High-throughput metabolomics identifies new biomarkers for cervical cancer

Abstract

Background: Cervical cancer (CC) is a danger to women's health, especially in many developing countries. Metabolomics can make the connection between genotypes and phenotypes. It provides a wide spectrum profile of biological processes under pathological or physiological conditions.

Method: In this study, we conducted plasma metabolomics of healthy volunteers and CC patients and integratively analyzed them with public CC tissue transcriptomics from Gene Expression Omnibus (GEO).

Result: Here, we screened out a panel of 5 metabolites to precisely distinguish CC patients from healthy volunteers. Furthermore, we utilized multi-omics approaches to explore patients with stage I-IIA1 and IIA2-IV4 CC and comprehensively analyzed the dysregulation of genes and metabolites in CC progression. We identified that plasma levels of trimethylamine N-oxide (TMAO) were associated with tumor size and regarded as a risk factor for CC. Moreover, we demonstrated that TMAO could promote HeLa cell proliferation in vitro. In this study, we delineated metabolic profiling in healthy volunteers and CC patients and revealed that TMAO was a potential biomarker to discriminate between I-IIA1 and IIA2-IV patients to indicate CC deterioration.

Conclusion: Our study identified a diagnostic model consisting of five metabolites in plasma that can effectively distinguish CC from healthy volunteers. Furthermore, we proposed that TMAO was associated with CC progression and might serve as a potential non-invasive biomarker to predict CC substage.

Impact: These findings provided evidence of the important role of metabolic molecules in the progression of cervical cancer disease, as well as their ability as potential biomarkers.

Keywords: Cervical cancer; Metabolites; Prognosis; Trimethylamine N-oxide.

Fig. 1

Metabolic molecular landscape in cervical cancer and control group. PCA (A) and OPLS-DA (B) score plot of metabolomics data. The ellipses display 95% confidence intervals. C Heatmap of differentially expressed metabolites using scaling peak intensity. D SMPDB (Small Molecule Pathway Database) enrichment analysis of altered metabolites. E Correlation plot of differentially expressed metabolites

Fig. 2

Identified metabolic biomarkers distinguishing cervical cancer from control. A Metabolites selection via LASSO regression analysis. ln(λ) is plotted on the x-axis while binomial deviance is plotted on the y-axis. B LASSO coefficient profiles of the 51 metabolites against ln(λ). C ROC curve for the training cohort. The AUC were 0.993 (95% CI 0.9808–1). D ROC curve for the test cohort. The AUC was 1 (95% CI 1–1). E ROC curve for the validation cohort. The AUC was 1 (95% CI 0.9279–1). F–J Boxplots of peak intensities of 5 potential biomarkers including cyclohexylamine, L-carnitine, Val-Thr, sinigrin, 5,6,7,8-Tetrahydro-2-Naphthoic acid with healthy volunteers and CC patients. The Student's t-test was used to evaluate the significance of the difference between the two groups

Fig. 3

Differential metabolic landscape between I-IIA1 and IIA2-IV stages in cervical cancer. A–C GSEA analysis results of IIA2-IV compared to I-IIA1. D, E PCA score plot and heatmap of differentially expressed metabolites. F Correlation between differentially expressed metabolites and clinical indexes

Fig. 4

Identification of TMAO as a biomarker for different stages of cervical cancer patients. A Odds ratios with 95% CI and p-value, of TMAO and CA125 in predicting CC progression from I-IIA1 to IIA2-IV for all patients. B ROC curve of TMAO for the training cohort. The AUC was 0.872 (95% CI 0.7338–1). C ROC curve of TMAO for the test cohort. The AUC was 0.875 (95% CI 0.6578–1). D Boxplot of concentration of TMAO with I-IIA1 and IIA2-IV patients in derivation and validation cohorts. E Schematic diagram of sources of TMAO, showing increased TMAO and unchanged TMAO precursors in plasma of IIA2-IV patients compared to those of I-IIA1 patients

Fig. 5

TMAO functional analysis in vitro. A TMAO promoted HeLa cell proliferation in a dose-dependent manner. B HeLa cells treated with 400 μM TMAO were tested for LDH assay to assess TMAO cytotoxicity. C HeLa cells treated with 400 μM TMAO were tested for EdU assay. Representative immunofluorescence images of Hoechst, EdU, and the merged