Leveraging sequence-based faecal microbial community survey data to identify a composite biomarker for colorectal...

Area of Science:

• Microbiome research
• Cancer diagnostics
• Bioinformatics

Background:

• Colorectal cancer (CRC) is a leading cause of cancer mortality in the USA.
• The fecal microbiome may offer non-invasive biomarkers for CRC detection.
• Identifying reliable microbial markers is crucial for early diagnosis and understanding cancer progression.

Purpose of the Study:

• To identify a composite and generalizable microbial marker for colorectal cancer (CRC).
• To re-analyze existing raw sequence and metadata from multiple studies uniformly.
• To assess the diagnostic capabilities of microbial biomarkers for CRC.

Main Methods:

• Processed raw 16S rRNA gene sequence data from nine studies using two pipelines: QIIME closed reference (QIIME-CR) and a strain-specific method (SS-UP).
• Analyzed a total of 509 samples, including colorectal adenoma, CRC, and control groups.
• Employed differential abundance, meta-analysis random effects regression, and machine learning for biomarker identification and classification.

Main Results:

• Identified specific taxa, including *Parvimonas micra* and *Streptococcus anginosus*, significantly increased in CRC patients' stools.
• Microbiome-based CRC classification achieved an AUROC of 76.6% (QIIME-CR) and 80.3% (SS-UP).
• Combined microbiome and clinical markers yielded higher diagnostic AUROC: 83.3% (QIIME-CR) and 91.3% (SS-UP).

Conclusions:

• Key microbial markers can be important for classifying CRC cases, potentially forming a universal diagnostic tool.
• The choice of bioinformatics pipeline significantly influences the accuracy of CRC classification.
• Strain-resolved microbial markers show promise for developing a precise microbial diagnostic for colorectal cancer.