Plasma protein changes reflect colorectal cancer development and associated inflammation

Abstract

Introduction: Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of death worldwide. Efficient non-invasive blood-based biomarkers for CRC early detection and prognosis are urgently needed.

Methods: To identify novel potential plasma biomarkers, we applied a proximity extension assay (PEA), an antibody-based proteomics strategy to quantify the abundance of plasma proteins in CRC development and cancer-associated inflammation from few μL of plasma sample.

Results: Among the 690 quantified proteins, levels of 202 plasma proteins were significantly changed in CRC patients compared to age-and-sex-matched healthy subjects. We identified novel protein changes involved in Th17 activity, oncogenic pathways, and cancer-related inflammation with potential implications in the CRC diagnosis. Moreover, the interferon γ (IFNG), interleukin (IL) 32, and IL17C were identified as associated with the early stages of CRC, whereas lysophosphatidic acid phosphatase type 6 (ACP6), Fms-related tyrosine kinase 4 (FLT4), and MANSC domain-containing protein 1 (MANSC1) were correlated with the late-stages of CRC.

Discussion: Further study to characterize the newly identified plasma protein changes from larger cohorts will facilitate the identification of potential novel diagnostic, prognostic biomarkers for CRC.

Keywords: biomarker; colorectal cancer; cytokines; early detection; inflammation; plasma proteomics; prognosis; proximity extension assay.

Figure 1

Colorectal cancer (CRC) development causes cytokine and oncogenic signaling pathway changes in plasma. (A) Volcano plot of statistical significance against fold-change of proteins between CRC patients and healthy controls. Dots indicate individual proteins and the red and blue dots represent significant up-regulation and down-regulation in patients, respectively. (B) Network of KEGG pathway enrichment analysis combined with STRING protein-protein interaction network analysis. Green and red proteins indicate significant up-regulation and down-regulation, respectively. (C) Box and whisker plots of selected DEPs not previously reported associated with CRC. * indicates statistically significant with an adjusted p-value < 0.05, ** indicates an adjusted p-value < 0.01, and *** indicates an adjusted p-value < 0.001. DEP, differentially expressed protein; FC, fold change, NPX; normalized protein expression.

Figure 2

Plasma protein changes induced by cancer-associated inflammation in CRC patients. (A) Heatmap of proteins with significant correlation with inflammation status. Protein expression is transformed with a z-score by row normalization and distributed by hierarchical clustering. The correlation coefficients (right) indicate a positive/negative correlation for each protein. (B) Volcano plot of statistical significance against fold-change of proteins between CRC patients with inflammation and without inflammation. Dots indicate individual proteins and the red and blue dots represent significant up-regulation and down-regulation in CRC patients with inflammation, respectively. (C) Box and whisker plots of selected DEPs not previously associated with cancer-related inflammation in CRC patients. *: adjusted p-value < 0.05, **: adjusted p-value < 0.01.

Figure 3

Plasma protein expression differences between early and late stages of CRC. (A) Heatmap of proteins with significant correlation with tumor stage. Protein expression is transformed with a z-score by row normalization and distributed by hierarchical clustering. The correlation coefficients (right) indicate a positive/negative correlation for each protein. (B) Volcano plot of statistical significance against fold-change of proteins between CRC patients with early tumor stage and with late tumor stage. Dots indicate individual protein and the red and blue dots represent significant up-regulation and down-regulation in CRC patients with late tumor stage, respectively. (C) Box and whisker plots of DEPs that are novel potential prognostic biomarkers associated with cancer stages in CRC patients. *: adjusted p-value < 0.05. (D) Venn diagram with the differentially expressed proteins for each comparison: CRC patients vs. control, Inflammation vs. Non-inflammation, and Early vs. Late. Black arrows indicate the proteins of interest that are in common between comparisons. Red and blue arrows indicate up-regulation and down-regulation for the specified group, respectively. C, control; Inf., inflammation; Non-Inf., non-inflammation; P, patient.

Figure 4

Validation of potential candidate biomarkers. (A) Plots with the concentrations of CSF3, IFNG, IL6, CXCL9, and CCL23 in CRC patients (P) and healthy controls (HC) (mean ± SEM) detected by Luminex. (B) Plot with the concentrations of ACP6 detected by ELISA in CRC patients with early and late stages, respectively (mean ± SEM). T test was used for statistical analysis. *: p-value < 0.05, ****: p-value < 0.0001, NS: non-significance.