Risk-Based Breast Cancer Screening: Studies Suggest Alternatives to Age-Based Guidelines

By Caroline McNeil
January 15, 2014, Volume 5, Issue 1

Figures:

Karla Kerlikowske, MD

Shelley S. Tworoger, PhD

Measuring certain hormone levels could help determine a woman’s risk for breast cancer and add a key factor to current risk-prediction models, according to investigators from Harvard Medical School. Their new study results were presented at the American Association for Cancer Research (AACR) Annual International Conference on Frontiers in Cancer Prevention Research, held in National Harbor, Maryland, in October.1

The findings come at a time when risk-based breast cancer screening is a major focus of research groups around the country and in Europe. Results like these could lead to more personalized risk profiles to help women and their physicians—and guideline developers—decide when and how often an individual should have a screening mammogram.

Speaking at an AACR session on mammography,2 Karla Kerlikowske, MD, Professor in the Departments of Medicine and Epidemiology/Biostatistics at the University of California, San Francisco (UCSF), said that risk-based screening would allow those with a high-risk profile to have earlier and more frequent screening and/or screening with more sensitive imaging modalities. Women with low risk could be screened less often and avoid the potential harms, such as the false-positives that lead to unnecessary biopsies.

Current Guidelines

Current mammography guidelines focus on age, with major national organizations issuing opposing recommendations; some say to begin at age 40, some say 50. But now many studies, like those presented at the AACR prevention meeting, suggest that the controversy may give way to an approach based less on age and more on an expanded list of risk factors that are emerging from research centers around the country.

Personalized Recommendations

“The randomized trials [of mammography] were performed according to age at trial entry, and since randomized controlled trial results are considered the gold standard, recommendations for mammography have primarily been based on age,” said Dr. Kerlikowske. “But people in their 40s can have the risk of a 50-year-old, and some people in their 50s are actually at very low risk,” she said.

The San Francisco Mammography Registry is one of seven centers in the Breast Cancer Surveillance Consortium (BCSC), a National Cancer Institute (NCI)-funded network of mammography registries linked to pathology and/or tumor outcomes. One BCSC focus is identifying and validating risk factors and risk models that could help personalize screening recommendations.

“We’re now able to move the discussion about who should be screened to a more risk-specific context, trying to identify people who will really benefit from screening and those who would not benefit—for whom the harms would outweigh the benefits,” said Dr. Kerlikowske, who is also Principal Investigator of the San Francisco Mammography Registry.

Risk-Prediction Models

Research on breast cancer risk models has been underway for years. The earliest and probably best known model, developed by Mitchell Gail at the NCI, is based on a woman’s medical and family history of breast cancer, previous biopsies, and reproductive history including age at menarche. Race was added to the Gail Model after research showed that African Americans were at increased risk.

More recently developed models have added other factors such as breast density (BCSC model), and body mass index, alcohol use, and age at menopause (Rosner-Colditz model). Researchers continue to investigate these factors as well as a range of others, including genetic variants, sex hormone levels, types of breast density, biopsy findings, and combinations of various risk factors.

In the hormone study presented at the AACR meeting, Shelley S. Tworoger, PhD,1 Associate Professor in the Channing Division of Network Medicine at Brigham and Women’s Hospital and Harvard Medical School, Boston, and her colleagues looked at levels of half a dozen sex hormones in 437 postmenopausal women with invasive breast cancer and 770 matching controls in the Nurses’ Health Study. Using blood samples taken before diagnosis, they analyzed whether any of the hormones improved prediction of invasive breast cancer when added to either the Gail or the Rosner-Colditz risk model.

They found that measuring estrone sulfate, testosterone, and prolactin levels offered the biggest improvement in risk prediction. Dr. Tworoger said in a statement that “the improvement in prediction when adding circulating hormone levels was better than the improvement observed by other studies that included mammographic density and genetic factors.”

Breast Density and Imaging

Up to now, mammographic breast density has been one of the most studied risk factors. Hundreds of studies have shown that it’s a factor in both short-term and long-term risk, Dr. Kerlikowske said in an interview. In the short term, it plays a role because it makes it harder to see tumors with standard mammography. In the long term, it appears to heighten biologic risk, although why some women with dense breasts develop cancer and others don’t remains unknown.

“Breast density is not all vanilla,” she said. “Most women with high breast density don’t develop breast cancer, and understanding what type of breast density contributes to cancer development could improve risk prediction.”

The awareness that mammography is not equally sensitive in all women has raised new questions about the best imaging modalities to use with different risk profiles; MRI is the most sensitive modality, but other options, such as ultrasound, tomosynthesis, and computed tomography, are under study. One focus of the BCSC is how to add to or improve on its current risk model by showing which risk profiles would benefit from other imaging modalities.

Single Nucleotide Polymorphisms

Other leading candidates to add to risk prediction models are the genetic variants known as single nucleotide polymorphisms or SNPs, some of which are known to contribute modestly to breast cancer risk. In one large NCI study, published in The New England Journal of Medicine in 2010,3 a panel of 10 SNPs, added to the Gail model, did a slightly better job of predicting which women would develop breast cancer than the standard model alone. The researchers, led by Sholom Wacholder, PhD, Senior Investigator in the Division of Cancer Epidemiology & Genetics at NCI, concluded that the improvement was too small to influence clinical decision-making.

But research on SNPs continues. Dr. Kerlikowske said that the BCSC is looking at 76 SNPs in combination with breast density. Also under investigation by the consortium are sex hormone levels, SNPs, and breast density combined. Other candidates for further study are benign biopsy findings such as hyperplasia, she said. ■

Disclosure: Drs. Kerlikowske, Tworoger, and Wacholder reported no potential conflicts of interest.

References

1. Tworoger SS: Inclusion of endogenous hormone levels in risk prediction models of postmenopausal breast cancer. 12th Annual AACR International Conference on Frontiers in Cancer Prevention Research. Presented October 29, 2013.

2. Kerlikowske KM: Risk-based breast cancer screening. 12th Annual AACR International Conference on Frontiers in Cancer Prevention Research. Presented October 29, 2013.

3. Wacholder S, Hartge P, Prentice R, et al: Performance of common genetic variants in breast-cancer risk models. N Engl J Med 362:986-993, 2010.