Robust methylation-based classification of brain tumors using nanopore sequencing

Abstract

DNA methylation-based classification of cancer provides a comprehensive molecular approach to diagnose tumors. In fact, DNA methylation profiling of human brain tumors already profoundly impacts clinical neuro-oncology. However, current implementations using hybridization microarrays are time-consuming and costly. We recently reported on shallow nanopore whole-genome sequencing for rapid and cost-effective generation of genome-wide 5-methylcytosine profiles as input to supervised classification using random forests complemented by a medium-resolution copy number profile derived from the same raw data. Here, we demonstrate that this approach allows to discriminate a wide spectrum of primary brain tumors using public reference data of 82 distinct tumor entities. We developed a pseudo-probability score as a confidence score for interpretation in a clinical context. Using bootstrap sampling in a discovery cohort of N = 56 cases, we find that a minimum set of 1,000 random CpG features is sufficient for high-confidence classification by ad hoc random forests for most cases and demonstrate robustness across laboratories with matching results in 13/13 cases. When applying the confidence score threshold to an independent validation series (N = 111), the method demonstrated 100% specificity for the remaining 93 cases. In a prospective benchmarking (N = 15), median time to results was 21.1 hours. In conclusion, nanopore sequencing allows robust and rapid methylation-based classification across the full spectrum of brain tumors. The integrated confidence score facilitates possible clinical implementation, while requiring further prospective evaluation.