Salivary microRNAs identified by small RNA sequencing and machine learning as potential biomarkers of alcohol dependence.

Aim: Salivary miRNA can be easily accessible biomarkers of alcohol dependence (AD). Materials & methods: The miRNA transcriptome in the saliva of 56 African-Americans (AAs; 28 AD patients/28 controls) and 64 European-Americans (EAs; 32 AD patients/32 controls) was profiled using small RNA sequencing. Differentially expressed miRNAs were identified. Salivary miRNAs were used to predict the AD presence using machine learning with Random Forests. Results: Seven miRNAs were differentially expressed in AA AD patients, and five miRNAs were differentially expressed in EA AD patients. The AD prediction accuracy based on top five miRNAs (ranked by Gini index) was 79.1 and 72.2% in AAs and EAs, respectively. Conclusion: This study provided the first evidence that salivary miRNAs are AD biomarkers.