EPITalk: Behind the Paper
This stimulating podcast series from the Annals of Epidemiology takes you behind the scenes of groundbreaking articles recently published in the journal. Join Editor-in-Chief, Patrick Sullivan, and journal authors for thought-provoking conversations on the latest findings and developments in epidemiologic and methodologic research.
EPITalk: Behind the Paper
The Obesity Paradox: Childhood Body Size & Breast Cancer Risk
Dr. Dorthe Pedersen unpacks the paradoxical relationship between childhood body size, breast density, and postmenopausal breast cancer risk among a large cohort of Danish women. Her article, “Associations of early life body size and pubertal timing with breast density and postmenopausal breast cancer risk: A mediation analysis,” can be found in the February 2025 (Vol. 102) issue of Annals of Epidemiology.
Read the full article here:
https://www.sciencedirect.com/science/article/pii/S1047279725000109
Episode Credits:
- Executive Producer: Sabrina Debas
- Technical Producer: Paula Burrows
- Annals of Epidemiology is published by Elsevier.
Hello, you're listening to EPITalk: Behind the Paper, a monthly podcast from the Annals of Epidemiology. I'm Patrick Sullivan, Editor-in-Chief of the journal, and in this series we take you behind the scenes of some of the latest epidemiologic research featured in our journal. Today we're here with Dr. Dorthe Pedersen to discuss her article "Associations of Early Life Body Size and Puberty timing with breast density and postmenopausal breast cancer risk: A mediation analysis. You can read the full article online in the February 2025 issue of the journal at www. annalsofepidemiology. org. So I'd like to introduce our guest. Dr. Dorthe C. Pedersen is a research associate at the Center for Clinical Research and Prevention, Copenhagen University Hospital. Her research focuses on women's health in a life course perspective, with an emphasis on how early life factors impact later reproductive health. Dr. Pedersen, thank you so much for joining us today.
Dorthe Pedersen:Thank you for inviting me onto the podcast. I'm very excited to be here.
Patrick Sullivan:So I'd first like to ask you just to explain a little bit about the purpose of the study. What question were you trying to answer with this work?
Dorthe Pedersen:Yeah, actually it's because we know from previous studies that the early life body size is associated with risk of post-menopausal breast cancer, but we don't know why. Also, we know that the breast density, that is, the amount of non-fatty tissue in the breast, is an important marker of later breast cancer risk. But it's also associated with early life body size. So we wanted to investigate whether breast density might mediate some of the observed associations between body size and breast cancer risk. So that was the overall aim of the study.
Patrick Sullivan:So great. So, with that research question, just walk us through the study design and, particularly like, the goal here was one of the substantive pieces is this mediation analysis. So what's the study design? How does the mediation analysis get you to that goal? And just a little bit about why this is the right methodology for this question.
Dorthe Pedersen:Oh yeah, so I think I will start out with that, to do this study, we took advantage of a large population-based cohort of Danish schoolchildren who underwent legally mandated health examinations in school and, because we are in Denmark, we have a personal identification number so we can link individuals to various registers.
Dorthe Pedersen:o this is what we did and obtained information from breast cancer screening programs on the breast density, and also we linked with the Danish breast cancer database to obtain information on those women who later developed breast cancer.
Dorthe Pedersen:First of all, the overall method is to see whether the childhood body size is associated with both the breast density but also the breast cancer risk. So this is part of the paper where we examine first the association with breast density using generalized linear models to obtain risk ratios in order to try to get results that are easily interpretable instead of logistic regression, and then we examine also association with breast cancer risk using survival analysis. And then for the mediation analysis, which is really the interesting part, where you decompose the associations between the childhood body size and the two factors the mediator and the outcome into direct and indirect effects. So we chose to do this with the national effect model that has been adapted for survival analysis so we could get the results out on the hazard ratios. But this method we used also allowed us to quantify the amount of the association that might be mediated,
Patrick Sullivan:S o I want to comment on a couple things. One is just this idea that one of the benefits of having a healthcare structure like you have which we might call healthcare, you know, like that there's the consistent identifier across healthcare settings and some of the early life surveys even really gives you a kind of administrative and healthcare data to be able to do this. So I think you know the journal is based in the United States and so the methods that people might use here often don't have the same confidence in the linkages or have to be more ecological because we don't have that through line. So I just want to make a case, and maybe the listeners of the podcast would be on board with this. But you know about all the values of providing health care that you know in a systematic way and in a way that's coordinated across the country and with other kinds of identifiers. So that's my editorial comment.
Patrick Sullivan:But in terms of the methodology, I think it's interesting a couple of pieces that you brought out here. One is that you're really kind of choosing the models and the form and the outcome for more direct interpretability, which I know is really appreciated when it comes to reporting the results of the study. And then just this idea of testing independently identifying the relationship between childhood BMI and the breast density, and the breast density and the breast cancer are the components of this. Did I sort of get that right, yeah? So, given that, can you describe the main findings?
Dorthe Pedersen:Of course. So in the analysis we included both information on birth weight, initial birth weight that will be self-reported in our cohort but we have tested it against the National Register, so it's highly correlated. And then we also tested the childhood BMI, childhood height and then timing of puberty, because all of these have shown in not only in our cohort but also other cohorts, that there are associations with the breast cancer risk. But in our cohort we only found association between childhood BMI and height, with both breast density and breast cancer risk. So conducting the mediation analysis, of course we found that where there was no association there was no mediation either. But for the childhood BMI we found that parts of the association was actually mediated by breast density, whereas the positive associations we see with childhood height and breast cancer risk was in fact not mediated by breast density. So that gives us an indication of that other mechanisms might be in play for these associations, but yeah.
Patrick Sullivan:Great. So what were some of the limitations or potential sources of bias in the data that you use and, conversely, what do you think some of the strengths were?
Dorthe Pedersen:So let's start with the limitations. Of course there are always limitations when we do studies. In this study it was that even though we have nearly no selection bias into the cohort because it includes virtually all Copenhagen schools, in either public or private schools then, even though breast cancer screening is offered free of charge to all Danish women, not all women choose to participate, so of course there is some selection in those who participate in the breast cancer screening. And also women have to of course, survive from childhood to adulthood to be included in our studies. So I think the main limitation is that we might have some sort of a healthy population we are looking at and we cannot preclude some socio-economic selection into the study at least.
Dorthe Pedersen:For the strengths, it is the size of the study population. Even though we have around 200,000 girls in the cohort, we ended up with around 33,000 in the study. That's also quite a large number, I think. And the strength is also that we have the measured weight and height in childhood. So this really increases precision and reduces risk of misclassification. And also when we get to the breast cancer, this is also in a nationwide register and reporting is mandatory. So there's really a low risk of undetected breast cancer patients. So yeah, that's what's the strengths.
Patrick Sullivan:Thank you. Can you say a little bit about how administratively difficult it is to link up these databases? Is it really just a human subjects approval process, or are there a lot of other steps and complications involved in being able to link these great national data resources that you use?
Dorthe Pedersen:The cohort of the school children is a cohort we have located where I work. So of course we had all permissions in place to have this cohort. And then when we link, we need permission from our national agency for conducting research. But in Denmark when you conduct register-based research you don't need ethical approval, so that's rather easy, but still someone needs to approve that you are asking a relevant question for science and for society.
Dorthe Pedersen:Then for this study we obtained the information from the breast cancer screening program. That was actually different, because this is information from patient journals going into the hospital. So we had to request for a waiver for not contacting the participants by promising that we will not contact them and we will not look at their individual data. But this is for the register-based research. And then we got the waiver and then could conduct a study. But I think for time spent on collecting this data that was at least six months. So it's not-e ven though size data and all the data is available, it's still a time-consuming process to have all the permissions in place and so on.
Patrick Sullivan:Yeah, it's interesting because I think it's a reminder that, even when these data systems exist and I think we would all say it's a good thing that at each step there's a question about what's individually identifiable, what's not, what's coded, and under what circumstances and with what understanding were the data collected and informs that. So thanks for reviewing that for us Again, like I'm interested in the topic, but also in just the potential of these national data sources that are linked, so thanks for sharing a little bit more about that with us. So I want to wrap up this part by asking what you think the implications of your findings are for breast cancer prevention.
Patrick Sullivan:What new recommendations would you make? Or sometimes our research just raises new questions that get us on that path. So what do you think the implications are here?
Dorthe Pedersen:I think, first of all, I would like to say, because we find that a higher BMI in childhood is actually seems to be protective against breast cancer risk.
Dorthe Pedersen:I think it's important to emphasize that we do not recommend girls to gain weight in order to prevent breast cancer, because all in all obesity is associated with multiple other adverse health outcomes and also several other cancer forms. So let us be clear that we are not recommending this. But in addition to this, I think our results are first step. So I think more studies are definitely needed before we can start talking about how this can feed into other preventive initiatives, unfortunately. But I definitely think that, because our results showed that some of the associated was mediated by breast density, there also appears to be a direct effect of the BMI on the breast cancer risk itself, which is also supported by a mendelian randomization study conducted in the UK Biobank. So I think it's really important to look into these mechanisms before we can start talking about preventive efforts. But yeah, perhaps it will make us smarter in the future.
Patrick Sullivan:I think that is the goal. You know, like everything's one step at a time, yeah, so I want to turn now to a little bit about some of the human side of you know how we do our work, how we come up with our questions and our identities as people who do research. And maybe, starting just to ask what motivated you to think about this potential relationship between childhood body size, pubertal timing, postmenopausal breast cancer? And so, just from a scientific point of view, what informs that? But also, how did you choose breast density as the potential mediator? So how do you put all this together before you started?
Dorthe Pedersen:First of all, I'm lucky to be in a research group where I can build on others efforts. So my primary supervisor, my mentor, had this research project where we were looking into different associations with childhood body size in relation to breast cancer. So we also did other studies, but for this particular one we were definitely motivated by the paradox- obesity paradox in breast cancer. That is, that we see that a high BMI in adulthood is associated with a high risk of breast cancer, postmenopausal breast cancer. But when we look in childhood we see that a high BMI in childhood appears to be protective against postmenopausal breast cancer and normally we think that BMI tracks from childhood to adulthood. So this is truly a paradox.
Dorthe Pedersen:We have also, before this, conducted another study where we were able to generate BMI trajectories across the life course in a smaller subsample and from this we saw signals that it's also the timing that matters for the breast cancer risk. So those who gain weight early on have a lower risk compared to those who gain weight later in life and those who gain weight early in life don't get as high up as those who gain later in life, if that makes sense. But of course this is just signals, but it motivated us to look into what could then explain the association. And the research group I was working with have previously conducted a study in a different subset using a different measure of breast density, where they also find some indication of that. It does play a role and because I think it's also biologically possible that having excess weight around the time where the breast develops that can interfere with breast development. So that was also, it was also because of the biology we thought of this.
Dorthe Pedersen:So yes that was how it was put together, and then the mediation has not fully been examined in the previous study. So that really was what we wanted to do in this, also with being able to try to quantify the amount mediated, so yeah.
Patrick Sullivan:Thank you. So did anything that you found in this study sort of point you in a future direction or raise the next question for you?
Dorthe Pedersen:Yeah, for me personally I would say no, not really the breast cancer. I'm mostly an epidemiologist and so for me it's more- I would like to conduct more mediation analysis in other parts of my research. But I definitely think that I hope that other researchers will pick up on these results, especially those more into mechanisms, and try to disentangle how it is that we see these associations not only with the breast entity but also the more direct effects. I think that might be something with the genetics, but yeah, I don't think we have the full overview of how genetics works yet, but.
Patrick Sullivan:And thanks for sharing that and this kind of behind the paper, just part of the discussion-
Patrick Sullivan:I think it is really interesting what you identify, which is that there are some folks who, as epidemiologists, have a deep interest in, develop, have or develop these deep interest in methodology.
Patrick Sullivan:That is a source of passion, which is amazing. And there are other folks who may start with a topical interest and say, like I'm going to pick and choose my collaborators or what I'm going to learn about to serve my interest in, like I'm an infectious disease epidemiologist, so that and that's sort of what what drives me, and I just think it's such a part of the richness of our profession that there are people who sort of get on both parts of that path, you know, and then can bring those skills to other research questions. So great to see this as like something that sparks an interest and a passion and that you're going to carry forward. I'm always interested in how people get to what they do and you've said a little bit about this. But I wonder who, or maybe what you think has been the biggest influence in your career that sort of led you to what you do now and to asking these kinds of questions. What put you on this path?
Dorthe Pedersen:Yeah, that's a tough question, I think, because it would be very difficult for me to pinpoint one person or one thing. I think a lot of things have influenced my career and also, you know, choosing to become a researcher, like my parents, reading books, and also I've been very fortunate to have great mentors around me who have got a dedicated time to guide me. But I think for my interest in women's health, that actually traced back to once I was traveling in Tanzania where I met some local women and you know I had a conversation with them and in less than 10 minutes they had really taught me about what privilege is and how easily you can be blinded by this, because they had a very different view of women's health and women's rights than I came with. So I think that really put on a spark.
Patrick Sullivan:Yeah, that was a spark for your research. Yeah, so what do you think are some misconceptions about doing research or being a researcher?
Dorthe Pedersen:That's a great question. I think two things come to my mind, and one is that people often think that you have to be bright or very talented to become a researcher. But in other sectors I think it requires that you are curious and very persistent and you have a willingness to learn from those around you. And the other thing is that I often hear so you're doing research. It's all about new things, testing new hypotheses, getting new ideas, but actually doing research is a very iterative process, I think, so you keep repeating the same steps over and over again to refine your hypothesis and or approach right. So I think people will be surprised that we do a lot of repetitive work actually in research.
Patrick Sullivan:Yeah, I think this idea about developing hypotheses is interesting because it speaks to the different pieces of our sort of professional identities that are needed to do impactful science right, because you do need the mathematical, the ability to work on that code until it runs, to interpret the coefficients, to experimentiate things or like whatever.
Patrick Sullivan:You know how are we getting to the estimates?
Patrick Sullivan:But this idea of iterating ideas and our social scientific networks, that are the sounding boards for us to say, you know, to articulate ideas or articulate questions, and what an amazing professional resource it is.
Patrick Sullivan:When you find yourself in a place that you can put out a question and I know for myself, like I always have, there's some my mind runs a little bit and says like I don't want to ask this if somebody has already answered it and I'm not sure I've thought through, like if you could even get that data, and so it can be this kind of self-censoring around our own feelings of vulnerability or like recognizing what we don't know. But that is the richness of being in a research group, that is a richness of being, you know, with colleagues, where you develop that level of comfort, where you can really iterate together and I think about how many times like I've come up with an idea that I thought was good and then threw it out, and maybe then I decided it wasn't so good, but by the time the discussion was over, like we ended up in a place that was far better than what I would have come up with, and so that sort of scientific village, you know.
Dorthe Pedersen:Yeah, I know that process. It's a hard one but it's rewarding when you get to the, when you actually get an idea you think is much better than the initial one.
Patrick Sullivan:Yeah, and then you know related but maybe a little bit different what advice would you have for people who are thinking about being a researcher either, in terms of how to prepare, how to decide things you wish you'd thought about early on?
Dorthe Pedersen:Ah, I think my overall is just to go for it, actually. I think, although it's hard work at times, it's also very rewarding, as we just talked about, when you finally get that idea that is perhaps new or perhaps helps you crack something, a problem, or something that's fulfilling.
Patrick Sullivan:Sometimes we call that the aha moment, like after thinking and thinking and turning the gears, a little moment of like aha, like that's there actually is a way to do that. Yeah, exactly, I find that I get that more often like hanging out with colleagues than in moments of solitude. But I think different people do different ways. I definitely know people who retreat and come back out three days later with everything mapped out and I sort of work well in a social scientific setting where we brainstorm and all different ways to do it for sure.
Dorthe Pedersen:Yeah, but I think for me it's both. I think I need some solitude sometimes to really process all the information, and then, of course, I need to test it with someone else. I think learning is something you do in collaboration with others.
Dorthe Pedersen:I think it's difficult to do by yourself. So, yeah, at some point it does require some input. But also for advice for perhaps people wanting to pursue a research career. I also just want to say that they should remember that doing research is a team effort, as we just discussed, so they should always act with this in mind, be kind to people and share credit and stuff like this, and try to find a supervisor or mentor who's also on board with the research as a team effort and not, yeah, just wanted to be lonely genius.
Patrick Sullivan:That sounds like an amazing place to wrap this up. Just the role role of being kind, seeking mentorship, being humble, takes us so much farther than other ways to get at these problems. So thank you, thank you for all that you shared with us and for that great sort of thought in wrapping up that brings us to the end of this episode. Thank you again, Dr. Pedersen, for joining us today. It's been such a pleasure to have you on the podcast.
Dorthe Pedersen:Thank you for inviting me on the podcast. It was a pleasure to participate and I like the discussion or the great thoughtful questions.
Patrick Sullivan:I'm your host, Patrick Sullivan. Thanks for tuning in to this episode and see you next time on EPITalk. Brought to you by Annals of Epidemiology, the official journal of the American College of Epidemiology. For a transcript of this podcast or to read the article featured on this episode and more from the journal, you can visit us online at www. annalsofepidemiology. org.
