Dr. Tosoian on screening approaches and tools for prostate cancer

Jeffrey Tosoian, MD, MPH

In this interview, Jeffrey Tosoian, MD, MPH, discusses the recent JAMA Network editorial “A Pragmatic Approach to Prostate Cancer Screening”¹ as well as the JAMA Oncology paper “Development and Validation of an 18-Gene Urine Test for High-Grade Prostate Cancer.”² Tosoian is an assistant professor of urology and director of translational cancer research in the division of urologic oncology at Vanderbilt University Medical Center in Nashville, Tennessee.

Please describe the background for this editorial.

The serum PSA test remains the first-line screening test to identify patients at increased risk of prostate cancer. And for many years, patients with a PSA greater than 3 [ng/mL] underwent a prostate biopsy. Yet, we learned that under that approach, a really large percentage of those biopsies were negative or detected the indolent, low-grade prostate cancers that can be monitored on active surveillance. Ultimately, a large proportion of the biopsies were unnecessary. And so in recent years, our guidelines have recommended the use of MRI and/or biomarkers to better define the risk of higher grade cancers before biopsy, as higher grade cancers are now the focus of diagnostic efforts. Yet, when and how exactly to use these tools - biomarkers and MRI - either in sequence or in combination, has really not been well defined, particularly in any large, prospective studies. The ProScreen trial³ used an approach in which men with an elevated PSA test first underwent a biomarker test, then those with a positive biomarker test proceeded to an MRI, and those with a positive MRI then underwent prostate biopsy. It was a large, randomized, well-controlled trial in which a specific approach to use of these tools was evaluated, particularly use of the biomarker tests first followed by MRI.

Could you discuss the ProScreen trial, including its strengths and limitations?

Patients were randomized to either a control arm, which did not undergo any specific approach to screening, or a screening arm, which underwent the PSA, the biomarker test—in this case, the 4k score—then MRI approach, as we discussed. Among patients that participated in screening, around 10% had an elevated PSA level of greater than 3 [ng/mL], and of those, 30% had a negative biomarker test. The remaining 70% went on to MRI; of those that had an MRI, around 40% had a positive MRI and underwent biopsy. Of those who underwent biopsy, 54% were found to have a clinically significant prostate cancer. That represents a remarkably high yield of biopsy, particularly as compared with early screening studies such as the ERSPC, the large European study, in which only around 12% of patients that underwent a biopsy were found to have clinically significant cancer. This was under the prior approach, in which elevated PSA led to biopsy; only around 12% of those biopsies found significant cancer as compared with 54% with the more modern approach that was used in the ProScreen trial. At the same time, the number of clinically insignificant prostate cancers—Grade Group 1, Gleason score 6—diagnosed was reduced by around eight fold—3.2% of the population in ERSPC to 0.4% in the ProScreen study. And so the ProScreen study really did show, under 1 of the potential approaches to use of biomarkers and MRI, the remarkable improvement that can be attained on a population level with use of the newer tools, by reducing greatly the proportion of patients that undergo a biopsy (3% in this first round of ProScreen vs. 22% in the early ERSPC experience) and the proportion of patients that are found to have a low-grade indolent cancer, while having quite similar yield in terms of the proportion found to have clinically significant cancers. It was 1.7% in this first round of the ProScreen trial compared to 1.8% in the early rounds of the ERSPC. And so, the fact that this was a prospective, well-controlled, clearly defined approach to testing and yielded such promising results was encouraging, certainly, for the state of the field. We did mention in the editorial that 1 potential limitation was just the lack of racial diversity in the study. Of course, the Finnish population is largely Finnish, and so there was not much racial or ethnic diversity in the study. That will certainly be an area for future studies.

What research areas or advancements in prostate cancer screening do you believe hold the most promise for improving patient outcomes?

I think in terms of the 2 large tool sets we have, biomarker testing through urine or blood tests and imaging. I think both are only going to get better. As our technology improves, the ability to focus these tools on the truly high-grade, potentially lethal cancers is only going to increase. At the same time, it will enhance our ability to offer them at reduced costs, which will be a positive for the overall health care system. I really do think that, as in all fields, the emergence of AI and/or machine learning algorithms will undoubtedly contribute. One potential application that comes to mind right away would be limiting one of the potential pitfalls of MRI, which, right now, is that it remains very dependent on who exactly is reviewing the MRI. We know that from site to site, and even among sites, radiologist to radiologist, there are substantial differences in the performance of MRI. And so perhaps integrating AI to assist in the review of images and assigning a risk value will hopefully add quite a bit to the field. So I think as all these tools continue to improve, more studies, such as ProScreen, will be essential to defining and optimizing how we use these tools.

What is an ideal approach to prostate cancer diagnosis?

This is something that we discuss in the editorial. As I mentioned, MRI is a phenomenal tool. In patients with a positive MRI, it allows us to guide our biopsies to specifically target those areas of abnormality, and that has been shown to improve the detection of the clinically significant cancers we seek to detect early. The limitation, as mentioned, is, of course, that the performance of MRI can vary quite a bit from site to site, among radiologists, and ultimately, we know that sites that are not using MRI as frequently may not have expert focused radiologists to interpret those, so the predictive value of MRI is often substantially reduced relative to higher volume expert centers. I do think that MRI could be a great initial tool at those sites where the performance of MRI is well known and ideally published, so that we can interpret those with confidence. And so at sites where that is the case, where the negative predictive value of a negative MRI is 90% or higher, it's reasonable [to use as a] first-line test after PSA. In my opinion, at other sites, where perhaps the performance of MRI isn't as strong or where it's not really known, I do think there's value to be had in first-line use of one of these blood- [or] urine-based biomarkers that can rule out the need for an MRI or a biopsy. Data have shown roughly 30% to 50% of the population can rule out the need for MRI or biopsy with a high degree of reliability or confidence, as reflected by the negative predictive value. And for those who cannot be ruled out, MRI provides great value as a second test to then optimize the yield of biopsy. I think that would be the ideal approach. And I think that's 1 reason why I found the ProScreen trial to be very informative to the field. And I think all of these approaches will benefit from additional analyses in terms of cost effectiveness in terms of which tests make the most sense when we think about the population on the whole.

Please provide an overview of MyProstateScore 2.0 and how it works.

Currently, we have a number of biomarkers that are available for use in men with elevated PSA to better define the likelihood of them having a clinically significant prostate cancer. And those have all been shown in multiple studies to improve upon PSA in terms of accuracy for the higher grade cancers that we need to detect early. They are, at the same time, limited in that the existing biomarkers improved upon PSA in that they are specific for cancer, meaning that the markers are more highly expressed or measured at a higher level in the presence of cancer as compared with benign prostate tissue. Because we know that, in contrast, PSA is not specific for cancer. And so those all provided great improvement on PSA. But our thought in approaching the work that led to the new test was that if we could identify markers that are specifically elevated in the higher grade cancers at the focus of diagnostic efforts, that would greatly improve our ability of the test to give a stronger predictive value for those very higher grade cancers. And so, we used RNA sequencing-based tissue data to identify several new markers of high-grade cancer and to confirm some known existing markers of cancer...and ultimately identified a large candidate panel of 54 genes. These were all detectable in urine. And so we measured those in more than 700 patients that underwent a prostate biopsy, and found that an optimal combination of 18 of those markers led to improved prediction of the higher grade prostate cancers that we seek to detect. And so we then applied that to an external cohort. We were able to compare this new test to PSA...as well as some of the existing biomarkers that are offered in our clinical guidelines. And we found that the new test, as we had hoped, did improve upon existing tests in terms of accuracy. Clinically, what exactly does that mean? We took an approach in which, if every patient were to undergo a biopsy, as they did during the study, we know how many clinically significant prostate cancers are detected. And so we took an approach in which we said, "okay, let's ensure that we continue to detect 95% of those clinically significant cancers but also rule out the need to undergo a biopsy at all in as many patients as possible." And so, we found that use of the test could rule out the need for 35% to 51% of unnecessary biopsies that would have been performed if, per the traditional approach, all patients with an elevated PSA underwent biopsy—while still maintaining detection of 95% of the clinically-significant cancer. Promisingly, the new test did show improvement relative to the existing tests. It was particularly striking in men who had a history of a negative biopsy; the improvement was on the level of roughly 40% more patients could avoid an unnecessary biopsy.

How will this test improve screening and detection for patients with prostate cancer?

The aim is that the use of this test in patients with an elevated PSA, whether or not they had undergone a biopsy at some point in the past, would allow us to substantially reduce the proportion of patients that required to undergo additional tests like MRI or biopsy. And so to do that, we took the approach in which patients with negative MPS 2 tests could avoid MRI or biopsy altogether, and we found that around 35% to 50% of unnecessary biopsies could be avoided, while maintaining detection of 95% of the clinically significant cancers that are detected under an approach in which everyone had a biopsy.And so really, it's eliminating the need for additional testing, in roughly a third to one half of patients, while maintaining detection of 95% of the clinically significant cancers.

REFERENCES

1. Tosoian JJ, Penson DF, Chinnaiyan AM. A pragmatic approach to prostate cancer screening. JAMA. Published online April 6, 2024. doi:10.1001/jama.2024.4089

2. Tosoian JJ, Zhang Y, Xiao L, et al. Development and validation of an 18-gene urine test for high-grade prostate cancer. JAMA Oncol. Published online April 18, 2024. Accessed April 26, 2024. doi:10.1001/jamaoncol.2024.0455

3. Auvinen A, Tammela TLJ, Mirtti T, et al. Prostate cancer screening with PSA, Kallikrein Panel, and MRI: The ProScreen randomized trial. JAMA. Apr 6:e243841. doi:10.1001/jama.2024.3841