Sensitive detection of copy number alterations in samples with low circulating tumor DNA fraction

Abstract

Copy number analysis is an important aspect of cancer genomics that enables identification of activated oncogenes, inactivated tumor suppressor genes and genome-wide signatures such as homologous recombination deficiency and the tandem duplication phenotype. Despite continuous development of copy number algorithms, the current sensitivity to detect clinically relevant focal alterations is poor if the cancer DNA fraction is low. This is particularly challenging for analysis of circulating tumor DNA (ctDNA) as it is not possible to know the cancer DNA fraction in advance or, as for tissue, macrodissect to increase the cancer DNA fraction. Here, we combine a novel algorithm (Jumble) with a tailored gene panel design and selected reference samples that achieve sensitive and highly specific detection of clinically relevant copy number alterations with limits of detection at 1-2% ctDNA fraction for amplifications and 4-8% for homozygous deletions. Jumble lowers the ctDNA fraction required for detection of homozygous deletions 3-6 times compared to commercial alternatives. Jumble is freely available as an R script and container, ready for integration into bioinformatic pipelines.