LAUSR

Dear Editor, Most cancers have altered metabolism with increased uptake of glucose (i.e. the “Warburg effect”) attributed to defective mitochondria. In addition, mitochondria are associated with multiple key processes linked to tumourigenesis including apoptosis, cell cycle, cell growth, and signalling. Multiple myeloma (MM) is essentially an incurable haematological malignancy, with most patients developing resistance to treatment and eventually dying from relapse. Recent studies have proposed mitochondria dysfunction is important in defining chemotherapy resistance and disease progression in MM. Such an assertion is supported by pre-clinical studies, which have suggested agents targeting mitochondria in relapsed MM can improve patient outcome. Thus far, the spectrum of mitochondrial DNA (mtDNA) mutations and their functional implications in MM have not however been well characterised, partly due to limited sample size and whole-exome sequencing depth. Furthermore, the paucity of MM representation in pan-cancer analyses has not allowed an appraisal of MMspecific mitochondrial mutations. By analysing wholegenome sequencing (WGS) data from the Myeloma XI trial, we have sought to address these shortcomings, characterising the somatic mutation landscape, mutation selection at relapse, nuclear genome integration, and copy number of MM mitochondria. To investigate mtDNA mutations in MM, we analysed WGS data on 80 matched tumour and normal samples from newly diagnosed patients, of which 25 also had matched relapsed tumours. Owing to high cellular copy number of mtDNA genomes, we obtained far greater mtDNA genome coverage (normals: median 2149×, range 1015×–7777×; primary tumours: median 7836×, range 2376×–7938×; relapsed tumours: median 7826×, range 4678×–7929×) compared to the nuclear genome (Supplementary Table 1). We identified 210 mtDNA single nucleotide variants (SNVs) in the 80 primary tumours (median 3 SNVs/ tumour). These showed strong replicative strand bias, predominantly C>T on heavy strand and T>C on light strand (Supplementary Fig. 1), which has previously been ascribed to replication-coupled process partly due to the lack of transcriptional strand bias. Examining the sequence context of mutations revealed the contribution of defective transcription-coupled DNA repair COSMIC signatures 12 (16%), 21 (15%), 23 (11%), and 26 (48%) (Fig. 1a). We observed transcriptional strand bias across all genes (Fig. 1b), with the strongest signal for C>T, where transcribed strand are more frequently repaired. The weaker transcriptional strand bias for T>C is likely due to the neutralising effects from COSMIC signatures with opposing transcriptional strand biases (Supplementary Fig. 2). To validate these observations, we repeated the analysis of mtDNA mutational spectra using WGS data from 850 newly diagnosed MM generated by The Relating Clinical Outcomes in Multiple Myeloma to Personal Assessment of Genetic Profile Study (CoMMpass; tumour and normal sample median read depth of 869× and 661×, respectively). The mutational spectra and strand biases observed in the Myeloma XI samples were also apparent in CoMMpass (Supplementary Fig. 3). Transcriptional strand biases in the CoMMpass samples persist when considering the 22 tRNA genes (14 light strand and 8 heavy strand) separately (Supplementary Fig. 3d). Collectively, these findings are consistent with the contribution of transcription-coupled DNA repair defects in MM mtDNA.