Do PSA dynamics enhance pretreatment prediction of outcome — or not?


We have commented several times on the continuing controversy between two research groups on this key topic.

Catalona and his collaborators argue fiercely that PSA velocity and PSA doubling times can be used to improve the accuracy of outcome prediction in patients undergoing first-line treatment for localized prostate cancer. By contrast, Vickers and his colleagues believe that PSA dynamics add nothing to the predictive accuracy available based on the data used in the standard prostate cancer calculator (including the PSA level at a point in time).

O’Brien et al. (i.e., the pro-Vickers group) have now assessed data from a cohort of 2,938 patients with two or more PSA values before radical prostatectomy. Biochemical recurrence  occurred in 384 patients, and metastases occurred in 63 patients. Median follow-up for patients without biochemical recurrence was 2.1 years.

On univariate analysis:

  • 2/12 definitions of PSADT and 4/10 definitions of PSAV were univariately associated with both BCR and metastasis.
  • 1/12 definitions of PSADT and 1/10 definitions of PSAV had a higher predictive accuracy for BCR over PSA alone.
  • 4/12 definitions of PSAV improved prediction of metastasis.

However, the improvements in predictive accuracy were small, associated with wide confidence intervals, and markedly reduced if additional predictors of stage and grade were included alongside PSA. In other words, they may be a signal, but not enough to light a fire.

The authors claim that, yet again, they have been unable to show any clear evidence that any definition of PSA dynamics substantially enhances the predictive accuracy of a single pretreatment PSA value. We do not expact Dr. Catalona and his supporters to “roll over” based on these data.

What is just silly about all this is that if Dr. Catalona’s data were examined using Vickers’ analytical model under the supervision of appropriately neutral statisticians, there would seem to be no doubt this issue could be simply resolved (in the best interests of patients and clinicians) and we could move beyond what seems to be a storm in an academic teacup.

4 Responses

  1. Yet we move forward with treatments that show a small but measurable benefit. So what is the alternative? It would be nice if physicians and patients could have information that would help them in decision making. If Catalona is correct them maybe more men would be comfortable with AS as a treatment, but based on these results if a man is diagnosed then maybe he needs to be more cautious about pursuing AS?

  2. Kathy:

    The controversy between Catalona and Vickers and their respective colleagues is not about treatment at all. It is exclusively about whether a patient should consider having a biopsy which might lead to a diagnosis and how he and his doctors should reach that decision. The issue of treatment is quite separate from this issue.

  3. If it is prebiopsy, can you explain what is meant by, “However, the improvements in predictive accuracy were small, associated with wide confidence intervals, and markedly reduced if additional predictors of stage and grade were included alongside PSA.”? And “that PSA velocity and PSA doubling times can be used to improve the accuracy of outcome prediction in patients undergoing first-line treatment for localized prostate cancer”? That makes it sound as if it should be used as a treatment decision guideline? I am confused.

  4. Kathy:

    This entire discussion is about whether adding PSA velocity or PSA doubling time data to age, ethnicity, family history, DRE result, and whether you have had a prior negative biopsy as opposed to just the most recent PSA result as a means to assess whether you are likely to have a positive biopsy (should you decide to have one). Catalona et al. say it does; Vickers et al. say it doesn’t. In other words, Vickers et al. say that what Catalona et al. are suggesting does not improve on the results one currently obtains if one uses the PCPT-based PSA Calculator.

    Vickers et al. are talking about the validation of their argument using the historically available data from thousands of patients on which they know not only all of the relevant data but also the follow-up biopsy results. Their analysis shows that there may be minor improvements in accuracy in some sets of patients, but the results are not consistent and the overall result of using PSAV instead of PSA is not statsitically significant. They have now shown this on several occasions, and Catalona et al. have never published a comparable analysis that shows clearly that there is a statistical benefit to using PSAV. Nor has any independent third-party ever validated Calatona’s hypothesis (as far as I am aware).

    QUITE SEPARATELY, Catalona et al. have also been arguing that presurgical PSAV can be used to predict outcomes to treatment — along with other potential prognostic data. As far as I am aware, yet again, there is no evidence that pre-treatment PSA velocity improves the ability to predict outcomes compared to the simple PSA value at the time of treatment (as used in the Kattan nomograms and the Partin tables).

    Mike

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: