Tumour mutation burden (TMB) is a biomarker for cancer immune checkpoint blockade (ICB) response, and mutation signatures provide a “life history” of tumour evolution. Researchers have repurposed cancer gene panels (CGPs) to cheaply estimate these parameters. The low coverage of CGPs, however, may be problematic in the exploration of TMB as a biomarker for ICB. Restriction enzyme associated DNA sequencing (RADseq) uses restriction enzymes to achieve coverage at random loci throughout the genome. It was hypothesised that RADseq could recapitulate mutation signatures and estimate TMB in breast cancer samples.
In silico RADseq libraries were generated by scanning the human genome for restriction enzyme cut sites and filtering for regions with adjacent cut sites. The resulting fragments were overlapped with mutations from whole genome sequencing (WGS) of breast cancer genomes. RADseq libraries were evaluated in terms of the accuracy of TMB estimation and mutation profile recapitulation, and compared to a CGP used for TMB estimation in terms of these parameters. A mouse breast cancer cell line was transfected with APOBEC enzymes, which are known to cause mutations in breast cancer, and profiled using RADseq.
Using cosine similarity (CS) to the WGS profile as a measure of mutation profile recapitulation quality, RADseq libraries were able to recapitulate WGS mutation signatures, while the CGP performed relatively poorly in this regard (e.g. enzyme_1 CS=0.90±0.06 vs CGP CS=0.28±0.19). The CGP had higher TMB estimation error than an enzyme library of comparable coverage (CGP range 0.0034-8.83 mutations/mb vs enzyme_2 range 0.0029-7.50, Wilcox p<0.001). Using RADseq, an increased TMB and the APOBEC mutation signature was detected in a mouse cell line transfected with APOBEC3A, both of which were absent in control cell lines.
CGP methods do not accurately capture “genome-wide” mutation parameters such as TMB and mutation signatures. This warrants investigation into whether these methods can be used in TMB estimation in ICB trials. By comparison, RADseq may offer a cheap, effective solution for TMB estimation. RADseq may also be of use to researchers seeking to study the evolution of mutation signatures in model systems, as we have demonstrated.
The authors.
Breast Cancer Foundation New Zealand.
All authors have declared no conflicts of interest.