A machine-learning classifier trained with microRNA ratios to distinguish melanomas from nevi

Abstract

The use of microRNAs as biomarkers has been proposed for many diseases including the diagnosis of melanoma. Although hundreds of microRNAs have been identified as differentially expressed in melanomas as compared to benign melanocytic lesions, limited consensus has been achieved across studies, constraining the effective use of these potentially useful markers. In this study we quantified microRNAs by next-generation sequencing from melanomas and their adjacent benign precursor nevi. We applied a machine learning-based pipeline to identify a microRNA signature that separated melanomas from nevi and was unaffected by confounding variables, such as patient age and tumor cell content. By employing the ratios of microRNAs that were either enriched or depleted in melanoma compared to nevi as a normalization strategy, the classifier performed similarly across multiple published microRNA datasets, obtained by microarray, small RNA sequencing, or RT-qPCR. Validation on separate cohorts of melanomas and nevi correctly classified lesions with 83% sensitivity and 71-83% specificity, independent of variation in tumor cell content of the sample or patient age.