Untargeted metabolomics characterization of the resectable pancreatic ductal adenocarcinoma

Abstract

Background: Diagnosis of pancreatic ductal adenocarcinoma (PDAC) is difficult due to the lack of specific symptoms and screening methods. Only less than 10% of PDAC patients are candidates for surgery at the time of diagnosis. Thus, there is a great global unmet need for valuable biomarkers that could improve the opportunity to detect PDAC at the resectable stage. This study aimed to develop a potential biomarker model for the detection of resectable PDAC by tissue and serum metabolomics.

Methods: Ultra-high-performance liquid chromatography and quadrupole time-of-flight mass spectrometry (UHPLC-QTOF-MS/MS) was performed for metabolome quantification in 98 serum samples (49 PDAC patients and 49 healthy controls (HCs)) and 20 pairs of matched pancreatic cancer tissues (PCTs) and adjacent noncancerous tissues (ANTs) from PDAC patients. Univariate and multivariate analyses were used to profile the differential metabolites between PDAC and HC.

Results: A total of 12 differential metabolites were present in both serum and tissue samples of PDAC. Among them, a total of eight differential metabolites showed the same expressional levels, including four upregulated and four downregulated metabolites. Finally, a panel of three metabolites including 16-hydroxypalmitic acid, phenylalanine, and norleucine was constructed by logistic regression analysis. Notably, the panel was capable of distinguishing resectable PDAC from HC with an AUC value of 0.942. Additionally, a multimarker model based on the 3-metabolites-based panel and CA19-9 showed a better performance than the metabolites panel or CA19-9 alone (AUC: 0.968 vs. 0.942, 0.850).

Conclusions: Taken together, the resectable early-stage PDAC has unique metabolic features in serum and tissue samples. The defined panel of three metabolites has the potential value for early screening of PDAC at the resectable stage.

Keywords: Pancreatic cancer; early diagnosis; metabolomics; serum; tissue.

Figure 1.

(A, B) Orthogonal partial least squares discriminant analysis (OPLS-DA) scatter plot between PDAC and healthy controls in serum samples in the positive and negative ion modes. (C, D) The statistical validation of the corresponding OPLS-DA models by permutation tests (200 times). Green plot indicates PDAC, blue plot indicated HC.

Figure 2.

(A, B) Orthogonal partial least squares discriminant analysis (OPLS-DA) scatter plot between paired pancreatic cancer tissues (PCT) and adjacent noncancerous tissues (ANT) in the positive and negative ion modes. (C, D) The statistical validation of the corresponding OPLS-DA models by permutation tests (200 times). Green plot indicated PCT, blue plot indicates ANT.

Figure 3.

(A) Heatmap visualization of the 12 overlapping metabolites from PDAC versus HC in serum samples. (B) Heatmap visualization of the 12 overlapping metabolites from paired pancreatic cancer tissues (PCT) versus adjacent noncancerous tissues (ANT). Blue color indicated lower level in PDAC. Red color indicated higher level of PDAC.

Figure 4.

(A) Correlation analysis of eight potential metabolites have the same changing patterns in both the serum and tissue samples of PDAC. LysoPC(18:0), 1-stearoyl-sn-glycerol-3-phosphocholine; LysoPE(16:0), 1-Palmitoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine. (B). Receiver-operating characteristic (ROC) curves analyses of the serum metabolites panel for differentiating PDCA patients from healthy controls. AUC, area under the curve; PDAC, pancreatic ductal adenocarcinoma. Blue line indicates CA19-9. Green line indicates metabolites panel. Red line indicates metabolites panel plus CA19-9.