Ovarian and Breast Cancer Risks Associated With Pathogenic Variants in RAD51C and RAD51D

Xin Yang, Honglin Song, Goska Leslie, Christoph Engel, Eric Hahnen, Bernd Auber, Judit Horváth, Karin Kast, Dieter Niederacher, Clare Turnbull, Richard S. Houlston, Helen Hanson, Chey Loveday, Jill S. Dolinsky, Holly LaDuca, Susan J. Ramus, Usha Menon, Adam N. Rosenthal, Ian Jacobs, Simon A. Gayther, Ed Dicks, Heli Nevanlinna, Kristiina Aittomäki, Liisa M. Pelttari, Hans Ehrencrona, Åke Borg, Anders Kvist, Bárbara Rivera, Thomas van Overeem Hansen, Malene Djursby, Andrew Lee, Joe Dennis, David D.l. Bowtell, Nadia Traficante, Orland Dı́ez, Judith Balmañà, Stephen B. Gruber, Georgia Chenevix‐Trench, kConFab Investigators, Allan Jensen, Susanne K. Kjær, Estrid Høgdall, Laurent Castéra, Judy E. Garber, Ramūnas Janavičius, Ana Osório, Lisa Golmard, Ana Vega, Fergus J. Couch, Mark E. Robson, Jacek Gronwald, Susan M. Domchek, Julie O. Culver, Miguel de la Hoya, Douglas F. Easton, William D. Foulkes, Marc Tischkowitz, Alfons Meindl, Rita K. Schmutzler, Paul D.p. Pharoah, Antonis C. AntoniouJNCI 2020 · DOI

TL;DR

This study provides the most comprehensive estimates to date of tubo-ovarian carcinoma (TOC) and breast cancer (BC) risks for individuals carrying pathogenic variants in the RAD51C and RAD51D genes. These findings are crucial for refining genetic counseling and improving cancer risk prediction models for these gene carriers.

What Problem Does This Paper Address?

Genetic testing for hereditary cancer predisposition is increasingly common, with multigene panels now routinely assessing risk for various genes. While BRCA1 and BRCA2 are well-established high-risk genes, the precise cancer risks associated with other genes, such as RAD51C and RAD51D, are less clear. Previous studies have yielded a wide range of risk estimates, often based on case-control designs that can be prone to biases like population stratification and confounding due to family history. This ambiguity makes it challenging for clinicians to provide accurate genetic counseling and for individuals to make informed decisions about cancer surveillance and prevention strategies. This paper aims to address this by providing more precise, family-based estimates of TOC and BC risks for RAD51C and RAD51D pathogenic variant carriers.

Methods

The researchers employed a complex segregation analysis using pedigree data from a large international consortium of families. This approach models cancer inheritance patterns within families to estimate relative risks (RRs) for carriers of pathogenic variants in RAD51C or RAD51D compared to the general population. They fitted various genetic models, including major-gene models and models incorporating a polygenic component (representing unobserved genetic or environmental factors shared within families), to best capture the observed cancer incidence. The analysis also accounted for the ascertainment process of each family to minimize bias. Age-specific risks were modeled using constant relative risks, age-decade specific risks, and piecewise linear functions to capture age-dependent risk patterns.

Datasets

The study analyzed data from 6178 families for RAD51C and 6690 families for RAD51D. These families were recruited between 1996 and 2017 from 28 study centers across Europe and North America. Families were ascertained through various means, including screening of individuals with multiple cancer-affected relatives and screening of unselected cancer patients. Population variant frequencies for RAD51C and RAD51D were estimated from the UK Biobank exome sequencing dataset.

Key Insights and Ideas

  • Family history matters: The study strongly suggests that cancer family history significantly modifies the risk of TOC and BC for RAD51C and RAD51D carriers, highlighting the need to incorporate this information into risk assessment.
  • Age-dependent risk: For TOC, the relative risk associated with RAD51C and RAD51D pathogenic variants appears to increase with age until around 50-60 years and then decrease, suggesting a potential window for intervention.
  • Polygenic component: The best-fitting models included a residual polygenic component, indicating that other unobserved genetic or environmental factors, beyond the primary gene variant, play a role in modifying cancer risk.
  • Refined risk estimates: By using family-based data and sophisticated statistical modeling, the study provides more precise and reliable risk estimates than previous case-control studies.

Key Results

  • TOC Risk: Pathogenic variants in RAD51C and RAD51D are associated with a significantly increased risk of TOC. The estimated cumulative risk of developing TOC by age 80 years is 11% (95% CI: 6%-21%) for RAD51C carriers and 13% (95% CI: 7%-23%) for RAD51D carriers, assuming no family history of TOC.
  • BC Risk: These variants also confer an increased risk of breast cancer. The cumulative risk of developing BC by age 80 years is 21% (95% CI: 15%-29%) for RAD51C carriers and 20% (95% CI: 14%-28%) for RAD51D carriers, assuming no family history of BC.
  • Impact of Family History: For carriers with two first-degree relatives diagnosed with TOC, the cumulative risk by age 80 can be as high as 32-36%. For BC, in carriers with two first-degree relatives diagnosed with BC, the cumulative risk can reach 44-46%.

Limitations and Future Work

The authors acknowledge several limitations:

  • Variant Type and Location: The study could not assess variations in risk based on the specific type or location of the pathogenic variant within the RAD51C or RAD51D genes.
  • Small Numbers in Age Groups: While the study is large, the number of cases in certain age groups remains small, leading to larger standard errors for age-specific risk estimates.
  • Cancer Subtypes: The analysis did not include cancer subtype information (e.g., ER-positive vs. ER-negative breast cancer), which may be important as some studies suggest RAD51C/D variants are more strongly associated with specific subtypes.
  • Geographic Restriction: The analysis was primarily restricted to studies from Europe and North America, limiting generalizability to other populations.

Future work could involve investigating the impact of specific variant types and locations, conducting subtype-specific analyses if tumor pathology data become available, and expanding studies to include diverse populations. The findings strongly support the inclusion of RAD51C and RAD51D in gene panel testing and risk prediction models for improved cancer surveillance and prevention strategies.

Summary updated

Abstract

These estimates will facilitate the genetic counseling of RAD51C and RAD51D pathogenic variant carriers and justify the incorporation of RAD51C and RAD51D into cancer risk prediction models.
BibTeX
@inproceedings{yang2020ovarian,
  author = {Xin Yang and Honglin Song and Goska Leslie and Christoph Engel and Eric Hahnen and Bernd Auber and Judit Horváth and Karin Kast and Dieter Niederacher and Clare Turnbull and Richard S. Houlston and Helen Hanson and Chey Loveday and Jill S. Dolinsky and Holly LaDuca and Susan J. Ramus and Usha Menon and Adam N. Rosenthal and Ian Jacobs and Simon A. Gayther and Ed Dicks and Heli Nevanlinna and Kristiina Aittomäki and Liisa M. Pelttari and Hans Ehrencrona and Åke Borg and Anders Kvist and Bárbara Rivera and Thomas van Overeem Hansen and Malene Djursby and Andrew Lee and Joe Dennis and David D.L. Bowtell and Nadia Traficante and Orland Dı́ez and Judith Balmañà and Stephen B. Gruber and Georgia Chenevix‐Trench and kConFab Investigators and Allan Jensen and Susanne K. Kjær and Estrid Høgdall and Laurent Castéra and Judy E. Garber and Ramūnas Janavičius and Ana Osório and Lisa Golmard and Ana Vega and Fergus J. Couch and Mark E. Robson and Jacek Gronwald and Susan M. Domchek and Julie O. Culver and Miguel de la Hoya and Douglas F. Easton and William D. Foulkes and Marc Tischkowitz and Alfons Meindl and Rita K. Schmutzler and Paul D.P. Pharoah and Antonis C. Antoniou},
  title = {{Ovarian and Breast Cancer Risks Associated With Pathogenic Variants in <i>RAD51C</i> and <i>RAD51D</i>}},
  booktitle = {JNCI},
  year = {2020},
  doi = {10.1093/jnci/djaa030},
}
Figure