The data behind the $3,400 test for prostate cancer recurrence
Posted Mar 06 2010 12:00am
Some days ago we commented on a media release issued by Myriad Genetics which stated that their new PROLARIS™ test “identifies patients at low risk of disease recurrence with 95% certainty.”
We have now been able to see the actual data presented by Swanson et al. at the Genitourinary Cancers Symposium in San Francisco. It would appear from that data that the above-mentioned statement may not be quite the whole truth.
According to Swanson et al., this test “dramatically improves the prediction of recurrence” of prostate cancer after radical prostatectomy “in men with clinically defined low-risk disease” when used in combination with other commonly employed post-surgical clinical parameters. The bold italic type has been added for emphasis for the very simple reason that this is a group of men that would seem to encompass many patients who may never have needed treatment in the first place, based on the current NCCN guidelines for prostate cancer management, and who were (presumably) therefore at minimal risk for any clinically significant disease!
Now we know that some 30-40 percent of men who have a radical prostatectomy (RP) do indeed have biochemical recurrence. However, many of those who have biochemical recurrence may also not need treatment for that recurrence (because they are at minimal risk for any clinical consequences of biochemical recurrence). We also know that being able to predict risk for recurrence for some patients would indeed help us to ensure appropriate early intervention for high risk patients, and spare those for whom no further treatment is warranted.
So what did Swanson et al. actually do?
They used data from a retrospective cohort of 442 post-RP patients to assess the level of expression level of a series of genes involved with cell cycle progression (CCP). We assume that there was a total of 46 genes involved, as stated in the original media release, but Swanson et al. do not specifically say that. The patients had been followed for a median of 9.5 years.
The 442-patient cohort was divided into two groups: a “training” group (n = 195) and a “validation” group (n = 247). Then they further categorized that patients as low-risk or high-risk using such traditional traditional factors as whether disease was organ confined, the patients’ PSA levels, and their Gleason scores. Again, the precise categorization criteria are not provided by Swanson eta l.) Only then did the authors seek to determine how the new 46-gene “signature” predicted outcome in each group, with the following results:
The CCP signature alone was, most certainly, a significant predictor of biochemical recurrence in the training group (p = 0.01) and in the validation group (p = 5.8 × 10-8).
After adjusting for clinical parameters, the CCP signature alone was a significant predictor of disease outcome for low-risk patients in both the training group (p = 0.0071) and the validation group (p = 1.9×10-4) with a validated negative predictive value of 0.95.
Patients in the low-risk group with a low CCP score had a 5 percent, 10-year recurrence rate
Patients in the low-risk group with a high CCP score had a 22 percent, 10-year recurrence rate.
Patients in the high-risk group with a low CCP score had a 36 percent, 10 year recurrence rate.
Patients in the high-risk group with a high CCP score had a 70 percent, 10-year recurrence rate.
In the validation cohort only, the CCP signature was prognostic in high-risk patients (p-value = 0.0026).
What this appears to tell The “New” Prostate Cancer InfoLink is that the new PROLARIS test is indeed capable of predicting prostate cancer recurrence with a high degree of accuracy but only in the men least likely to have such a recurrence and who can therefore be followed very adequately with regular PSA tests. By comparison, the PROLARIS test appears to be a good deal less effective in predicting recurrence in those men who could benefit most from a test of this type the men at greater risk for early biochemical recurrence.
Futhermore, the available evidence is all based on retrospective data. There are no prospective data to support the use of this test as yet. In other words, there has been no clinical trial in which the PROLARIS test has been given to men immediately before or after their RP, for whom a prediction of their risk for biochemical progression has been made, and for whom the accuracy of the prediction has been assessed “in the real world” by following the patients for a significant period of time.
Absent such data, The “New” Prostate Cancer Infolink again raises the question of how many insurers will be willing to cover the $3,400 cost of the PROLARIS test (as compared to regular post-surgical PSA tests). There may well be patients willing to pay for this test out of their own pockets but most of us don’t have a spare $3,400 lying around the house these days!