Polygenic Risk Score: Validation of a Breast Cancer Risk Prediction Model in a Large Prospective Cohort Thomas Slavin, MD
When we think about polygenic risk score, it is fairly new. We are doing a lot of the innovation at Myriad. We’ve been doing a lot to restructure the company focused down on women’s health, oncology, mental We’re thinking very differently. We’re trying to be more collaborative, unique access and equity advancements and continue to innovate really heavily in that area. And this is really one of those examples where we saw amazing science going on in the field that had been really going on for the last few decades.
I actually did my post-doctoral research in genome-wide association studies. So this has been, ripening over. And if you think of BRACA1 and 2 in the hereditary context, as I, I think of that as version one of all of this in hereditary cancer, version , I think, of multi-gene panels. So where we started going beyond, just a few really strong high penetrant genes and bringing in some of the moderate penetrant genes to take care of patients. And this is really version three where, Everybody can get a genomically informed result. So we launched this in 2017 in breast cancer. And it’s been wildly successful. We’ve tested at this point hundreds of thousands of women. So this is very mainstream. However, over the last. Few years we’ve had to actually make a lot of advancements to make it equitable for all. Because the original data on adding in these little genomic factors and adding them up to establish risk stratification was really built off European populations. So we had to do a lot of work, which really led to, what we’re seeing here at San Antonio to adapt that information and make it a available for women of all ancestors.
What is the standard of care in this disease state, and why choose to pursue this trial?
So the standard of care is really either no risk assessment, sadly, I mean there is recommendations to do risk assessment for women with breast cancer. And there’s different tools and models out there to take in clinical and family history variables. One of the most used tools. In the United States really the population that we focus risk score on is the Tyrer-Cuzik model. It’s a clinical and family history tool that is freely available online where people can go and put in information. We license that model and actually added in then our polygenic risk score component on top of that. So now with every my wrist test for hereditary cancer, you get the 48. That are, like BRACA1 and two that say, do you have a gene mutation? But that’s only about 5% of people. So the other 95% of unaffected women say they’re wondering about their breast cancer risk, they would otherwise have just gotten a negative result. However, now with risk score, we take in all those background genetic factors, we put them together into a model and we’re able to add that personalized risk assessment. And it’s really an advancement on the field.
Can you please tell us about the trial design and why it was set up this way?
This is, of the studies that we’ve done to date the most impactful bar none, because what we actually did was say, okay, we’ve had hundreds of thousands of women who have had this test. Let’s actually just follow them over time. And look at medical claims de-identified matching data and see if they actually got breast cancer or not, and let’s compare that to if we just gave them a Tyrer-Cuzik risk estimate. So just on the clinical and family history variables and not adding in the genomic component. And what we found was, we followed about 130,000 women. Then comprised over 150,000 patient years and we saw about 340 incident breast cancers over that time span. And really the observed over expected for the Tyrer-Cuzik model plus the genomic factors, which we call risk score commercially really had almost a perfect fit, it looked fantastic. The observed over expected ratio when we looked, however, at the Tyrer-Cuzik, when you don’t add the genomic factors in we see a drop off, particularly in the higher risk individuals where it was overestimating risk pretty substantially. So you do need that genetic component to really write the model for Tyrer-Cuzik. It’s really hopefully great real world data for the field and people can be much more comfortable that what we say on the test is accurately predicting what’s going on.
Can you give us significant data from this trial?
So when I look at the data, the things that jump out to me the best are the observed over expected ratio, which is almost one for risk guard. I So it really was a fantastic predictor of predicting that a woman was, actually gonna develop breast cancer and particularly compared to Tyrer-Cuzik. So that was a big one, the other main one that really comes out and we have a figure in the poster that people can see, and we’re actually working to put all this together for publication. But when you look at women that got a high risk score result, verse, those that got a low risk score result the odds of them getting breast cancer are about four times higher. So it’s very significant when you’re trying to take care of a woman with breast cancer. Also, when we looked at women that had risk score and look at their Tyrer-Cuzik, at the same time we actually see that their development of breast cancer was much more consistent with the risk score than it was with the Tyrer-Cuzik model over a five year time period.
What are the most common questions you get from your colleagues about this study?
Unlike BRACA1 and 2 risk score is a little bit more of a black box. What I do in medical affairs is really, the head of medical affairs in the company is really try to break down that black box polygenic risk are sometimes things in genes, and that’s actually new to a lot of people. We think of these sometimes as these single nucleotype polymorphisms that are just all over the genome, but actually when you actually look what’s in our risk score and the single nucleotype polymorphisms, not all are actually single nucleotide polymorphisms. Some of the ones with the highest odds ratio in there are actually single nucleotide variants in genes like CHEK2 and BRCA2 and ESR1. So very important genes that you really do need to take into account. And one example of that is actually a paper, a few years ago where we looked at the Women’s Health Initiative data and we looked at just Tyrer-Cuzik alone across different ancestors. This was in prep work for the risk score work that we did, and what we saw was Tyrer-Cuzik alone actually works pretty well to predict, that women’s health initiative really followed women over time. And if you look at their Tyrer-Cuzik at one point in time and say, okay, did they actually develop breast cancer or not? You can actually answer that question in that prospective cohort. What we saw was particularly in Hispanic individuals, it didn’t really work that well, and that’s where the uniqueness of bringing in the genetic factors is really needed because for instance, there’s a lot of individuals of Latin American ancestry that carry a ESR1 single nucleotide variant that’s actually really protective and it keeps them from getting breast cancer. So that’s a core component and really helps write the model. When we look at risk score particularly in the Hispanic population.
Watch and Share the Video Here: https://oncologytube.com/v/41905
But where are their primary endpoints of this trial and where they met?
We were really looking at the endpoint of breast cancer. Did the risk or model itself predict breast cancer like we thought it would? And the answer was absolutely overwhelmingly yes. We were actually honestly all looking at the data very. Press when we first saw it come out of the work that we were doing on the claims database. It really holds up very strongly, and, being a clinician, I’ve also, and talking with other clinicians I think there is an understanding in the field that a lot of people think that Tyrer-Cuzik may overestimate, particularly on the higher risk end of things. And what we see is, When we dive deeper into the data, that when you add that polygenic risk component, it actually decreases a lot of people’s lifetime risk around a 20% threshold rather than actually increasing which helps adjust for that overestimate sometimes from Tyrer-Cuzik.
What are the key takeaways from this research and data?
The main takeaways from this are, this is primetime, this is ready, it’s out there in clinic. We’re testing patients and it’s best for patient care as a model. Compared to clinical and family history models, for instance, like Tyrer-Cuzik that do not bring in a genomic component. So I am a strong advocate looking at all the data, particularly, and not just from our group, but also data, amongst other groups over the last particularly last 5 years or so. It is very clear that if done well, a polygenic risk score can be a very strong risk stratifier. Now, when I say if done well, you do need to. For those nuances in ancestry because you can’t take a European model and apply it directly to non-European ancestry. So we just had another paper in JCO Precision Oncology that really showed how we developed our model using ancestry and. Formative markers in the background to really help weight each breast cancer snip, and it’s really a transformative game-changing model for the field that can be applied beyond breast cancer. Really to any sort of multifactorial disease you’re gonna be looking at with polygenic risk.
What is the Polygenic Risk Score in Breast Cancer?
The Polygenic Risk Score (PRS) is a genetic test that determines an individual’s breast cancer risk based on the combination of several genetic variations. Even in the absence of known high-risk mutations such as BRCA1 and BRCA2, the PRS (Polygenic Risk Score) test can assist physicians in identifying patients at elevated risk of getting breast cancer.
For breast cancer treatment, the PRS (Polygenic Risk Score) test can be utilized to guide decisions regarding preventative treatments, such as increased screening or prophylactic surgery. Those with a high PRS (Polygenic Risk Score) may be offered more frequent breast cancer screening or preventive surgery, such as mastectomy or oophorectomy.
In addition, the PRS (Polygenic Risk Score) test may help breast cancer patients identify the most appropriate treatment options. To lower the likelihood of recurrence, a patient with a high PRS (Polygenic Risk Score) may benefit from more aggressive treatment, such as chemotherapy. A patient with a low Polygenic Risk Score (PRS) may be able to avoid unneeded therapy and its associated negative effects.
Overall, the PRS (Polygenic Risk Score) test provides doctors with significant information that may improve patient outcomes and quality of life by allowing them to personalize breast cancer therapy and prevention methods to each patient’s risk level.
What is the Tyrer-Cuzik model?
The Tyrer-Cuzick model, also known as the IBIS model, is a breast cancer risk assessment instrument that evaluates a woman’s lifetime risk of acquiring breast cancer. The model takes into consideration a range of risk variables for breast cancer, including age, breast cancer in the family, reproductive history, and breast density.
The Tyrer-Cuzick model is intended to provide a more thorough risk assessment than other models, such as the Gail model, which only takes a limited number of risk factors into account. In addition, the Tyrer-Cuzick model accounts for the presence of certain genetic abnormalities, such as BRCA1 and BRCA2, which can substantially increase the risk of breast cancer.
The model estimates a woman’s 10-year and lifetime risk of acquiring breast cancer based on her unique risk factors. This information can aid in breast cancer screening and prevention decisions, such as the use of chemoprevention and prophylactic surgery.
Overall, the Tyrer-Cuzick model is a useful tool for measuring breast cancer risk and can assist women and their healthcare providers in making informed decisions regarding their breast health.
Thomas Slavin, MD – About The Author, Credentials, and Affiliations
Thomas Slavin, MD, is a board-certified doctor and medical geneticist who works at Myriad Genetics, a leading molecular diagnostic company that specializes in genetic testing and personalized medicine. He received his medical degree from the University of California, San Francisco School of Medicine, and completed his residency in internal medicine at Brigham and Women’s Hospital. Dr. Slavin then went to the Harvard Medical School Genetics Training Program for a fellowship in medical genetics.
Dr. Slavin has a lot of clinical and research experience in medical genetics. He has also written and spoken about a wide range of genetic disorders in many peer-reviewed publications and presentations. At Myriad Genetics, he is an expert at interpreting and analyzing complex genetic data, such as next-generation sequencing, chromosomal microarray analysis, and other genetic testing technologies.
Dr. Slavin is committed to giving his patients personalized care, and he works closely with other health care providers to make sure that genetic testing and counseling services are of the highest quality. He is committed to improving patient care and moving the field of medical genetics forward through ongoing research and education.