文献速递骨质疏松与代谢性骨病

利用液相色谱-串联质谱代谢组学方法发现骨质疏松潜在生物标志物

Discovery of potential biomarkers for osteoporosis using LC-MS/MS metabolomic methods.
PubMed 文献 · Osteoporos Int · 2019-02-18

研究纳入上海地区320名参与者(138名男性和182名绝经后女性),检测骨转换标志物及221种血清代谢物,并与腰椎、股骨颈和全髋骨密度关联。在男性中,骨转换标志物模型AUC为0.729,加入所选代谢物后模型1的AUC升至0.828(模型1与骨转换标志物模型比较P=0.0158);绝经后女性中骨转换标志物模型AUC为0.638,模型2的AUC为0.741(比较P=0.0182)。作者认为代谢物联合传统骨转换标志物在区分骨质疏松方面可能优于单用骨转换标志物。

要点

  • 男性中骨转换标志物模型AUC 0.729,加入代谢物后模型1 AUC 0.828,比较P=0.0158。
  • 绝经后女性中骨转换标志物模型AUC 0.638,加入代谢物后模型2 AUC 0.741,比较P=0.0182。
  • 代谢物联合传统骨转换标志物可能优于单用骨转换标志物区分骨质疏松。

核心结论

  • 代谢物联合传统骨转换标志物在男性和绝经后女性中区分骨质疏松可能优于单用骨转换标志物。

研究设计与样本

  • 研究招募了320名参与者,包括138名男性和182名绝经后女性,来自上海地区。
  • 检测骨转换标志物(骨钙素、PINP和β-CTX)及其他生化指标。
  • 采用双能X线吸收法测定腰椎(L1-4)、股骨颈和全髋骨密度。
  • 通过质谱评估包含221种代谢物的血清代谢组谱,涵盖酰基肉碱、氨基酸、生物胺、甘油磷脂、鞘脂和己糖五组。

主要发现

  • 主成分分析或偏最小二乘判别分析模型中,不同骨密度组间代谢谱未见视觉分离。
  • 在男性和绝经后女性中分别比较三种不同骨密度水平的代谢物,并通过基于随机森林的特征选择进一步筛选。
  • 筛选出重要性最高(≥5%)的代谢物(男性5种,绝经后女性9种)用于构建骨质疏松分类模型。
  • 男性中,加入所选代谢物后模型1的AUC为0.828(95%CI 0.754-0.888,P<0.0001),骨转换标志物模型AUC为0.729(95%CI 0.647-0.802,P<0.0001),模型1与骨转换标志物模型比较P=0.0158。
  • 绝经后女性中,模型2的AUC为0.741(95%CI 0.669-0.803,P<0.0001),骨转换标志物模型AUC为0.638(95%CI 0.562-0.708,P=0.0025),模型1与骨转换标志物模型比较P=0.0182。

关键数值与推荐

  • 男性:骨转换标志物模型AUC 0.729(95%CI 0.647-0.802,P<0.0001),模型1 AUC 0.828(95%CI 0.754-0.888,P<0.0001),模型1与骨转换标志物模型比较P=0.0158。
  • 绝经后女性:骨转换标志物模型AUC 0.638(95%CI 0.562-0.708,P=0.0025),模型2 AUC 0.741(95%CI 0.669-0.803,P<0.0001),模型1与骨转换标志物模型比较P=0.0182。
  • 所选代谢物重要性阈值≥5%,男性5种,绝经后女性9种。

临床怎么用

  • 原文未据此给出推荐。
  • 作者认为代谢物联合传统骨转换标志物可能对骨质疏松的早期诊断有意义。

局限性

  • 原文未报告局限性。

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

UNLABELLED: Our study focused on the associations of metabolites with BMD and osteoporosis, finding that several metabolites are associated with BMD, and metabolites combined with bone turnover markers tend to be more sensitive in distinguishing osteoporosis in both males and postmenopausal females, which might be meaningful for the early diagnosis of osteoporosis. INTRODUCTION: Our study aimed to evaluate the association of metabolites with bone, trying to find new metabolic markers that are distinguishing for low bone mineral density (BMD). METHODS: Our study recruited 320 participants, including 138 males and 182 postmenopausal females from the Shanghai area. Bone turnover markers (BTMs), including osteocalcin, PINP and β-CTX, and other biochemical traits were tested. BMD values of the lumber spine (L1-4), femoral neck and total hip were determined using dual-energy X-ray absorptiometry and the serum metabolome profiles including 221 metabolites from five groups (acylcarnitines, amino acids, biogenic amines, glycerophospholipids, sphingolipids and hexose) were assessed by mass spectrometry. RESULTS: No visual separation in the metabolic profiles between different BMD groups was observed in principal component analysis (PCA) or partial least squares discriminant analysis (PLS-DA) models. We compared metabolites in three groups with different BMD levels in males and postmenopausal females separately and further filtering these metabolites via random forest-based feature selection, a commonly applied machine learning algorithm which could select the features with the greatest impact on osteoporosis, then metabolites with the highest importance (≥ 5%) (5 in males and 9 in postmenopausal females) were selected to construct better models for osteoporosis classification. After adding these selected metabolites to the model, the area under the curve (AUC) of receiver operating characteristic (ROC) curves increased significantly (BTMs: AUC 0.729, 95% CI 0.647-0.802, p < 0.0001, model 1: AUC = 0.828, 95% CI 0.754-0.888, p < 0.0001; model 1 versus model of BTMs: p = 0.0158) compared to the AUC of the BTM-only model in males. Similar results were also observed in postmenopausal females (BTMs: AUC = 0.638, 95% CI 0.562-0.708, p = 0.0025; model 2: AUC = 0.741, 95% CI 0.669-0.803, p < 0.0001; model 1 versus model of BTMs: p = 0.0182). CONCLUSION: Metabolites combined with traditional BTMs tend to better markers for distinguishing osteoporosis in both males and postmenopausal females than BTMs alone.

来源:Osteoporos Int,查看原文。版权归原作者所有。

引用

  1. Wang J, Yan D, Zhao A, et al. Discovery of potential biomarkers for osteoporosis using LC-MS/MS metabolomic methods. Osteoporos Int. 2019. PMID: 30778642. doi:10.1007/s00198-019-04892-0

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。