The goal of an educator is to help transform learners into experts and leaders of tomorrow. One Georgetown University professor witnessed this concept come full circle as her student's engineering project developed into an oncological research tool with the potential to yield greater breast density cancer insights.
Studying breast density images in mice, students theorized that a two-pronged approach could provide additional aid in the detection of cancer cells and masses. If this theory is also found to be true in humans, the research could help provide better direction on breast screening best practices.
According to Priscilla A. Furth, MD, Professor of Oncology and Medicine at Georgetown University, breast cancer density in women has been known to be associated with breast cancer development for several decades. She explained that, at a population level, women with dense breasts have a higher percentage of breast cancer development. However, the issue is that the association between dense breasts and cancer development is very nonspecific. Furth believes this is because most collected data has been from populations and not from individual women.
"You can have an individual woman who actually might have dense breasts, but she never develops breast cancer. So it's more of a statistical thing that's difficult to apply in the clinic in terms of providing people useful information on the risk of them developing breast cancer," Furth explained.
Also in terms of statistics, the Centers for Disease Control and Prevention (CDC) reports that about half of women who are 40 years old or older have dense breasts. Women whose breasts are almost entirely fatty (about 10% of women), or who have a few areas of dense tissue scattered through the breasts (about 40% of women), are said to have low-density, non-dense, or fatty breasts.
Alternatively, women with breasts that are evenly dense throughout (about 40% of women) or who have extremely dense breasts (about 10% of women) are said to have high-density or dense breasts. The CDC also confirms that women with dense breasts have a higher chance of getting breast cancer-the more dense your breasts, the higher the risk.
Study Details
To better understand the connection between breast density and cancer development, Furth has studied mouse models across her preclinical translational research. Specifically, her research has focused on the computer grip and digital reading of mammograms to determine if patterns of breast density provide additional insight into the recognition of cancer that may ultimately be useful when scanning individual women.
"In other words, can we look beyond just the broad picture of breast cancer density and look more particularly at the patterns that people have on their mammograms or patterns over time?" questioned Furth.
Helping her uncover the answer, Furth encountered undergraduate student Brendon Rooney in the lab during her breast cancer prevention research. Using mouse mammary gland whole mount preparation and analysis (modeling potential mammogram image results), Rooney asked if he could develop an IT program to better identify patterns of growth that are more predictive of later breast cancer development than those that rely on doing so by eyesight.
"He went ahead and researched it, and he came out with a program. At that time, he was looking at rather younger mice under age 12 months, and the mammary glands look sort of lobular in that age group," Furth said. "He used a program called Gaussian Denoising, which looks at those lobular structures. As he was finishing his undergraduate career, our research program was evolving."
Being able to play such a role in Rooney's research and education, Furth emphasized that it's always a pleasure to have a student come into the lab and feel comfortable suggesting new ideas. She explained that, through encouragement, she always welcomes opportunities for her students to want to discover new ideas and concepts. Further, she said that in the case of Rooney's research specifically, she found his ability to take charge and work toward this academic accomplishment very inspirational.
She added that her team's research (which Rooney utilized) was looking at older mice because they made a model that was more reflective of the kind of breast cancers that women get after the age of menopause. In mice up to and after the age of 20 months, the glands structurally look different, she said.
This is the same in women, Furth explained, as women's gland structure changes at the time of menopause. While most standard programs cannot follow the density, computer programs have paradigms that track along and look at what's the growth structure and the stroma or fat behind it.
"[Rooney] realized what we had were more linear type structures, and that he needed a different kind of program, and this is the ridge filter. He set out to develop a program deciding which path it would go down, so the science really brought us to using this bifurcated approach."
Furth shared that this type of research has become increasingly important as scientists continue to conclude that the structure of the mammary gland is non-uniform and that different imaging approaches may need to be applied. To acquire more exacting information, she said imaging cannot be thought of as a "one-stop shop," as the breast and the mammary gland do have variability within the structures. Engineers and researchers must acknowledge this in their design when developing a digital or computerized program.
"[Using mice], our particular program won't be directly analogous to what humans do in terms of the particular code that we write. But the concept is that it shows that, if you do allow your program to pick up different details, you can then isolate the kind of structures that are actually more specifically linked to later cancer development," Furth shared.
Using this information, Furth questions, "What could we eventually find out?" She believes the answer involves how older mice change and become more linear as they go through a process like involution where the mammary gland recedes a little bit after they finish reproductive senescence (reflective of menopause in women). According to Furth, if there's an area that doesn't recede, it becomes lobular and looks more like a piece that might belong to a younger mouse. It shouldn't be there at that age.
"So this lobular thing then becomes a potential focus perhaps for being the kind of fertile soil where cancer could later develop?" Furth asked. "The idea is: can you look for that kind of anomaly in structure or pattern?"
She added that most of the studies people had done on mice before have actually been in younger mice prior to reproductive senescence. For this reason, her team found that it was when they were really looking at the older mice that they needed the flexibility to find abnormal patterns.
"The imaging field in humans is quite broad, and it extends beyond mammograms. It extends to many different types of imaging procedures," Furth noted. "I believe some of those procedures have different ways in which they analyze their images."
Moving forward, Furth believes an interesting next research project would be to use comprehensive mammogram data to start to look at programs being used to assess the density and ask how they can improve each program to actually have more predictive value for an individual woman. If a woman's digital breast density was followed for 5 years, or even a single reading, she suspects that professionals could become better at interpreting what it means when something's dense.
"In other words, we're trying to get the kind of thing we can do with the mice where we can have the histology available to match an image to what's actually going on at the tissue level," she explained. "I think there would be experimental ways for people to approach that in women with mammograms-sort of doing what we did in the mice, but in humans. However, this would require a lot of work and funding."
Today, Rooney's Python program and its code are open and available to all individuals. Furth said this was a deliberate choice on their part as they wanted other researchers to have public access to the tool, which is included in the supplemental file that's in their research paper. Thus, anyone can download the program and place it on their computer, and she hopes the work continues toward making an impact on how breast imaging is conducted.
"One would hope maybe this work inspires insurance companies to start covering digital mammography in a way that starts looking a little deeper into the structure of breasts that we can get from mammograms, and then think about how we can do better predictively with individual patients."
Lindsey Nolen is a contributing writer.