NEW YORK (GenomeWeb) – Researchers from the Wellcome Trust Sanger Institute have sequenced the whole genomes of tumor and normal tissue from 560 breast cancer patients and have identified 93 driver genes, as well as mutational signatures correlated with deficient DNA repair and the function of BRCA1 and BRCA2.
The research, which was described today in two studies in Nature and Nature Communications, is the most comprehensive analysis of breast cancer genomes and sheds light on the mechanisms driving breast cancer and the genes responsible for its genesis.
In the Nature paper, the researchers — led by Sanger Institute Director Michael Stratton and Cancer Genome Project Group Leader Serena Nik-Zainal — sequenced tumor/normal pairs from 556 females and four males with breast cancer, and also obtained transcriptome sequence, microRNA expression, array-based copy number, and methylation data from a subset of cases.
"This huge study, the largest of any one cancer type to date, gives insights into which genetic variations exist, and where they are in the genome," Stratton said in a statement. "Unpicking the genetic variations between cancers is crucial to developing improved therapies."
Sequencing uncovered 3,479,652 somatic base substitutions, 371,993 small indels, and 77,695 rearrangements. In order to identify new driver genes, the researchers used the whole-genome sequence data they generated along with previously published sequence data from an additional 772 breast cancer genomes, and searched for genes that exhibited a higher than expected-by-chance cluster of mutations. That turned up five previously unidentified driver genes.
When they combined data from base substitutions, indels, and rearrangements that affected genes, the researchers identified 1,628 likely driver mutations in 93 genes. Ninety five percent of all genomes had at least one of those driver genes. The 10 most frequently mutated genes were TP53, PIK3CA, MYC, CCND1, PTEN, ERBB2, the ZNF703/FGFR1 locus, GATA3, RB1, and MAP3K1, accounting for 62 percent of drivers.
Interestingly, although the researchers identified a number of in-frame fusion genes that could potentially activate driver genes, they found recurrent gene fusions were rarely expressed. Rather, they found that rearrangements that interrupted the "gene footprints" of a handful of genes — CDKN2A, RB1, MAP3K1, PTEN, MAP2K4, ARID1B, FBXW7, MLLT4 and TP53 — were more common.
They then extended their analysis to the noncoding portion of the genomes, again looking for regions that contained more mutations than expected by chance. They found recurrent mutations in the PLEKHS1, TBC1D12, and WDR74promoters, as well as in the MALAT1 and NEAT1 long non-coding RNAs.
Different mutational processes cause different types of mutational signatures. For instance, there are specific mutational signatures associated with DNA damage from UV radiation and smoking. Previous studies have identified five mutational signatures in breast cancer based on base substitution patterns. For this new study, the researchers used mathematical modeling to look for additional signatures in the cancer genomes, and found a total of 12, including the five previously described. Of the seven new signatures, five had been identified in other cancer types, while two were completely novel.
Next, the team performed a completely novel analysis, looking at mutational signatures based on rearrangements. They identified six rearrangement signatures. Signatures 1 and 3 were characterized by tandem duplications. The signature 3 group included 91 percent of cancers with BRCA1 mutations and was also enriched for basal-like, triple negative, and cancers with a high homologous recombination deficiency index. Rearrangement signature 5 was characterized by deletions less than 100 kb, while signature 2 was characterized by deletions greater than 100 kb, and was enriched for estrogen receptor-positive cancers.
Ninety of the 560 cases had BRCA1 or BRCA2 mutations, including 60 cases with germline mutations and 14 with somatic mutations. Because genomes with deficient BRCA1 and BRCA2 genes correlated only with either signature 3 or 5, the researchers wrote that identifying these mutational signatures "may be better biomarkers of defective homologous-recombination-based DNA double-strand break repair and responsiveness to [cisplatin and PARP inhibitors] than BRCA1/2 mutations or promoter methylation alone, and thus may constitute the basis of future diagnostics."
In the Nature Communications paper, the researchers delved further into the mutational signatures, enabling them in some cases to identify the associated mutational process as well as whether the signatures were biased to one of the DNA strands. They found that some signatures were age-related and dominated by deamination of cytosine, while others were related to APOBEC, mismatch repair deficiency, or homologous recombination deficiency, and some had unknown etiology.
Ewan Birney, director of the European Bioinformatics Institute at the European Molecular Biology Laboratory, led this portion of the analysis. "We know that genetic changes and their position in the cancer genome influence how a person responds to a cancer therapy," he said in a statement, adding that the study "gave us an unexpected way to characterize the types of mutations that happen in certain breast cancers."