Abstract:Background and objective: The association between human intestinal microbes and disease is an emerging topic in microbiome research, with potential implications in cancer diagnosis. Meanwhile, immune checkpoint inhibitor (ICI) therapy has been extensively applied to cancer patients but with relatively low overall response rate owing to pathological complexity. Thus, accurate prediction of cancer onset as well as patients' response to ICI treatment through non-invasive methods, such as gut microbiome profiling, could greatly facilitate cancer diagnosis and treatment practice, which essentially improves patients' quality of life. With the rapid accumulation of microbiome sequencing data and advanced machine learning algorithms, the use of gut microbiome as predictive biomarkers presents a feasible approach to address these problems. Methods and results: The first model we built is a predictive model for classifying healthy people from patients with colon cancer (CRC). The data (originated from Curated Metagenomics Data of Human Microbiome) consists of 166 fecal samples from healthy population and equal number of CRC patients from two studies (PRJEB24748, PRJEB6070). The data were divided into training and testing sets with a 3:1 ratio. Using the training dataset, we developed a Gradient Boosting Decision Tree model. To demonstrate the model's generalizability, we evaluated it using the testing dataset and the model achieved an AUC score of 0.93 (accuracy: 89%). Moreover, using an external independent cohort (which included 52 CRC patients from PRJEB12449 and 52 healthy individuals), we obtained an AUC score of 0.94 (accuracy: 88%). We also detected bacteria highly related to CRC pathology including Solobacterium_moorei, which could be potential biomarkers for CRC diagnosis. In addition, to predict CRC patients' response to ICI, we built a random forest model based on patients' fecal microbiome data before the ICI treatment. In total, stool metagenomics data from 63 CRC patients were used, including 30 patients responding to ICI treatment and 33 non-responders. With five-fold cross-validation, we achieved training and testing AUC score of 0.89 and 0.72, respectively. With an external data with 11 CRC patients' samples, the classifier achieved an AUC of 0.78. Several microbial taxa were listed as important features contributing to the classification, e.g. Eubacterium hallii – a butyrate producer, which previously report to have a positive association with ICI clinical benefits. Conclusion: Using the non-invasive human gut microbiome data and advanced machine learning techniques, we present here two models with great performance to predict the occurrence of CRC and clinical response to ICI treatment in CRC patients. Similar strategy could be readily transferred to other cancer types to improve early diagnosis and patient stratification. Moreover, the microbial taxa identified to be essential features in the model could be developed as potential therapeutic targets. Citation Format: Die Dai, Xiaomin Xu, Wei Shen, Zhengnong Zhu, Fang Li, Xiaochen Yin, Yan Kou. Gut microbiome as a promising biomarker for colorectal cancer diagnosis and immunotherapy response prediction [abstract]. In: Proceedings of the AACR Special Conference on Colorectal Cancer; 2022 Oct 1-4; Portland, OR. Philadelphia (PA): AACR; Cancer Res 2022;82(23 Suppl_1) nr A030.

Leveraging Scheme for Cross-Study Microbiome Machine Learning Prediction and Feature Evaluations

Integrating clinical and cross-cohort metagenomic features: a stable and non-invasive colorectal cancer and adenoma diagnostic model

Systematic Comparisons for Composition Profiles, Taxonomic Levels, and Machine Learning Methods for Microbiome-Based Disease Prediction

Leveraging pre-trained language models for mining microbiome-disease relationships

Performance of Gut Microbiome as an Independent Diagnostic Tool for 20 Diseases: Cross-Cohort Validation of Machine-Learning Classifiers

Human disease prediction from microbiome data by multiple feature fusion and deep learning

Effects of Data Transformation and Model Selection on Feature Importance in Microbiome Classification Data

Machine learning approaches in microbiome research: challenges and best practices

Improve the Colorectal Cancer Diagnosis Using Gut Microbiome Data

Abstract A030: Gut microbiome as a promising biomarker for colorectal cancer diagnosis and immunotherapy response prediction

Faecal microbiome-based machine learning for multi-class disease diagnosis

Large-scale microbiome data integration enables robust biomarker identification

Using Machine Learning Methods to Study Colorectal Cancer Tumor Micro-Environment and Its Biomarkers.

Machine learning for data integration in human gut microbiome

Longitudinal Microbiome-based Interpretable Machine Learning for Identification of Time-Varying Biomarkers in Early Prediction of Disease Outcomes

Comprehensive assessment of machine learning methods for diagnosing gastrointestinal diseases through whole metagenome sequencing data

DeepMicro: Deep Representation Learning for Disease Prediction Based on Microbiome Data

Integration of multiview microbiome data for deciphering microbiome-metabolome-disease pathways

Predicting the Associations Between Microbes and Diseases by Integrating Multiple Data Sources and Path-Based HeteSim Scores

P070 Machine learning approaches to identify IBD biomarkers from longitudinal microbiome data

Establishment and evaluation of prediction model for multiple disease classification based on gut microbial data