1. Nolen, Lindsey

Article Content

When early breast cancer risk prediction tools were developed, the hope was that they would help identify higher-than-average risk across all populations of women. However, it was soon clear that these models tended to underestimate breast cancer risk in U.S. Black women, who are known to be more likely to have breast cancer at earlier ages and with a worse prognosis than White women.

breast cancer patien... - Click to enlarge in new windowbreast cancer patient. breast cancer patient

A team of researchers from Boston University's Slone Epidemiology Center set out to develop and evaluate a new risk prediction tool for breast cancer in U.S. Black women, one that would be suitable for use across primary care settings.


"Risk prediction models are used in two main situations. One is to help women and their clinicians make an informed decision about when or how often they go for mammographic screening or other types of screening," explained the research's corresponding author, Julie Palmer, ScD, Director of the Slone Epidemiology Center and the Karin Grunebaum Professor in Cancer Research at Boston University School of Medicine. "The other way it's used is that every so often there's a new medication developed that might possibly reduce a woman's risk of developing breast cancer. During a trial, investigators usually want to only enroll women at high risk of the disease, because if they enrolled a general cross-section of the population, they would have to wait too long for enough of the disease to occur."


In developing their own evaluation tools for the risk prediction model, Palmer and her colleagues used epidemiologic data from three case-control studies of Black women from various regions of the U.S. Then, 15 years of follow-up data from 51,798 participants in the Boston University Black Women's Health Study was used to test the model-which was found to be well-calibrated.


According to the research, discriminatory accuracy, which reflects how well a model predicts risk for an individual woman, was "similar to that of the most frequently used questionnaire-based breast cancer risk prediction models in White women and was best for women under age 40."


Other appeals of this new prediction tool are that the model is simple, can be easily used by primary care providers to guide screening recommendations and/or referral for genetic testing, and all the information required can be obtained from the women themselves with a few questions. The hope is that these features help lead to earlier diagnosis and reduced mortality in young Black women.


"We had for a long time been acutely aware of disparities in the incidence of disease and the care that people were receiving in this country. We had started a prospective cohort study called the Black Women's Health Study 25 years ago, funded by the National Cancer Institute, of which the main purpose was to get at the difference in breast cancer mortality between Black women and White women in the United States," Palmer explained. "The Black Women's Health Study enrolled 59,000 Black women from all parts of the U.S., and those participants were enrolled by completing a very long questionnaire in 1995, about all different parts of their life, health, medication, and activities."


She continued that, since then, most have continued to complete questionnaires every 2 years. Over that time, as with any population, some developed diseases and some did not. Through this data collection, Palmer and her team have been able to study many other conditions, with a main focus on breast cancer. As about 3,000 women in the study developed breast cancer, the research team became very aware of the problem with the current models-underestimating risk of breast cancer in Black women.


"We wondered if that had something to do with the different distribution of the two main types of breast cancer, cancers that are hormonally responsive, which we call hormone receptor- (or estrogen receptor-) positive. Those are the ones that are related to most of the breast cancer risk factors we know about," Palmer said. "Breast cancers that are estrogen receptor-negative make up about a third of breast cancers in Black women."


She further explained that this third is twice as big as the proportion of breast cancers that are estrogen receptor-negative among White women or Asian women in the U.S. Researchers don't know as much about risk factors for ER-negative breast cancer. According to Palmer, the risk prediction models that were developed in White women are essentially risk prediction models for ER-positive breast cancer, because that's what most of the breast cancers were.


"We waited until there was enough data from several different studies being conducted at different institutions across the U.S., in addition to our own Black Women's Health Study. This was so we would have a large enough sample with Black women who have breast cancer and Black women without cancer who could serve as controls with extensive data on these potential risk factors," Palmer said. "We then used this information to determine which risk factors were statistically retained in a prediction model based on the typical distribution of estrogen receptor-positive and estrogen receptor-negative breast cancers in U.S. Black women.


For that reason, the research team's model isn't as weighted towards estrogen receptor-positive breast cancer as other models are. Palmer believes this may be why her team's results and measures of discriminatory accuracy were higher than in previous models.


Moving forward, Palmer and her team want to evaluate the acceptability of this tool among physicians. They plan to start this evaluation in the Boston area with neighborhood health centers to see if there's anything her team can do to make the model more useful for the clinicians.


"We're also working with investigators at other institutions, looking at the other end of the spectrum in both Black women and White women, trying to evaluate risk prediction for older women-those older than 75," Palmer shared. "We're looking at what the best risk prediction model is, not just for breast cancer, but for taking into account dying of other causes so that older women can make their decision about screening based not just on their chances of getting breast cancer in the next few years, but also on whether finding out about breast cancer early would make a beneficial difference in their lives."


She stressed that, ultimately, it's important for women to have good information on their own risk of breast cancer at a given age so that they can make informed decisions not just about when to start screening, but also about when to stop.


Lindsey Nolen is a contributing writer.