Understanding patterns in prostate genomic testing using SEER and GPS assay

In a recent study presented at the 2021 AUA Annual Meeting, Scarlett Gomez, MPH, PhD, and co-authors investigated the determinants of selecting active surveillance/watchful waiting as a treatment path for patients with localized, low-risk prostate cancer.

As precision medicine becomes more abundant in today’s health care system, clinicians find it important to perform genomic testing to move forward with the treatment decision process.

In a recent study presented at the 2021 American Urological Association Annual Meeting,1 Scarlett Gomez, MPH, PhD, and co-authors investigated the determinants of selecting active surveillance/watchful waiting as a treatment path for patients with localized, low-risk prostate cancer. They linked data from the National Cancer Institute’s (NCI’s) Surveillance Epidemiology, and End Results (SEER) program with the Oncotype DX Genomic Prostate Score (GPS) assay. Gomez is a professor in the department of epidemiology and biostatistics at the University of California, San Francisco.

Please discuss the background for this study.

In terms of the background for the study, we know and appreciate that we are in the era of precision medicine. We're increasingly seeing the addition of molecular and genetic tests to guide us in terms of the cancer treatment that we provide our patients. This trend really motivates us to better understand who's getting these tests and whether there are potential gaps in terms of who's receiving them out in the real world. To address that larger question, we undertook a project, which is a public-private collaboration between Exact Sciences Corporation, the NCI's SEER program, as well as the Greater Bay Area Cancer Registry, which I represent.

As some background, NCI SEER is the population-based cancer registry that compiles cancer incidence, initial treatment, and survival data. What's notable about this resource is that it covers 35% of the US population, and it's considered population-based in that it covers within that catchment area, everybody who's been diagnosed with cancer. Exact Sciences produces the Oncotype DX genomic prostate score, or GPS score. This assay measures the expression levels of 17 genes, 12 of which are cancer related and then 5 reference genes taken from prostate biopsies. This is used to help guide treatment decisions at the time of diagnosis.

And so, what we did in this unique collaboration was that we linked the results from this prostate GPS assay with SEER cancer registry data. This is our very first report using this novel linked database. We looked descriptively at data of over 120,000 men diagnosed with prostate cancer. Specifically, we were interested to look at the factors including, in particular, the role of receiving a GPS test and the results of the test that may be associated with selecting active surveillance or watchful waiting for clinically localized low-risk prostate cancer.

What were some of the notable findings? Were any of them surprising to you and your co-authors?

Some notable findings from the first initial analysis of this novel linked data set were that once we linked the GPS assay data to over 120,000 men from the SEER database, we found that about 5,500, or 4.6%, had linked in the sense that they had GPS assay results. When we did our multivariable logistic regression analysis, we found that having a GPS result versus not having a GPS result was independently and very strongly associated with receiving active surveillance and watchful waiting, and this was more than 2.5 times higher.

We also looked at the GPS score itself in relation to the receipt of active surveillance and watchful waiting treatment and found that the percent of patients utilizing active surveillance and watchful waiting actually decreased in a dose-dependent manner with increasing GPS score. In other words, of those with a GPS score in the low range, from 0 to 20, 58% had active surveillance or watchful waiting compared to 47% with a score from the mid-range, or 21 to 40. And then 27% for those with scores at a higher range—41 to 101.

One final thing that was surprising to us in the results was that we noted pretty substantial geographic variations in having a GPS result when looking across the SEER regions. This ranged from 6.7% in the Eastern states of New Jersey, New York, and Connecticut, to less than 1% in the states of Utah and Idaho. Interestedly, we noted that these state-level differences were not associated with differences geographically in the utilization of active surveillance and watchful waiting.

How does this research build on previous studies of Oncotype DX?

Oncotype DX for prostate cancer has been around for a long time. It has been studied in 5000 men and has been recorded in 15 publications to date. It's the only genomic test for prostate cancer that's been looked at with 20-year outcomes. This public-private collaboration builds on a prior collaboration with Oncotype DX for breast cancer, which had been linked to SEER data for breast cancer patients. That particular linkage also had resulted in some really important findings regarding patterns in the real world in terms of receipt of testing and its use to guide treatments.

What is the take-home message for the practicing urologist?

The increase in active surveillance and watchful waiting over the past 10 or so years is well documented. We know by now that this pattern exists, but the key finding here is that the GPS assay result was independently and strongly associated with active surveillance and watchful waiting use . This pattern we saw was also consistent across different patient subpopulations whether we stratified by age, by race/ethnicity, or by [National Comprehensive Cancer Network] risk group.

I think that a second take-home message from this study and what's really important here is that by integrating 2 population-based data resources, this type of private-public collaboration resulted in a novel, real-world data set that allows us to understand the patterns in prostate genomic testing use across the US population.

Is there anything else you feel our audience should know about the findings?

Some other key points that are important for the audience to know is that there are potential disparities in active surveillance and watchful waiting use by patient factors and social determinant factors, such as race/ethnicity and other social risk categories. This type of collaboration, using real-world, population-based data sets, helps us to illuminate where those disparities are, and ultimately, to ensure that all men have the opportunity to benefit from precision medicine advances. And so, I think that this type of collaboration is a really nice illustration of the types of projects that we should be undertaking in the future.


1. Aboushwareb T, Bennett J, Yuan Y, et al. Active surveillance for watchful waiting in clinically low-risk prostate cancer patients in the SEER database with and without an Oncotype DX genomic prostate score assay. Paper presented at: 2021 American Urological Association Annual Meeting; September 10-13, 2021; virtual. Abstract MP62-06