McConnell Yarrow J, Farshidfar Farshad, Weljie Aalim M, Kopciuk Karen A, Dixon Elijah, Ball Chad G, Sutherland Francis R, Vogel Hans J, Bathe Oliver F
Department of Oncology, University of Calgary, Calgary, AB T2N 4N2, Canada.
Department of Surgery, University of Calgary, Calgary, AB T2N 4N2, Canada.
Metabolites. 2017 Jan 13;7(1):3. doi: 10.3390/metabo7010003.
Previous work demonstrated that serum metabolomics can distinguish pancreatic cancer from benign disease. However, in the clinic, non-pancreatic periampullary cancers are difficult to distinguish from pancreatic cancer. Therefore, to test the clinical utility of this technology, we determined whether pancreatic and periampullary adenocarcinoma could be distinguished from benign masses and biliary strictures. Sera from 157 patients with malignant and benign pancreatic and periampullary lesions were analyzed using proton nuclear magnetic resonance (¹H-NMR) spectroscopy and gas chromatography-mass spectrometry (GC-MS). Multivariate projection modeling using SIMCA-P+ software in training datasets ( = 80) was used to generate the best models to differentiate disease states. Models were validated in test datasets ( = 77). The final ¹H-NMR spectroscopy and GC-MS metabolomic profiles consisted of 14 and 18 compounds, with AUROC values of 0.74 (SE 0.06) and 0.62 (SE 0.08), respectively. The combination of ¹H-NMR spectroscopy and GC-MS metabolites did not substantially improve this performance (AUROC 0.66, SE 0.08). In patients with adenocarcinoma, glutamate levels were consistently higher, while glutamine and alanine levels were consistently lower. Pancreatic and periampullary adenocarcinomas can be distinguished from benign lesions. To further enhance the discriminatory power of metabolomics in this setting, it will be important to identify the metabolomic changes that characterize each of the subclasses of this heterogeneous group of cancers.
先前的研究表明，血清代谢组学能够区分胰腺癌与良性疾病。然而，在临床中，非胰腺壶腹周围癌很难与胰腺癌区分开来。因此，为了测试这项技术的临床实用性，我们确定了胰腺和壶腹周围腺癌是否能够与良性肿块及胆管狭窄区分开来。我们使用质子核磁共振（¹H-NMR）光谱法和气相色谱-质谱联用（GC-MS）分析法，对157例患有恶性和良性胰腺及壶腹周围病变的患者的血清进行了分析。在训练数据集（n = 80）中，使用SIMCA-P+软件进行多变量投影建模，以生成区分疾病状态的最佳模型。在测试数据集（n = 77）中对模型进行了验证。最终的¹H-NMR光谱和GC-MS代谢组学图谱分别由14种和18种化合物组成，曲线下面积（AUROC）值分别为0.74（标准误0.06）和0.62（标准误0.08）。¹H-NMR光谱和GC-MS代谢物的组合并未显著提高这一性能（AUROC 0.66，标准误0.08）。在腺癌患者中，谷氨酸水平一直较高，而谷氨酰胺和丙氨酸水平一直较低。胰腺和壶腹周围腺癌可以与良性病变区分开来。为了进一步提高代谢组学在这种情况下的鉴别能力，识别出这一异质性癌症群体各亚类所特有的代谢组学变化将很重要。
