Signatures of ionising radiation in second malignancies
1Wellcome Trust Sanger Institute, Hinxton, UK,2Harvard Medical School, Boston, UK,3University of Tampere and Fimlab Laboratories, Tampere, Finland,4University College London, London, UK
Buried within the genomes of radiation-associated second malignancies are the somatic mutations generated by ionising radiation. It is conceivable that some of these are discernible as distinct mutation signatures, which we aimed to define.
We performed whole genome sequencing of 4 different types of radiation-associated second malignancies: 4 osteosarcomas; 3 breast cancers; 2 spindle cell sarcomas; 3 angiosarcomas. We defined all classes of somatic mutations using the analysis pipeline of the Cancer Genome Project, supplemented by bespoke pattern and statistical analyses. We compared our findings to 319 radiation-naïve tumours processed by the same analysis pipeline. We extended our findings to metastatic prostate tumours from 10 patients, half of whom received ionising radiation treatment prior to metastases formation.
Overall the tumour genomes were diverse and displayed somatic changes specific to each tumour type. Against this backdrop of genomic diversity, we identified two mutation signatures that characterised radiation-associated second malignancies, irrespective of tumour type. We found a significant enrichment of deletions in radiation-associated malignancies (p=1.93 x 10-15, linear mixed effects model). Compared to radiation-naïve deletions, these were of larger size and distributed more evenly across the genome. Further, radiation-associated genomes harboured a significant excess of balanced inversions, a rare type of rearrangement (p = 2×10-16, generalised linear model). We were able to extend these findings into metastatic prostate tumours, comparing radiation-treated with radiation-naïve patients.
In tumours associated with ionising radiation we identified a highly significant enrichment of two mutation signatures, an excess of deletions and of balanced inversions. As these were present across tumour types, it seems likely that they represent genomic signatures of ionising radiation. A future challenge will be to explore the utility of our findings as genomic markers of radiation exposure, for example, as diagnostic adjuncts or in determining the burden of radiation-associated tumours at population level.