A BRCA1-mutation associated DNA methylation signature in blood cells predicts sporadic breast cancer incidence and survival

Abstract

BackgroundBRCA1 mutation carriers have an 85% risk of developing breast cancer but the risk of developing non-hereditary breast cancer is difficult to assess. Our objective is to test whether a DNA methylation (DNAme) signature derived from BRCA1 mutation carriers is able to predict non-hereditary breast cancer.MethodsIn a case/control setting (72 BRCA1 mutation carriers and 72 BRCA1/2 wild type controls) blood cell DNA samples were profiled on the Illumina 27 k methylation array. Using the Elastic Net classification algorithm, a BRCA1-mutation DNAme signature was derived and tested in two cohorts: (1) The NSHD (19 breast cancers developed within 12 years after sample donation and 77 controls) and (2) the UKCTOCS trial (119 oestrogen receptor positive breast cancers developed within 5 years after sample donation and 122 controls).ResultsWe found that our blood-based BRCA1-mutation DNAme signature applied to blood cell DNA from women in the NSHD resulted in a receiver operating characteristics (ROC) area under the curve (AUC) of 0.65 (95% CI 0.51 to 0.78, P = 0.02) which did not validate in buccal cells from the same individuals. Applying the signature in blood DNA from UKCTOCS volunteers resulted in AUC of 0.57 (95% CI 0.50 to 0.64; P = 0.03) and is independent of family history or any other known risk factors. Importantly the BRCA1-mutation DNAme signature was able to predict breast cancer mortality (AUC = 0.67; 95% CI 0.51 to 0.83; P = 0.02). We also found that the 1,074 CpGs which are hypermethylated in BRCA1 mutation carriers are significantly enriched for stem cell polycomb group target genes (P <10-20).ConclusionsA DNAme signature derived from BRCA1 carriers is able to predict breast cancer risk and death years in advance of diagnosis. Future studies may need to focus on DNAme profiles in epithelial cells in order to reach the AUC thresholds required of preventative measures or early detection strategies.

Integrative Analysis Identifies Transcription Factor–DNA Methylation Relationships and Introduces New Avenues for Translating Cancer Epigenetics into the Clinic

“…For instance, a recent study Anjum et al identifi ed a BRCA1-mutation-associated DNA methylation signature that was predictive of breast cancer incidence and survival [ 37 ]. Despite these preliminary results, the potential of DNA methylation markers is still not fully realized.…”

Systems Epigenomics and Applications to Ageing and Cancer

“…smoking, sunlight exposure, inflammation, viral infection) mediate cancer risk by increasing intra-sample epigenetic (and genetic) heterogeneity. Although evidence is mounting that genetic and epigenetic heterogeneity both play a role (Anjum et al 2014;Dumanski et al 2015;Feinberg and Irizarry 2010;Forsberg et al 2014;Genovese et al 2014;Issa 2011;Jacobs et al 2012;Laurie et al 2012;Maley et al 2006;Xie et al 2014), clonal mosaicism generally may increase the risk of cancer, since a more heterogeneous cell population is more likely to give rise to a future clone with neoplastic properties. We posited that at the tipping point of the emergence of such a neoplastic clone, the intra-sample molecular heterogeneity would be specially high and this may also drive a high inter-sample variability, as observed between unrelated individuals who are all at the same prior disease stage .…”

“…However, predicting the risk of a given cancer requires, in principle, access to a relevant tissue, the most relevant one being the normal cell of origin that gives rise to the cancer. In the case of blood-borne cancers, this is, in principle, feasible, and indeed recent studies indicate that prospective risk prediction of haematological cancers may be possible (Anjum et al 2014;Dumanski et al 2015;Forsberg et al 2014;Genovese et al 2014;Jacobs et al 2012;Laurie et al 2012;Xie et al 2014;Xu et al 2013). However, the case of epithelial cancers is far more challenging given that the normal cell of origin is usually not readily accessible.…”

“…The emphasis in cancer genomics is slowly shifting from prognosis and treatment to early detection and risk prediction (Anjum et al 2014;Dumanski et al 2015;Forsberg et al 2014;Genovese et al 2014;Jacobs et al 2012;Laurie et al 2012;Xie et al 2014;Xu et al 2013). This shift is happening for various reasons.…”

A General Strategy for Inter-sample Variability Assessment and Normalisation

“…We consider an Illumina 450k data set, encompassing 152 whole blood samples from women for which extensive epidemiological information is available (Anjum et al 2014). Of particular interest is to identify CpGs whose DNA methylation level correlates with smoking pack-years, an epidemiological indicator of an individual's smoking history.…”