In patients with early-stage cancer, ctDNA detection rates can be low due to the presence of few or no copies of any individual mutation in each sample. Sensitivity may be increased by collecting larger plasma volumes, but this is not feasible in practice. Although cancers typically have thousands of mutations in their genome, previous analyses measured only individual or up to 32 tumour-specific mutations in plasma.

Here, we demonstrate that sensitivity can be greatly enhanced for any given input DNA mass by analysing a large number of mutations via sequencing. We sequenced in plasma 10^2-10^4 mutated loci per patient, using custom capture panels, whole exome (WES) or whole genome sequencing (WGS). We developed a method for INtegration of VAriant Reads (INVAR) that aggregates reads carrying tumour mutations across multiple mutant loci, and assigns confidence to error-suppressed reads based on mutation context, fragment length and tumour representation. This workflow combines a number of concepts in a novel approach in order to quantify ctDNA with maximal sensitivity.

We applied INVAR to plasma sequencing data from 45 patients with stage II-IV melanoma and 26 healthy individuals. ctDNA was detected to 1 mutant molecule per million, and tumour volumes of ~1cm^3. We show that this algorithm is applicable across targeted and untargeted sequencing methods. In patients with stage II-III melanoma who relapsed after resection, ctDNA was detected within 6 months post-surgery and prior to relapse in 50% of cases, compared to 16% in a similar cohort analysed with digital PCR. In addition, INVAR may enhance detection of ctDNA in samples with limited input or sequencing coverage by aggregating signal across a large number of mutations. Using low-depth WGS (0.6x), ctDNA was detected to 1 mutant per 10,000 molecules. Given that 60 genome copies of cfDNA may be obtained from 1 drop of blood (50-75μL), we suggest that INVAR may enable cancer monitoring from limited samples volumes.

As tumour sequencing becomes more widespread allowing identification of a large number of mutations per patient, this method has potential to quantify ctDNA with enhanced sensitivity, and to enable routine cancer monitoring using low-depth sequencing, potentially from low-volume blood samples that might be self-collected.