Abstract
Several studies have ascertained differences in salivary microbiota between patients with type 2 diabetes mellitus (T2DM) and healthy populations. However, the predictive accuracy and reproducibility of these 16S rRNA sequencing analyses when applied to other cohorts remain enigmatic. A comprehensive analysis was conducted on the included 470 samples from five researches in publicly available databases. The discrepancy and predictive accuracy of salivary microbiota between T2DM patients and healthy populations were evaluated from multiple perspectives, followed by the identification of salivary biomarkers for DM. Next, a classification model (areas under the curves = 0.92) was developed based on a large sample. The model could be used for clinical diagnosis and prognostic monitoring and as a basis for hypothesis-driven mechanistic researches. Furthermore, the research heterogeneity across geographic regions suggested that microbiological markers might not become a uniform clinical standard in human beings. They rather identify abnormal alterations under the microbiological characteristics of a specific population.