Abstract 3579: Racial disparity in microbial risk score for predicting subsequent risk of head and neck squamous cell cancer

Abstract Background: The association between oral bacteria and head and neck squamous cell cancer (HNSCC) risk has been investigated in recent years. Traditional approaches often explore individual microbes, overlooking the interaction and compositional profiles within microbial communities. We propose the use of a Microbial Risk Score (MRS) to examine the collective effects of bacterial communities in predicting HNSCC risk, with a focus on racial disparities. Method: We conducted a prospective nested case-control study within 3 large US cohorts: the ACS-Cancer Prevention Study-II Nutrition Cohort, the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, and the Southern Community Cohort Study. Among 159, 840 cohort participants who provided oral wash samples, we selected HNSCC cases and their matched controls (2:1 frequency matching on cohort, age, sex, race/ethnicity, and time since oral wash sample collection) during a mean of 5.1 years of follow-up. Our study population included 440 White participants (149 HNSCC cases and 291 controls) and 222 Black participants (76 HNSCC cases and 146 controls). We characterized the oral bacterial microbiome using whole-genome shotgun sequencing. We explored the probability of MRS to predict HNSCC risk across different racial groups and at different p-value(p) thresholds for bacterial community selection. The MRS was calculated as a weighted sum of relative abundance of the selected bacterial species, with weights assigned according to their effect size determined via Analysis of Compositions of Microbiomes with Bias Correction. The association between MRS and HNSCC risk was estimated using logistic regression after adjusting for age, sex, cohort, smoking, drinking, and oral human papillomavirus 16 status. Results: The MRS derived from Whites at p<0.01 yielded an OR in Whites of 1.38 (95% CI: 1.07-1.80) per 1 SD increase in the MRS, while this MRS yielded a weaker OR of 1.23 (0.86-1.75) in Blacks. The MRS derived from Blacks at p<0.01 yielded an OR of 1.54 (1.01-2.33) in Blacks, while it yielded a weaker OR of 1.19 (0.92-1.53) in Whites. Decreasing the stringency of bacterial community selection (e.g., to p<0.20) tended to yield overall stronger associations which were also more closely comparable for the counterpart racial groups; MRS derived from Whites at p<0.20 showed an OR of 1.74 (1.20-2.52) in Blacks, and the MRS derived from Blacks showed an OR of 1.39 (1.07-1.79) in Whites. These associations between MRS and HNSCC tended to hold for oral cavity, pharynx, and larynx, and were unrelated to tobacco and alcohol use. Conclusion: Our findings suggest that while the MRS tends to perform better within racial groups, race-specific MRS do associate with disease risk in another racial group, particularly under less stringent selection of bacteria that comprise the score. The MRS offers a promising approach for HNSCC risk prediction across diverse racial groups. Citation Format: Soyoung Kwak, Chan Wang, Mykhaylo Usyk, Feng Wu, Steve C. Moore, Wen-Yi Huang, Caroline Y. Um, Martha J. Shrubsole, Qiuyin Cai, Huilin Li, Jiyoung Ahn, Richard B. Hayes. Racial disparity in microbial risk score for predicting subsequent risk of head and neck squamous cell cancer. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3579.