School of Chemistry and Biochemistry , Georgia Institute of Technology , Atlanta , Georgia 30332 , United States.
Centro de Investigaciones en Bionanociencias (CIBION) , Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) , Godoy Cruz 2390 , C1425FQD, Ciudad de Buenos Aires , Argentina.
J Proteome Res. 2019 Mar 1;18(3):1316-1327. doi: 10.1021/acs.jproteome.8b00926. Epub 2019 Feb 20.
Technological advances in mass spectrometry (MS), liquid chromatography (LC) separations, nuclear magnetic resonance (NMR) spectroscopy, and big data analytics have made possible studying metabolism at an "omics" or systems level. Here, we applied a multiplatform (NMR + LC-MS) metabolomics approach to the study of preoperative metabolic alterations associated with prostate cancer recurrence. Thus far, predicting which patients will recur even after radical prostatectomy has not been possible. Correlation analysis on metabolite abundances detected on serum samples collected prior to surgery from prostate cancer patients ( n = 40 remission vs n = 40 recurrence) showed significant alterations in a number of pathways, including amino acid metabolism, purine and pyrimidine synthesis, tricarboxylic acid cycle, tryptophan catabolism, glucose, and lactate. Lipidomics experiments indicated higher lipid abundances on recurrent patients for a number of classes that included triglycerides, lysophosphatidylcholines, phosphatidylethanolamines, phosphatidylinositols, diglycerides, acyl carnitines, and ceramides. Machine learning approaches led to the selection of a 20-metabolite panel from a single preoperative blood sample that enabled prediction of recurrence with 92.6% accuracy, 94.4% sensitivity, and 91.9% specificity under cross-validation conditions.
质谱 (MS)、液相色谱 (LC) 分离、核磁共振 (NMR) 波谱和大数据分析等技术的进步使得研究代谢物组学或系统水平的代谢成为可能。在这里，我们应用多平台（NMR+LC-MS）代谢组学方法来研究与前列腺癌复发相关的术前代谢变化。到目前为止，预测哪些患者即使在根治性前列腺切除术后也会复发仍然是不可能的。对术前采集的前列腺癌患者血清样本（n=40 缓解 vs n=40 复发）中的代谢物丰度进行相关分析，结果显示许多途径的代谢物发生了显著变化，包括氨基酸代谢、嘌呤和嘧啶合成、三羧酸循环、色氨酸分解代谢、葡萄糖和乳酸。脂质组学实验表明，复发患者的许多脂质类别的丰度更高，包括甘油三酯、溶血磷脂酰胆碱、磷脂酰乙醇胺、磷脂酰肌醇、二甘油酯、酰基肉碱和神经酰胺。机器学习方法从单个术前血样中选择了一个 20 代谢物的检测面板，能够以 92.6%的准确率、94.4%的灵敏度和 91.9%的特异性进行交叉验证条件下的复发预测。
