Integrating genetic predictors into subsequent breast cancer risk prediction in survivors of childhood cancer.

Combination of a polygenic risk score validated in the general population with germline pathogenic variants in breast cancer susceptibility genes

This study included females surviving at least 5 years after a childhood cancer diagnosis, from the Childhood Cancer Survivor Study and the St. Jude Lifetime Cohort, with array data (n = 2,096, including 218 subsequent breast cancers) or whole-genome sequencing (n = 3,292, including 101 subsequent cancers) — 5,388 survivors and 319 subsequent breast cancers overall. Ninety-nine externally validated polygenic risk scores were computed: 90.9% were nominally associated with risk (p < 0.05) but with substantially varying effect sizes. The most discriminating scores had broader genome-wide coverage (6.4 million-variant score, HR per standard deviation = 1.71; 95% CI 1.43-2.05; p = 4.2 × 10⁻⁹) and 7.7-fold higher odds of including variants across multiple DNA damage repair pathways. Among survivors with whole-genome sequencing, 1.6% carried a pathogenic or likely pathogenic variant in a clinical testing panel gene, associated with a 7.4-fold greater risk (95% CI 3.16-17.19); adding genetic factors improved risk prediction by age 40 compared with treatment exposures alone (p < 0.001).

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

The study answers a concrete question: in a woman irradiated in childhood, does constitutional genetic risk remain informative once treatment exposure is accounted for? The answer is yes, and the most immediately usable finding is not the polygenic score but the 7.4-fold risk associated with pathogenic variants in panel genes, which argues for germline testing in this population. The polygenic score, with a hazard ratio of 1.71 per standard deviation, remains a cohort-level stratification variable rather than an individual decision tool, and the spread across the 99 scores tested shows that none is standardised. The ancestry composition, overwhelmingly European (4,752 of 5,388), further limits transferability of these scores to other populations.

Analysis by Dr Thibaut Benquey