Introduction
The identification of missense variants in the BRCA1 and BRCA2 genes presents a significant clinical challenge. While certain mutations are known to drastically increase the risk of breast and ovarian cancer, many others are classified as "variants of uncertain significance" (VUS). Traditional functional assays are often limited in scope, focusing on single protein functions. This study, originally published in PLoS Genetics, investigates whether a systems-level approach—analyzing global gene expression profiles in response to DNA damage—can distinguish between pathogenic mutation carriers and individuals with familial breast cancer of unknown genetic origin (BRCAX).
Methodology
Researchers utilized Lymphoblastoid Cell Lines (LCLs) derived from 72 women in high-risk families. The cohort included 23 BRCA1 carriers (17 truncating, 6 missense), 22 BRCA2 carriers (19 truncating, 3 missense), and 27 BRCAX individuals. To stimulate the DNA damage response pathways where BRCA1 and BRCA2 are critical, cells were exposed to 10Gy of ionizing radiation (IR).
Gene expression was measured 30 minutes post-irradiation using Illumina microarrays (~46,000 probes). The data were analyzed using unsupervised hierarchical clustering and two machine learning approaches: Gaussian Process Classifiers (GPC) for multi-way status prediction and Support Vector Machines (SVM) to identify specific gene signatures associated with mutation status.
Key Findings
The Impact of "Low Clinical Significance" Variants
A pivotal discovery was the behavior of BRCAX individuals carrying rare variants previously classified as having "low clinical significance" (LCS). These individuals exhibited gene expression profiles more similar to BRCA1/2 mutation carriers than to other BRCAX cases. This suggests that these supposedly neutral variants may exert subtle functional effects or contribute to moderate cancer risk, potentially confounding diagnostic assays that do not account for them.
Mutation Type and Classification Accuracy
The study found that the "molecular fingerprint" of a mutation depends heavily on its type. Truncating mutations produced different signatures than missense mutations. When using SVMs with optimized gene lists, the researchers achieved high accuracy (82–94%) in identifying BRCAX individuals without LCS variants. However, accuracy dropped significantly (to 40–50%) when LCS variants were included in the BRCAX group.
Furthermore, BRCA1 and BRCA2 carriers showed remarkably similar expression profiles shortly after irradiation, making them difficult to distinguish from one another, though they remained distinct from the majority of the BRCAX cohort. The GPC achieved a maximum accuracy of 62.26% when distinguishing between the three groups (excluding missense and LCS samples).
Conclusion
This research demonstrates that gene expression profiling in irradiated LCLs can capture the functional consequences of BRCA1/2 mutations. However, it also highlights the complexity of the genetic landscape in familial breast cancer. Future array-based diagnostic tools must be sophisticated enough to account for the specific nature of the mutation (missense vs. truncating) and the presence of rare, low-risk variants that can mimic the signatures of known pathogenic mutations. These findings underscore the importance of considering "mutation effect" in the development of clinical pathogenicity assays.
Summary updated