Join Us

Breast Cancer Risk and Race
Lola Fayanju, M.D.
August 6, 2020

Save as Favorite
Sign in to receive recommendations (Learn more)
Fayanju lola preferred small

Dr. Lola Fayanju is assistant professor of surgery at the Duke University School of Medicine. In addition to treating people with breast cancer surgery, Dr. Fayanju’s research interests include using big data and sophisticated analyses to reduce disparities in outcomes after breast cancer diagnoses and to improve the value of breast cancer care. When she was a general surgery resident at Washington University in St. Louis, her research found that women treated by safety-net primary care doctors in the greater St. Louis area were more likely to be diagnosed with more advanced-stage breast cancer than women who had private insurance. She also looked at the reasons behind this disparity, and her work led to an overhaul of the referral process for underserved women in the St. Louis area.

In a June 2020 New England Journal of Medicine article, at least two widely used tools estimating breast cancer risk have been found to offer lower risk estimates for women of color.

Listen to the podcast to hear Dr. Fayanju explain:

  • how these tools were created
  • why an artificially lower risk of breast cancer can be harmful for women of color
  • the factors that she thinks need to be incorporated into risk calculators for minority women
  • how women of color can accurately figure out their risk of breast cancer and develop an appropriate screening plan with their doctor

Running time: 25:12

Thank you for listening to the podcast. Please subscribe on iTunes, Stitcher, Spotify, TuneIn, Google Play, or wherever you listen to podcasts. To share your thoughts about this or any episode, leave feedback on the podcast episode landing page on our website.

Show Full Transcript

Jamie DePolo: Hello. Thanks for listening. Our guest today is Dr. Lola Fayanju, assistant professor of surgery at the Duke University School of Medicine. In addition to treating people with breast cancer surgery, Dr. Fayanju’s research interests include using big data and sophisticated analyses to reduce disparities in outcomes after breast cancer diagnoses and to improve the value of breast cancer care.

When she was a general surgery resident at Washington University in St. Louis, her research found that women treated by safety-net primary care doctors in the greater St. Louis area were more likely to be diagnosed with more advanced-stage breast cancer than women who had private insurance. She also looked at the reasons behind this disparity, and her work led to an overhaul of the referral process for underserved women in the St. Louis area.

An article published June 17, 2020, in the New England Journal of Medicine looking at how race was used in risk assessment tools for different diseases found that two breast cancer risk assessment tools offered lower risk estimates for non-white women.

Dr. Fayanju joins us today to talk about how these lower risk estimates can potentially harm non-white women. We’ll also discuss the blog post she wrote about disparities in cancer care after COVID-19. Dr. Fayanju, welcome to the podcast. It’s so nice to talk to you again.

Lola Fayanju, M.D.: Thanks. Glad to be back.

Jamie DePolo: So, let’s talk about the New England Journal of Medicine article first. The scientists looked at a number of tools that predict a person’s risk of various types of cancer, including two for breast cancer: the National Cancer Institute Breast Cancer Risk Assessment Tool and the Breast Cancer Surveillance Consortium Risk Calculator. So, can you explain to us how these tools are used, to start?

Lola Fayanju, M.D.: Sure. So, both of these tools offer an opportunity for clinicians to assess a woman’s likelihood of being diagnosed with breast cancer in the ensuing 5 or 10 years, and the tools are each a little bit different and incorporate different amounts of information to make these assessments.

So, the Breast Cancer Risk Assessment Tool is actually based on something called the Gail Model, which was developed by a senior biostatistician at the Cancer Epidemiology and Genetics Division of the NCI. And the Gail Model incorporates age; start of menstruation or menarche; age at first live birth; whether or not someone has a first-degree relative with breast cancer, that is a mother, daughter, or sister; whether you’ve had previous biopsies, even if they’re benign, and how many; and then whether or not you have what’s called atypia, or abnormal cells that are not cancerous.

The Gail Model, which as I said is a risk assessment tool, acknowledges the limitations of the extent which it can accurately predict future breast cancer risk for women of color. So, in particular, they acknowledge that for women who are African American that have had previous biopsies, there is probably some risk of underestimation. And, likewise, for Hispanic women who were born outside the United States, there’s a likely risk of underestimation. And we do know that Hispanic women have a higher risk of breast cancer if they are born in the U.S. — that’s kind of an interesting side note.

So, anyway, when you look at the data that was used to develop the Gail Model, you recognize that the number of subjects used to build the model are relatively small for the groups of underrepresented minorities. And so, in particular, you see that African Americans were validated using a population that included about 1,600 women with invasive breast cancer and about 1,600 women who didn’t have it. The population used to validate the model in Asian women included about 600 women who had breast cancer and about 1,000 women who did not have breast cancer. And likewise, for Hispanic women, they included a population that had about 1,000 women with breast cancer and 1,400 women without breast cancer. So, you’re working on the order of a thousand, while in contrast, 280,000 white women were included in the validation population. So, as you can imagine, the larger the number you have in which you test the model, the more discrimination you have to see what characteristics are associated with development of breast cancer.

And so that’s just an example of how because there were insufficient numbers of people of color in, presumably, both the testing as well as the validation set of data, that you have this potential to have lack of discriminatory power in assessing who is likely to get breast cancer and who is not if you’re a woman of color.

The other tool, the Breast Cancer Surveillance Consortium Risk Calculator, is a little bit different. It offers risk assessment for the ensuing 5 or 10 years, but it covers a slightly smaller age range from age 35 to 74, while the government-based model that’s based on the Gail Model actually covers women up to the age of 90. And this model was acknowledged also to be tested in a large group, but it was a model where there’s a population of 1.1 million individuals undergoing mammography, of which 18,000 had invasive breast cancer. And it was further validated in a group by a study done by the Mayo Clinic.

And so, again, you have to question what proportion of those patients were people of color and whether there was sufficient sample size to adequately power discrimination. And while the caveat about potentially underestimating the risk of breast cancer in African American and Hispanic women is mentioned as part of both the Gail Model and the government-based breast cancer risk tool that it is based on, that caveat isn’t as much publicized on the page regarding the Breast Cancer Surveillance Consortium Risk Calculator. So, women have to take that with a grain of salt, and providers have to take that with a grain of salt when they’re communicating to patients what their risk of breast cancer is.

Jamie DePolo: Okay. And the concern with underestimating, obviously, is that these women may not be recommended for more aggressive screening or starting screening at an earlier age, which would mean if a breast cancer did develop, it could be missed for a while and not diagnosed until it was more advanced.

Lola Fayanju, M.D.: Right. So, various breast cancer risk prediction tools are investigating not only whether someone would benefit from earlier or more intensive screening — that is, more frequent screening or screening that incorporates MRI as well — but it also looks at whether women would potentially benefit from chemo prevention, which is taking a medication that can reduce risk of getting breast cancer. So, that is another reason some of these tools are used.

And different tools look at different things. So, a woman also needs to be aware that one tool may be used to, again, assess the likelihood of needing chemo or benefitting from chemo prevention, medical chemo prevention, while other tools may be used to assess the potential benefit of getting MRI screening or having more frequent screening.

Jamie DePolo: Okay. Now, I have read about other tools — and I do not know how widely they are used — but risk assessment tools, say, specifically for African American women, specifically for Latinas. Can you talk about those a little bit? Are they widely used? Are they more accurate?

Lola Fayanju, M.D.: You know, I think that there has been really a strong attempt to try and use those tools in people, in populations of people of color. I personally have not seen them being widely used, but that may reflect also just kind of institutional cultural practice patterns. I definitely think there is benefit to improving dissemination of those tools amongst clinicians so that we have a better sense of whether people of color are at risk for breast cancer.

One of the other challenges about assessing breast cancer risk is that there are differences between groups. And of course, these groups remain heterogenous entities, so these are not blanket statements for everyone, but there are some relative differences with regards to how much patients know or want to communicate about their family history, which is actually a very important contributor to assessing breast cancer risk.

Sometimes, people from different cultural backgrounds are more private about collecting family history information, and they might know that someone in their family, their mother or their grandmother, had “a woman’s cancer,” but they may not know if that cancer was ovarian or cervical, which is really important because ovarian cancer is often associated with some type of predisposing gene, while cervical cancer, we know in the vast majority of cases, is caused by a virus, the human papilloma virus. And so knowing whether or not grandma had a hysterectomy because of ovarian cancer or for cervical cancer is incredibly important, and people don’t always have that information. And that may also prevent us from adequately and correctly assessing their likelihood of getting breast cancer in the future based on a relevant family history.

Jamie DePolo: Okay. So, in your opinion, do we need a tool for each specific ethnic group? Would that be more helpful to people?

Lola Fayanju, M.D.: I’m not sure that that’s the exact right approach, but I think that we have to really improve our shared decision making around the tools that we have, and then potentially in the future, I do think it would be great if we can incorporate better information about genetic ancestry into assessing who needs screening.

And I recognize we’re some ways away from this, but a lot of the recommendations to deescalate screening are based on studies conducted in environments that are quite different from the United States, such as Scandinavia, where they not only have a much more homogenous population, but also a healthcare system that is more widely available to patients. They have probably lower burden of comorbidities across the country, and so to extrapolate that screening can be less frequent or can start later in the U.S. population based on studies done in populations that are incredibly different, I think is dangerous. And that’s something that I think a lot of us in the oncology community have pushed back against.

So, I would argue that we need to better inform women about the sources of our data for recommendations and have them engage in shared decision-making prophecies that take into account not only a desire to avoid overdiagnosis of indolent, or nonaggressive, breast cancers or even unnecessary biopsies of things that look like they might be cancer but turn out to be benign on pathology; we need to balance our informing about those things with known increased prevalence of certain aggressive forms of cancer in certain groups, such as triple-negative breast cancer amongst African American women and also at higher rates among Hispanic women.

So, I think at this point, what we really need to do is improve our contextualization of data we have and the tools we have. And then, I think we need to ultimately move toward 1) trying to revalidate models and incorporate larger portions of people of color, and then 2) ultimately use genetic ancestry rather than racial categorization to drive screening, because that’s really going to be the difference. It’s not about race, which is a social construct, but it incorporates many, many things besides, you know, the genes of where your ancestors came from. That is much more complex, and using that as a blunt risk stratifier, I think, is not going to be the best way to go forward.

Jamie DePolo: Okay. And it sounds also like maybe there needs to be some sort of outreach based on culture, because if people are reluctant to talk to their relatives or relatives are reluctant to disclose what sorts of illnesses they have, that can be huge gaps in the knowledge.

Lola Fayanju, M.D.: Absolutely. Sometimes certain healthcare sites will require a patient who presents for genetic counseling or screening, if they don’t have a history of cancer themselves, they’ll require them to go away and find out which of their relatives had cancer, and if they’re alive, they’ll ask for those relatives to get tested first before they have this index individual tested. And that can be really challenging if you’re not having a good relationship with your relative, if that relative is unable to afford genetic testing at this point because they’re long beyond the period during which they may have had cancer, if that person is just unwilling to do it because they have mistrust of anyone collecting samples from them and finding out about their genetic background.

So, I think that we need to be more generous in terms of meeting patients where they are, and people who present for screening, I think we really need to take seriously. They may not be able to go away and get the perfect kind of family tree and have everyone in that family tree conform with our recommendations to get testing. We need to work with who we have and what they can provide in that moment. I’ve definitely seen delays in people getting testing because they’ve been asked to go away and collect information from family members from whom that information cannot always be obtained.

Jamie DePolo: Okay. Thank you. So now, I want to move on to your blog post where you made some predictions about cancer care. And the first one specifically focused on breast cancer, and you said that many women will either have a mammogram later than they should or won’t have a mammogram at all this year because of COVID-19. And your prediction, I believe, was very correct, because I just watched some presentations at the American Association for Cancer Research meeting on COVID and cancer, and there were survey results presented showing that that has, indeed, happened.

So, from your viewpoint in your practice and in your area, was it more than just facilities closing down that led to this delay or skipping mammograms?

Lola Fayanju, M.D.: Well, so, throughout the country, roughly the period spanning the end of March to the middle of May, mammography was widely closed — that is, screening mammography. So, we were only really doing mammography and ultrasound, MRI, for individuals who either presented with a symptom or sign that was concerning for breast cancer or who were already undergoing treatment for breast cancer and needed some type of follow-up imaging to assess their response to treatment or a possibility of recurrence. And so we have this 2-month window during which, essentially, screening across the country was paused, and that window may be wider or smaller in certain individual pockets depending on how hard-hit they were by COVID-19.

And then, afterwards, what happened is that women that had to, in most areas, reestablish an appointment, and if they had a history of cancer, reestablish an appointment with their oncologist at the same time, because all of these things would have been deferred. And the concern is that a lot changed in 2 months. Many people would have lost their job, and suddenly they no longer had insurance through their job that would allow for them to get screening covered. Other people found that they needed to work additional jobs just to make ends meet, given that many businesses were either firing people or even if they kept them on, cutting their wages, cutting their hours in order to accommodate their financial straits.

And so many other conditions got worse over the course of the time when people weren’t accessing medical care for a variety of reasons. And other specialties were also seeing that people are presenting with more severe disease: worse limb ischemia that may ultimately lead to people needing amputations at higher rates, worse erosion of cartilage in joints such that people are presenting with worse levels of osteoarthritis and the need for potential joint replacement.

So, we’ve seen basically a lot of people with different conditions having to, when they do present for care, prioritize things other than breast cancer screening. So, what that means is that a lot of women who, perhaps, already had more precarious financial and employment situations before the COVID-19 pandemic simply have other competing financial, logistic, familial, and health concerns that mean that screening is something that they will not prioritize at this time.

And the concern is that the people who are most likely to be affected in this way, as I mentioned, women who are in a socioeconomically precarious situation, but also women who live in rural areas, and we also think potentially women of color, given that we know the COVID-19 pandemic has had an outsized effect on both morbidity and mortality from COVID-19, as well as the socioeconomic downstream sequelae of the COVID-19 pandemic.

Jamie DePolo: Okay. And just to sort of put that together, too, the concern about the disparities is more women of color, say, fall at or below the poverty line, so may now be more likely to not have insurance than a white woman.

Lola Fayanju, M.D.: Yeah, more likely not to have insurance, more likely to simply not be able to come in for a mammogram during daytime hours because they’re working other jobs or they are also dealing with having to watch children who can’t be in school and don’t have additional people in their lives who might be able to watch those children or keep them home.

So there are a lot of reasons why women will not get tested, and the reason that’s concerning is that we worry about people of color being disproportionately affected by what we call interval cancers, that is, cancers that are found between screens. And the thing that’s challenging for women in general is that if you’ve had a normal mammogram in, let’s say, January, and then in May, you notice a new lump or a new nipple discharge or a change in the nipple, a lot of women are reluctant to then say, “Well, I need another mammogram,” because they think, “I just had one, and it was good.” But it is entirely possible that cancer as developed in that period or at the time of their original screen was too small to be detected. And so, we always have to encourage them and to say, “You know what? A new symptom, a new sign, warrants a new test, and you shouldn’t rely on the fact that you were ‘clean’ on your mammogram 4 months ago, 5 months ago to not come in.”

And what we’ve found in other research that I’ve done looking at distress and stressors among women with breast cancer is that often, women of color have multiple familial obligations that make it harder for them to come in and get care for themselves, and so they put everyone ahead of themselves instead of themselves. And as a result, let’s say that you do notice a lump during this time and you decide to skip a screening mammogram this year. Well, you’re also potentially taking care of your mother and you’re also potentially taking care of your partner and you’re also potentially taking care of your child. You may simply put off getting that lump looked at because there are other people you’re taking care of, other things you want to worry about. So, that is something we’re very concerned about, and we still need some time to see whether these concerns ultimately play out.

Jamie DePolo: Okay. And I do want to ask, too, I’ve heard some researchers talk about, we need to develop messaging to get women back in for screening, but no one has specifically talked about aiming those messages particularly at women of color to, perhaps, overcome some of these disparities. Have you heard anything about that?

Lola Fayanju, M.D.: One of the heartening things about the kind of breast cancer story in the United States is that we actually have quite high rates of screening across racial and ethnic groups except amongst Native Americans. But if you look at data over the past 20 years, the difference between the rates at which Black, Latina, white, and Asian women get screened between the ages of, let’s say, 50 and 64, they really aren’t that different. People are actually going for screening, and so the targeting of communities and the messaging has been fairly successful.

The challenge often is what to do if that screen comes back abnormal. And that is where we are concerned we’re losing people. That you go in for the screen on the assumption that things will be fine. Sometimes, people do it with their sister, with their friend, so you go to the mammovan together or you go to the clinic together, and it’s kind of a partnership. You have a buddy. But if one of you comes back with what they call a BI-RADS 0 estimate, which means it’s an imaging result that indicates you need further workup, are we losing people there? Are we losing people and not having them come back for that second look? And that’s something that I’m interested in studying, and I know that there is some interest in studying that across the literature, but we don’t have great answers about that yet.

Jamie DePolo: Okay. Okay. Well, I definitely would want to talk to you when you get some results on that. Also, in your blog post, you mentioned that you hope for a refinement of screening guidelines. Can you talk about that a little bit and tell us what you mean?

Lola Fayanju, M.D.: So, there is a lot of conflicting recommendations about screening, and some say start at 40. Others say start at 45. Others say start at 40 but talk to your clinician about whether or not that’s right for you. So, I think that we need to continue having these more nuanced conversations with patients, but I think, in particular, we must be very careful not to minimize risk of breast cancer amongst women of color who are known to present with higher rates of premenopausal breast cancer and more aggressive pathology and biomarker subtype when they do present.

So, in terms of refinement, I suspect what it will mean is that we are less eager to try and move later the date, or the age, at which someone gets screened if they’re a person of color, recognizing in the population-level increased risk for early age at breast cancer diagnosis and aggressiveness at diagnosis. So, by refinement, I mean that we need to not just assume, “Okay, unless you have a strong family history or unless you have a gene, you can just start at 45 or 50.” I think we need to say, “For women of color, given population-level concern, we need to start having a conversation about when you should start getting screening earlier,” and not simply use guidelines that were based, as I said, on data collected in populations and countries that are very unlike the United States and had essentially no Black people.

Jamie DePolo: Okay. And finally, to sort of give people an action step or a direction, if a woman is Black, Latina, Asian, Native American — essentially all non-white women — how would you recommend that she approach figuring out her own personal risk of breast cancer and developing a screening plan? Obviously, it has to be done in conjunction with her doctor, but how should she proceed?

Lola Fayanju, M.D.: I think one thing all women need to do is have very honest conversations with their family members about family history. We really need to move away from being vague or circumspect about what people have been through. I have personally treated family members who did not know that the other person had breast cancer, and that’s just a heartbreaking situation, 1) because you can’t provide each other with the support that you need, but also, you don’t know the extent to which that other person’s cancer may affect the trajectory of other people in your family because now you have multiple individuals who have breast cancer in the family, not just one of you.

And so, I would encourage women to 1) get a primary care provider so that you have someone who’s looking out for these things and can advise you and get a mammogram and follow up on the results with you. And 2) women need to really have honest conversations with their family members about their family history and know it really well for which they can be informed advocates for themselves. And then 3) they need to, I would say, start having conversations with their primary care provider as early as 35 in terms of thinking about whether or not they’re someone who would benefit from early breast cancer screening or more intensive breast cancer screening. Do conversations start earlier? Then there’s a good chance that a woman who, for various reasons, is at potentially increased risk can catch a cancer earlier than she otherwise would have if she had been a passive player in her own health decision making.

Jamie DePolo: Okay. Thank you so much. I really appreciate your time. This has been really great information.

Hide Transcript

Was this article helpful? Yes / No
Rn icon

Can we help guide you?

Create a profile for better recommendations

How does this work? Learn more
Are these recommendations helpful? Take a quick survey
Fy22oct sidebarad v02
Back to Top