Oncotype DX GPS Test Shows Predictive Benefit in Unfavorable Intermediate-Risk Prostate Cancer


Jennifer Cullen, MPH, PhD, discusses the genomic test as a predictor of outcomes in patients with unfavorable intermediate-risk prostate cancer and next steps for research.

Jennifer Cullen, MPH, PhD

The 17-Gene Oncotype DX Genomic Prostate Score (GPS) has been found to be a strong independent predictor of biochemical recurrence, distant metastasis, and prostate-specific death in patients with unfavorable intermediate-risk prostate cancer, according to Jennifer Cullen, MPH, PhD, although more research is needed to examine its use in more racial and ethnic populations.

Investigators collected GPS results from 2 previously published studies which focused on patients treated with radical prostatectomy: the Center for Prostate Disease Research (CPDR) and the Kaiser Permanente Northern California System. These patients were assessed to identify associations of the test result with biochemical recurrence, distant metastases, and prostate-specific death in those with unfavorable intermediate risk prostate cancer. A total of 299 patients with intermediate-risk prostate cancer, 175 of whom had unfavorable intermediate-risk disease, were included in the analysis.1,2

Results showed that GPS was a significant predictor of biochemical recurrence (HR 7.1; 95% CI, 5.7-8.8; P < .0001), distant metastasis (HR 5.4; 95% CI, 3.8-7.8; P < .0001), and prostate-specific death (HR 3.4; 95% CI, 1.5-8.9; P = .006). Although these findings illustrate the potential benefit of GPS in men with unfavorable intermediate-risk prostate cancer, Cullen explained that additional research is still needed.

“We always want to err on the side of caution when we guide patients in treatment decision-making. There's this very careful balance that we must strike between preserving quality of life (QOL), and not just oncologic outcomes,” said Cullen. “These data are unique and lend support for guiding men with unfavorable intermediate-risk disease into a path of less aggressive treatment. However, as with any study, even a very rich one like this with 2 very large cohorts that both support the same findings, additional data would be of value, especially in more racial and ethnic groups.”

In an interview with OncLive, Cullen, a James T. Pardee – Carl A. Gerstacker Professorship in Cancer Research and an associate professor in the Department of Population and Quantitative Health Sciences, School of Medicine, as well as the associate director for Cancer Population Sciences at Case Comprehensive Cancer Center, further discussed the genomic test as a predictor of outcomes in patients with unfavorable intermediate-risk prostate cancer and next steps for research.

OncLive: Could you first discuss the different methods that were used in this study?

Cullen: Two cohorts were examined in this study; one was from the CPDR and the other was the Kaiser Permanente Northern California System. For the CPDR [cohort], this was a military cohort that we examined of men who had undergone radical prostatectomy between 1990 and 2011 and who had archival biopsy specimens, as well as pathology specimens, from the time of surgery available for examination and assessment of the GPS score.

In the 2 studies, 1 key difference is that the CPDR cohort has an overrepresentation of African American men and that's a function of the demographics of the military; the difference was 19% versus 14% in Kaiser. The other key difference was that the study at CPDR was a retrospective cohort where we just [included] all eligible men who met the study criteria and who were determined to have specimens available.

In the Kaiser study, because there were more patients who would have met the eligibility criteria, investigators performed a stratified sampling design where the eligible patients were evaluated. Then, from strata of calendar year and other key patient factors, they did a random probability sampling and sliced out a section of each group of men who met the criteria from every calendar year; this was done so that they wouldn't have too many patients on the study.

In this particular paper, we wanted to focus on a group of men who we did not focus on previously: those who meet the National Comprehensive Cancer Network (NCCN) criteria for unfavorable intermediate-risk disease. The importance of this is that the NCCN guidelines currently recommend GPS as a test that clinicians can use when they're counseling their patients on treatment decision-making for [those with] either very low-, low-, and favorable intermediate-risk disease. What has been previously unknown is whether the GPS will be a useful treatment decision guidepost for those with unfavorable intermediate-risk disease; that is what was so novel about the study.

What were the results of this analysis?

I should mention that the CPDR study was only able to model 1 of the 3 key study end points, which included biochemical recurrence, distant metastasis, and prostate cancer–specific death. CPDR had 2 events of metastasis or death, so the focus in the CPDR data was on the biochemical failure. For the Kaiser population, we were able to examine all 3 study end points because they had more events to observe.

The key findings were that for all 3 end points, whether it was short-term failure, long-term metastasis, or ultimately dying of prostate cancer, the GPS score was predictive of all 3 study end points. [This was] such that patients who were deemed to have a score above a cut point of 40, which was validated in previous work, those patients resembled, had clinical end points, and behaved more like a high-risk population. Whereas those who were below a cut point of 40 were behaving much more like a low- or a favorable intermediate-risk population. This was demonstrated most nicely in the Kaplan Meier curves, where a sharp separation in the curves [was observed] early on in the patient follow-up. This tells us that those with the higher GPS scores behave much more poorly and have poorer clinical end points than those with the lower GPS cut point.

Does this genomic test show any utility in high-risk prostate cancer?

That has not been examined yet. Although, the dilemma in the field is really in the intermediate-risk group. We need a better understanding on how aggressive we need to be with these patients with regard to treatment. For a high-risk patient, you already kind of know the answer to that question. Clinically, you're going to need to give them adjuvant and possibly salvage therapies because their disease is going to behave badly.

However, in this murkier area [of intermediate-risk patients], what do we do? It is a critical question to ask because you don't want to overtreat a patient whose disease might behave favorably and that's where the concerns have really rested. We don't know if we're over treating [in these patients]. For high-risk patients, we do know that a more aggressive treatment is necessary and so that question is less [difficult] for clinicians.

How will this test affect future genetic testing in prostate cancer?

This is obviously a very nascent field. There's a lot of buzz at national meetings about how to take findings from the literature and actually act on them and apply them to a clinical setting. It's still a difficult time to know what exactly are all the right genomic factors to consider. However, again, the NCCN has recognized the GPS as a useful tool in guiding patients and clinicians in treatment decision-making.

We feel that the value and the contribution of this particular study is now further characterized by another subset or slice of patients where, previously, it was harder to know what the right treatment path would be. Now, we have contributed data that help support another group of men for whom the GPS is going to be a tool of value in both counseling and treatment decision-making.

Are any next steps planned for this research?

The most valuable next step would be expanding into other racial ethnic groups because, to date, we've had the value of looking at racial diversity with respect to the African American race. This is so important given that African American men have 1.5 times the incidence of other racial and ethnic groups in terms of developing prostate cancer incidence. Some studies have shown that [these patients] are also more likely to die [from their disease]. As such, that particular racial group is very important to examine as a starting point. We should also expand [our efforts to examine this] among other ethnic groups, as well, [such as] Asians and Pacific Islanders. If other academics around the country have the ability to look at those types of populations, that would be of great value.

Was there anything else about this study that you wanted to add?

I do want to point out [the] very long-term follow-up [that we had]. It's very difficult to have available specimens that go back far enough in time and allow you to follow them and have an adequate number of end points in a cohort. To appreciate the dilemma, you need to know that many of these end points are very uncommon. Most men today are diagnosed with low-risk disease and it makes studying an end point like distant metastasis and prostate cancer–specific death very challenging because we just don't see many of these events unless we follow a man 10, 15, or 20 years. Having this access to patient databases, such as the CPDR and the Kaiser Permanente cohorts, is incredibly valuable because they go back so far in time and they have an infrastructure of saving patient samples over time that can be used to generate the GPS findings.

Are any other trials examining GPS in this space that you would like to highlight?

The Canary Prostate Program’s Prostate Cancer Active Surveillance Study (PASS) is a consortium effort of over 10 medical centers nationwide that is trying to answer the question of what happens to men once they go on active surveillance, which was one of the original goals of the GPS­: To identify good candidates for active surveillance. This is the watch-and-wait approach, where we monitor a patient closely when he has low-risk or very low-risk disease and we delay treatment, so his QOL can be preserved longer, knowing that he might ultimately need treatment, but today or tomorrow he might not.

The Canary PASS study has been incorporating GPS data into their data collection efforts. That cohort will be a very rich source of information on what becomes of patients on active surveillance and whether or not the GPS is able to prognosticate early on who might have failed in the future and had that disease, even though, originally, when they were clinically presenting with prostate cancer, their [outcome] was favorable.

That's going to be an incredible resource in the years to come. It's still a somewhat new effort; it’s only about 10 years old, which sounds probably like a very long time, but in the world of prostate cancer, these end points do take a long time to happen. You need a lot of follow-up time. However, that's going to be a very special study that will provide insight into how the GPS can help inform the future outcomes of patients placed on active surveillance.


  1. Cullen J, Kuo HC, Shan J, et al. The 17-gene genomic prostate score test as a predictor of outcomes in men with unfavorable intermediate risk prostate cancer. Urology. 2020;143:103-111. doi:10.1016/j.urology.2020.05.045
  2. Oncotype DX Genomic Prostate Score result shows for the first time the value of the test to guide treatment plans for unfavorable intermediate-risk prostate cancer patients. News release. Exact Sciences Corp. June 16, 2020. Accessed September 10, 2020. https://prn.to/3m7nvq6
Related Videos
David R. Wise, MD, PhD
Howard I. Scher, MD
Benjamin H. Lowentritt, MD, FACS
David Morris, MD, FACS
E. David Crawford, MD
Daniel Spratt, MD
Joshua Hurwitz
Expert on prostate cancer
Kevin Kayvan Zarrabi, MD, MS, FACP
Clara Hwang, MD