“Screen smarter” and treat less is the message

The March 2013 issue of the AUA News carries a strong message from two leading experts on the early diagnosis and treatment of localized prostate cancer regarding the appropriate use of PSA testing for risk of prostate cancer and consequent follow-up.

The article — by Peter Scardino and Andrew Vickers of the Memorial Sloan-Kettering Cancer Center in New York — is entitled “Stop PSA screening, defend PSA to the death, or screen smarter?” and can be found on pages 27-28 of the March issue of AUA News.

Drs. Scardino and Vickers come down strongly on the side of “screen smarter”. They state that while one can disagree with the recommendation made by the U.S. Preventive Services Task Force (USPSTF) against the use of mass, population-based PSA screening, this does not mean that the USPSTF hasn’t made some very clear points that the urology community needs to accept and understand.

The authors argue that, while screening does lead to a reduction in mortality from prostate cancer in men who would otherwise not have been screened, there are three key steps to the smarter use of PSA testing to assess risk for prostate cancer, as follows:

  • Do not test men who will not benefit from testing. In addressing this issue, Scardino and Vickers note that we have clear data to indicate that
    • Widespread PSA screening for men > 70 years of age is just not beneficial.
    • Surgery is of little benefit to men > 65 years of age with low-risk disease.
    • Men of > 60 years of age with a PSA of < 1 ng/ml are at high risk for over-diagnosis and over-treatment.
  • Do not biopsy men without good reason. In support of this statement, Scardino and Vickers point out that
    • Most men with an elevated PSA do not have prostate cancer.
    • PSA levels vary under normal circumstances, and most men should have a repeat PSA test (after about 6 to 12 weeks) before consideration of a biopsy.
    • New tests are now available to assess whether a biopsy is appropriate.
    • Biopsies are rarely necessary in men with a PSA level of < 3 ng/ml.
  • Don’t treat unless you have to. In making this statement Scardino and Vickers observe that
    • Active surveillance should be considered for management of any low-risk cancer (and for intermediate-risk cancers in the elderly).
    • All definitive treatments for localized prostate cancer carry risks for side effects that may significantly impact patient quality of life.
    • When treatment is needed, outcomes are generally better when treatment is delivered by high volume practitioners or at high volume centers.

In a strong message to their peers in the urology community, Scardino and Vickers state very clearly that:

Unless we accept that PSA testing may lead to unintentional harm, we will not modify the [indiscriminate] use that led to the USPSTF recommendation to stop screening, and primary care physicians may stop testing altogether.

This article by Scardino and Vickers correlates closely to the presentation given by Dr. Ian Thompson at the Genitourinary Cancers Symposium in Orlando in February.

The “New” Prostate Cancer InfoLink thinks it is going to take a while for this set of messages to really sink in and be acted on in an appropriate manner, but the weight of evidence is clearly growing to suggest that greater care about who to test, who to biopsy, and who to treat will be in the best interests of both patients and physicians over time.

8 Responses

  1. Men of > 60 years of age with a PSA of 3 to cause any sort of concern at all?

  2. I cannot speak for Drs. Scardino and Vickers, but I think that what they would tell you is that any man who is > 60 years of age with a PSA of about 3 ng/ml should get a repeat total PSA test and (probably) at least a %free PSA test before anyone decided to biopsy him (unless he had a clear family history of prostate cancer and/or a positive digital rectal exam). They are not recommending “cookbook medicine”. They are saying that every patient needs to be assessed as an individual but “first do no harm”.


    I’ve read the piece by Drs. Scardino and Vickers, and I’m hopeful the strong and wise stand of these well-known prostate cancer experts will influence more of the urologic surgery community toward smart screening. I’m still looking at some of the details, but they have made good points.

    I also looked at their reference #2, the paper by Dr. Fritz Schröder, J. Hugosson and colleagues in the March 2012 edition of the New England Journal of Medicine (“Prostate-cancer mortality at 11 years of follow-up”) which is available for free online. This is the latest update of the European Randomized Screening for Prostate Cancer trial (ERSPC), one of the two key studies upon which the USPSTF relied, the other being the US Prostate, Lung, Colorectal and Ovarian cancer trial (PLCO). At the time of publication in the NEJM back in 2009, many of us were stunned by the prematurity of the data upon which conclusions and summary figures rested, as well as numerous serious flaws from the perspective of whether it was appropriate to publish research at that time. In addition to prematurity, another key issue, among several, included a disturbingly large percentage of non-compliance by “research participants” (meaning subjects, a.k.a. patients in the study) with randomization assignments. To the research-oriented among us, the relatively large percentage of non-compliance called the effectiveness of randomization and the reliability of statistics requiring effective randomization into very serious question.

    Having noted that reported “follow-up” was not started from the “date of diagnosis,” a starting point that is customary and that many have assumed applied to these studies, but rather from the often much earlier date of the patient’s assignment by randomization to either the screening or non-screening arm, we thought that survival result comparisons based on follow-up from diagnosis of less than 10 years would likely tell us very little. (In other words, follow-up of 11 years from randomization is really much shorter than follow-up from diagnosis, no doubt well short of 10 years.) That’s because survival in the US at that 10-year point is so high, with Europe presumably not that far behind. In any disease where 10-year survival is very high, a researcher simply is not going to be able to count many deaths at the 10-year point; he will need more years to see if an intervention, such as screening, is making a difference. Therefore, apparently unlike the ERSPC researchers themselves, we expected that the survival advantages of screening in the ERSPC trial, were likely to surge in prominence as years of follow-up grew.

    Well, drum roll here, that’s exactly what the updated ERSPC results are showing, though I’m a bit puzzled that the advantages are surging so quickly! (I have a hunch about that.) Results are showing the valuable advantages of screening even at the still quite premature point of 11 years since randomization (clearly well under 10 years since diagnosis). The new study has a very informative and lengthy supplement section with an abundance of statistical detail, and that’s where you can see the advantages of screening emerging. The supplement includes tables comparing the screening and non-screening groups such as Table (or figure) 2a, showing a clinical stage comparison (p. 18); 2b, showing a Gleason score comparison (p. 18); 2c, showing a risk group comparision; 3A, showing the number needed to invite to participate to prevent 1 death (p. 20); a table showing the numbers needed to detect to prevent one death (p. 20); and 9A treatments (enabling comparison of treatments normally used for early stage disease versus those used for later stage disease, during the period when the men would most likely have been treated.) Unfortunately, I did not find data on the time periods from diagnosis to an event (such as death from prostate cancer, the primary scoring event in the ERSPC).

    I really hope we can get into the specifics, but to me the evidence looks persuasive toward the position that screening makes a clear and valuable difference in men’s lives, with the magnitude of the advantages in the ERSPC screening group only likely to increase substantially as years of follow-up accumulate. Just for instance, Table 3A regarding the numbers needed to be invited to participate in screening to prevent 1 death plunged from 2,111 at 9 years of follow-up, to 1,055, half, with just one more year of follow-up at 10 years, to 936 at 11 years. That is happening despite the fact that the “invitation” statistic does not consider the impact of compliance. In other words, substantial portions of men deemed to be in the screening group were not screened in fact, and vice versa, a circumstance that weakens the extent of difference seen in a trial as it is executed as compared to what was planned. As other reported statistics strongly suggest, those “numbers needed to invite” would drop substantially if compliance were considered, even with follow-up since randomization still at an anemic 11 years. How misleading and tragic it was that such a high number as 2,111 was reported and echoed around the medical and media worlds as if true! To me, it is still premature to publish any such “numbers needed to” figures except to show early trends.

    Also, we should bear in mind that “screening” in the ERSPC trial was far looser than the screening those of us in the US are accustomed to. Rather than the annual screening we might assume for the trial, as per protocol, the screening interval was 4 years (2 years in Sweden only), and the actual median screening interval was 4.02 years. That “loose” definition of screening, as I see it, is likely to substantially downgrade the impact of screening measured by survival. That’s because a screened man without prostate cancer at the initial screen would give cancer nearly a 4-year head start if he developed the disease shortly after the initial screen. For the vast majority of us, that would not be critical as our disease would be slow-growing, but for those like me with short PSA doubling times and a real threat of lethality, loss of earlier detection would be more likely to prove deadly. Just try running the numbers starting at a PSA of only 1 with a doubling time of 3 months. (In just 2.5 years the PSA exceeds 1,000 and is leaping upwards.) In short, I’m convinced that a shorter screening interval would have substantially boosted the already favorable results toward screening by saving more men in the screening group from early death.

    I expect updated PLCO results will take a similar path, but I’m hopeful the USPSTF will feel honor bound to reconsider its position.

  4. Dear Jim:

    I do understand your argument, but the whole point of a prostate cancer screening trial is to measure time from initiation of screening to prostate cancer-specific mortality, not time from diagnosis. The whole concept of the trial is based on the idea that screening will reduce risk for prostate cancer-specific mortality, regardless of when diagnosis occurs!

    The times from initiation of screening to diagnosis and from diagnosis to prostate cancer-specific death are at best secondary endpoints in a screening trial … otherwise it wouldn’t be a screening trial!

  5. Dear Sitemaster:

    I’m responding to your reply to me of 4:37 pm, and I certainly agree with your comment.

    But we need to bear in mind the vital point that follow-up time must be adequate in order for any real differences to have time to appear. Not to beat a dead horse, but, for example, a follow-up of, let’s say 3 years would obviously be too short, as surely all would agree. It’s likely we could get a strong consensus that 5 years would be too short. Based on studied survival times achieved in the recent past of 10 years since diagnosis for the vast majority of men in the US, and recognizing there will be a lag of likely several years at least between randomization and diagnosis in the screening trial, it seems clear that we should be thinking at least around 10 years (at least) plus a few more years, or perhaps 13 to 15 years since randomization before we would expect to see substantial mortality numbers from the disease. Even at that point, we would expect that mortality from the disease would continue to rise substantially for some years.

    If that is a sound perspective, then we still are looking at very early mortality figures in the recent report of the ERSPC because the average reported follow-up from randomization was just 11 years. That’s why the period from diagnosis to mortality looms as so important to me. If we were now at, let’s say, the 20-year point since randomization, I would be far more comfortable that we were getting a good grip on meaningful mortality figures, and I probably wouldn’t be concerned about the time since diagnosis to mortality.

    That’s a long way of saying that I think we are seeing the issues the same way.

  6. Dear Jim:

    I’m actually not sure we are seeing the issue the same way at all. In my view (as I have said several times before) both the ERSPC and the PLCO studies were so poorly designed and/or implemented as to be meaningless.

    The ERSCP study doesn’t even meet the criteria for a single randomized trial. It is simply a meta-analysis. When one starts quibbling over the theoretical implications of studies like these, the whole discussion (in my humble opinion) becomes pointless for the simple reason that the data on which the whole discussion is based are fatally flawed.

    The recent, final, 30-year follow-up data on the 223 Swedish men managed by watchful waiting shows that the actual prostate cancer-specific mortality rate in these men was only 17%. The implication is that among men being diagnosed early enough to receive curative therapy, it might actually take a follow-up of 40+ years from initiation of a screening trial to demonstrate whether there was or was not a real survival benefit from screening, and that in doing this study one would know from day 1 that 83% of the men who got curative treatment would (arguably) never have needed that treatment!

    My entirely personal opinion is that that is not a trial that I would want to participate in! It might even be unethical. My risk (from side effects and complications of treatment) would potentially far outweigh any possible benefit compared to active surveillance. This implies that in the unlikely setting that anyone would fund such a 40+ year study, it would now have to be done with active surveillance as the first-line management for all participants diagnosed with low-risk disease, and that is a very different trial from the PLCO or the ERSCP studies!

  7. CONTINUING THE DISCUSSION OF THE ERSPC (European Randomized Screening for Prostate Cancer trial),

    Hi again Sitemaster. Your reports so often help me think, as do the ongoing discussions, and this discussion is certainly typical of that. Thanks. Also thank you for reading all our replies (sometimes long, sorry) and thoughtfully responding.

    This picks up the discussion from my first response covering the ERSPC of 4/5 1:29 PM, your response of 4:37 PM, my reply of 9:19 PM, and your response of 10:11 PM. (Doesn’t anyone else want to join in?)

    I actually think we are very much in agreement on the key issues regarding smart screening but likely in disagreement about what can be helpful as evidence.

    I too see the ERSPC as so flawed as to be meaningless for “inferential” statistical analysis claims, including reliable means, standard deviations, confidence limits, and p values. The USPSTF is apparently addicted to evidence that supports such indicators, and it seems to me they cannot discern, at least for prostate cancer, when their evidence is so flawed that it does not support those claims.

    Your point about the follow-up needed led me to think I have been too accommodating in thinking that follow-up of, let’s say, around 15 to 20 years would give meaningful information in a well-done trial. I’m now seeing that much longer follow-up would be needed. However, I’m thinking that shorter periods would give us some potentially useful interim information. I too would be fully in favor of active surveillance for low-risk patients in the contemplated (though impractical) screening trial.

    EVIDENCE. I believe you and I may have different views of what constitutes useful evidence for clinical decisions, possibly including advice about screening. While we all would like to see evidence from perfectly done Phase III trials to support clinical action, I am willing to favor action based on evidence with more uncertainty and flaws, if that evidence, thoughtfully considered, seems to have some worth in helping us figure out what to do. If I were in charge of the USPSTF, it would take this view too, at least for the issue of prostate cancer and screening, in part because of the point you make about the seeming impracticality of executing a really sound trial. I recognize that the USPSTF has a role that differs from that of a doctor (and patient) who must make clinical decisions even where information is far from perfect; in contrast, the USPSTF is focused on reliable guidelines and needs to have a body of convincing evidence before it can apply its official seal of approval or disapproval. I also recognize a point you have often made that using such less formal and technically elegant evidence leaves open a potentially substantial possibility of error. But, putting this in an action perspective, the question is whether there is a better alternative.

    Throughout my career I was involved in forecasting outcomes (usually cost outcomes for R&D projects in the context of quality and schedule requirements). We were sometimes pleasantly surprised to find nice, neat, informative evidence and formulas that resolved all issues. However, the typical available evidence was uncertain, often flawed, and often with evident bias. Yet so often key and meaningful clues were available in that evidence, especially when seasoned heavily with mathematics, tested with logic, and sifted through the filters of consistency and comparison to other evidence (like an optometrist presenting lenses for our vision). Those clues gave us confidence that our choices were better than if they had been guided merely by flips of a coin. My experience as a patient has been similar; there simply was no definitive evidence available for key decisions, but there was an abundance of clues from somewhat flawed evidence.

    I am far from an expert in medical history, but I’m aware of breakthroughs by men like Ignaz Semmelweiss (child bed fever), Louis Pasteur (germ theory, anthrax, Pasteurization, etc.), Edward Jenner (proof of smallpox vaccine), and Alexander Flemming (penicillin), to name some standouts known to many of us. Obviously, these men were not relying on Phase III trials, but rather they achieved their breakthroughs by observing clues and using their minds. I’m most grateful we now have Phase III trials, but I am convinced we sometimes disregard valuable evidence if it does not meet Phase III standards. To me, that is a critical failing in the approach of the USPSTF.

    There is a most legitimate concern about over commitment to actions based on the less than perfect evidence for which I am advocating. That concern is one reason I hold back on advocating as a primary therapy for certain patients a short course of about a year of triple androgen deprivation therapy plus ongoing maintenance with a 5-ARI drug. I’m quite impressed with the clues I see and have personally experienced, but I am concerned about the incompleteness of the evidence, both for efficacy and side effects. Similarly, as another example, a prominent medical oncologist dedicated to prostate cancer held off for years on advocating that much recurrent metastatic prostate cancer involves just a few metastatic sites and is potentially curable. Now, with the emergence of some published data (and his own personal success with his challenging case of metastatic prostate cancer), he is explaining the approach publicly and having patients treated with curative intent, recognizing that the approach is still investigational. (I am one of them, though, surprisingly, advanced scanning revealed no detectable metastases last year.) While I’m convinced the evidence for screening is not yet conclusive, I feel greater confidence about recommending smart screening, based on an abundance of clues and the tests of consistency, logic, what we know about the disease, and so on. I do not feel we are over committing ourselves.

    I hope to see or post here some of the clues I see in the interim 11-year ERSPC results about screening. Please let us know if you feel we have beat this topic to death.

  8. Dear Jim:

    Respectfully, I think everyone has beaten this topic to death!

    There is no doubt at all that regular testing has found aggressive prostate cancers early in some men that has led or will lead to them not dying from those cancers. There is also no doubt at all that early diagnosis has led to hundreds of thousands of men getting over-treated.

    The extreme position taken by the USPSTF is arguably just as crazy as the extreme position taken by others (largely in the urology community) that all men should receive annual PSA tests after age 50. From specific points of view, both are arguable — but probably not justifiable because there are no good data to support either position, and there are unlikely ever to be any good data unless we can find a much more sophisticated indicator for clinically significant prostate cancer than anything I know of at this time.

    In the interim, the concept put forward by Scardino and Vickers, which is not dissimilar to a concept I also aired a while back, seems reasonable. It also meets the criteria of most of the guideline-issuing bodies because it invokes the necessity of a meaningful discussion between a patient and his doctor. I know it ain’t easy, but it does meet what I would describe as the “clinical common sense” criterion. The core question is only whether the urology community is willing to practice against it (or something like it). There is, at least, a movement in this direction.

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: