Inference on the Genetic Architecture of Breast Cancer Risk

Cancer Epidemiol Biomarkers Prev. 2023 Jan 18;EPI-22-1073. doi: 10.1158/1055-9965.EPI-22-1073. Online ahead of print.


Background: What are the major determinants of women's breast cancer risk? Rare mutations such as those in the BRCA1/2 genes, polygenic scores of common alleles identified by genome-wide association studies, or non-genetic factors?

Methods: The population-based Nordic Twin Study of Cancer, with 3,933 breast cancer cases among 21,054 monozygotic and 30,939 dizygotic female twin pairs, provides three key clues to this question: (1) the average lifetime risk, approximately 8%, does not differ by twin zygosity; (2) the mean time interval between diagnoses when both twins develop disease (i.e., disease concordance) also does not differ by zygosity; but, (3) conditioning on one twin having developed disease, the incidence rate in the co-twin is approximately 1% per year if the pair is monozygotic and 0.5% per year if dizygotic.

Results: Assuming that non-genetic risk factors are shared similarly between twins regardless of zygosity, we can draw two conclusions from (1)-(3).

Conclusions: First, (1) and (3) imply that the chief determinant of risk is in the germline DNA, because the conditional incidence rate is several-fold higher than the average risk (8% lifetime) in monozygotic twins but only half as much in dizygotic twins. Second, the seeming inconsistency between the two-fold conditional incidence rate (3) and the equality of the mean inter-twin disease intervals in disease concordance (2) can be resolved if the risk factors in the germline DNA are rare variants, not common variants.

Impact: This paper details simple deductive reasoning for these conclusions and draws a critical inference regarding breast cancer etiology.