377.5 - Microbial multidimensional signature assessment reveals microbial SNVs as the superior non-invasive biomarkers for early detection of colorectal cancer

Wenxing Gao (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Ruicong Sun (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Lixin Zhu (Guangdong Institute of Gastroenterology, Guangdong Provincial Key Laboratory of Colorectal and Pelvic Floor Diseases, the Sixth Affiliated Hospital, Sun Yat-sen University), Sheng Gao (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Zhongsheng Feng (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Dingfeng Wu (National Clinical Research Center for Child Health, the Children’s Hospital, Zhejiang University School of Medicine), Xiang Gao (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Rui Zhao (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Ye Yao (Center for Life Science Education, College of Art and Sciences, The Ohio State University, Columbus, OHIO, 43210, USA), Zhanjiu Liu (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Ruixin Zhu (Center for IBD Research, The Shanghai Tenth People’s Hospital, Department of Bioinformatics, School of Life Sciences and Technology, Tongji University), Na Jiao (National Clinical Research Center for Child Health, the Children’s Hospital, Zhejiang University School of Medicine)

Background:  The early detection of colorectal cancer (CRC), especially at precancerous adenoma stage, significantly reduces its incidence. Gut microbiome has become a promising non-invasive tool for CRC screening, whereas the potential of microbial multidimensional signatures remains poorly understood. Here we performed a cross-cohort analysis to evaluate the capability of microbial multidimensional biomarkers for adenoma detection. 

Design:  Whole metagenome sequencing data of four public datasets including 183 adenoma patients and 439 healthy controls were reprocessed consistently to obtain taxonomic-, functional- and single-nucleotide variants (SNVs)-profiling. With MMUPHin, differential multidimensional signatures were identified after adjusting confounders, based on which random forest (RF) models were constructed and then optimized by recursive feature elimination. Finally, internal validation and external validation with in-house dataset (10 controls and 6 adenomas) and five resampled datasets were conducted to further assess the robustness of the best biomarker panel. 

Results:  The integrated analysis identified 103 multi-kingdom differential species between adenoma and control group, of which 15 optimal species were selected to construct a RF model achieving an AUC of 0.75. Meanwhile, the model constructed with 31 optimal biomarkers out of 386 differential KO genes reached an AUC of 0.74. Notably, the diagnostic model with 75 SNVs from 10 species showed superior accuracy (AUC = 0.85) with high specificity to adenoma. Co-abundance analysis revealed intensive bacterial-fungal associations in line with functional abnormalities related to microbial quorum sensing, purine and butanoate metabolism. 

Conclusion:  Microbial SNV biomarkers outperform other biomarkers and display high specificity to adenoma, which may serve as a novel non-invasive tool for early detection of CRC. Furthermore, multidimensional signatures provide potential therapeutic targets for adenoma. 

Keywords:

colorectal adenoma, microbial multidimensional biomarkers, early detection, non-invasive

National Natural Science Foundation of China grant numbers 81774152, 82170542 to RXZ, 81770571 to LXZ, 82000536 to NJ,81630017,91942312 to ZL.