A new urine test to assess risk for prostate cancer?


Apparently a newly developed type of urine test — an RNA-based “urine exosome gene expression assay” — has the potential to differentiate between risk for high-grade prostate cancer (Gleason score ≥ 7) and low-grade prostate cancer (Gleason score 6) or benign disease.

To get the details, have a look at the abstract of this article by McKiernan et al. in JAMA Oncology and/or this information on the ScienceDaily web site. It also appears that this test is going to be available for your doctor to prescribe in  the not too distant future — as the so-called ExoDx™ Prostate(IntelliScore) test from a company called Exsome Diagnostics.

Basically, the new test is non-invasive and just requires a standard urine sample (with no need for an aggressive digital rectal manipulation of the prostate beforehand, in contrast to the PCA3 test).

McKiernan et al. set out to investigate the accuracy of the new test along with information about the patients’ PSA level, age, race, and family history as against the accuracy of the patients’ PSA level, age, race, and family history (“standard of care” or SOC) to see how well each could be used to predict risk for high-grade prostate cancer (Gleason score ≥ 7) as compared to risk for low-grade prostate cancer (Gleason score 6) or benign disease in two large sets of patients.

First of all, they used a “training set” of patients with PSA levels of between 2 and 20 ng/ml to develop a “prognostic score” for the new test. Then they validated that prognostic score in a second set of patients from 22 community practices and academic urology practices across the US. Eligible participants included men who were all 50 years or older, and who were scheduled for an initial or repeat prostate biopsy because their PSA level was between 2.9 and 20.0 ng/ml and/or they had a  suspicious rectal examination.

Here is what they found among 255 men in the training set:

  • Average (median) age was 62 years
  • Average (median) PSA level was 5.9 ng/ml
  • The new test + SOC was better than SOC alone at finding Gleason ≥ 7 disease.
    • AUROC was 0.77 for men with prostate cancer and a Gleason score of ≥ 7.
    • AUROC was 0.66 for men with either prostate cancer and a Gleason score of 6 or benign disease.
    • The difference was statistically significant (P < 0.001).

(Note that the so-called AUROC score would be 1.0 for a test that was perfect and identified the correct result with 100 percent accuracy; the AUROC score would be 0.5 for a test that did just as well as flipping a coin or guessing.)

The authors used these data to set a composite cut-point score of 15.6 (for the new test + SOC data) that was able to identify > 90 percent of the men found to have cancer with a Gleason score of 3 + 4 = 7 or higher.

And here is what they found in 519 men in the validation set:

  • The new test + SOC was again better than SOC alone at finding Gleason ≥ 7 disease.
    • AUROC was 0.73 for men with prostate cancer and a Gleason score of ≥ 7.
    • AUROC was 0.63 for men with either prostate cancer and a Gleason score of 6 or benign disease.
    • The difference was statistically significant (P < 0.001).
  • When they used the pre-defined cut-point score of 15.6
    • 138/519 men (27 percent) could have been advised to avoid a biopsy.
  • Among the men with a cut-point score below 15.6
    • 126/138 men (91 percent) had no cancer or low-grade cancer.
    • 12/138 men (9 percent) had a Gleason score of 7 or higher.
    • 7/138 men (5 percent) had some amount of Gleason 4 + 3 = 7 or 4 + 4 = 8 prostate cancer.
    • 9/12 men had moderately aggressive cancer.
    • 3/12 men had a higher risk of cancer.

So, first and foremost, the test seems to work when used in combination with other data (the patients’ PSA level, age, race, and family history), but it isn’t perfect. But, it does miss some higher-risk disease (as does every other test tried to date).

Is this test good enough to use to tell a man he doesn’t need to have a biopsy? Well, maybe. It probably depends on some other factors too, like how worried he is about his risk. On the other hand, if this test along with another test, like the Prostate Health Index (phi) test or the 4KScore test both (or all) said he didn’t need a biopsy, then maybe a biopsy really could be avoided safe;y in 99.5 percent of patients. (There is always going to be the odd exception to every rule like this.)

What is really important in all of this is that we are close to becoming really, really good at deciding who actually needs to be biopsied and who doesn’t. This would allow us a much better way not only to decide who needs and initial biopsy but also how to monitor over time the men who are on active surveillance without the need for as many additional biopsies!

What we do not know yet is how much Exsome Diagnostics is hoping to charge Medicare and the commercial insurance industry for running their new test on a urine sample! We figure we’ll get to know that soon enough too!

 

4 Responses

  1. “However, the test also predicted high-grade cancer in 66 percent of men whose biopsies revealed low-grade or no cancer.” from ScienceDaily web site.

    Find the design a little curious; Given the built in error in biopsies I would have thought the control would have been or at least included actual prostatectomies.

  2. One of the markers is PCA3 mRNA as I read the linked information.

    PCA3 urine collection asks for a pre-collection “attentive” DRE (not prostate massage). Wonder why this combo of urine RNA markers does not?

  3. This looks like a landmark development! Thanks for your review.

    I intend to read the complete paper, looking especially to see if the PSA test added any independent value for that small proportion of high-risk patients who were not picked up with the new test.

  4. The ScienceDaily reference says, “the test also predicted high-grade cancer in 66 percent of men whose biopsies revealed low-grade or no cancer.” This is a very low specificity test – only 34% at that cutpoint. This very high rate of false positives means that 66% of those with a score over 15.6 were sent for an unnecessary biopsy. I don’t understand the next statement that “In clinical practice, use of the test would have spared 27 percent of men from having an unnecessary prostate biopsy.” I suppose they are comparing it to biopsy recommendations based on PSA alone at some PSA cutpoint.

    For comparison, at a cutpoint as low as 1.0 ng/ml for PSA, its sensitivity is 83% (vs 92% for ExoDx Prostate IntelliScore) and its specificity is 39% (vs 34% for ExoDx Prostate IntelliScore) — the new test was a little better at detecting cancer and a little worse in regard to unnecessary biopsies, but not all that different. Granted, these ROCs are for high grade prostate cancer with the new test, while PSA is for any grade prostate cancer. But what we really want in a test is one that has a cut point that simultaneously has good sensitivity and good specificity. Neither PSA nor %free PSA has that quality. The Prostate Health Index or phi test, at a cutpoint of 34, has 65% sensitivity and 65% specificity — reasonably good, but not great. (It is also better at detecting higher grade cancer.) PCA3 (with prostate massage) also has reasonably good sensitivity (64%) and specificity (76%) at a cut point of 35. It’s impossible to tell if this new test has that quality from the information provided.

    I don’t know the ROCs for ERG (one of the three components of this test). TMPRSS2:ERG FUSION at a cut point of 65 has very high specificity (84%) but very low sensitivity (34%) because so few types of prostate cancer express it. I know little about the third component — the SPDEF gene.

    I agree with the Sitemaster that we need more information (and cost information too!) before we can decide whether this test or some combination of others is a better detection method.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

%d bloggers like this: