Fecal microbial biomarkers combined with multi-target stool DNA test improve diagnostic accuracy for colorectal cancer

Abstract

Background: Colorectal cancer (CRC) is a major global health burden. The current diagnostic tests have shortcomings of being invasive and low accuracy.

Aim: To explore the combination of intestinal microbiome composition and multi-target stool DNA (MT-sDNA) test in the diagnosis of CRC.

Methods: We assessed the performance of the MT-sDNA test based on a hospital clinical trial. The intestinal microbiota was tested using 16S rRNA gene sequencing. This case-control study enrolled 54 CRC patients and 51 healthy controls. We identified biomarkers of bacterial structure, analyzed the relationship between different tumor markers and the relative abundance of related flora components, and distinguished CRC patients from healthy subjects by the linear discriminant analysis effect size, redundancy analysis, and random forest analysis.

Results: MT-sDNA was associated with Bacteroides. MT-sDNA and carcinoembryonic antigen (CEA) were positively correlated with the existence of Parabacteroides, and alpha-fetoprotein (AFP) was positively associated with Faecalibacterium and Megamonas. In the random forest model, the existence of Streptococcus, Escherichia, Chitinophaga, Parasutterella, Lachnospira, and Romboutsia can distinguish CRC from health controls. The diagnostic accuracy of MT-sDNA combined with the six genera and CEA in the diagnosis of CRC was 97.1%, with a sensitivity and specificity of 98.1% and 92.3%, respectively.

Conclusion: There is a positive correlation of MT-sDNA, CEA, and AFP with intestinal microbiome. Eight biomarkers including six genera of gut microbiota, MT-sDNA, and CEA showed a prominent sensitivity and specificity for CRC prediction, which could be used as a non-invasive method for improving the diagnostic accuracy for this malignancy.

Keywords: Colorectal cancer; Diagnostic model; Gut microbiome; Multi-target stool DNA test; Tumor biomarker.

Figure 1

Flow chart of participant selection. CRC: Colorectal cancer; MT-sDNA: Multi-target stool DNA; FMDI: Fecal microbiota diagnostic index; CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein; CA199: Carbohydrate antigen 199; PCA: Principal component analysis; RDA: Redundancy analysis; LEfSe: Linear discriminant analysis effect size.

Figure 2

Differences in fecal microbiota between colorectal cancer patients and healthy controls. A: The clustering tree with red areas and green areas represents different groups. The red nodes in the branches represent microbial groups that play an important role in the red group, the green nodes represent microbial groups that play an important role in the green group, and the yellow nodes represent microbial groups that do not play an important role in either group. The genus names are shown in the legend on the right; B: Linear discriminant analysis (LDA) score histogram to identify diverse bacterial genus (LDA score ≥ 4, P < 0.05). CRC: Colorectal cancer.

Figure 3

Redundancy analysis in bacterial community characteristics (genus level) and tumor markers. Dots in the figure represent sample names; arrows indicate tumor biomarkers; inverted triangle represents species. CRC: Colorectal cancer; CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein; CA199: Carbohydrate antigen 199; PCA: Principal component analysis; RDA: Redundancy analysis.

Figure 4

The biomarker identification results by random forest model. The lattice plots show the identified biomarkers and their importance, with retention importance higher than 1.5.

Figure 5

Receiver operating characteristic curves of different detection methods to assess colorectal cancer. CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein; CA199: Carbohydrate antigen 199; PCA: Principal component analysis; AUC: Area under the receiver operating characteristic curve.