Cleveland researchers support Stamey’s hypothesis on utility of the PSA test


A recent article in Cancer seems to further support Stamey’s contention (published in 2004) that — at least in the USA — the PSA test has a great deal less value as a screening test for prostate cancer today than it had when it was initially introduced in the 1980s and 1990s.

Bott et al. have compared trends in PSA level, age, and prostate cancer detection among black and white men in Cleveland, Ohio, and the surrounding area during a 16-year period from 1990 to 2006. Their study is based on a retrospective analysis of patient charts and electronic medical records from 5,570 patients diagnosed with prostate cancer (including 911 black patients).

The basic findings of this study are that:

  • The patients’ mean age at the time of diagnosis did not change in either the white or the black group of patients, “despite what we had believed about the effects of patient education and screening campaigns.”
  • Prostate cancer detection rates have decreased during the time period studied.
  • There have been significant decreases in the sensitivity and specificity of PSA as a screening tool over this time period.
  • Analysis of more recent cases demonstrated a positive predictive value for PSA levels as a predictor for prostate cancer that was only about 50:50 (“comparable to a coin toss”).
  • Although the relative incidence of specific Gleason scores has remained relatively stable, reporting of prostate intraepithelial neoplasia (PIN) and inflammation has increased.

The authors draw the conclusion that the use of lower PSA thresholds as a criterion for biopsy (e.g., 2.5 ng/ml as opposed to 4.0 ng/ml), promoting PSA testing at a younger age, and increasing efforts to educate the public have not seemed to influence age at time of diagnostic testing. They suggest that this fact may reflect such matters as the perceived value of screening, physician referral patterns, patient compliance, and other sociodemographic issues.

These results are far from surprising to The “New” Prostate Cancer InfoLink, and only go to confirm what Stamey and his colleagues noted in their 2004 paper: that we urgently need a better test that will be much more specific for prostate cancer and that will — ideally — allow us to dintinguish between clinically significant and indolent disease.

9 Responses

  1. Many of us found Dr. Stamey’s initial statement upsetting and unwise in that it likely would deter some men from getting screened. I actually found his address and wrote him a letter stating that, while also thanking him for his great role in advancing the field. I’m hoping that the findings in this current study are put in the proper context, with the result that screening is encouraged.

    The decreases in PC diagnosis, PSA sensitivity, and PSA specificity, as well as the increased findings of PIN and inflammation (likely potential precursers) are all what we should have expected as consequences of successful screening efforts early in the PSA era. Essentially, a great many men were screened, and a great many long-existing cases of prostate cancer were detected. That took care of the “low hanging fruit.” Following that, we should have expected the declines reported, and sure enough, we can now see them in trend data. I would like to see the actual data on the “relatively stable” Gleason scores. I’ll bet it shows a rather sharp drop in the early years of the PSA era, with relative stability since, after the cases caught late were smoked out. Again, that’s what we would expect based on effective screening thanks to PSA.

    When we strive to educate the public about screening, it is so important to present the PSA test as a multiple purpose sentry that guards against several adverse conditions of the prostate, particularly BPH, then infection, and least commonly but most importantly prostate cancer. To me it is not helpful to complain that PSA is an imperfect test for prostate cancer when it should be seen as a multipurpose test.

    I too hope we someday have a more sensitive and specific test for prostate cancer, but, now that we have well endorsed active surveillance as the “go to” or at least preferred option for very low risk and low risk cases, we have taken care of the overtreatment issue and are able, with some follow-up effort and expense to be sure, to smoke out stealthy aggressive prostate cancer from indolent cancer.

    I am awed by what great superiority this confers in both treatment and research of prostate cancer compared to other cancers, such as breast cancer, where there is no marker even close to the effectiveness of PSA. We prostate cancer survivors should feel a profound joy and sense of gratitude that our 5- and 10-year survival rates are so superior compared to other cancers, largely due to PSA screening, in my layman survivor’s opinion. I am discouraged when the head of the ACS and others tell Congress that PSA is very imperfect. They are badly off the mark!

    Yes, I would like to see a better test than PSA, but to me our sense of urgency is far better focused on educating the public to appreciate the coupling of screening with active surveillance for very low and low risk patients. I’ll play the devil’s advocate here and declare that I wish we would cease calling attention to supposed inadequacies of the PSA test!

  2. “Over time, we did not observe a correlation between PSA elevation and positive prostate-biopsy results. Indeed, as demonstrated by the ROC curves plotted over time, PSA elevation was as effective as a coin toss in predicting subsequent cancer detection on biopsy.”

    This was taken from the text of the main article and is also mentioned in the abstract of the link given in Sitemaster’s post. I’ve said here before that I strongly suspected that PSA “screening” was doing very little actual screening, but rather was primarily only providing an excuse to perform a biopsy. This data supports that suspicion. This was why I asked about the number of biopsies performed in the Göteborg study. In the 7,578 in the control group who were actually screened in that study, 2,298 (30%) were biopsied at least once with a total 4,153 biopsies were performed overall. That is a lot of biopsies (a procedure that is not trivial) to extend the life of 34. Is it worth it? Again, different people will arrive at different conclusions. I’ve also said here that I would find a study where the control was to randomly biopsy the same number of (age-matched) men in the control group as are biopsied in the screening group very illuminating, but I doubt that will ever happen.

    Richard Ablin received a great deal of criticism after his recent New York Times editorial in which he said “the [PSA] test is hardly more effective than a coin toss.” Perhaps these results will encourage those who criticized to consider the actual data before giving an emotional reaction.

  3. Dear Jim: Both the Prostate Cancer Prevention Trial and the REDUCE study did in fact biopsy all participating patients at least once during each trial. Although neither of these studies could be categorized as “screening” trials, since they were not population-based, they do give a very clear indication of the probability of finding cancer on biopsy in a random population of older men.

  4. Hi Jim West (fellow Scientist↔Survivor Program buddy this year at the annual meeting of the American Association for Cancer Research – both of us representing prostate cancer survivors).

    Here’s another line on the “coin-toss” view, and why it is not helpful in educating men and physicians about the role and value of PSA.

    Since we are focusing on prostate cancer, other unhealthy conditions of the prostate can be viewed as “noise” in determining whether we are getting a signal from prostate cancer. Now we know that PSA is a multipurpose marker for not only prostate cancer but especially BPH and prostatitis as well, with each contributing to the overall PSA “signal.” Early in the PSA era, there was a lot of previously undetected prostate cancer, much of it relatively advanced and producing a strong PSA signal dominated by prostate cancer. Over time, with a lot of screening in the US, we smoked out many of those prostate cancer cases, including the vast majority of advanced cases. That inevitably led to a decline in the amount of PSA signal due to prostate cancer in the US population, and it led to relative increases in PSA signals due to BPH and prostatitis, which, for our purpose, is noise. Hence the coin-toss results.

    However, why should this development decrease our estimation of the value of PSA? As a multipurpose sentry, it alerts us that there is something that needs to be checked out, and we have a number of ways to home in on the possibility of prostate cancer (PSA velocity; PSA doubling time; PSA density; free PSA, PCA3; DRE; individual risk with age; race, and family history considered; challenge with 5-alpha reductase inhibitors); and trying to rule out other causes with cultures, antibiotic challenge, and determining the size of the prostate. All of these can be done before resorting to a biopsy.

    How helpful it would have been if the New York Times editorialist had presented that view!

  5. I just finished reading Richard Ablin’s commentary in the NYT cited by Jim West, above.

    Near the beginning of his article, Dr. Ablin states that he discovered PSA back in 1970, so his voice deserves consideration.

    However, it is readily apparent that he is not thinking through the issues with a sound critical approach. Here’s just one fact that shows that: his uncritical endorsement of the two deeply flawed studies published in the New England Journal of Medicine on screening last year.

  6. Sitemaster:

    Please correct me if I am wrong, but I believe that the two studies you mention did not track the results of follow-up treatment based on the outcomes of those biopsies. What I am proposing (very hypothetically because I am certain it will never be done) is a study where for every man who is biopsied in the study arm, another man in another arm of the study in the same age bracket is randomly selected and biopsied. After that point, they are treated exactly the same, with any subsequent treatment determined by the outcome of the biopsy and any other tests that may become relevant. This would determine if any observed reduction in disease-specific mortality resulted because the men were selected for biopsy based on PSA (that is, the PSA test is doing the screening), or because they were selected for biopsy for any reason (the biopsies are doing the screening). (A third arm, the control, which is neither screened by PSA nor given random biopsies would be needed, so I misspoke a bit above.) The data such as that given in the Cleveland Clinic study (and echoed by Stamey and Ablin) suggest that the PSA test itself is providing little useful direct information on who should be biopsied.

    Jim Waldenfels:

    I agree that the PSA test is useful in detecting advanced prostate cancer, which makes it useful as a diagnostic tool. However, the goal of screening is not to detect advanced cases (at which point it is too late), but to identify early stage cases that will ultimately advance. The data indicate that for this use PSA is little better than a coin toss, and therefore is of little use as a screening tool. Its use was based on the faulty logic that if advanced prostate cancer implies a greatly elevated PSA then a slightly elevated PSA must imply an early stage cancer that will ultimately advance. This has been shown to be invalid. As for the other conditions you mentioned, they do not need to be treated until they cause symptoms anyway, so what is the point in screening for them in asymptomatic men?

    You stated that “Over time, with a lot of screening in the US, we smoked out many of those prostate cancer cases, including the vast majority of advanced cases.” Is that true? As was just discussed a few days ago by Sitemaster and Terry Herbert in the thread on the Göteborg study, the total number of men diagnosed with late-stage cancer has not changed dramatically with the increase in screening. What has changed is that many more men are being diagnosed with insignificant cancers, so the proportion of late-stage diagnoses out of all diagnoses is smaller. But that is a meaningless statistic if the total number of late-stage cases is not being reduced. And finally, even if some late-stage cases really are being prevented (as is finally suggested by the Göteborg study after decades of unproven use), that is achieved only at the very high cost of needlessly treating many to benefit a few. Everyone needs to fully understand that rather than simply being told “Screening saves lives!”

  7. Jim:

    (1) No, the PCPT and REDUCE studies did not follow patients after treatment. But that wasn’t my point. What they do tell us is the probability of a positive biopsy regardless of the value of a PSA test (which is a key item your trial is trying to establish — except that you trial is certainly unethical and probably impractical).

    (2) I have already had one urologist suggest to me — on the basis of the Goteborg data — that PSA testing is a waste of time, and (as you imply) we should just start biopsying everyone when they are 45 or 50. He was only half serious — but that was still half!

    (3) There are many risk factors for prostate cancer. A PSA value is just one of them. And there are many options for management when cancer is found in the prostate. I have never been an advocate for mass, population-based screening. IMHO, selective testing, which includes PSA testing, and which is based on careful, individual risk assessment, has long been a better option.

    (4) The bottom line to all of this is still the same. Until we have a test that can disciminate, reliably, between men with a very high probability of indolent disease and men with a very high probability of clinically significant disease, we are stuck in a trap of our own making — and we are going to go on over-treating men who don’t need to be treated, because of fear and ignorance.

  8. Gentlemen,

    Frankly, I’m a little weary of the constant, somewhat debilitating academic debate about the worthiness of the PSA as a diagnostic tool in the discovery of prostate cancer. You can chronically debate the subject forever, between yourselves, but in my opinion anything that discourages the PSA test as a screening device for identifying prostate problems is a dis-service to patients.

    The significant improvement that is cited in prostate cancer mortality in most studies has been achieved throughout the PSA era, largely as a result of the use of PSA warning signals. A PSA result is not a diagnostic aid that is intended to be used in isolation from other diagnostic and clinical findings. It is one segment of the TOTAL information available to the doctor and patient, that should help determine the path to be taken in diagnostic follow-up.

    John@newPCa.org (aka) az4peaks

  9. Sitemaster,

    I’ve miscommunicated here, and I apologize for that. The motivation for my mostly (but I must admit not entirely) in-jest suggestion for the trial is my frustration over the misrepresentations of the significance of the Gutenberg results that are appearing all across the web now, many by people who should know better. It is being claimed to prove that PSA testing reduces mortality by half, proves that the benefits of PSA testing outweigh the harms, and, most annoyingly, proves that all men should be routinely tested. One advocacy group has even demanded that the American Cancer Society apologize for their current guidelines, even though authors of the study agreed that the ACS guidelines are best course of action.

    PSA being marketed as a binary “pass-fail” test despite the fact that data such as Cleveland Clinic, PCPT, and REDUCE studies (as noted by Ablin and Stamey) demonstrate that the PSA test is at best only a small contributor to the observed benefits. The amount of harm that has been done by these well-meaning but misguided advocates is large. The reason I said I would find the trial I proposed “very illuminating” (intended to carry an ironic inflection that clearly did not come across!) is that it would be the only way to clearly demonstrate to all involved how few of the observed benefits are due to the results of the initial PSA testing itself.

    As an aside, I do wonder if choosing who to biopsy by chance in one arm of study is really that much more unethical than choosing who to biopsy based on PSA levels that have been shown to be only slightly better than chance, or even performing a biopsy on everyone in the PCPT or REDUCE trials, but that is an irrelevant question since it would never be done in the first place. (I will state flat-out that I find the “free PSA screening” events for the past two decades despite the total lack of trial data showing a positive benefit-to-cost ratio to be highly unethical, but again that discussion is for another time.)

    Also please understand that I would never advocate that all men be routinely biopsied. But not only has your urologist colleague half-jokingly suggested it, it was once seriously proposed in the medical literature. I’m trying to find the correct keywords to google to find the link once again, but the suggestion was made that since they would already be fully prepped for a prostate biopsy and would already in an appropriate “position”, men undergoing a routine colonoscopy should also receive a prostate biopsy. At least the reaction to the suggestion was quite negative, but it shows how the judgment of some physicians can be clouded in the goal to prevent prostate cancer deaths.

    And finally, I am well aware that this site does not advocate routine, population-based screening for prostate cancer (by any means). That is in fact why I come to this site … you provide information and not blind advocacy, and I am always willing to reconsider my decision not to be screened for prostate cancer when new information becomes available. Unfortunately, the case is still extremely uncompelling. I am in full agreement that a better test is needed. (On the other hand, I am in routinely screened for colorectal cancer since the benefit-to-cost ratio has been shown to be far better, particularly for someone with a family history.) I greatly appreciate your efforts, and I’m sorry that my comments may have been misunderstood.

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: