Machine Learning-Driven Detection of Colorectal Neoplasia from 16S rRNA Gut Microbiome Profiles: A Compositional Approach
Main Article Content
Abstract
The early detection of colorectal cancer (CRC) and advanced adenomas through non-invasive methods remains limited due to the lack of sensitivity of traditional fecal-based techniques. In order to solve this challenge, 16S rRNA gene amplicon sequencing data derived from 59 stool samples were analyzed using optimized machine learning classifiers using compositional data analysis techniques. CLR transformations were applied on 208 genera to overcome issues of compositionality and sparseness. With respect to 5-fold stratified cross-validation, the Random Forest algorithm proved to be superior, exhibiting an average AUC of 0.81 ± 0.15 in binary CRC classification as well as an average accuracy of 0.67 ± 0.08 for multi-class classification. The predictive models suggested an ecological dichotomy, where the oral pathogens, Fusobacterium and Parvimonas, were found significantly abundant in CRC patients compared to healthy controls, while the protective and beneficial bacteria producing butyrate, Roseburia and Faecalibacterium, showed progressive reduction from adenoma to carcinoma. The stability of feature selection was determined using a Jaccard index of more than 0.80. This suggests the development of a reliable computational framework for improving diagnostic accuracy. Nevertheless, the clinical relevance of these results awaits confirmation in large-scale studies.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
References
[1] H. Sung et al., “Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA. Cancer J. Clin., vol. 71, no. 3, pp. 209–249, May 2021, doi: 10.3322/caac.21660.
[2] D. J. Robertson et al., “Recommendations on Fecal Immunochemical Testing to Screen for Colorectal Neoplasia: A Consensus Statement by the US Multi-Society Task Force on Colorectal Cancer,” Gastroenterology, vol. 152, no. 5, pp. 1217-1237.e3, Apr. 2017, doi: 10.1053/j.gastro.2016.08.053.
[3] T. F. Imperiale et al., “Multitarget Stool DNA Testing for Colorectal-Cancer Screening,” N. Engl. J. Med., vol. 370, no. 14, pp. 1287–1297, Apr. 2014, doi: 10.1056/NEJMoa1311194.
[4] H. Tjalsma, A. Boleij, J. R. Marchesi, and B. E. Dutilh, “A bacterial driver–passenger model for colorectal cancer: beyond the usual suspects,” Nat. Rev. Microbiol., vol. 10, no. 8, pp. 575–582, Aug. 2012, doi: 10.1038/nrmicro2819.
[5] W. S. Garrett, “The gut microbiota and colon cancer,” Science (80-. )., vol. 364, no. 6446, pp. 1133–1135, Jun. 2019, doi: 10.1126/science.aaw2367.
[6] J. Yu et al., “Metagenomic analysis of faecal microbiome as a tool towards targeted non-invasive biomarkers for colorectal cancer,” Gut, vol. 66, no. 1, pp. 70–78, Jan. 2017, doi: 10.1136/gutjnl-2015-309800.
[7] J. Wirbel et al., “Meta-analysis of fecal metagenomes reveals global microbial signatures that are specific for colorectal cancer,” Nat. Med., vol. 25, no. 4, pp. 679–689, Apr. 2019, doi: 10.1038/s41591-019-0406-6.
[8] A. M. Thomas et al., “Metagenomic analysis of colorectal cancer datasets identifies cross-cohort microbial diagnostic signatures and a link with choline degradation,” Nat. Med., vol. 25, no. 4, pp. 667–678, Apr. 2019, doi: 10.1038/s41591-019-0405-7.
[9] S. Yachida et al., “Metagenomic and metabolomic analyses reveal distinct stage-specific phenotypes of the gut microbiota in colorectal cancer,” Nat. Med., vol. 25, no. 6, pp. 968–976, Jun. 2019, doi: 10.1038/s41591-019-0458-7.
[10] M. R. Rubinstein, X. Wang, W. Liu, Y. Hao, G. Cai, and Y. W. Han, “Fusobacterium nucleatum Promotes Colorectal Carcinogenesis by Modulating E-Cadherin/β-Catenin Signaling via its FadA Adhesin,” Cell Host Microbe, vol. 14, no. 2, pp. 195–206, Aug. 2013, doi: 10.1016/j.chom.2013.07.012.
[11] C. Gur et al., “Binding of the Fap2 Protein of Fusobacterium nucleatum to Human Inhibitory Receptor TIGIT Protects Tumors from Immune Cell Attack,” Immunity, vol. 42, no. 2, pp. 344–355, Feb. 2015, doi: 10.1016/j.immuni.2015.01.010.
[12] D. R. Donohoe et al., “The Microbiome and Butyrate Regulate Energy Metabolism and Autophagy in the Mammalian Colon,” Cell Metab., vol. 13, no. 5, pp. 517–526, May 2011, doi: 10.1016/j.cmet.2011.02.018.
[13] P. Louis, G. L. Hold, and H. J. Flint, “The gut microbiota, bacterial metabolites and colorectal cancer,” Nat. Rev. Microbiol., vol. 12, no. 10, pp. 661–672, Oct. 2014, doi: 10.1038/nrmicro3344.
[14] G. B. Gloor, J. M. Macklaim, V. Pawlowsky-Glahn, and J. J. Egozcue, “Microbiome Datasets Are Compositional: And This Is Not Optional,” Front. Microbiol., vol. 8, Nov. 2017, doi: 10.3389/fmicb.2017.02224.
[15] M. C. B. Tsilimigras and A. A. Fodor, “Compositional data analysis of the microbiome: fundamentals, tools, and challenges,” Ann. Epidemiol., vol. 26, no. 5, pp. 330–335, May 2016, doi: 10.1016/j.annepidem.2016.03.002.
[16] J. Aitchison, “The Statistical Analysis of Compositional Data,” J. R. Stat. Soc. Ser. B, vol. 44, no. 2, pp. 139–177, Jun. 1982, [Online]. Available: http://www.jstor.org/stable/2345821
[17] T. P. Quinn, I. Erb, M. F. Richardson, and T. M. Crowley, “Understanding sequencing data as compositions: an outlook and review,” Bioinformatics, vol. 34, no. 16, pp. 2870–2878, Aug. 2018, doi: 10.1093/bioinformatics/bty175.
[18] B. D. Topçuoğlu, N. A. Lesniak, M. T. Ruffin, J. Wiens, and P. D. Schloss, “A Framework for Effective Application of Machine Learning to Microbiome-Based Classification Problems,” MBio, vol. 11, no. 3, Jun. 2020, doi: 10.1128/mBio.00434-20.
[19] A.-L. Boulesteix and C. Strobl, “Optimal classifier selection and negative bias in error rate estimation: an empirical study on high-dimensional prediction,” BMC Med. Res. Methodol., vol. 9, no. 1, p. 85, Dec. 2009, doi: 10.1186/1471-2288-9-85.
[20] J. A. Martín-Fernández, K. Hron, M. Templ, P. Filzmoser, and J. Palarea-Albaladejo, “Model-based replacement of rounded zeros in compositional data: Classical and robust approaches,” Comput. Stat. Data Anal., vol. 56, no. 9, pp. 2688–2704, Sep. 2012, doi: 10.1016/j.csda.2012.02.012.
[21] A. D. Fernandes, J. N. Reid, J. M. Macklaim, T. A. McMurrough, D. R. Edgell, and G. B. Gloor, “Unifying the analysis of high-throughput sequencing datasets: characterizing RNA-seq, 16S rRNA gene sequencing and selective growth experiments by compositional data analysis,” Microbiome, vol. 2, no. 1, p. 15, Dec. 2014, doi: 10.1186/2049-2618-2-15.
[22] H. Zou and T. Hastie, “Regularization and Variable Selection Via the Elastic Net,” J. R. Stat. Soc. Ser. B Stat. Methodol., vol. 67, no. 2, pp. 301–320, Apr. 2005, doi: 10.1111/j.1467-9868.2005.00503.x.
[23] L. Breiman, “Random Forests,” Mach. Learn., vol. 45, no. 1, pp. 5–32, Oct. 2001, doi: 10.1023/A:1010933404324.
[24] T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning. in Springer Series in Statistics. New York, NY: Springer New York, 2009. doi: 10.1007/978-0-387-84858-7.
[25] N. T. Baxter, M. T. Ruffin, M. A. M. Rogers, and P. D. Schloss, “Microbiota-based model improves the sensitivity of fecal immunochemical test for detecting colonic lesions,” Genome Med., vol. 8, no. 1, p. 37, Dec. 2016, doi: 10.1186/s13073-016-0290-3.
[26] G. S. Collins, J. B. Reitsma, D. G. Altman, and K. G. M. Moons, “Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement,” BMJ, vol. 350, no. jan07 4, pp. g7594–g7594, Jan. 2015, doi: 10.1136/bmj.g7594.
[27] B. Flemer et al., “The oral microbiota in colorectal cancer is distinctive and predictive,” Gut, vol. 67, no. 8, pp. 1454–1463, Aug. 2018, doi: 10.1136/gutjnl-2017-314814.
[28] B. J. Callahan, P. J. McMurdie, M. J. Rosen, A. W. Han, A. J. A. Johnson, and S. P. Holmes, “DADA2: High-resolution sample inference from Illumina amplicon data,” Nat. Methods, vol. 13, no. 7, pp. 581–583, Jul. 2016, doi: 10.1038/nmeth.3869.
[29] C. Quast et al., “The SILVA ribosomal RNA gene database project: improved data processing and web-based tools,” Nucleic Acids Res., vol. 41, no. D1, pp. D590–D596, Nov. 2012, doi: 10.1093/nar/gks1219.
[30] F. Pedregosa et al., “Scikit-learn: Machine Learning in Python,” J. Mach. Learn. Res., vol. 12, no. null, pp. 2825–2830, Nov. 2011.
[31] H. Lin and S. Das Peddada, “Analysis of compositions of microbiomes with bias correction,” Nat. Commun., vol. 11, no. 1, p. 3514, Jul. 2020, doi: 10.1038/s41467-020-17041-7.
[32] G. M. Douglas et al., “PICRUSt2 for prediction of metagenome functions,” Nat. Biotechnol., vol. 38, no. 6, pp. 685–688, Jun. 2020, doi: 10.1038/s41587-020-0548-6.
[33] Y. Yang et al., “Prospective study of oral microbiome and colorectal cancer risk in low‐income and African American populations,” Int. J. Cancer, vol. 144, no. 10, pp. 2381–2389, May 2019, doi: 10.1002/ijc.31941.
[34] Y. Komiya et al., “Patients with colorectal cancer have identical strains of Fusobacterium nucleatum in their colorectal cancer and oral cavity,” Gut, vol. 68, no. 7, pp. 1335–1337, Jul. 2019, doi: 10.1136/gutjnl-2018-316661.
[35] A. Hiergeist, U. Reischl, and A. Gessner, “Multicenter quality assessment of 16S ribosomal DNA-sequencing for microbiome analyses reveals high inter-center variability,” Int. J. Med. Microbiol., vol. 306, no. 5, pp. 334–342, Aug. 2016, doi: 10.1016/j.ijmm.2016.03.005.
[36] E. D. Sonnenburg and J. L. Sonnenburg, “Starving our microbial self: the deleterious consequences of a diet deficient in microbiota-accessible carbohydrates.,” Cell Metab., vol. 20, no. 5, pp. 779–786, Nov. 2014, doi: 10.1016/j.cmet.2014.07.003.
[37] G. Zeller et al., “Potential of fecal microbiota for early‐stage detection of colorectal cancer,” Mol. Syst. Biol., vol. 10, no. 11, p. MSB145645, Nov. 2014, doi: 10.15252/msb.20145645.