Serum Metabolomic Profiles for Human Pancreatic Cancer Discrimination.
Serum Metabolomic Profiles for Human Pancreatic Cancer Discrimination.
复制标题
DOI：
10.3390/ijms18040767
复制
发表时间：
2017-04-04
影响因子：
5.6
通讯作者：
Sunamura M
中科院分区：
文献类型：
--
作者：
Itoi T;Sugimoto M;Umeda J;Sofuni A;Tsuchiya T;Tsuji S;Tanaka R;Tonozuka R;Honjo M;Moriyasu F;Kasuya K;Nagakawa Y;Abe Y;Takano K;Kawachi S;Shimazu M;Soga T;Tomita M;Sunamura M
本研究评估了血清代谢组学在区分恶性癌症（包括胰腺癌（PC））与恶性疾病（如胆道癌（BTC）、导管内乳头状粘液癌（IPMC）和各种良性胰胆疾病）的临床应用。毛细管电泳质谱用于分析带电代谢物。我们反复分析不同储存时间的血清样本 (n = 41)，以鉴定具有高定量重现性的代谢物，并随后分析所有样本 (n = 140)。总体而言，对 189 种代谢物进行了定量，66 种代谢物的变异系数为 20%，其中 24 种代谢物在对照组、良性组和恶性组之间显示出显着差异（p < 0.05；Steel-Dwass 检验）。开发了四种多重逻辑回归模型 (MLR)，其中一种 MLR 模型可以清楚地区分所有疾病患者与健康对照，受试者工作特征曲线下面积 (AUC) 为 0.970（95% 置信区间 (CI)，0.946–0.994，p < 0.0001）。另一种区分 PC 与 BTC 和 IPMC 的模型的 AUC = 0.831（95% CI，0.650-1.01，p = 0.0020），与癌胚抗原 (CEA)、碳水化合物抗原 19-9 (CA19-9)、胰腺癌相关抗原 (DUPAN2) 和 s-pancreas-1 抗原 (SPAN1) 等肿瘤标志物相比，准确性更高。代谢组学谱的变化可用于筛查恶性癌症以及区分 PC 和其他恶性疾病。
This study evaluated the clinical use of serum metabolomics to discriminate malignant cancers including pancreatic cancer (PC) from malignant diseases, such as biliary tract cancer (BTC), intraductal papillary mucinous carcinoma (IPMC), and various benign pancreaticobiliary diseases. Capillary electrophoresis−mass spectrometry was used to analyze charged metabolites. We repeatedly analyzed serum samples (n = 41) of different storage durations to identify metabolites showing high quantitative reproducibility, and subsequently analyzed all samples (n = 140). Overall, 189 metabolites were quantified and 66 metabolites had a 20% coefficient of variation and, of these, 24 metabolites showed significant differences among control, benign, and malignant groups (p < 0.05; Steel–Dwass test). Four multiple logistic regression models (MLR) were developed and one MLR model clearly discriminated all disease patients from healthy controls with an area under receiver operating characteristic curve (AUC) of 0.970 (95% confidential interval (CI), 0.946–0.994, p < 0.0001). Another model to discriminate PC from BTC and IPMC yielded AUC = 0.831 (95% CI, 0.650–1.01, p = 0.0020) with higher accuracy compared with tumor markers including carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), pancreatic cancer-associated antigen (DUPAN2) and s-pancreas-1 antigen (SPAN1). Changes in metabolomic profiles might be used to screen for malignant cancers as well as to differentiate between PC and other malignant diseases.