Prostate Cancer Modeling: The CISNET Prostate Group - Ruth Etzioni

May 9, 2021

Ruth Etzioni, PhD joins Matthew Cooperberg, MD MPH in a discussion about the role and history of prostate cancer modeling based upon PSA.  Cancer Intervention and Surveillance Modeling Network (CISNET) is a consortium of NCI-sponsored investigators who use simulation modeling to improve our understanding of cancer control interventions in prevention, screening, and treatment and their effects on population trends in incidence and mortality. These models can be used to guide public health research and priorities, and they can aid in the development of optimal cancer control strategies. The CISNET Prostate Working Group (PWG) has focused on modeling prostate cancer progression, screening, and treatment in the U.S. population, assuming biologically plausible assumptions validated against various data sources. The most effective ways to personalize the prevention and treatment of prostate cancer using data from an individual patient are unknown. The PWG applies a unique interdisciplinary approach that combines available data with modern statistical techniques to understand the long-term effects of prostate cancer interventions. In the current round, the PWG will evaluate personalizing patient care, including screening high-risk men more frequently, using biomarkers and imaging tests to select patients for biopsy, using patient and cancer features to determine when and which kinds of treatment to offer, and practical approaches for reducing racial disparities. This work will advance the evidence necessary to make informed decisions about individualized screening and treatment for this most common cancer in men. The specific aims are:
1. Precision early detection, including risk-stratified screening and biopsy using genetic tests, novel biomarkers, and imaging technology
2. Precision active surveillance, including adaptive biopsy intervals and imaging technology
3. Precision treatment, including type and timing of initial and salvage therapies.
4. Targeting screening, biopsy, and treatment policies to reduce racial disparities
5. Prioritizing screening and treatment interventions in international settings.


Ruth Etzioni, Ph.D., Professor, Public Health Sciences Division, Rosalie and Harold Rea Brown Endowed Chair, Fred Hutchinson Cancer Research Center, Seattle, WA

Matthew Cooperberg, MD, MPH, FACS, Professor of Urology; Epidemiology & Biostatistics, Helen Diller Family Chair in Urology, The University of California, San Francisco, UCSF

Read the Full Video Transcript

Matthew Cooperberg: Morning. I'm Matt Cooperberg from the Department of Urology and Epidemiology Biostatistics at the University of California, San Francisco. Welcome to another in our series of localized prostate cancer-focused interviews for the UroToday Center of Excellence. It is a real pleasure today to welcome Ruth Etzioni, who is the Rosalie and Harold Rea Brown Endowed Chair in Public Health Sciences at Fred Hutch Institute in Seattle. I've had the real honor and privilege of working with Ruth for a number of years through the CISNET Prostate Group, and I think it's safe to say there are very few people who have had as much influence on the question of PSA screening and what we really do about prostate cancer at the public health level than Ruth. So, really looking forward to the conversation, and Ruth, welcome.

Ruth Etzioni: Thank you. And back at you. It's been great working with you as well. You're my walking encyclopedia actually.

Matthew Cooperberg: So I'd love to honestly just start with a little bit of background for the audience. Tell us a little bit about CISNET and tell us about your journey on this path to where we are today.

Ruth Etzioni: Yeah, I'd love to. Well, I actually began my career in ovarian cancer screening because I was very interested in biomarkers. I realized that this was a new frontier. Until the ovarian cancer biomarker, we'd really only had tests that were positive or negative. Ovarian cancer is very rare though, and I'm a modeler, and modelers need data. And I realized that there was another biomarker that was rapidly approaching. It was not on the horizon anymore, which was PSA. And so I started modeling prostate cancer. And we'll talk a little more about what that means.

But right about that time, I got a phone call from the National Cancer Institute. I'd been doing a little modeling in prostate cancer. It was 1997, and they had just gotten the 1994 SEER data. And the 1994 data showed that prostate cancer mortality had been declining for two years. PSA was now being used in the population. Everyone was asking whether that decline could have been due to screening.

So I started working with the National Cancer Institute on that specific question, and it then expanded to other cancers. This question of, why are cancer trends doing what they're doing? Breast cancer mortality was going down too. Was it screening? Was it treatment? It then expanded to lung cancer. Was it smoking? And so we now have a consortium of six cancers where the focus was to try and explain trends, but we ended up going into many more questions that we could answer with modeling.

Matthew Cooperberg: And what is your process in the prostate group? You've got three different centers with different models. How does it all kind of come together?

Ruth Etzioni: Yeah, it's the same in all the groups, really. The idea is that you have multiple groups. We actually just added a fourth group that has been focused on risk stratification, looking at targeted screening for different risk groups. But the idea is that the groups together tackle the same questions, but they've developed their models independently. And the idea is to get a measure of validity and consistency across different models, addressing the same question, so that we really can give the field a reliable answer.

The models don't always agree, and then investigating why is educational in its own right. And so we learn from each other. It's been a very educational and productive partnership, and it's been going for 20 years.

Matthew Cooperberg: How's the new site, aside from Michigan and Erasmus?

Ruth Etzioni: Yes, Michigan and Erasmus have always been in there. The new site is from University College London. The principal investigator is Nora Pashayan. And as you know, in the UK, there's a very robust group of polygenic risk scholars and investigators. And so she's been working with them.

Matthew Cooperberg: So what are the most important lessons you think that you've learned over the last 20 years of working on this always interesting and never settled question about PSA screening?

Ruth Etzioni: Yeah. So I love that question, and you have to stop me if I talk too long, but I think there are three main ones. The first one is that you cannot always hang your hat on trials, particularly when it comes to screening trials. Screening trials are so complex and they take so long to do. I could spend the rest of the time just talking about trials. So that's the first thing.

You can't hang your hat on just the empirical results of trials, but you can learn a lot from trials, about the disease, about how it progresses, and about how screening might generate a mortality benefit. So that's the first one.

I think the second one that we've learned, and I think we've learned that very early, is that, in the case of prostate cancer, it's definitely possible to screen too much. And what does that mean for our recommendations in practice?

And the third one is I'll say, and maybe it's a little self-serving, modeling can be very useful in going beyond the limited empirical data to understand the dynamics of disease. And then to take that forward in virtual trials or in silico studies to address questions that normally would need a trial, but a trial simply could not be done. For example, the use of new biomarkers, and are they to be useful or not, or should we be doing targeted screening? Should we stop screening at certain ages? We just can't do trials to address all of those. But if we understand enough about disease and how it progresses, we can generate evidence to address those questions via modeling.

Matthew Cooperberg: Do you see a path to get the US Preventive Services Task Force to engage a little bit more directly with modeling groups and incorporate modeling results a little bit more formally into the process?

Ruth Etzioni: Well, they have. I mean, I can't fathom why they won't talk to us in prostate because they have worked with CISNET. They've worked with the breast group, the cervical group, and the lung group, and the colorectal group. So why are they not engaging with the prostate group? And I think it's become very clear that the US Preventive Services Task Force has this notion of level one evidence from trials. It's become very clear that the prostate trials are only telling part of the story.

The closest they've come is, they did an evidence review for their last prostate cancer guideline. And they did a review of models, and they consulted us for that, but they don't want to use the models. And I think it's because they only are going to go forward if trials give them a significant result.

Anyway, I'm getting circular here. So maybe one of them will be listening and we can take it forward. But definitely over the years, they've become much more willing to look at models and they have engaged deeply with other CISNET groups.

Matthew Cooperberg: What do you think we've learned from the trials? You guys did a really nice study trying to reconcile the PLCO and The European Randomized Study of Screening for Prostate Cancer (ERSPC).  What do you think the final take-home message is from the two or three big trials if you include [inaudible] as a separate one.

Ruth Etzioni: Yeah. I think that what we knew from the trials we knew before the models, which was that the US trial was not going to be informative about benefit. The comparison was not between screening and no screening. The comparison was between a lot of screening and almost as much screening. So we already knew that. And we did quite an intensive analysis of the trials to tell the quantitative side of this well-known non-quantitative story about the US trial.

So I think there have been concerns voiced about the European trial from the US side. But I think it's quite clear that if you take the trials and you do an analysis that really amounts to screening versus no screening comparison, that there is a statistically and clinically significant benefit of screening that increases with the earliness of detection and is incontrovertible.

Matthew Cooperberg: So what else do we need in terms of evidence? Do you think the Malmö study and the HPFS and the Southern Cohort studies of early baseline testing add to our understanding of how to optimize screening? Or do you think we need new studies and not necessarily new trials? What are the missing pieces and how do we fill them in?

Ruth Etzioni: Well, there's one piece that's not missing from those studies that you didn't mention, which was the stopping screening. The very clear evidence that leads to a recommendation to stop screening between age 60 and 70 for a low-baseline PSA. And I think that we don't need a trial for that. I think it's very clear because we've had 30 years of learning from PSA screening. If you have a very low PSA, the chance that you will survive for the PSA to grow and disease to progress to a life-threatening state is absolutely minimal. So that could prevent us from screening a large swath of older men. And that's where the problems of over-diagnosis and over intervention come in.

What else do we need to know? I think it's not so much what we need to know, but what we need our providers and patients to know. We need them to understand that there is such a thing as too much screening, that many prostate cancers don't need to be treated. And that if they're willing not to treat, they should definitely be screening, because it could save their lives.

Matthew Cooperberg: So getting onto that topic of active surveillance, of course, you started to work a little more in the last few years on what we do with prostate cancer after diagnosis. What unique insights has your group brought in terms of reconciling the different surveillance cohorts and where we need to be going in terms of managing these men with low-risk disease that we find through screening?

Ruth Etzioni: So I think that different surveillance cohorts are for us a very interesting question. They tend to give different results. And this has led us into really a whole new area in my group of trying to reconcile these kinds of data, because differences can be due to underlying differences in risks, or they can simply be due to differences in surveillance and compliance with surveillance and biopsy recommendations in the cohorts.

So I think the first thing, again, is we have to go beyond the empirical data. This is why modeling is so useful to understand what the real underlying risk of an adverse outcome is. And I think that our findings are similar to others, which is that we don't need to do surveillance very often. That annual biopsies are just not necessary, it's very taxing for the patient.

What we haven't done enough of yet, which the data are evolving, is looking at MRI, looking at alternatives to the invasive biopsy procedure, to make surveillance more acceptable and more feasible to men. Because there's a lot of dropout of surveillance, which obviously counteracts the intent. If we can make surveillance work better for men. And within our group, we have really one of the real-world leaders working on surveillance data to try to develop recommendations that are targeted to men based on their history. So depending on your history, you may be able to wait three years for a biopsy, or depending on your history, you may need to come in for your scheduled biopsy. And I think that that could also be helpful if men are willing, again, to accept a black box in their medical record recommendation.

Matthew Cooperberg: And the last topic that I want to talk about is this question of prostate cancer disparity, and there are all these throwaway statements in the task force guideline that we need more trials in African-American men, which of course are never going to happen. So how do we address that problem? How do you fill in the missing pieces where African-American men have been underrepresented in not only the trials but most studies, to try to make cogent recommendations about whether we should be screening differently or whether we should be managing differently in men who are at higher risk? How are we going to go forward in this area?

Ruth Etzioni: I'm so happy that you've asked about this because we have never been more acutely aware of how social determinants of health are hurting our minorities as this time of COVID. And prostate cancer is the cancer with the greatest racial disparity. I'm talking about the disparity between Black men and the general population, between Black men and white men, where white men have such a much higher risk of diagnosis and an even higher risk of mortality relative to the general population.

So you're right. We're not going to be able to do trials. So we've taken an approach to this that we've taken just generally, to try to understand the disease process. How was it different in Black men? And we've come to the conclusion that the increase in incidence is very consistent with just more disease in this population. And that higher mortality relative to incidence is related to some difference in treatment or surveillance. And we don't understand that well enough yet.

Black men have as much screening as white men. They also have similar biopsy rates. So it's not in the detection. It may be partly in early treatment. It may be partly in surveillance and later treatment, for which we are making great efforts to get data because that is just not as well understood as it needs to be, what decisions Black men are taking later relative to white men, and whether they're getting the surveillance that they need.

So I think that once we understand that there's this difference between Black men, it becomes clear that if we're going to try to close the gap on mortality, we have to be doing more for Black men. And we have an editorial just out in JNCI calling on guideline panels to stop asking for more trials and recognize that with our work and the work of others, we actually have enough to know what to recommend for this population.

And really this brings me back to a point that I made earlier about modeling, that it really helps to fill out the evidence-base beyond what we're seeing in trials. It brings together everything that we've seen. We've had 30 years of PSA experience and gathering of data, and that data used judiciously can tell us a lot more than trials can, and can help to create an evidence base for informed decision-making, both at the individual and the policy level.

Matthew Cooperberg: What does that tell us today? Your group had a paper a couple of years ago, suggesting another five-year earlier latency onset for African-American men. I mean, should we start screening five years earlier? Is it as simple as that? Or are there more nuances?

Ruth Etzioni: There's another nuance. We definitely recommend starting earlier, five to 10 years earlier. But the issue with having a higher burden of incidence means that later, in older men, where over-diagnosis happens most, it'll happen more for Black men. And so we really have to be thinking on both ends of the age range. So start earlier, but then make sure that you have some safeguards in place later, either not screening until elderly ages, or just having a conservative criterion for biopsy among older men, so that you can balance harm and benefit.

But we do make a concrete recommendation and we're going to continue to do work that speaks to that to try to move the ticker on the guideline so that there's really an explicit guideline for Black men.

Matthew Cooperberg: Terrific. I think this will have a huge impact going forward, and hopefully, the policymakers will listen on this topic and many others to your group. So, any other concluding thoughts you want to share with the listeners?

Ruth Etzioni: I'm so glad to have this opportunity to talk with you and to distill all these years of work into 10 minutes. It's a great pleasure. And we continue to keep our eye on evolving practices because what happens is that new technologies diffuse into the population before they're able to be tested, and we have to watch how they're diffusing and be sure that we're gathering the data that we need so that we can help to guide recommendations. And I'm thinking right now about imaging technologies that I think are going to change the face of prostate cancer diagnosis, potentially active surveillance and diagnosis of metastasis. And these kinds of things create artifact. And people might think, for example, that we're diagnosing more late-stage prostate cancers because we have better imaging, but really we're just able to find them. And we need to keep making sure that the field is getting the numbers right. And that's really our mission in my group, is making sure that we get the numbers right.

Matthew Cooperberg: Terrific. Could not agree more. I will just conclude by saying thank you so much for all the work you've done over all the years. I personally hope to continue to keep working with you over many years to come.

Matthew Cooperberg: And thanks for sharing a few minutes with us today.

Ruth Etzioni: Same here. Thank you so much, Matt.

Matthew Cooperberg: Hope I'll have you back on soon.