Investigating What Drives Recurrence and Metastasis with Dr. Christina Curtis
By BCRF | October 17, 2022
By BCRF | October 17, 2022
About 30 percent of people diagnosed with early-stage breast cancer will experience a recurrence and develop metastatic disease. To end deaths from breast cancer, we need to better understand both processes: recurrence (when breast cancer comes back) and metastasis (when it spreads to other areas of the body beyond the breast and lymph nodes).
Why recurrence and metastasis happen and how we might then personalize breast cancer care and prediction is what Dr. Christina Curtis and her colleagues are working to uncover. With BCRF support, Dr. Curtis and her team are specifically looking to develop more effective treatments for patients with metastatic estrogen receptor-positive breast cancer—and to develop novel ways to prevent recurrence in people with this form who are at a high risk of recurrence.
A BCRF investigator since 2011, Dr. Curtis is an endowed professor of medicine and genetics at Stanford University, where she leads the Cancer Computational and Systems Biology group and serves as the director of Breast Cancer Translational Research and co-director of the Molecular Tumor Board at the Stanford Cancer Institute.
Chris Riback: Dr. Curtis, thanks for joining. I appreciate your time.
Dr. Christina Curtis: Absolutely. It's a pleasure to be here.
Chris Riback: I want to get into your science and approach to care, for you first to explain. I think I understand what molecular biology is, but what is computational biology and why is it so central to your work? And I think the real question that I want to understand here is, do you self-identify more as a science nerd or math geek? Where are you?
Dr. Christina Curtis: Ooh, that's tough. I'll say I'm a science nerd because I'm really motivated by the biology that we can unpack and understand using computational and mathematical techniques, but it's really the biology that drives me and the tools which can be computational and experimental in nature, often both are a means to an end.
Chris Riback: Computational biology is the science or the practice of applying math or math skills or computational skills, specifically to biology, for biological outcomes. Is that kind of the mashup of the terms?
Dr. Christina Curtis: That's right. It sort of emerged and actually there's been recent celebrations of sort of the 40-year anniversary of computational biology, it was just a couple weeks ago. And this has really been motivated by the vast amounts of data we can now generate using high-throughput technologies. And so where this comes to play is, what are the methods, the algorithms that we need to dissect this data and then how do we interpret it, really? How do you interpret and distill hundreds of thousands, millions of data points, to uncover biological patterns?
Chris Riback: Yes. With very, very big computers, I would think, or maybe increasingly small, but still very, very powerful. Maybe let's dig into the two big questions that I believe, in reading about you, drive your life's work, which is, why does cancer return and why does cancer spread or metastasize?
Dr. Christina Curtis: Yes, so we're really interested in the origins of cancer and what allows some cancers to be particularly aggressive, and in breast cancer, we know that there's subgroups that we've helped define, subgroups of women who can recur at different points in their trajectory. And we're interested in understanding why there might be lapses in time, why some patients might recur early versus late. And these late recurrences, we can talk more about, but have a characteristic of being dormant. And by dormant, I mean, they could lie silently and they're undetected for many years, and we really want to intercept those tumors before they recur and present at different organ sites in the body because it's of course much harder to treat them at that point.
Chris Riback: That is the part that I would like to get into now and some of that data that you've collected in your studies in these subgroups. But just listening to what you just said. Historically, and maybe it's literally before your discoveries and other colleagues, but a dormant tumor. For example, I know that some of your discovery, which you'll talk about, shows how to identify areas that might have high relapse in as late as 20 years. Before we get kind of specifically to that, historically, a tumor that was to use your word dormant, for some extended period, 5, 10, 20 years, did folks believe historically like, "Oh, I'm cured." And then all of a sudden, 10, 15, 20 years later, whammo, they got hit with something out of the blue. Is that kind of what you were up against?
Dr. Christina Curtis: That's right. So I think our understanding of metastasis is probably still, it's been limited and that's in part because it takes new tools to study that. But yes, I think the challenge is these are patients who think, "Okay, I'm five years out, I might be disease free. I've made it." And then five, as you say, past five, often 10, 15, 20, and beyond, the tumors are emerging again. And of course this is alarming, and we're seeing this at an increased rate to the point that this actually has been recognized as one of the great clinical challenges of our time.
And part of that's of course, because the primary tumors are being treated more effectively. These patients are living much longer, so we see them. So this is also sort of fueled by the many advances that we have achieved in breast cancer, and I want to make that clear. But of course we want to treat patients through the duration of their journey. And I think first knowing that this happens, recognizing it, and then being able to identify those patients, is the next critical step that we're, I think, making a dent in.
Chris Riback: Why is it so important and why is it so revolutionary to be able to identify patients whose tumors express the estrogen receptor, but not the HER2 receptor?
Dr. Christina Curtis: These are the sort of markers that we use routinely in clinical practice. They in fact, guide therapy, of course, for estrogen receptor positive patients, that's been indicative of the use of anti-estrogen therapies that have really, in a way, one of the first targeted therapeutic approaches. And on the other hand, we have HER2-positive disease, we can of course have ER-positive and HER2- positive tumors and those HER2-positive tumors again, also represent an archetype type of precision medicine, where we've just seen tremendous advances in being able to target HER2, starting with the development of Herceptin and now many, many other FDA approved agents.
But the patients that are ER-positive, HER2-negative and have active hormone receptor signaling, the standard of care has been endocrine therapy. And that works very well for many, many patients. And I want to acknowledge, the vast majority of patients are indeed hormone receptor–positive, some 75 to 80 percent. Typically, we actually think of those tumors as being somewhat less aggressive because we can treat many of them well with anti-estrogen therapies. But there is this appreciation that there's a subset of that population, roughly a quarter.
Chris Riback: I was going to say, if 75 percent are in one direction, does that necessarily mean that a quarter are in this other segment, and it sounds like, yes.
Dr. Christina Curtis: Right. To zoom out, about 75 percent, 80 percent are ER-positive, the total breast cancer population. And then within that 75 percent, I would say that there's a 75 percent that tend to respond quite well to therapy, but don't recur so late. And yet, there's been observations that there's about a quarter of the ER-positive subset, that can recur well beyond five years and work from many groups, has shown that these populations exist and that they may not have the completely typical characteristics. These patients might actually have no lymph node involvement, so they look like they're very low risk tumors. And yet, despite having no lymph nodes at the time of diagnosis implicated that are malignant, some of these patients are recurring very late. And so that tells us that our standard approach of just looking at clinical covariates such as tumor size and grade lymph node status, may not tell us everything we need to know about the patients that are recurring and that we really need to understand the biology, the makeup, the genetics of those tumors.
Chris Riback: Were these initial results in 2019?
Dr. Christina Curtis: That's right. So our study came out in 2019 describing the sort of late recurrence, but there had been a meta-analysis of a large number of studies by multiple groups in 2017, showing that we have patients who are recurring late, it's about a quarter of ER-positive, HER2-negatives. But in that study, which was incredibly powerful, there wasn't detailed molecular information. So we couldn't further refine who those patients were, what their biological makeup was and what the 2019 study, where we followed on from our analysis of the METABRIC Cohort, is molecular taxonomy of breast cancer, international consortium. It was a study of 2000 women with very detailed molecular profiling. So the genome and the transcriptome and beyond. So really looking inside the cells at the genetic makeup of the tumor, we were able to follow those patients for many years.
In fact, such that we would have, up to 20 years of clinical follow-up, which linked to the molecular data, then allowed us to go back and ask, "Okay, there's a subset of patients that are recurring late, who are these patients?" And it turned out that really, they mapped entirely onto the subgroups that we had defined years in the past, back in 2012, using an unsupervised approach. And by unsupervised, I just mean that we really let the biology speak to us. We weren't trying to prescribe how we define subgroups. We asked how many subgroups there were based on the molecular features. And so that was a big kind of insight into what the molecular features of these high-risk patients might be.
Chris Riback: And am I understanding correctly, the data that you're examining and the patients, 20 years looking at this, this is the centralized data that the consortium was able to pull together, and then groups like yours, is able to access that?
Dr. Christina Curtis: Right. So I was the first author of the first study that led that analysis back in 2012 that combined these data sets, it was a consortium effort, a major effort across five hospitals, led, at the University of Cambridge and British Columbia, Vancouver. And there were hospitals from multiple sites. So it was this collective effort that really allowed us to amass enough samples, enough data points. And at the time when we did the deep molecular characterization, it remains really the largest study of its kind, with this kind of molecular information and outcome. And we've made it available to the public. It's been used broadly by breast cancer researchers around the world. And so I'm really pleased that we were able to allow others to build on that. And then we went back and built on it when we were able to get additional follow-up information for these patients. We've been following them and following them. But of course, then there's a collation effort to say, "Aha, we've got the follow-up time we need to go back and do these statistical analyses."
Chris Riback: Yes. I've spoken, had the privilege to speak with other researchers who dig into some of that shared data, it may be other consortia, but the power that comes from people like you identifying, collecting that data, but then pooling it and finding ways to make it shareable, it's become clear to me as a layperson, how important that is. I wonder, is now the time when I should ask you what you think about scientists who studied at the University of Cambridge?
Dr. Christina Curtis: Whenever, I have warm, warm memories and had a brilliant time there. It's such an intellectually stimulating environment, so yes, always happy to talk about different environments.
Chris Riback: Yes. It must be fantastic there. What other subgroups did you identify and why was identifying those so significant?
Dr. Christina Curtis: Sure. We had originally described, there's 11 integrative subtypes, is how we refer to them and they're integrative because what we did in this first study that was quite novel, was to combine not only transcriptome information, which is how the genes are expressed in our cells, but also they're copy number state. And by copy number, is a particular gene amplified, are there are multiple copies in a cell. And typically, genes that drive cancer, might be represented in more copies, than in a normal cell. And it fuels this kind of overdrive of a particular signal in cascade.
By bringing in the copy of HER2 information, which hadn't been done before, we found that there's a number of subgroups that actually look a lot like HER2-positive breast cancer, with different genomic drivers, and HER2 is really an archetype of, that gene can be amplified and over expressed many, many times, and we target that ONCA gene. But as a field, kind of as a whole, we haven't really followed up on targeting copy number drivers. We go after mutations. And so that's a bit nuanced, but what we really defined here was new subgroups that we think might be targetable in a similar fashion to HER2.
And that, knowing how transformative that has been for the field, I think that gives us a lot of optimism. And it happens to be that there are four subgroups amongst these ER-positive, HER2 negative patients, that account for this 25 percent that are at high risk. There's four different groups. They have different molecular aberrations or mutations. And so really the interest comes in, how do we go about targeting those specific subgroups in a way that we can ultimately, circumvent relapse. And when a relapse has occurred, perhaps treat it more effectively. And so, the notion is that perhaps, endocrine therapy alone isn't enough for these patients. And there's of course, other agents in development as well, and a lot of exciting work in the past few years to bring forward new therapies, but that's really the premise.
Chris Riback: Which leads, I think, to an area that I'm certain, that you spend a lot of time thinking about: personalized breast cancer treatment and risk prediction. From your point of view, why is personalization of care and prediction, both so important and so hard, and where do you see personalization going?
Dr. Christina Curtis: Yes, that's a great question. I think this is really where we need to move to and where there's already been a lot of developments, but we can certainly do more. I think personalization’s so critical because we want to make sure that we're delivering the right therapy to the right patient at the right time, also in the right dose. And that we're sparing them any unnecessary toxicity, and that's a whole other area of focus within my group is, when can we deescalate or reduce the therapy that we're giving, perhaps chemotherapy in favor of a targeted agent and when do we need to escalate? And this is an example of escalation where we think we need more than just endocrine therapy to prevent these recurrences. And part of that may be because such patients are intrinsically resistant to those endocrine therapy. When we think about personalization, we really want to take as much information into account as possible, about the likelihood that this tumor is going to be an aggressive one and the likelihood that it would respond to particular agents.
We want to be predictive with respect to the treatment response as well. And we know that molecular features can influence the vulnerabilities of those tumor cells, so we want to go after them and target the tumor cells and hopefully not the rest of the cells, want to really mitigate that. There's a lot that can be done with this information. It does require a lot of data and it requires these kinds of computational modeling approaches to say, "How do we categorize patients? How do we group them together? And when do we need more refinement in that grouping?" And there's multiple layers to that.
Chris Riback: You started to segue then into my next question, which is, what's next regarding your study, or were the areas that you were just talking about, not the areas that you plan to attack next?
Dr. Christina Curtis: No, they absolutely are. We think that this framework that we've described, really provides a lot of potential for how we can tailor these therapies because it turns out, I think, what's really exciting about this discovery is that, it's not just that we can define which patients are at risk, we can do that. But the very fact that we can group them actually the way that we've grouped them, uncovers novel vulnerabilities, that we can then target potentially therapeutically. And so of course, bringing new therapies forward is always a long road. There's a lot of testing to make sure that these are safe and efficacious for the population. And we've been doing a lot of experimental work in the laboratory to sort of evaluate these therapeutic strategies and then to design new clinical trials.
And we're really, I think, excited about the fact that these new biomarkers that we've developed predict risk, potentially predict benefit from particular therapies. And so we're looking to essentially deliver trials to early stage patients, where we can really understand the unique biology and doing trials in the early stage setting, can be more challenging in some set. We often start in the more advanced setting and move drugs back up the pipeline, and sometimes that works and sometimes it doesn't work. And our rationale around this is that we really need to understand that unique biology. And if we can start by looking how these therapeutic interventions work in these high risk populations, we can learn a lot about how to do better for these patients. And so we're embarking on clinical trials to do just that, to do biomarker stratification.
And these are ambitious in many respects, because we're doing very comprehensive molecular profiling up front. But I think that this also sets the stage for the wave of the future and how we can personalize therapy at the very beginning. And the reason I think that ultimately is so important and of course there's great standard of care options, so that's where the challenge is, we've come so far in our therapies for breast cancer patients. So to do better, the bar is very high. And I want to make that clear. The bar is very, very high. But I think that if we can do more sooner and set patients on the right course for the beginning, with therapies that are tailored to their risk and we can monitor them, because we also need to follow them. It's not, as I said, some of these patients may recur later and we want to continuously track that risk, then I think we will be achieving better outcomes. And we don't want to wait to just the end to do it when it's later than we would like, and the efficacy may be lower.
Chris Riback: Tell me about you. How did you get into this? Going back, where did you grow up? Was it always science and math for you? Did you ever think perhaps, fiction novelist, world class skier. How'd you get into this?
Dr. Christina Curtis: Yes, well, I still have ambitions to be a world class skier. I don't know. I think my kids have surpassed me by now.
Chris Riback: Probably, yes.
Dr. Christina Curtis: Yes. It was always science. It was always this from the time I was really a high school student, I determined I would do a PhD in genetics and to focus on cancer, but I wanted to understand the genetic basis of cancer. And of course, I didn't know what tools I would need to do that. That was predated, the sort of genomic revolution that we have now lived through. But, a family history of cancer and I thought, "This is encoded in our cells. I need to get to the bottom of this." And it's really been the driving force for everything I've done. And I feel so fortunate that that passion was sparked really early, because in some sense, it made my path really clear. And as these technologies emerge to allow us to probe hundreds, millions of molecules simultaneously, it suddenly was also obvious that we needed computational tools to interpret this.
So in many ways I feel like I was in the right place at the right time, where these programs to train people in this new field that was kind of largely unrecognized, a merger of folks from computer science and math and biology, to really get to train in that space, I think has been just a huge privilege. And now I think it's hard to even think about some of the questions we address in biology without some of those tools, so it's certainly changed, but I do feel really fortunate. And I think, this is just such an amazing time to be doing this kind of work as well. The pace at which discoveries are happening and clinical translation is happening, that I think we can really move the needle so much faster, and I tell that to my trainees, that this is the golden era to be doing this.
Chris Riback: And incredible to me that you were able to develop such a passion in high school. That was pre-CRISPR. There were aspects from the human genome project, obviously that had come to light. And so they were in society to a certain extent. But pre a lot of the exciting, more popular innovations that many of the rest of us have heard about, and yet even back in high school, you were finding it inspiring. That's pretty terrific. What about BCRF? What role has BCRF played in your research?
Dr. Christina Curtis: BCRF has been instrumental. They funded me early on when I was just getting going as an assistant professor and taking my first ideas to the lab, some of which were bold and probably would not have been funded by conventional means. And the funding allowed us to get some of our first federal funding, federal awards, but I absolutely would not be doing the work that I'm doing today, and I think that I wouldn't be as close to translating it to the clinic, which is what I find to be of immense reward. I always hoped when I started doing this back, even early on that this would, some of the discoveries we would make would eventually reach patients. You never know if that will happen, how long that will take and some of it's serendipity and some of it is, we're starting with very basic fundamental questions that may not have an immediate path.
But I would say some of the most rewarding things that I'm pursuing right now is how do we deliver on this new knowledge and how do we deliver that to patients? And though I'm not a clinician, I really enjoy being able to see how this can move that needle. And yes, I'm kind of hooked on it. I think now, passions emerge along the way and it's like, "How do we do better? And how do we do better, faster? How do we accelerate this?"
Chris Riback: That seems to be a theme of yours. Just in listening to your answer, how do you balance bold versus practical, whatever the opposite of bold is? How do you balance that both in how you run your lab, how you think about the problems that you want to attack? What's your philosophy on that?
Dr. Christina Curtis: Yes. I guess part of this is I think that couple fold, one is we need to be multidisciplinary in our approach and my lab certainly is that. We are using computational techniques. We're using experimental techniques. I don't think we can fully move this needle without doing that. And we work very closely with clinicians to make sure that we're really thinking about this from where is the biggest bang for buck for the patient? How do we move that? But I also feel like, I don't know, some sense because I also trained in these other fields, I was a bit of an outsider, and so sometimes my perspective that come in with is different.
And I just am sort of asking these questions, "Well, why do we think this works this way? Why is that? Do we know that it works that way?" And maybe I'm a bit of a skeptic, but as a result, you don't always know when you start asking these questions, that what you're doing might challenge dogma. You don't know, you get going and then you realize, "Oh, this was the status quo," and we are seeing this, and this is what we see, and I'm a big believer that we have to let the data speak to us. And sometimes that leads you in bold places and new places.
But I think I'm not really afraid to take risks because I think that if we want to get there faster, we have to do some risky things. And there's great value in continuing to build on what our knowledge is. But sometimes I think we have to look at it through a different lens and see what's there and accept that occasionally, that maybe more often than not, that will be a dead end. But sometimes we'll find things that allow us to really understand the biology differently and maybe treat patients differently as a result.
Chris Riback: And I'm certain that the team members in your lab are thrilled to get the question from Dr. Curtis, how do we know that? I'm sure they don't panic at all when they get that question, do they?
Dr. Christina Curtis: We have a lot of fun in lab and I'm always learning from them. I think that's the amazing thing is the great thing about this job, the best job in the world is that, you never stop learning. And it does mean that it can be all consuming at times, but it is a great privilege.
Chris Riback: You're all fighting this and trying to get these answers for patients and for people together, and we thank you for that. Dr. Curtis, thank you. Thank you for your time. Thank you for the work that you and your team do every day.
Dr. Christina Curtis: Thank you so much. It's a privilege and yes, I really enjoyed our discussion.
When you give to BCRF, you're funding critical hours in the lab. More time for research means longer, healthier lives for the ones we love.