Genomic data and prostate cancer risk classification: an update


Three different molecular tests have been validated as providing useful data that can help a man and his doctors to come to decisions about the management and treatment of localized prostate cancer. These three tests are the Decipher test (from Genome Dx Biosciences), the Oncotype Dx test (from Genomic Health), and the Prolaris test (from Myriad Genetic Laboratories).

The other way to think about decision  about the management and treatment of localized prostate cancer is through the use of the NCCN risk classification system, which subdivides men with localized prostate cancer into six groups as follows:

  • Very low risk (clinical stage T1c and Grade Group 1 and PSA < 10 ng/ml and PSA density < 0.15 and < 3 possible biopsy fragments/cores positive for cancer with < 50 percent cancer in each fragment/core)
  • Low risk (clinical stage T1c or T2a and Grade Group 1 and PSA < 10 ng/ml)
  • Intermediate risk (one or more intermediate-risk feature [IRF], including clinical stage T2b or T2c, Grade Group 2 or 3, and PSA 10 to 20 ng/ml, and no high-risk features)
    • Favorable intermediate risk (1 IRF and Grade Group 1 or 2 and < 50 percent of biopsy cores positive)
    • Unfavorable intermediate risk (2 or 3 IRFs and/or Grade Group 3 and/or ≥ 50% of biopsy cores positive)
  • High risk (clinical stage T3a or Grade Grouo 4 or 5 or PSA ≥ 20 ng/ml)
  • Very high risk (clinical stage T3b or 4 or primary Gleason pattern 5 or PSA ≥ 20 ng/ml)

What we do not have as yet is any completely objective data as to the relative value of these three different molecular tests — ideally from a trial in which a large group of men are given all three of the tests and are then followed for about 10 years.

However, what we do now have are data from a small study carried out by Syed et al. at Hartford Hospital in Hartford, CT.

Syed and his colleagues carried out a retrospective analysis of data from a total of 22 patients, each of whom were given at least two of the three different molecular tests between 2014 and 2017.

Here are their core findings:

  • For 12 patients who received both the Decipher and the Prolaris tests,
    • The percentage agreement between the two tests was 66.7 percent.
    • The pair-wise agreement between the two tests (using Chen’s kappa score) was κ = 0.31 (p = 0.276).
  • For 8 patients who received both the Prolaris and Oncotype DX tests,
    • The percentage agreement between the two tests was 75.0 percent.
    • The pair-wise agreement between the two tests was κ = 0.39 (p = 0.168).
  • For 2 patients who received both the Decipher and the Oncotype DX tests,
    • The percentage agreement between the two tests was 50.0 percent.
    • The pair-wise agreement was not calculable.
  • For 10 patients who had the Prolaris test, as compared to the NCCN risk classification,
    • The percentage agreement was 75.0 percent.
    • The pair-wise agreement was κ = 0.21 (p = 0.117).
  • For 15 patients who had the Decipher test, as compared to the NCCN risk classification,
    • The percentage agreement was 60.0 percent.
    • The pair-wise agreement was κ = 0.15 (p = 0.268).
  • For 10 patients who had the Oncotype DX test, as compared to the NCCN risk classification,
    • The percentage agreement was 50.0 percent.
    • The pair-wise agreement was not calculable.

What these data appear to show us is that there is no clear correlation from test to test between the data provided by these tests and that there is also no clear correlation between the data provided by any of these tests and the data implied by the NCCN risk classification system.

From your sitemaster’s perspective, this study suggests a compelling need for data from a large enough prospective study to show us clearly which of these three molecular tests offers the most accurate additional data about risk for clinically significant prostate cancer and how we can then build that data into the NCCN risk classification system.

2 Responses

  1. I also think the NCCN risk classification system needs to be updated to use multiparametric MRI results. My clinical stage was T1c. By mpMRI it was T3a. Pathological stage after radical prostatectomy was T3a.

  2. Dear David:

    The problem with trying to use mpMRI data to upgrade the NCCN risk classification system is that it appears (at this time) to be insufficiently accurate on a consistent basis and is too dependent on the skills of the radiologist reading the scans. Furthermore, most patients are not given an mpMRI scan prior to a biopsy.

    While the inclusion of scan data in evaluation of risk is customary in those patients for whom scans are given, the whole point of the NCCN risk classification system is that it is based on simple tests that the vast majority of men are given as part of their initial diagnosis.

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: