| 代清池,习聪,李爽,等.基于大鼠血浆辐射敏感代谢物绝对定量的生物剂量估算与分类模型构建与验证[J].中华放射医学与防护杂志,2026,46(3):230-237.Dai Qingchi,Xi Cong,Li Shuang,et al.Construction and validation of biodosimetry models for dose estimation and classification based on absolute quantification of radiation-sensitive metabolites in rat plasma[J].Chin J Radiol Med Prot,2026,46(3):230-237 |
| 基于大鼠血浆辐射敏感代谢物绝对定量的生物剂量估算与分类模型构建与验证 |
| Construction and validation of biodosimetry models for dose estimation and classification based on absolute quantification of radiation-sensitive metabolites in rat plasma |
| 投稿时间：2025-10-28 |
| DOI：10.3760/cma.j.cn112271-20251028-00382 |
| 中文关键词: 代谢组学 电离辐射 生物标志物 剂量估算 受试者工作特征曲线 |
| 英文关键词:Metabolomics Ionizing radiation Biomarker Dose estimation Receiver operating characteristic (ROC) |
| 基金项目:国家自然科学基金(82003393);中国疾控中心辐射安全所青年科学研究所长基金(2023-02) |
|
| 摘要点击次数: 3310 |
| 全文下载次数: 68 |
| 中文摘要: |
| 目的 对15种辐射敏感代谢物进行绝对定量分析,建立并验证大鼠血浆样本的辐射剂量估算与分类模型,以探讨辐射敏感代谢物作为辐射生物标志物的可行性。方法 用60Co γ射线对50只大鼠进行全身照射,按随机数表法分为6组,照射剂量分别为0、1、2、3、5、8 Gy,照射后3 d采集血浆,采用超高效液相色谱-三重四极杆质谱(UPLC-MS/MS)联用技术对血浆中15个辐射敏感代谢物进行绝对定量。基于定量结果建立15个单变量一般线性模型和代谢物组合的多元线性回归模型。另取25只大鼠,按随机数表法分为5组,每组5只,照射剂量分别为0、0.5、2.5、4、6 Gy,同样于照射后3 d采集血浆并对15个辐射敏感代谢物进行绝对定量,用于评估所建模型的预测准确性及稳健性。结果 照射后3 d,大鼠血浆中苯丙氨酸、牛磺酸浓度呈剂量依赖性上调(R2＞0.8, P <0.05),丁酰肉碱、瓜氨酸、N-乙酰鸟氨酸、脯氨酸浓度呈剂量依赖性下调(R2＞0.8, P <0.05)。建立多元线性回归模型y=-0.237x1+0.046x2+0.058x3-3.246x4+5.126(x1:瓜氨酸,x2:苏氨酸,x3:肌酸,x4:N-乙酰鸟氨酸,R2 = 0.941)。模型验证显示,在受照后0~6 Gy剂量区间内,瓜氨酸单变量模型与多变量模型中均有48%样本预测相对标准偏差(RSD)< 40%,多变量模型具有更好的稳定性。瓜氨酸、脯氨酸、丁酰肉碱、己酰肉碱代谢物组合区分0 Gy与> 0 Gy样本的AUC值在训练集和验证集中分别为0.923和0.910,瓜氨酸、脯氨酸、丁酰肉碱、肌酸、己酰肉碱、棕榈酰肉碱代谢物组合区分<2 Gy与≥2 Gy样本的AUC值在训练集和验证集中分别为0.994和1.000。结论 瓜氨酸、苏氨酸、肌酸、N-乙酰鸟氨酸代谢物组合与瓜氨酸、脯氨酸、丁酰肉碱、肌酸、己酰肉碱、棕榈酰肉碱代谢物组合分别显示出用于辐射特定剂量估算与分类的标志物潜力。 |
| 英文摘要: |
| Objective To explore the feasibility of radiation-sensitive metabolites as biomarkers, this study established and validated radiation dose estimation and classification models for rat plasma samples by performing absolute quantification of 15 radiation-sensitive metabolites. Methods Fifty rats were randomly divided into 6 groups(n = 8 or 9) and exposed to total-body irradiation (TBI) using 60Co γ-rays at doses of 0, 1, 2, 3, 5, and 8 Gy, respectively. Plasma samples were collected 3 d post-irradiation, and absolute quantification of 15 radiation-sensitive metabolites was performed by ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). Based on the quantification result, fifteen univariate general linear models and a multivariate linear regression model incorporating metabolite panel were developed. For external validation, an additional cohort of 25 rats were radomly divided into 5 groups(n = 5) and irradiated at doses of 0, 0.5, 2.5, 4, and 6 Gy, respectively. Following the same protocol, plasma was collected 3 d after exposure and the same 15 metabolites were absolutely quantified to evaluate the predictive accuracy and robustness of the established models. Results At 3 d post-irradiation, we observed a significant dose-dependent increase in metabolites phenylalanine and taurine (R2 > 0.8, P < 0.05), while the metabolites butyrylcarnitine, citrulline, N-acetylornithine, and proline exhibited a significant dose-dependent decrease (R2 > 0.8, P < 0.05). The general linear models for citrulline, phenylalanine, butyrylcarnitine, N-acetylornithine, proline, and taurine all achieved R2 > 0.8 in the univariate analysis. The final multivariate linear regression model was constructed as: y = -0.237x1+0.046x2+0.058x3-3.246x4+5.126 (x1: citrulline, x2:threonine, x3: creatine, x4:N-acetylornithine, R2 = 0.941). Model validation within 0-6 Gy irradiation revealed that 48% of sample predictions had a relative standard deviation (RSD) below 40% in both the univariate model of citrulline and the multivariate model, with the latter demonstrating superior stability. The combined metabolite panel (citrulline, proline, butyrylcarnitine, creatine, hexanoylcarnitine, palmitoylcarnitine) achieved AUC values of 0.923 and 0.910 in the training and validation sets, respectively, for discriminating between 0 Gy and >0 Gy samples. Similarly, for distinguishing <2 Gy from ≥2 Gy samples, the panel yielded AUCs of 0.994 and 1.000 in the training and validation sets. Conclusions The metabolite panels consisting of citrulline, threonine, creatine and N-acetylornithine, as well as the combination of citrulline, proline, butyrylcarnitine, creatine, hexanoylcarnitine, and palmitoylcarnitine, show potential as biomarkers for specific radiation dose estimation and classification, respectively. |
| HTML 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|