New multi-biomarker test accurate in detection of clinically significant prostate cancer


A newly published study in The Lancet Oncology has provided some very interesting data on the accuracy of a new, multi-biomarker test in screening 50- to 69-year-old men for risk of prostate cancer.

The new paper by Grönberg et al. describes the initial results of the so-called Stockholm 3 or STHLM3 study of the ability of a combination of plasma protein biomarkers (PSA, free PSA, intact PSA, hK2, MSMB, MIC1), genetic polymorphisms (232 single nucleotide polymorphisms or SNPs), and clinical variables (age, family, history, previous prostate biopsy, prostate exam) to identify men at risk for clinically significant prostate cancer (i.e., prostate cancer with a Gleason score of 7 or higher) as compared to a simple PSA test. Additional information is available on the ScienceDaily web site.

The Stockholm 3 study was a prospective, population-based, paired, screen-positive, diagnostic study of men aged between 50 and 69 years and no known history of prostate cancer. The men were randomly invited, based on their dates of birth, as recorded in the Swedish Population Register kept by the Swedish Tax Agency. Men found to have prostate cancer at time of enrollment were excluded from the study.

The goal was to see if the multi-biomarker test could increase diagnostic specificity compared with PSA without decreasing the sensitivity for diagnosis of prostate cancers with a Gleason score of 7 or higher.

Here are the basic study findings:

  • 58,818 men from Stockholm, Sweden, aged between 50 to 69 years, were enrolled and tested between 2012 and 2014.
  • The multi-biomarker model performed better than PSA alone for detection of cancers with a Gleason score of ≥ 7 (p < 0.0001)
  • The so-called “area under the curve” or AUC was 0.56 with PSA alone and 0.74 using the multi-biomarker model.
  • All variables used in the multi-biomarker model were significantly associated with prostate cancers with a Gleason score of at least 7 (p < 0.05) in a multiple logistic regression model.
  • At the same level of sensitivity as the PSA test, using a PSA cutoff of ≥ 3 ng/ml to diagnose higher risk prostate cancers, the multi-biomarker model could
    • Reduce the number of biopsies by 32 percent
    • Avoid 44 percent of biopsies exhibiting benign pathology.

Grönberg and his colleagues conclude that their multi-biomarker model

could reduce unnecessary biopsies without compromising the ability to diagnose prostate cancer with a Gleason score of at least 7, and could be a step towards personalized, risk-based prostate cancer diagnostic programs.

We are all well aware that the PSA test comes with risk for a high false-positive rate that can lead to  unnecessary prostate biopsies and unnecessary diagnosis of low- and very low-risk prostate cancers. This new test appears to be able to take yet another step toward more sophisticated diagnosis of clinically significant as opposed to clinically insignificant prostate cancer.

We note that the Stockholm 3 study was conducted by research team from the Karolinska Institutet “in collaboration with Thermo Fisher Scientific, which provided the protein and genetic marker assays used in the clinical study.” Actual funding for the study came from Stockholm County Council. One cannot help wondering, however, what a commercial version of a test like this might cost.

On the other hand, a sophisticated commercial test like this, which could cut down significantly on costs of the millions of biopsies carried out in the USA every year, might be well worth the price if it was of the order of a few hundred dollars per test. This test is clearly “the next step up” from the 4KScore test currently available here in the USA, and that also raises the question of whether it is really much more accurate than the 4KScore test. It sounds as though it might be … but is it really, in actual clinical practice as opposed to in a clinical trial like this one? The 4KScore test currently costs around $395.00 and is covered by Medicare.

8 Responses

  1. “Men found to have prostate cancer at time of enrollment were excluded from the study.” Did the men enrolling in the study have to start out with a biopsy? If so, hats off to them for being willing to participate.

  2. One caveat I would add to this is that the population tested is different, and probably far more homogeneous, than the US population. I would also like to see the ROC curve for levels of PSA above 3.0 ng/ml and for PHI before concluding that the presumably more expensive genetic testing will add value. That being said, while I think we are moving beyond the era of PSA alone, I hope we have not lost the gains that prostate screening has brought.

    Different topic: prostate pathologist humor.

  3. Doug:

    No, what this means is that there may have been men whose available medical records in the Swedish databases did not show them as having been diagnosed with prostate cancer at the time the invitations to screening were sent out, but who had been diagnosed by the time they responded to the screening requests.

  4. Dear Les:

    I think you might be surprised at how non-homogeneous the population of Stockholm is today. When you get out of Stockholm, into more rural parts of Sweden, there is still a great deal of the genetic homogeneity one might expect, but in Stockholm itself … think more like Geneva or London.

  5. I’m confused on the site master’s statement’s. Please clarify.

    You state the so-called “area under the curve” or AUC was 0.74 using the multi-biomarker model STHLM3.

    You then state that, ‘This test is clearly “the next step up’ from the 4KScore test currently available here in the USA, and that also raises the question of whether it is really much more accurate than the 4KScore test. It sounds as though it might be.”

    One 4Kscore study; Parekh, D et al 2014 submitted, AUC = 0.82
    2008 BMC Medicine 2008 6:19 AUC = 0.90
    2010 BMC Cancer 2010 10:635 AUC = 0.87
    J Clin Oncol 2010 28:2493 AUC = 0.84
    Eur Urol 2013 64:693 AUC = 0.81
    Cancer 2010 116 (11) 2612 AUC = 0.83
    Clin Can Res 2010 16(12) 3232 AUC = 0.80
    BJC 2010 103:708 AUC = 0.87

    As you can see the AUC for the 4Kscore clearly surpasses the AUC for STHLM3.

    So my question is, why in your opinion would the STHLM3 sound or appear more accurate than the 4Kscore? The AUC on the 4KScore through clinical studies surpass the AUC on the STHLM3.

    Ref#
    http://clinical.opko.com/hs-fs/hub/375787/file-1968687361-pdf/documents/OPKO_4Kscore_Presentation_10-16-2014.pdf?t=1431702334703

    Also it’s important to note, the 4Kscore was presented to the Pathology Coding Caucus, an advisory group the AMA CPT Editorial Board and to the AMA CPT Editorial Board itself, an application for category 1 CPT growth for their 4Kscore test. The PCC Advisory Group recommended the issuance of a category 1 CPT code for the 4Kscore.

  6. Dear Sitemaster,

    Where did you hear that 4KScore is covered by Medicare? The OPKO web site says it’s not FDA approved, and I thought that was a precondition for Medicare coverage. But the ins and outs of Medicare coverage remain mysterious to me.

    I agree with Les that phi has a better AUC, is cheaper, is approved by FDA, and is covered by Medicare and insurance. It doesn’t appear that all the extra genetic info really improved its sensitivity or specificity.

  7. Dear Mr or Dr. Judge:

    My opinion about the potential of the new test as opposed to the 4KScore test is based not on the AUC scores (which are — as I understand it — only meaningful in an actual head to head test on the same test samples) but on the fact that the new test includes all of the same markers as those included in the 4KScore test (and two more) as well as a series of 232 genetic mutation markers too.

    I was extremely careful to point out, however, that my speculative opinion doesn’t make anything true. I wrote, with great care, “that also raises the question of whether it is really much more accurate than the 4KScore test. It sounds as though it might be … but is it really, in actual clinical practice as opposed to in a clinical trial like this one?”

    You might wish to note that other correspondents aren’t willing to accept the idea that the 4KScore test is necessarily any better than the phi test either!

    We both know exactly what would need to be done to resolve all such questions one way or the other.

  8. Dear Allen:

    I appear to be mistaken about coverage of the OPKO test by Medicare but I thought I had seen such approval given several weeks ago. A test like the 4KScore test which must be carried out at a central clinical laboratory (as opposed be being made available as a test kit that can be acquired and used by any laboratory) does not have to be approved by the FDA. That is an option that OPKO could consider if it wanted to, but it is not required and it is not a necessary condition for Medicare or insurance coverage.

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 )

Google+ photo

You are commenting using your Google+ 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 )

Connecting to %s

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

%d bloggers like this: