清氨基酸和细胞因子构建儿童抑郁障碍风险预测模型

金伟峰1, 李丹2, 王梦霞2, 潘若璇2, 张泓1(), 林萍2()   

  1. 1.上海交通大学医学院附属儿童医院 上海市儿童医院检验科,上海 200040
    2.上海交通大学医学院附属精神卫生中心 上海市精神卫生中心检验科,上海 200030
  • 收稿日期:2024-11-20 修回日期:2025-04-10 出版日期:2025-07-30 发布日期:2025-07-28
  • 通讯作者: 张泓,E-mail:zhanghong3010@VIP.126.com;林萍,E-mail:linpingsun2000@aliyun.com
  • 作者简介:林萍,E-mail:linpingsun2000@aliyun.com
    张泓,E-mail:zhanghong3010@VIP.126.com
    金伟峰,男,1985年生,学士,副主任技师,主要从事临床质谱检测工作。
  • 基金资助:
    上海市卫生健康委员会资助项目(GWVI-3)

A risk prediction model for depressive disorders in children based on serum amino acids and cytokines

JIN Weifeng1, LI Dan2, WANG Mengxia2, PAN Nuoxuan2, ZHANG Hong1(), LIN Ping2()   

  1. 1. Department of Clinical Laboratory,Shanghai Children's Hospital,Shanghai Jiao Tong University School of Medicine,Shanghai 200040,China
    2. Department of Clinical Laboratory,Shanghai Mental Health Center,Shanghai Jiao Tong University School of Medicine,Shanghai 200030,China
  • Received:2024-11-20 Revised:2025-04-10 Online:2025-07-30 Published:2025-07-28

摘要:

目的 基于血清氨基酸和细胞因子构建预测儿童抑郁障碍风险的列线图模型,并评价其临床效能。方法 选取2024年1—10月上海市精神卫生中心抑郁障碍患儿103例(抑郁障碍组)和同期上海市儿童医院健康体检儿童54名(正常对照组)。分别采用流式细胞术法和液相色谱串联质谱法检测所有研究对象血清中10种细胞因子和23种氨基酸的水平。采用LASSO回归分析和多因素Logistic回归分析筛选变量。根据筛选出的变量建立列线图预测模型。采用受试者工作特征(ROC)曲线、Hosmer-Lemeshow拟合优度检验和临床决策曲线评估模型的临床效能。结果 抑郁障碍组和正常对照组之间血清色氨酸、苯丙氨酸、苏氨酸、精氨酸、甲硫氨酸、丙氨酸、天冬酰胺、谷氨酸、亮氨酸、赖氨酸、脯氨酸、酪氨酸、缬氨酸、鸟氨酸、5-羟色胺(5-HT)、犬尿氨酸、5-羟吲哚乙酸(5-HIAA)、α-干扰素 IFN-α)、白细胞介素(IL)-5、IL-1β、 γ-干扰素(IFN-γ)、肿瘤坏死因子-α(TNF-α)、IL-10、IL-6、IL-2水平差异均有统计学意义(P<0.1)。采用LASSO回归分析筛选出5个变量(精氨酸、天冬酰胺、谷氨酸、5-HT、IL-2)。精氨酸、天冬酰胺、5-HT降低和谷氨酸、IL-2升高均是儿童发生抑郁障碍的独立危险因素(P<0.05)。基于筛选出的5个变量构建的列线图预测模型诊断抑郁障碍的曲线下面积(AUC)为0.961,内部验证的AUC为0.939。列线图模型的预测概率与实际概率拟合度良好(Hosmer-Lemeshow检验P=0.510 6),在阈值概率0.02%~98.00%范围内有较高的净获益率,净获益率最大值为0.6。结论 基于精氨酸、天冬酰胺、谷氨酸、5-HT、IL-2构建的列线图预测模型在儿童发生抑郁障碍的诊断中有较高的效能,或可作为儿童抑郁障碍辅助诊断的工具。

关键词: 氨基酸, 细胞因子, 列线图预测模型, 抑郁障碍, 儿童

Abstract:

Objective To establish a nomogram model for predicting the risk of depressive disorders in children based on serum amino acids and cytokines,and to evaluate its clinical efficacy. Methods A total of 103 children with depressive disorders (depressive disorder group) from Shanghai Mental Health Center from January to October 2024 and 54 healthy children undergoing physical examinations (healthy control group) were enrolled. Ten cytokines and 23 amino acids in serum of all the subjects were determined by flow cytometry and liquid chromatography-tandem mass spectrometry,respectively. Variates were screened by LASSO regression analysis and multivariate Logistic regression analysis. A nomogram model based on the screened variates was established. The clinical efficacy was evaluated by receiver operating characteristic (ROC) curve,Hosmer-Lemeshow goodness-of-fit test and clinical decision curve. Results Serum levels of tryptophan,phenylalanine,threonine,arginine,methionine,alanine,asparagine,glutamic acid,leucine,lysine,proline,tyrosine,valine,ornithine,5-hydroxytryptamine(5-HT),kynurenine,5-hydroxyindoleacetic acid(5-HIAA),interferon-alpha(IFN-α),interleukin-5(IL-5),interleukin-1beta(IL-1β),interferon-gamma(IFN-γ),tumor necrosis factor-alpha(TNF-α),interleukin-10(IL-10),interleukin-6(IL-6)and interleukin-2(IL-2) showed statistical significance between depressive disorder group and healthy control group(P<0.1). Five variates(arginine,asparagine,glutamic acid,5-HT and IL-2)were selected using LASSO regression analysis. Decreased levels of arginine,asparagine and 5-HT,along with elevated levels of glutamic acid and IL-2,were all independent risk factors for the development of depressive disorders in children(P<0.05). The nomogram prediction model constructed based on the 5 selected variates had an area under curve(AUC) of 0.961 for diagnosing depressive disorder,with an internal validation AUC of 0.939. The predicted probability of the nomogram model showed good agreement with the actual probability(Hosmer-Lemeshow test P=0.510 6). The model demonstrated high net benefit rates within the threshold probability range of 0.02% to 98.00%,with a maximum net benefit rate of 0.6. Conclusions The nomogram prediction model based on arginine,asparagine,glutamic acid,5-HT and IL-2 demonstrates high efficacy in the diagnosis of childhood depressive disorders and may serve as an auxiliary tool for diagnosing depression in children.

Key words: Amino acid, Cytokine, Nomogram prediction model, Depressive disorder, Children

中图分类号: