What does a very high PSA at diagnosis tell us?


Probably because of high level of research funding as a consequence of the Movember Foundation‘s initiatives, Australia is increasingly providing us with interesting data on the diagnosis, management, and outcomes of men with prostate cancer.

In another recent paper in BJU International, Ang et al. have reported on research into the outcomes of patients initially diagnosed with prostate cancer and with a PSA level of > 100 ng/ml at time of diagnosis.

They looked at data on > 5,500 men, all diagnosed with prostate cancer in South Australia between January 1998 and August 2013, and whose clinical history and outcomes could be identified in the database of the South Australian Prostate Cancer Clinical Outcomes Collaborative (SA-PCCOC).

They categorized these men into five groups based on their PSA levels alone:

  • Men with PSA levels of < 20 ng/ml at diagnosis (Group A)
  • Men with PSA levels of 20 to ≤ 100 ng/ml at diagnosis (Group B)
  • Men with PSA levels of 100 to ≤ 200  ng/ml at diagnosis (Group C)
  • Men with PSA levels of 200 to ≤ 500 ng/ml at diagnosis (Group D)
  • Men with PSA levels of >500  ng/ml at diagnosis (Group E)

They then set out to determine which of the following were prognostic for prostate cancer-specific and overall mortality in the men whose PSA levels were > 100 at diagnosis (i.e., the men in Groups C, D, and E combined).

Here is what they found:

  • 5,716 men were identified in the SA-PCCOC database for whom relevant data were available.
  • Just 241/5,716 patients (4.2 percent) had a diagnostic PSA level >100 ng/ml (such that the patients fell into Groups C, D, and E above).
  • Overall survival rates at 5 years from diagnosis were
    • 87.0 percent among men in Group A
    • 36.7 percent among men in Group B
    • 29.1 percent among men in Groups C, D, and E combined
  • Overall survival rates at 10 years from diagnosis were
    • 70.7 percent among men in Group A
    • 36.7 percent among men in Group B
    • 18.2 percent among men in Groups C, D, and E combined
  • For the men in  Groups C, D, and E combined
    • Prostate cancer-specific mortality was associated with Gleason score and age at diagnosis.
    • Prostate cancer-specific mortality was not associated with PSA level at diagnosis.
    • Overall survival was associated with Gleason score and age at diagnosis.
    • Overall survival was also associated with PSA level at diagnosis.
  • Overall survival decreased as PSA increased (in a linear manner) until a diagnostic PSA level of 200 ng/ml, but there was no association thereafter.

Quite what this means is hard to tell, but it is not an expected result, and it seems to imply that, when one ignores things like Gleason score and age, significantly elevated PSA levels at diagnosis may be telling us as much about a man’s overall state of health as it is about his prostate cancer-specific risks.

3 Responses

  1. Likely Explanation Why PSA Level Was Not Increasingly Predictive of Prostate Cancer Mortality Among Men In Groups C, D, and E (Each Group With PSA of 100 or More at Diagnosis)

    As a veteran member of this overall group (PSA 113.6 at diagnosis), I am quite familiar with this territory. I have a very strong hunch that the reason PSAs of 100 to 200, versus 200 to 500, 500 or greater did not make a difference is because this overall group is very likely dominated by men with short PSA doubling times combined with PSAs that were already at at least 100 at diagnosis.

    Consider men with a PSA of 100 and a doubling time of 3 months (mine turned out to be 3 to 4 months). In three months the PSA would be 200, in six months 400, and in nine months 800, so the patient would have moved from Group C to Group E in less than a year. Even with a doubling time of 6 months (at the slower end of the spectrum of rapid doubling times), PSA would have gone from 100 to 200 in 6 months, from 200 to 400 in one year, and from 400 to 800 in a year and a half.

    Therefore, prostate cancer specific survival time would not be much different – likely a matter of months – for men starting with a PSA of at least 100 regardless whether they were in Group C, Group D, or Group E.

    This line of thought also seems a strong contender for explaining the final finding noted above that “Overall survival decreased as PSA increased (in a linear manner) until a diagnostic PSA level of 200 ng/ml, but there was no association thereafter.” In other words, whether the PSA at diagnosis is 200 or higher is not going to make much difference. I suspect the difference between 100 and 200 is small.

  2. Dear Jim:

    Of course this group was dominated by men who had a PSA that was > 100 at diagnosis. That was the criterion for inclusion in the evaluation!

    I think you are missing the point of the article, which is that being diagnosed with a high PSA level of > 100 appears to have had little to no impact on risk for prostate cancer-specific mortality as compared to risk for overall mortality! In other words the statement in your third paragraph is wrong (based on the findings in this article).

  3. Withdrawing Point in Third and Fourth Paragraphs

    Sitemaster – One of the reasons I find this blog so helpful is the responses you provide as they help me think. I’ve been thinking about the issues here for several days now.

    I’m reaching your conclusion by a different line of thought. In the third paragraph, the point was that most men with PSAs at diagnosis in the ranges stated would soon overlap, based on the assumption of short PSA doubling times. On rethinking this, that is unlikely to hold true as most men diagnosed with such high PSAs would be highly motivated to start some kind of treatment to combat the cancer, and that treatment would likely reverse, halt, or at least slow the rise in PSA. In my case, I was on Lupron within 2 weeks of diagnosis. It seems likely that most men with such high PSAs would soon be on some form of androgen deprivation therapy.

    Therefore, I am now interested in the finding you report that “For the men in Groups C, D, and E combined [those with PSAs of 100 and higher at diagnosis] … Prostate cancer-specific mortality was not associated with PSA level at diagnosis.” This is in a context of the last finding: “Overall survival decreased as PSA increased (in a linear manner) until a diagnostic PSA level of 200 ng/ml, but there was no association thereafter.” I’m not sure what to make of this set of findings which indicates that diagnostic PSA above 200 does not matter much regarding overall survival.

    (Regarding the “dominance” point you noted, what I was trying to emphasize was that men diagnosed with very high PSAs would also be very likely to have short PSA doubling times, which was not addressed in the study abstract.)

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: