The Performance of an Oral Microbiome Biomarker Panel in Predicting Oral Cavity and Oropharyngeal Cancers

Abstract

The oral microbiome can play a role in the instigation and progression of oral diseases that can manifest into other systemic conditions. These associations encourage the exploration of oral dysbiosis leading to the pathogenesis of cancers. In this study, oral rinse was used to characterize the oral microbiome fluctuation associated with oral cavity cancer (OCC) and oropharyngeal cancers (OPC). The study cohort consists of normal healthy controls (n = 10, between 20 and 30 years of age; n = 10, above 50 years of age), high-risk individuals (n = 11, above 50 years of age with bad oral hygiene and/or oral diseases) and OCC and OPC patients (n = 31, HPV-positive; n = 21, HPV-negative). Oral rinse samples were analyzed using 16S rRNA gene amplicon sequencing on the MiSeq platform. Kruskal-Wallis rank test was used to identify genera associated with OCC and OPC. A logistic regression analysis was carried out to determine the performance of these genera as a biomarker panel to predict OCC and OPC. In addition, a two-fold cross-validation with a bootstrap procedure was carried out in R to investigate how well the panel would perform in an emulated clinical scenario. Our data indicate that the oral microbiome is able to predict the presence of OCC and OPC with sensitivity and specificity of 100 and 90%, respectively. With further validation, the panel could potentially be implemented into clinical diagnostic and prognostic workflows for OCC and OPC.

Keywords: biomarker; oral cancer; oral microbiome; oral rinse; saliva.

Figure 1

(A) Rarefaction curve of observed operational taxonomic unites against sequences per sample for normal healthy controls (n = 20), high-risk individuals (n = 11), and oral cavity and oropharyngeal cancer patients (n = 52) as well as (B) the comparison of Shannon index for each category using rank test. Significant differences are denoted with *P < 0.05 and **P < 0.01, respectively.

Figure 2

(A) Bubble plot of 20 most abundant bacterial genera detected in oral rinse samples of normal healthy controls (between 20 and 30 years of age, n = 10 and above 50 years of age, n = 10), high-risk individuals (n = 11) and oral cavity and oropharyngeal cancer patients (n = 52). (B) Partial least squares regression-discriminant analysis (PLS-DA) and (C) redundancy analysis of microbial communities (genus-level) in oral rinse samples, based on OTU frequencies within respective categories.

Figure 3

Performance of the oral microbiome panel in predicting oral cavity and oropharyngeal cancers. Carstensen's multivariate receiver-operating characteristics curve based on the abundance of Rothia, Haemophilus, Corynebacterium, Paludibacter, Porphyromonas, Oribacterium, and Capnocytophaga; comparing normal healthy controls (above 50 years of age, n = 10) with oral cavity and oropharyngeal cancer patients (n = 52).