Abstract 5908: Specific gut microbial community is associated with specific cancer types: A strategy for cancer detection and prevention

Abstract

Abstract Background: Cancer is the second leading cause of death in the world. Unfortunately, survival rates for cancer patients have stayed poor for many decades. Thus, there is an unmet need to identify novel strategies for early cancer detection and prevention. Recent studies suggest that the gut microbiome including bacteria may have a role in cancer initiation and progression. These bacteria play a vital role in maintaining homeostasis in the body. An imbalance in the bacterial composition may cause diseases including cancer. This evidence led us to hypothesize that the changes in the composition of the gut microbiome can be used as a biomarker for early cancer detection. To test this hypothesis, we focused on analyzing the microbial community of colorectal and blood cancers, which have poor survival rates. Methods and Results: To analyze the association of cancer with the microbiome, we used a published microbiome dataset of leukemic and healthy individuals that were collected by sequencing stool samples (Oliver et al., 2022). We identified that specific bacteria are either abundant, such as Faecalibacterium (1.17 fold), or reduced, such as Clostridium (0.32 fold, p value=0.0029), in leukemia patients compared to healthy controls. We observed similar correlations by analyzing the published colorectal cancer data (Flemer et al., 2017) in which Prevotella increased by 3.75 fold and Escherichia/Shigella increased by 4.0 fold in colorectal cancer while Blautia is reduced compared to healthy controls. Faecalibacterium, which is abundant in leukemia, is reduced in colorectal cancer. Further, we found that treatment of leukemia patients normalized the abundance of these bacteria to the level of healthy patients. To improve analysis efficiency, we applied machine learning algorithms to predict the presence of cancer in patients based on the composition of bacteria, age, gender, and symptoms. Among the tested algorithms, the decision tree and Random Forest both had very high false-negative rates, with 45.5% and 40% respectively, as well as having low sensitivities, with 50% and 60% respectively. The support Vector Machine had a suboptimal false-negative rate of 28.6% and good sensitivity of 80%. Finally, the Neural network had a good false-negative rate of 16.7% and great sensitivity of 90%. Overall, our results suggest that different cancers have an abundance of specific bacteria that can be used as biomarkers for cancer detection. Conclusion: We demonstrate that the increased abundance of certain bacteria in the microbiome can be used as a biomarker for cancer detection. Our data set was small for this project, but the machine learning approaches can be used to further predict cancer These are clinically relevant findings as a microbiome test would be non-invasive and easily accessible that can be performed at a routine checkup. Citation Format: Ekansh Mittal, Andrew Oliver, Kenza El Alaoui, Carolyn Haunschild, Julio Avelar-Barragan, Laura F. Mendez Luque, Katrine Whiteson, Angela G. Fleischman. Specific gut microbial community is associated with specific cancer types: A strategy for cancer detection and prevention [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5908.