[Integrated analysis of serum untargeted metabolomics and targeted bile acid metabolomics for identification of diagnostic biomarkers for colorectal cancer]

Abstract

Objective: To identify potential diagnostic biomarkers of colorectal cancer (CRC) using serum metabolomic technology for minimally invasive and efficient screening for CRC.

Methods: Serum samples from 79 healthy individuals and 82 CRC patients were analyzed by metabolomics using ultra-high-performance liquid chromatography-tandem highresolution mass spectrometry (UHPLC-HRMS). The differential metabolites between the two groups were analyzed using principal component analysis and orthogonal partial least squares discriminant analysis (OPLS-DA). Receiver operating characteristic curve (ROC) analysis was performed to identify the differential metabolites with good diagnostic performance (AUC>0.80) for CRC, and targeted bile acid metabolomics was used to verify the selected bile acids as biomarkers.

Results: Serum metabolic profiles differed significantly between the healthy individuals and CRC patients, and a total of 82 differential metabolites (mostly fatty acids and glycerophospholipids) were selected. ROC analysis identified 10 differential metabolites, including adenine, bilirubin, ACar 12:0, ACar 10:1, ACar 9:0, PC 18:2e, deoxycholic acid, chenodeoxycholic acid, ACar 14:1 and palmitoylcarnitine. One of these metabolites was significantly up-regulated and 9 were down-regulated in the serum of CRC patients (P < 0.05). Multivariate ROC analysis with support vector machine algorithm showed that the biomarker panel consisting of 7 differential metabolites had an AUC of 0.94 for CRC diagnosis. The results of targeted bile acid metabolomics were consistent with those of untargeted metabolomics. The serum levels of deoxycholic acid and chenodeoxycholic acid were significantly down-regulated in patients with CRC as compared with the healthy individuals (P < 0.05).

Conclusion: Metabolic disorders of fatty acids and glycerophospholipids are closely related wigh tumorigenesis of CRC. Ten differential metabolites show good performance for CRC diagnosis, and the panel consisting 7 of these metabolites has important diagnostic value for CRC. Deoxycholic acid and chenodeoxycholic acid may serve as potential diagnostic biomarkers of CRC.

Keywords: biomarkers; colorectal cancer; serum untargeted metabolomics; targeted bile acid metabolomics.

图 1

NR和CRC组血清样品的总离子流图 Total ion chromatograms (TICs) of serum samples from normal control (NR) and CRC groups. A: TIC of serum samples from NR group in ESI+mode. B: TIC of serum samples from NR group in ESI-mode. C: TIC of serum samples from CRC group in ESI+mode. D: TIC of serum samples from CRC group in ESI-mode. ESI: Electrospray ionization; RT: Retention time; MS: Mass spectrometry; NR: Normal control; CRC: Colorectal cancer.

图 2

NR组与CRC组在正、负离子模式下的PCA得分图，OPLS-DA得分图及其置换检验图 Principal component analysis (PCA) score plots, OPLS-DA score plots, and the corresponding permutation tests between NR and CRC groups in the ESI+ and ESI-modes. A: PCA score plot between two groups in the ESI+mode. B: PCA score plot between two groups in the ESI-mode. C: OPLS-DA score plot between two groups in the ESI+mode. D: OPLS-DA score plot between two groups in the ESI-mode. E: Permutation test between two groups in the ESI+ mode. F: Permutation test between two groups in the ESI-mode. OPLS-DA: Orthogonal partial least squares discriminant analysis; ESI: Electrospray ionization; QC: Quality control; NR: Normal control; CRC: Colorectal cancer.

图 3

NR组与CRC组间差异代谢物的构成分布 Component distribution of differential metabolites between NR and CRC groups.

图 4

NR组和CRC组间血清差异代谢物的单因素ROC分析及组间变化趋势、基于SVM算法的多元ROC分析及差异代谢物的选择频率 Univariate ROC analysis of serum differential metabolites between NR and CRC groups, their change trend, multivariate ROC analysis and selected frequency of the differential metabolites based on SVM algorithm. Differential metabolites with AUC>0.80 in ROC analysis between NR and CRC are included. A: Adenine. B: Bilirubin. C: ACar 12:0. D: ACar: 10:1. E: ACar 9:0. F: PC 18:2e. G: Deoxycholic acid. H: Chenodeoxycholic acid. I: ACar 14:1. J: Palmitoylcarnitine. Among them, the levels of differential metabolites of A, B, C, D, E, F, G, H and I show significant decrease, whereas J exhibits obvious increase in the patients with CRC compared to NR (P < 0.05). K: Multivariate ROC analysis of the differential metabolites with high differentiating performance for NR and CRC groups. L: The selected frequency of differential metabolites during multivariate ROC analysis by SVM algorithm. ROC: Receiver operating characteristic curve; SVM: Support vector machine; ACar: Acetyl carnitine; PC: Phosphatidylcholine; Var: Variate; AUC: Area under the curve; Cl: Confidence interval; NR: Normal control; CRC: Colorectal cancer. ^*P < 0.05 vs NR group.

图 5

4种对CRC具有高诊断效能的差异代谢物的LC-MS/MS验证 Identification of 4 differential metabolites with high diagnostic ability for CRC by LC-MS/MS method. A: Chenodeoxycholic acid; B: Deoxycholic acid; C: Palmitoylcarnitine; D: Bilirubin. CRC: Colorectal cancer; RT: Retention time; m/z: Mass to charge ratio.

图 6

基于靶向胆汁酸代谢组学测定的7种差异胆汁酸及其组间变化趋势 Seven differential bile acids between NR and CRC groups based on targeted bile acid metabolomics and their change trend. A: CA. B: DCA. C: CDCA. D: HDCA. E: GCA. F: GCDCA. G: UDCA. The levels of differential BAs of A, B, C, D and G showed significant decrease, whereas E and F exhibited obvious increase in the CRC patients compared to the NR (P < 0.05). BAs: Bile acids; NR: Normal control; CRC: Colorectal cancer; Conc.: Concentration; CA: Cholic acid; DCA: Deoxycholic acid; CDCA: Chenodeoxycholic acid; HDCA: Hyodeoxycholic acid; GCA: Glycocholic acid; GCDCA: Glycochenodeoxycholic acid; UDCA: Ursodeoxycholic acid. ^*P < 0.05 vs NR group.