Urine testing for risk of prostate cancer: a current and future perspective

It is becoming very clear that over the next 5 to 10 years we will develop a whole new spectrum of urine-based tests for the assessment of risk for prostate cancer and whether a specific patient will actually need a biopsy and/or other tests to confirm diagnosis.

One such urine test has already been approved in the USA: the ExoDx Prostate Test. This test claims to be 92 percent accurate with regard to sensitivity for prostate cancer and 91 percent accurate with regard to negative predictive value (NPV).

A second such test is probably going to be approved by the US Food and Drug Administration (FDA) later this year: the miR Scientific Sentinel test, which claims a similar level of scientific accuracy but also claims to be able to categorize the results from thsi test as negative for prostate cancer, positive for low-risk forms of prostate cancer, positive for intermediate-risk forms of prostate cancer, and positive for high-risk forms of prostate cancer. We await the FDA’s approval of this test with great interest since it could have great value in initial decision-making about how biopsies need to be used and on how men on active surveillance could be managed in the future.

The degree to which tests of these types will be able to eliminate need for blood sample-based tests such as PSA, the Prostate Health Index (phi), and the 4KScore tests (not to mention prostate biopsies) is yet to be determined.

The other day, one of our readers brought a media release to our attention. This media release refers to South Korean research on yet another urine test for prostate cancer that has been used to develop

a technique for diagnosing prostate cancer from urine within only twenty minutes with almost 100% accuracy. The research team developed this technique by introducing a smart AI analysis method to an electrical-signal-based ultrasensitive biosensor.

The abstract of the paper by Kim et al. reporting this research can be found here, in a journal called ACS Nano (published by the American Chemical Society). And we wish to be clear that we are not disputing the data reported in this paper. However, we made sure that we had read the full text of the paper prior to commenting.

The bottom line is that this paper is a very complex research paper reporting the possibility that a commercializable test could hypothetically be developed, based on urinanalysis and artificial intelligence techniques, that has a very high specificity and NPV for the diagnosis of prostate cancer. The current data, however, is based on information from just 76 urine samples:

  • The urine samples were from three categories of men:
    • “Normal” urine from men who had no evidence of either prostate cancer or benign prostatic hyperplase (BPH)
    • Urine from men known to have prostate cancer taken before the men had a digital rectal exam
    • Urine from men known to have prostate cancer taken after the men had a digital rectal exam
  • The samples were then divided into a “training” group of samples (used to create the testing system) and a “test” group (sed to validate the testing system.
    • 53/76 samples (70 percent) were used in the training group.
    • 23/76 samples (30 percent) were used in the “test” group.

While this research is certainly interesting, the full text of this paper contains these two paragraphs, as follows:

Our [prostate cancer] screening platform showed exceptional prostate cancer screening performance using 76 urine specimens. However, out study is limited to Korean males. This leaves the opportunity for additional experiments with different races. Another useful study to validate our approach is measuring a large number of samples with controlled age, PSA level, and prostate volume. An effort to look for more efficient biomarkers and their combinations is also necessary.

With increased patient numbers, the accumulation of data will drive an evolution of our ML algorithm in the multimarker sensing platform. Therefore, after validation with a large amount of data, our multimarker sensing platform has the potential to shift the current [prostate cancer] screening paradigm. On-site and accurate [prostate cancer] screening in 40 min with a drop of naturally voided urine will directly impact the human healthcare system in the foreseeable future.

In other words, the authors are very well aware that their proposed test is nothing like ready for “prime time”. We also feel we need to point out that we could find no reference to published data from the Sentinel test under development by miR Scientific, which raises the question of whether the authors were aware of this test by the time this paper was accepted for publication.

What we can say with confidence is that we are in complete agreement with the last sentence in the second paragraph of the quotation above. As noted at the start of this commentary, we are convinced that accurate testing of urine for diagnosis of prostate cancer will be “coming to a laboratory near you” in the near future, and that the accuracy of such tests will probably increase over time. Which test turns out to be “the best” and the most cost-effective is going to take some time to work out.

Editorial commentary: The “New” Prostate Cancer InfoLink thanks a regular reader for bringing this paper to our attention. We also thank both ACS Publications and another of our regular readers for providing us with full text copies of the original paper by Kim et al. for review in the development of this commentary.


One Response

  1. Thank you for this interesting and thorough look at the field. It’s clear that urine testing is the future, it remains to be seen how distant a future!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: