Updated data from ERSPC trial still show no impact on all-cause mortality

A new article in the New England Journal of Medicine this week has updated the prostate cancer-specific and all-cause (overall) mortality data from the European Randomized Study of Screening for Prostate Cancer (ERSPC).

This latest analysis of data by Schröder et al., and based on a randomized comparison of screening (with regularly scheduled PSA tests) as opposed to non-screening, has shown the following results in the predefined, core group of men aged between 55 and 69 years at the time of enrollment:

  • The average (median) follow-up for men in the core group was 11 years.
  • There was no significant difference in all-cause mortality between the groups who were or were not screened for risk of prostate cancer.
  • Reductions in the risk for prostate cancer-specific mortality in men randomized to the screening group as compared to the unscreened group were
    • An absolute reduction of 0.10 prostate cancer deaths per 1,000 person-years
    • An absolute reduction of 1.07 prostate cancer deaths per 1,000 men who underwent screening
    • A relative overall reduction in the risk of prostate cancer deaths in the screening group of 21 percent
    • A relative overall reduction in the risk of prostate cancer deaths in the screening group of 29 percent after adjustment for non-compliance with screening
  • To prevent a single case of prostate cancer-specific mortality
    • 1,055 men would need to be invited to be screened
    • 37 cases of prostate cancer would need to be detected

The authors conclude that this analysis, which adds two additional years of follow-up data to the data originally published in early 2009, shows that “PSA-based screening significantly reduced mortality from prostate cancer but did not affect all-cause mortality.”

For those who are not familiar with this study (the largest study of screening for prostate cancer ever carried out), it enrolled 182,160 men (from eight European nations) between the ages of 50 and 74 years at study entry. It included a predefined core group of 162,388 men aged between 55 and 69 years of age. Men were randomly assigned to either the screening group (who were offered regular PSA-based screening tests) or to a control group (who were not offered such screening). The primary outcome was prostate cancer-specific mortality. The first report of data from this trial was published in early 2009 based on a median 9 years of follow-up.

These new data are unlikely to help to clarify the debate over the value of mass screening for prostate cancer. From one point of view one can use them to argue that screening can prevent between 20 and 30 percent of prostate cancer-specific deaths. From the alternative point of view, one can argue that screening a million men would indeed prevent about 950 prostate cancer-specific deaths, but would also lead to the potential over-treatment of 36 out of every 37 cases of prostate cancer identified, and would have no impact whatsoever on overall mortality.

77 Responses

  1. At the very least, no man should get tested without being informed about the high odds of getting a treatment that does not improve his survival. If, however, the small chance of lowering the risk of dying from prostate cancer is acceptable, then and only then should screening occur. Importantly, however, the editorial points out the potential problems with the European study, namely, that differences in treatment are the reason for their findings rather than the screening itself. Also, the US trial cannot be ignored.

  2. Too bad this research did not tease out the statistics showing the likely benefit of PSA screening in younger than the 55-69 age study population. It would seem logical that baseline PSA screening would catch those younger victims and early treatment would significantly reduce cancer-specific mortality in this younger age subset whose normal life expectancy would be much higher than those included here.

    We have two prostate cancer survivors in my support group who were PSA diagnosed with operable cancer in their 40s. Without that screening they would most likely be dead by now from prostate cancer.

  3. I’ve made this point before, but it is inappropriate to put any weight on the statistical insignificance of the effects of all-cause mortality. The European study simply does not have sufficient size to detect such effects.

    If one looks at Figure 1 in the paper, we find that 72,891 men were in the screening group, of whom 299 died from prostate cancer, and 13,917 died from all causes; 89,352 men were in the control group, of whom 462 died from prostate cancer, and 17,256 died from all causes.

    If one calculates rates: the screening group had 0.410% die from prostate cancer versus 0.517% who died in the control group, so the prostate cancer death rate declined by 0.107%. If one looks at overall mortality rate, the screening group had 19.093% die from all causes versus 19.312% in the control group. This is a decline of 0.219%, which is actually over twice the decline in prostate cancer deaths. In other words, if one believed the point estimates, there is no sign that non-prostate cancer mortality changes somehow “offset” the prostate cancer mortality effects.

    Yet the difference in prostate cancer death rates is statistically significant between the two groups, whereas the difference in overall mortality is not even close to being statistically significant. Why is this? The intuitive reason is that, statistically, it is much easier to detect the larger, relative change from a 0.517% death rate from prostate cancer to a 0.410% death rate from prostate cancer than the much smaller, relative change from a 19.312% death rate to a 19.093% death rate, even though the absolute change in mortality is actually somewhat larger in all-cause mortality.

    If I take these data, and the actual sample sizes, and ask the following question: what would be the probability in a sample of this size in the treatment and control groups, with a baseline all-cause mortality rate of 19.312%, of detecting a decline by 0.107% to 19.205%, at a statistical significance level of 95%, the answer is, the probability would be only about 8%. In other words, the sample sizes are simply not big enough to make it at all likely that we could detect statistically significant all-cause mortality declines of the size of the observed prostate cancer mortality declines.

    If I asked instead, how big a sample, with equal numbers in the treatment and control group, would I need to have an 80% probability of detecting an all-cause mortality decline from 19.312% to 19.205%, at a statistical significance level of 95%, what statisticians call a “power” of 80%, the answer is, I would need 2.35 million in the treatment group, and 2.35 million in the control group. This is far larger than any likely sample sizes we will ever get in prostate cancer screening studies, even if we do a meta-analysis to pool all results across all studies.

    Therefore, in debating this issue, the issue of statistically significant effects on all-cause mortality is a red herring, a distraction. The fact that they found nothing statistically significant in all-cause mortality means nothing because the study was simply not designed to be large enough to detect such effects at the likely magnitude that we might expect.

  4. It also is of interest to compare differences in prostate cancer mortality in the treatment vs. control groups vs. differences in actual treatments that could cause serious side-effects.

    Table 9B in the supplements to this paper shows that of the 6,963 men diagnosed with prostate cancer in the screening group, 1,600 received only “watchful waiting” as a treatment. In the control group, of the 5,396 diagnosed with prostate cancer, 861 received “watchful waiting” as a treatment. Presumably this is related to the fact that the screening group had, on average, prostate cancer diagnosed at an earlier stage. There is also a higher percentage of the control arm that received hormone therapy.

    Therefore, the percentage of men who actually received some treatment other than watchful waiting is (6,863 – 1,600)/72,891 = 7.358%. In the control arm, the percentage of men who actually received some treatment other than watchful waiting is (5,396 – 861)/89,352 = 5.075%. So an extra 2.273% received some treatment other than watchful waiting due to the screening assignment.

    The percentage of men who died from prostate cancer was 0.410% in the screening group versus 0.517% in the control group, a decline of 0.107%.

    If one calculates the ratio of (a) the extra percentage receiving treatment other than watchful waiting to (b) the extra percentage not dying from prostate cancer, the ratio is 21.4.

    In other words, for every man who does not die of prostate cancer during the follow-up period because of the screening random assignment, there are a little over 21 men who must be treated in some way other than watchful waiting, and who in many cases will get side effects. I think this gives a little different picture than the headline numbers from the latest European study paper. If we’re concerned about the ratio of treatment side effects to lives saved, we should focus on this ratio, rather than the ratio of prostate cancers detected to lives saved.

    These numbers reflect 23% of the screened group who had prostate cancer being assigned to watchful waiting only. But the European study also reports that 60% of the screening group with cancer was considered to have a low-risk cancer, versus 42% of the control group. If a higher percentage of the low-risk group could be persuaded to adopt active surveillance, I suspect the ratio of men treated to lives saved would drop.

  5. Setting aside the details of this study (or similar studies) for a minute, for every seven men diagnosed with prostate cancer in USA this year, one man will die of it unless we have significant changes in prostate cancer detection and / or treatment.

    This is based on the latest SEER data for prostate cancer available on their web site: viz. “It is estimated that 240,890 men will be diagnosed with and 33,720 men will die of cancer of the prostate in 2011.” The data table also shows that in 2008, 153.03 men were diagnosed with prostate cancer and 22.82 men died of prostate cancer per 100,000 men in USA.

    I do believe that we can improve our method of dealing with or treating prostate cancer to achieve results much better than this. Now, if we improve cancer detection and treatment so that only 1 person in 14 (for example) dies of prostate cancer, won’t that be a major improvement? Yes. We will have to treat 14 men better to reduce the death rate from 2 to 1, but won’t that be worth it?

  6. Dear akt1000:

    I think it is rather important to point out that of the estimated 240,890 men diagnosed with prostate cancer in 2011, significantly less than 1% were projected to die of the disease within the ensuing 12 months. Similarly, of the 33,720 men who were estimated to die of prostate cancer in 2011, almost all would have been diagnosed at least 5 and many 10 or even 20 years earlier. The time factor is important here.

    No one is arguing with your statement that a reduction in the risk of dying of prostate cancer would be good. The problem is that we have no idea (at present) about (a) how to accurately select men who really do need treatment because they have clinically significant disease that might lead to their deaths or (b) whether treatment of some of these men will, in fact, make any difference to whether they die of prostate cancer or not.

    At the present time, when men have demonstrably metastatic prostate cancer (whether evident on an MRI, a CT scan, or a bone scan or truly micrometastatic) our ability to cure any of these men is minimal. … All we can do is alleviate their suffering (maybe for as much as 20 years). Conversely, we know that many men with low-risk prostate cancer can live with this cancer without any clinical impact for 20 years and more, so they don’t actually need treatment … but we can’t tell accurately which men with low-risk disease actually have this indolent form of the disease. It will take a solution to one, the other, or ideally both of these problems if we are to reduce the risk of dying of prostate cancer and the risk of being pointlessly over-treated for indolent prostate cancer.

  7. I agree with everything you say.

    Yes. The SEER data does not say anything about the time when that 1 person (out of every 7 diagnosed of prostate cancer) is going to die of prostate cancer. Yet, I assume that at some time in the year before he dies of prostate cancer, he will begin to show metastatic prostate cancer. If it is too late to do anything good for him at that time, that means that we have to start treating him earlier to make his life better.

    As you said, we need to have better methods of identifying men with low grade prostate cancer who may not need to be treated. But the real answer lies in not only developing better methods for identifying such people, but also developing better methods for identifying prostate cancer earlier and treating prostate cancer more effectively with less side effects. My main problem with the way some of the trial study results are presented is that their message seems to be that even if we don’t do anything, it won’t make much difference.

  8. Dear akt1000:

    Just so we are clear, survival times today for men discovered to have micrometastatic or early metastatic disease are usually well in excess of 5 years from the time such men start hormonal therapy. There are exceptions to this among men with very aggressive forms of prostate cancer, but these men are the exceptions rather than the rule.

    Even for men with chemotherapy-naive, metastatic, castration-resistant prostate cancer, median survival times of around 18 to 24 months (1.5 to 2 years) are predictable, as demonstrated in completed Phase III clinical trials.

  9. None of this means anything until you or your loved one are diagnosed with prostate cancer. Then all options need to be reviewed. Death from prostate cancer and the metastatic course it takes is a horribly painful way to die. Yes, prostate cancer grows slowly. Age of diagnosis, expected life expectancy based upon general health, along with symptoms being experienced should be the guiding information for a patient and his physician to decide course of action. Thank heavens for the development of the PSA and its impact on prostate cancer survival.

  10. To RobC:

    One cannot assume that finding cancers in younger men will definitely be better. In all the breast cancer screening studies, women under age 40 who were diagnosed did not show an improved survival. The benefit was only found in the women who were over 50 at the time of their diagnosis.


    Sitemaster’s comment that these new results will not clarify the debate about screening is right on the money! Major flaws in the ERSPC have been noted by Sitemaster and a number of us above and previously, but once again we need to remember that these updated results are again simply premature, and that precludes meaningful estimates of impact on survival.

    In essence, these results are premature because the median follow-up of 11 years from trial enrollment for the core group in the ERSPC is too short in view of the fact that virtually all men diagnosed with prostate cancer, with a few clear exceptions, will be alive at the 10-year point from diagnosis, at least in the US. It is obvious that there will be a lag of some years from trial enrollment to diagnosis for many trial participants who are diagnosed, and thus the median follow-up from diagnosis in the ERSPC is well short of 11 years. This means an extremely high likelihood that the effects of late diagnosis due to absence of screening, in the case that screening does make a substantial difference in survival, will not be evident for a number of additional years. My impression, considering current progress in management and treatment of the disease, is that it could easily take about another 4 to 10 years before we would have some opportunity to see meaningful evidence of an effect of screening. However, as Sitemaster and others have observed, flaws in the trial are likely to leave us with large questions even at those points.

    I am at a loss to understand why apparently sincere and dedicated researchers and physicians associated with the ERSPC again ventured a calculation of the numbers of us needing to be screened and cancers detected to save one life. If they want to assert that they found that, let’s say for example based on my conjecture (informed by the Goteborg group results), at the eighth year of follow-up from diagnosis 1,055 men would need to be screened and 37 cancers detected to save one life from prostate cancer at that very early point, well and good, though not impressive — the kind of statement that generates a “duh” type response.

    I am also at a loss for understanding why more of us do not appreciate the criticality of inadequate follow-up in the ERSPC at this point. Maybe I’ve just become more sensitive to real odds of survival as a high-risk patient (bPSA 113.6 in December 1999 — first ever; Gleason 4 + 3 = 7 Epstein; stage III; all cores positive, most 100%) who has done very well on intermittent triple androgen deprivation therapy with 5-ARI maintenance and supporting drugs. Now in my thirteenthyear since diagnosis, I’m nearing the 2-year point of my third vacation from Lupron and Casodex/bicalutamide, Thanks to advances and improvement in my case, I even seem to have a reasonable shot at a cure, which I’m pursuing.

    The damage done by the ERSPC paper is that many physicians, researchers, organizations, policy makers, and media reporters will, once again, fail to appreciate that the 11-year results are still way premature, and their advice to patients, discussisons in papers, policies, and reports will again lead people to think that screening has been shown to be of questionable worth. This is the thinking that I tuned into for years before insisting on that first PSA test that was almost too late!


    Hi Dr. Chodak. I have a question regarding your post of March 15, the first response posted. You did not work “active surveillance” into your information to the patient as a way to reduce the odds, perhaps reducing the odds sharply, of getting unnecessary treatment. Do you feel that men considering screening should not be informed about active surveillance as a viable option for appropriate cases?

    What evidence can be cited of “high odds of getting a treatment that does not improve survival”? (I have a response pending expressing severe doubt that the ERSPC provides any such evidence at this point.)

  13. Analyzing surgery methods is looking for the best of the bad, and yes possibly for unnecessary surgery.

    If PSA testing is giving earlier diagnosis, but not better survival, then it is in fact, in many cases, giving more misery than relief. Does that mean more knowledge is bad or simply that knowledge is being badly used?

    Watch and wait is not the same as watch and manage. Would I take a little chemo, a little radiation, or the removal of just the bulk of the tumor in trade for fewer years of life with bodily functions? I would, and that should be my choice. My choice, which is not available to me.

  14. That’s right, Mike. And if men don’t want to be permanently maimed, they should have access to non-maiming and reversible treatments even if their disease is not metastatic. If it’s possible to control metastatic disease for 20 years with intermittent castration and chemotherapy, how long could slow-growing, non-aggressive tumors be managed? Perhaps with just one round of ADT and follow-up Avodart? Men are never going to have a shot at low impact, side effect-reversible disease control while surgeons rule the roost. How about putting the patients first?

  15. The problem is simple. People — as they always will — desire certainty in an uncertain world. Currently that is impossible with respect to screening for prostate cancer.

    There are many bad ways to exit this world and, frankly, prostate cancer is probably better than say MS or ALS or any number of other gruesome fates. That is life and that is death.

    All the European data demonstrate is that a particular demographic have a higher risk of dying from cancer of the prostate in Sweden and the Tyrol but not in Spain. Asians have a very low risk and African Americans a high risk. An autosomal dominant trait is estimated to account for 10% of the incidence. Generally absent from the risk side of the debate is the incidence of depression (10% after radical prostatectomy) — no doubt interrelated to erectile and urinary dysfunction.

    Dr Bartlik may be accurate with respect to the statistics, but it would be arrogant to assume that removal of a body part does not have unforeseen consequences with respect to other organs — and therefore to all-cause mortality.

    The road to hell is paved with good intentions and nothing should be assumed with regard to the utility of the prostate. Reflect for a moment on the following — Avandia, sibutramine, DePuy hip prostheses, calcium supplementation plus aledronate, Vioxx, hormone replacement therapy, the putative increase in colorectal cancer following cholecystectomy, and on it goes. The arrogance and ignorance of the medical profession is without end.

    Doctors have a duty to inform — but not to dictate to people. They can also say “I don’t know.” Interventions need to be titrated to individual psychology.

    I am more than a little disturbed by those who aggressively promote any intervention — it is tautological that they have lost sight of the fact that there is infinite variability in peoples’ world view and how they wish to live and die. In the communities I work with in Auckland, New Zealand, the happiest and most content ethnicities have the shorter lifespans and the highest consumers of antidepressants are the wealthiest.

  16. We all bring our biases to these discussions. If threatened by aggressive prostate cancer we are preoccupied with avoiding premature death and if less threatened we are perhaps more preoccupied with avoiding negative side effects from unnecessary over-treatment. I acknowledge my bias from the former situation.
    The aim of this exercise is surely to quantify (as best we can) any benefits that may flow from screening activities and also any risks and costs that may also flow from those activities. That information should be presented to patients who can then (in consultation with their doctors) make decisions on screening that are best suited to themselves.

    My comments below relate to better quantification of those benefits and risks. My bias affects my perspective but surely honest debate is the only way to thrash this thing through to a best compromise with the information we have.

    (1) I simply cannot understand this preoccupation with all-causes mortality versus prostate cancer-specific mortality. Every time there is an apparent benefit to prostate cancer-specific mortality from screening, the lack of proven benefit to all-causes mortality is wheeled out as a valid rebuttal — or perhaps even a better one. Tim Bartik (above) indicates that these screening trials simply do not have the statistical power to demonstrate a benefit or otherwise to all-causes mortality. I am not sure whether his statistical argument is valid or not but it does sound plausible. As far as I can see his argument has been roundly ignored in subsequent discussion. So is he right or not? Making progress in this involves addressing issues like this rather than ignoring them. Mortality patterns with prostate cancer are complex, with long lead times and substantial overlap with other mortality causes in older age groups. Most men do not die of prostate cancer, so demanding that a benefit in a minority condition like prostate cancer should be obvious in all-causes mortality stats is asking more than is reasonable. If there was a benefit in all-causes mortality, how confident would we be that a prostate-cancer mortality benefit was the reason in any case?

    (2) The PLCO study has the most contamination and has no prostate cancer-specific mortality benefit. ERSPC has intermediate contamination and has 21% benefit. Goteborg has the least contamination and has 40% benefit. No proof but I do detect a trend in these numbers. My understanding is that attempts to take account of the contamination increase the level of benefit (from 21% to 29% in the case of ERSPC). I refer you to the journal article “Mortality reductions produced by sustained prostate cancer screening have been underestimated” by James A. Hanley (a Canadian epidemiologist) in J Med Screen 2010; 17:147-151. He makes two points:
    (a) Prostate cancer-specific mortality benefit is increasing in these screening trials during the latter part of the trial so more time may further increase the benefit.
    (b) There was no prostate cancer-specific mortality benefit in the first half of the trial period but the benefit (e.g., 21% in ERSPC) was calculated by averaging over the full trial period. If it was averaged over the period where the benefit was actually accruing (the second half) the average benefit would double.
    Hanley also makes the point that all these screening trials recruited participants from a range of age groups (e.g., 50, 55 to 64, 69, 74). If they had recruited a younger end cohort (e.g., 50-55) the screening benefits would likely be greater. Is Hanley right in this? I have seen no rebuttals in the literature to date. Can you rebut his arguments? Or should we just ignore them? I do take the point that absolute numbers of men getting benefit seems low and that needs discussion (and perhaps more complete data).

    (3) Where do we stand on the quantification of over-treatment rates for prostate cancer. ACS and AUA hint at rates of 20-40%. While these rates seem plausible, I had difficulty following the possibly incomplete logic they used to arrive at them. I would be appreciative if anybody could add to my knowledge on this. The comment that potentially 36 out of 37 men diagnosed with prostate cancer are at risk of over-treatment is unbalanced in my view. There seems a line of thought that men diagnosed with prostate cancer either have indolent cancer that needs no treatment or they are dead men — palliative care fodder. This ignores the many men who are successfully treated and the large army of advanced prostate cancer guys who are still living rich and fulfilled lives. Can I suggest we ignore a bit less and consider all the aspects of this issue in an appropriate fashion.

  17. Tony:

    Your questions are valid. However, the base problem is that they are unanswerable given the available data. Tim is completely correct when he points out that to get really accurate answers to any of these questions would require studies in (probably) millions of men who were rigorously assessed and followed according to carefully prescribed protocols over something like 30+ years, and also carefully stratified by age and other factors. It ain’t gonna happen … and so everything becomes speculation based on assumptions and presumptions.

  18. On the all-cause mortality issue, you don’t have to take my word for it. All you have to do is look at the confidence intervals reported in the European study for all-cause mortality. It is fairly apparent from those confidence intervals that the study has nowhere near the sample size needed to detect all-cause mortality effects equal to the estimated decline in prostate cancer mortality.

    The European study, in appendix table 5a, reports total deaths in the screening group and control group, and prostate cancer deaths, as a percentage of “person-years”. The prostate cancer deaths per 1,000 person years are 0.39 in the screening group versus 0.50 in the control group, a decline of 0.11. For overall deaths, the control group has 18.49 deaths per 1000 person years. If overall deaths went down the same as prostate cancer deaths, we would expect to see a rate per 1,000 person years of all-cause mortality of 18.38 deaths per 1,000 person years (18.49 minus 0.11). Instead, we see a decline in all-cause mortality in the screening group to 18.21 deaths per 1000 person years, a decline of 0.28 deaths per 1,000 person-years. This is over twice as great a decline.

    The authors in their table focus on the ratios of death rates between the two groups, and the confidence intervals around the death rate. They report that the ratio of the prostate cancer death rate in the screening group versus the control group is 0.79 (from dividing 0.39 by 0.50, or actually dividing the unrounded values, as after rounding the ratio is 0.78). The 95% confidence interval around that ratio is from 0.68 to 0.91.

    On the other hand, they report that the ratio of the all-cause mortality death rate in the screening group to the control group is 0.99. This is the ratio of 18.21 to 18.49. (Actually, I get 0.9847, so I think there was some mistaken rounding up somewhere.) The 95% confidence interval for that ratio is from 0.97 to 1.01.

    If all-cause mortality declined by the same amount as prostate cancer mortality, we would expect to see this ratio be 0.9944 (= 18.38 divided by 18.49). The confidence interval reported by the study’s authors encompasses that value. Unfortunately, it also encompasses an all-cause mortality effect 5 or 6 times as great, to 0.97, and also no all-cause mortality effect at all, or even an increase in the all-cause mortality in the screening group.

    If we were going to tell whether 0.9944 was significantly difference from 1, we would have to have a confidence interval of plus or minus 0.0056 or less. Instead, the confidence interval in this study is plus or minus 0.0200 (e.g., 0.99 plus or minus 0.02 to give us interval from 0.97 to 1.01). This is about 4 times too high.

    A rule of thumb in statistics is that the precision of an estimate goes up inversely with the square root of the sample size. So, four times the sample size, you cut standard errors in half. To get a confidence interval that is four times smaller, you would need a sample about 16 times as big. So, to detect effects on all-cause mortality similar to the prostate cancer mortality reduction in the European study, we would need a sample size about 16 times as big as in the European study. They had about 160,000 men. You would need millions of men in a study to cut down the confidence interval to a range where you would have a reasonable chance of detecting effects on all-cause mortality.

    What people seem to forget is that the fact that some estimate is statistically insignificant is meaningless unless the estimate is sufficiently precise to tell you what you need to know. Here, what we “need to know” is whether there are any offsets in other-cause mortality of sufficient size to offset the reduction in prostate cancer mortality. But with the sample sizes we have, we do not have estimates sufficiently precise to tell us anything useful about this issue.

  19. Tony:

    Also thank you for the analysis. An improving curve is encouraging, rather frightening if there was no indication of improvement. Either improved treatments or earlier detection and certainly the combination of these should have this effect. The logical extension of no improvement from earlier detection and newer treatments would indicate a much darker logical conclusion. The assumption that traditional treatment itself has resulted in overall survival improvement is not entirely sound. Would not the decades-long tendency for earlier detection, even the earlier detection resulting simply from more common annual physicals and greater social acceptance of digital exams result in at least a statistical improvement resulting from traditional methods of treatment? Do I think it is the major factor? No. Is it a piece of contamination? Hard to believe it is not.

    On an individual basis the questions are simpler:

    Why is it watch and wait rather than watch and manage? This results (from my very limited viewpoint) from the all or nothing treatment tradition. It is a little like never weeding a garden unless you can then prove that finally weeding resulted in a crop improvement. The traditional methods and the current “improved methods” share one common trait, rip out the garden to try and get the weeds.

    The indications are determining the type of cancer is at least as critical as deciding the course of treatment. It is rather illogical that this step is not taken to the utmost detail possible prior to treatment; it is almost an afterthought from what I see. I would rather know my type than know my stage, especially given the variety in staging from the same slides.

    Improved treatments are diluted when applied to a broad rather than a targeted cancer. What improvements in survival would result if the sequence of treat and identify was reversed? It certainly would improve the ability of an individual to choose his path.


  20. Mike,

    Fully understand complete and accurate answers to to these questions are not available now and likely never will be. That is not the issue — there is some information available and we are trying to get the best answers we can with that. If everything is just speculation based on assumptions and presumptions we should all go away and do something more useful with our time. You do not seem to feel that is the situation — you make an argument based on the information available and with concerns for over-treatment and lack of screening impact on overall mortality rates. I make an argument based on the information available with concerns for minimizing premature death from prostate cancer and you dismiss the whole thing as unanswerable.

    If my issue is unanswerable then it is likely that yours is also.

    The fact is that both arguments are at least partly answerable with the information available and I would hope we could make an effort to engage on the issues and agree where we can and agree to disagree where we cannot. Robust debate never hurt anyone (except perhaps Socrates) and beats silence and ignorance every time in my book.

  21. But Tony, you are correct in your first statement. We should indeed “all go away and do something better with our time.”

    We should be spending much more time and money on learning how to test for and differentiate accurately between indolent, low-risk, and high-risk forms of prostate cancer (“cancer types” to use Mike’s term, although I am not sure what he meant by his use of that term, since we don’t know how to “type” prostate cancers yet in any meaningful way) at the time of diagnosis so that we stop wasting time and money diagnosing and treating cancers that are never going to have serious impact and so that we stop under-treating the cancers that are really serious.

    We have spent hundreds of millions of dollars doing bad studies (ERSPC and PLCO and others) that many of us knew at the time they were initiated were never (with 90%+ certainty) going to resolve the questions being asked. We have spent the last 3 years quibbling over what those data mean … but they don’t mean anything except that the study results did not substantiate the hypothesis. All the retrospective analysis in the world cannot resolve the fact that these are badly designed and (in part) badly executed trials.

    What I may believe and what these studies show or do not show actually have very little to do with each other. We don’t know how to accurately identify men who really need early treatment. We also don’t know how to accurately identify men who don’t need treatment. Least of all are we able to accurately differentiate between the two. On top of that, we have a socio-cultural problem, which is that most men diagnosed with prostate cancer are under the utterly misguided apprehension that it is going to kill them, so they rush off to do what is almost certainly the wrong thing in most cases (with the full support of a large subset of clinicians).

    You make a mistake in stating that I am arguing one way or another on this issue. What I am saying is only that the hypothesis is unproved by the studies conducted to date (which almost the entire medical community concurs with), that they were lousy studies to begin with, and that, as a consequence, we are pretty much still where we were in about 1994 … when an individual man needed to make a personal decision about what he wanted to do — as an individual — about his entirely personal risk for prostate cancer. If I was a black male of sub-Saharan origin with a family history of prostate cancer, I can assure you that my personal decisions would be very different from the ones I have taken as a Caucasian male with no family history of prostate cancer (and almost no family history of cancer at all). But those decisions have never been influenced or justified or negated by the information provided by PLCO or ERSPC. Two overly discussed, bad studies have done nothing to help us to make good decisions about screening for prostate cancer, and because those studies were badly designed and (at least in part) badly executed, we have lost the chance to actually get any type of meaningful answers.

    Frankly, we need to start again. We need better tests. We need to stop thinking of prostate cancer as one disease (when we know it isn’t one disease). And we need a public campaign to tell men that prostate cancer is a medical risk, not a death sentence, and that at the present time immediate, overly-aggressive treatment may and can be as dangerous for an individual as doing nothing if one has demonstrably low-risk disease.

    All I have ever said about PLCO and ERSCP is that the available data do not justify mass screening. That is hardly a surprise, since the available tests are so bad and the available treatments are nearly as bad. Breast cancer has exactly the same problem. At least when I go get a colonoscopy to check my risk for colon cancer, I know that someone can remove potentially risky polyps without any serious risk to the rest of my health or quality of life … and now I only actually seem to need such a test twice in my lifetime if those two tests are both negative.

  22. Tony and Sitemaster,

    Thank you both for providing additional clarity to my comments and further clarifying my own thoughts.

    I think of prostate cancer as a continuum. First, below critical mass; next, critical mass; then, highly critical mass, and finally, maybe we can pull of a come from behind win or at least send it into overtime.

    The good news from the resistance of the stats to improve is there must be a fairly large grouping of cancers with a fairly long window between critical mass and highly critical mass where conventional treatment is at least of some significant benefit. The low hanging fruit if you will. The downside of a large grouping showing progress is that it limits the view to the past.

    What do I mean by “types” of cancer? I wish there were better definitions available, but there are some “types” as I mean it. Those with DNA changes, hormone resistant, those that reach critical mass before they are detectable, almost Day 1. I understand that sometimes, but not always, surgery pauses to check for lymph node involvement. It brings to mind my surgeon recalling a false negative lymph node with a follow-up after complete prostate removal showing aggressive cancer. This on a close friend of his.

    This brings to my mind a question. What, are you people crazy? It is like sending in the swat team without knowing if there are hostages. If you don’t know what you are dealing with, find out first. Cancer cells won’t grow in a dish after biopsy to see how they respond? No way to harvest lymph cells without a non-stop trip to complete surgery? (When I say surgery I am not differentiating between the various methods, including radiation. If there is no predicable change in results, does the method really matter that much?) If this is still impossible or simply not practiced, then doctors are really throwing darts in the dark and calling it medicine.

    To a large extent I am simply saying that expecting to reach beyond the low hanging fruit and expecting significant improvement in results by simply refining of the existing methods is no longer logical. Not impossible, but not logical.

    Yes, I agree it is time to step back and regroup. I am also saying watch and manage is a relatively low-cost, low-risk path to try.

    Are there already rejected drugs, ineffective on late stage cancer, which are worth a second look to delay reaching the critical mass?

    Standard testing of new drugs has a built in flaw. A teaspoon of water is 100% effective in extinguishing a match. A bucket of water is 100% ineffective in extinguishing a house fire. Yet we continue to wait and watch to see if the fire gets big enough to warrant fighting. Then the first step in fighting the fire is to blow up the house.

    A completely unsupported observation: Two men go to the doctor with identical prostate symptoms. The first is diagnosed with BPH and is offered treatment. The second is diagnosed with a low grade cancer and is offered watch and wait. Not completely fair, but not completely untrue either.

    If I am, as a good friend used to say simply “beating a dead horse to death”, I apologize.

  23. Hi Tony and Tim,

    I keep thinking that the ERSPC and PLCO results are both so premature that I’m not going to get anything reliable from the results at this time. I outlined some key points about that on 3/22 at 1:54 AM. Unfortunately, as Mike has emphasized, serious problems with the trials may have made it impossible to draw meaningful lessons even in the future when we have much longer follow-up.

    Tony, in your response of 4/7 7:30 AM, at the end of your paragraph (2), you remarked that the absolute numbers benefiting from screening seemed low. Here’s what I suspect accounts for a substantial portion of the very few deaths that have occurred. First, in view of the premature follow-up, we would expect that there would be virtually no deaths from prostate cancer at this point for men with low- and intermediate-risk disease. In fact, we would expect almost no deaths among men with high-risk disease except for those with extremely high-risk disease, including, for instance, cases involving “small cell” and so called “endometrial” prostate cancer (the latter term used because of shape resemblance). The incidence of such cases is well under a single percent, as I understand it, but the prognosis (at least during the years of this trial that have been reported, hopefully with some more recent improvement) is often short — short enough to register despite the limited follow-up in the trials. I’m thinking that screening helped spot some of these extremely high-risk cases at an earlier point where treatment was effective enough to result in a survival advantage compared to those who were not screened and therefore were treated at a later and probably less-effective point, with the latter delay in treatment accounting for the small advantage for screening noted in the ERSPC trial. The PLCO and ERSPC teams surely have such data associating mortality with risk level, but it has not been published. On the other side of the coin, overall survival side effects of treatment are likely to be front-loaded, especially with surgery. For instance, mortality from surgery, including 90 days following surgery, is apparently about half a percent. There are some continuing low-level treatment mortality risks as well, of course.

    My hunch is the premature results we are now seeing from the PLCO and ERSPC trials balance the mortality impact among men with extremely high-risk cases against the generally front-loaded mortality from treatment (probably dominated by surgery mortality among low- and intermediate-risk patients as high-risk patients tend not to have surgery), both very small effects. As another 4 to 10 years of median follow-up are reported, we should see some mortality among men with intermediate-risk cases, and even limited mortality among men with low-risk cases. While I still hope some meaningful analysis can be done at that point, I share Mike’s pessimism.

  24. In reviewing what I just submitted, I would like to add, in the second to last sentence, mortality among men with high-risk cancer that is other than extremely high risk.

  25. Dear Tim Bartik:

    Your statistical analysis seems to have integrity but cannot be used to support population screening. This is because the sample — by your analysis — is too small to validate current conclusions re all-cause mortality. Thus the argument for screening is not advanced.

    Currently, those who favour screening are making assumptions predicated upon divergence in the Goteborg trial — the study demonstrating the apparent greatest benefit. The prevailing assumption is that the divergence will increase in favour of those receiving prostatectomy and that, in x years time, will lead to an NNT of 18 — still a difficult number of false positives and morbidity.

    I have problems with this presumption for two reasons. Firstly — as I alluded to previously — it is unwise to assume that the prostate gland does not in some way have beneficial effects on other organs as yet not scientifically articulated. Secondly, as the lead time to potential benefit increases, so does the time to die from some other cause and therefore a plateau must arise. It may already have done so, but nobody knows.

    Currently there are a cluster of known risk factors, e.g., ethnicity, genetics, and infection.

    As a GP I will continue to communicate what is known and let the patient decide.

  26. I feel that I should point out that it is my (entirely personal) opinion that Dr. Montgomery’s assessment of the proper behavior of physicians based on the available data today is highly appropriate: inform and educate the patient and then help him to come to a decision that feels appropriate for him as a single individual.

    As I have noted earlier, my own ethnicity and family history have quite certainly “colored” my personal behavior regarding PSA-based screening, and I have been completely frank about that for nearly 20 years. However, my decisions about what are “correct” for me do not necessarily mean that they are “correct” for others.

  27. Sitemaster,

    I agree that we should be putting every effort possible into getting better diagnostics to identify men with prostate cancer in danger of dying and those who are not. I have recently signed up as a consumer rep on several prostate cancer research projects that involve study of micro RNA. The researchers seem quite excited about the potential of micro RNA as a diagnostic biomarker or even perhaps as a therapeutic — but even if it all comes to pass it is likely years away. We keep hearing about possible new tests but they never seem to arrive. I think we have to accept the fact that we may be stuck with our current technology for some time to come.

    If that is so, we still have an obligation to provide our best information to men with prostate cancer to facilitate their decisions — we cannot abdicate that responsibility simply because the info is complex or incomplete.

    You make the point that you are not partisan about concern for over-treatment — yet most of your other statements do indicate that concern. It is perfectly OK to be concerned about it — but can we move ahead on that subject by addressing some of the questions I raised previously rather than ignoring them?

    Then there is the unmitigated pessimism about screening, diagnosis and treatment. The only people who have a problem are the ones with indolent cancer — yes they do have a problem but they are only part of the group — others in the group have a problem called dying. Balancing these two problems is the task ahead of us.

    You say the randomised screening trials are hopeless and useless and provide no useful info — that is incorrect. The quality of info may be much lower than we like but there is info to be had — a likely reduction in relative prostate cancer mortality after a long lead time but low absolute numbers to date. I do not support (and there is no move in that direction as far as I am aware) population wide PSA screening. On the other hand I was pleased to see the recent USPSTF proposals come to naught.

    Diagnosis involves a biopsy and Gleason score — know that Gleason is only a rough marker for mortality risk but it is better than nothing. Other advanced prostate cancer guys with Gleason 7 get holidays from hormone treatment, but with Gleason 8 I win the top prize — continuous HT.

    In summary your obvious concern for over-treatment is driving an over-simplified analysis that is ignoring important aspects of the situation. We don’t have all the information you would like and I would like but we do have some info and it is our duty to provide that in a digestible form to all men with prostate cancer or contemplating screening. Their decisions will continue to need to be made — better with some info than nothing.

  28. Dear Tony:

    There are so many things in your most recent comment that I find disturbing that I hardly know where to begin. So I will start with the personal:

    (1) I know of no data that support the idea that men with progressive, Gleason 8 disease must, de facto, be given continuous hormone therapy. Indeed, I know of men with that diagnosis who have been on IADT (on and off) for years. It appears to be effective for some men and not for others. It may not be appropriate for you for all sort of reasons, particularly if you have a low PSA doubling time if you come off ADT, but this is not some form of absolute truth. (Have you ever come off ADT and stayed on the Avodart to see what happens?)

    (2) You seem to be assuming that just because we don’t have good answers to all sorts of important questions, that I don’t think men should be educated and informed. I absolutely reject that idea. What we are disagreeing about is the nature of that education and that information. I believe that that education and information have to be placed in a highly individual context, and that to the greatest extent possible they should be based on demonstrable facts.

    I have been helping to educate and inform men at risk for and with prostate cancer for nearly 20 years. There has been one absolute constant to that education, which is that one has to distinguish between what applies, statistically, to large numbers of patients, and what is appropriate for the individual patient. As just one example, the fact that annual screening starting at age 40 does not appear to be appropriate for all men does not mean that it is inappropriate for definable individuals and subgroups. As a second example, I will always advise men with low-risk prostate cancer and men of 65 years and older that the risks associated with surgery need to be balanced against the limited benefits of active surgical intervention in these risk groups (recently proven in both the PIVOT trial and the Scandinavian trial). However, the degree of emphasis I place on how to think about the risks as opposed to the benefits is always “customized” to what I know about that individual, and at the end of the day I always make sure to emphasize that no one can make that decision for them except them (preferably with the assistance and support of their spouse or partner).

    (3) If you think that the USPSTF recommendation has “gone nowhere”, please let me disabuse you of that idea. It simply hasn’t been formalized yet. That battle is just on temporary hiatus.

    (4) The idea that I am “abdicating responsibility” is completely countermanded by the fact that there is a social network associated with this site on which I spend a massive amount of my time trying to help others to understand their options. However, those options are rarely supported by any good data. The only two published, really good attempts at a randomized clinical trial of two types of first-line therapy for the treatment of localized prostate cancer are a Canadian one that compared brachytherapy to cryotherapy and the Scandinavian trial of surgery vs. watchful waiting in men who were diagnosed without the benefit of the PSA test. Neither of these trials offers much insight into the decisions that must be taken today by a newly diagnosed patient of 57 with a PSA of 3.7, a Gleason score of 3 + 3 = 6, and clinical stage T1c! The data from the PIVOT trial is still not — to my disgust — published yet.

    (5) Most men with prostate cancer are still being diagnosed between their mid to late 50s and their early 70s. What is appropriate for such men is very different to what may be appropriate for men who, like you and my good friends Tony Crispino and Bill Manning and others, were diagnosed much younger. However, it has been a constant in the world of prostate cancer that many of the men who get diagnosed much younger seem to have a hard time accepting that they are a small minority of the global prostate cancer risk pool. (In America, only 0.5% of men are diagnosed between 35 and 44 years of age; 8.3% of all prostate cancer diagnoses are in men aged between 45 and 54 years of age, and 64% between the ages of 55 and 74 years.) I absolutely acknowledge that such younger men, faced with the very real potential for death from prostate cancer and/or a 20- to 40-year expectancy of significant impact of therapy on their quality of life are going to have to make some very difficult choices. However, I can also tell you that what is “truth” in their cases may well not be “truth” for an overweight, diabetic 68-year-old with a history of high blood pressure and high cholesterol. (Such men are far from uncommon here in the United States.)

    (6) Your comment that “The only people who have a problem are the ones with indolent cancer” is — I am sorry to say — rather naive. The people who have a problem are all the men who get diagnosed with prostate cancer who are going to die before they ever exhibit or suffer from symptoms of prostate cancer — even if they have no treatment whatsoever. This includes all the men with indolent disease and tens of thousands of men in America diagnosed each year with D’Amico low-risk disease. Taken together, the best available data suggest that this is well over 50% of all the men diagnosed in America each year today.

    You began this conversation by noting that individual perspectives color opinions. You appear to believe that my perspective biases my opinions (as opposed to coloring them). I will defend your right to think that, but it doesn’t make you right. You don’t appear to be allowing me the same degree of grace that you demand for yourself.

    You are completely entitled to disagree with my opinions and my perspectives, but to do this you have to be able to provide facts that back up your opinions. You read into what I say things that are not there because you don’t concur with my opinions and my perspectives, but I hear no fact-based, compelling arguments back from you. Indeed, the only actual “fact” that you offer in your prior comment is the one about men with Gleason 8 disease necessarily being ineligible for IADT. … If you know of data that actually confirm this supposed “fact” (as opposed to the opinions of specific physicians) please do let me know, because I have never seen these data.

    I shall watch with interest to see how and whether Tracy responds to your comments above (assuming she sees them). If you think that I am overly concerned about the risks associated with over-treatment, I can assure you that I’m nowhere near to being in Tracy’s league!

  29. Two comments:

    1. Dr. Montgomery commented that the screening experiments did not support screening because they did not show an effect on all-cause mortality. But if we were to use that criteria, then it is quite difficult for screening for any disease to pass that test in a sample with a 20% overall mortality rate.

    2. I know the European study was carefully designed to detect effects on prostate cancer-specific mortality. The European study was never intended to detect effects on all-cause mortality. As I mentioned, to detect all-cause mortality effects requires a sample size of millions for almost any disease that accounts for less than 5% of overall deaths and for which the screening might reduce disease-specific mortality by 20% or 30%.

  30. Tim:

    I recognized your statistical concerns. However, the only component of the data sets that comprise the entire ERSPC that showed an effect on prostate cancer-specific mortality effect was the data from the Goteborg trial that were wrapped into the ERSPC. If you carry out the analysis on all the data from the ERSPC except the Goteborg data, there is no prostate cancer-specific survival benefit.

    I am willing to entertain the idea that there is something about Swedes that may be relevant to this result, but don’t you find it odd that there is no survival benefit from the other 160,000 men, including the very large set of Finnish men (the largest in the trial)?

    Of course I still believe that it is better to characterize this study as a meta-analysis of several independent trials for the simple reason that every national element of the study did, in fact, have slightly differing executional components. In other words, there was no single, agreed protocol applied to all the patients who were enrolled into the ERSPC.


    Dr. Montgomery, Sitemaster and others –

    Thank you Dr. Montgomery for your concern for avoiding unnecessary harm to patients. As a patient, I appreciate that. However, there is another way to look at whether screening is worthwhile.

    I’m convinced we have enough information and satisfactory technology now to advocate that men be screened for prostate cancer with PSA tests and DREs, provided they understand that active surveillance is often their best option if case characteristics indicate low or very low risk. In the past few years, both the National Comprehensive Cancer Network and the American Urological Association have advocated a key role for active surveillance. We can look forward to better information and technology, but remain mired in uncertainty when what we have now is good enough?

    Yes, the usual clues PSA level, PSA doubling time, Gleason score, and stage are not enough to sort out who really needs treatment and who does not. Nor does the addition of other clues such as the size of the prostate, the number and percent of positive cores, the percentage of cancer in each positive core, etc., answer this question. However, well-conducted active surveillance appears to safely sort patients into those likely to be fine forever or for a long time from those with stealthy, aggressive cancer. From the pioneering Laurence Klotz series in Toronto, which has been in progress since 1996, and from series at major centers including (at least) Johns Hopkins, Memorial Sloan-Kettering, Erasmus Medical Center in the Netherlands, UCSF, and M. D. Anderson, we have by now learned that active surveillance patients who are later determined to need treatment have outcomes virtually as good as if they had had treatment shortly after diagnosis. The huge benefits are that a clear majority of active surveillance patients do very well long term, that even those determined to need treatment typically gain at least a couple of treatment-free and side-effect/complication free years, and that those who need treatment have strong evidence that they really do need treatment — they have the assurance of knowing they did not get unnecessary treatment.

    The strongest counter-argument I have seen has to do with the cost and side effects/complications from the periodic biopsies that are key to monitoring active surveillance patients. However, that argument is based mainly on the assumption that biopsies are frequent, such as on an annual basis, as is the case with some programs, including the one at Johns Hopkins. That argument loses force if the frequency of biopsies is sharply lower. Dr. Klotz, whose program employs a sharply lower frequency of biopsies, has provided data showing that biopsies in the first 2 or 3 years, plus perhaps one or two more biopsies at intervals of a few years up to age 80, are usually sufficient for effective monitoring when combined with case data viewed from the perspective of nomograms based on hundreds of cases.

    Dr. Klotz has effectively examined the role played by other causes of death for prostate cancer patients, and he has shown that that reality supports the use of active surveillance for appropriate patients as contrasted with immediate treatment.

    In sum, active surveillance appears sufficient to separate patients needing treatment from those that do not, and that success makes it generally wise to screen men who can appreciate the role of active surveillance for prostate cancer.

    I am following the Goteborg trial results as well as the other published results, but, mainly for reasons I’ve mentioned earlier under this topic, those results are at the periphery of my enthusiasm for screening. The main reason is that prostate cancer is a highly aggressive, deadly threat for some of us, and early detection enables therapy that appears to greatly improve outcomes, including cure. I am one of those patients. I’m convinced that my cancer — featuring a baseline, first ever PSA of 113.6, a Gleason score of 4 + 3 = 7 (Epstein), stage III, all cores positive, most 100% cancer, scans negative, PSA doubling time during the 5th to 10nth months of IADT vacations of about 3 months — would have killed me by now (or at least required very demanding treatment) if my symptomless cancer had not been picked up by the PSA test my doctor prescribed only at my insistence in December 1999. I am now, at the 13-year point, enjoying a decent quality of life in my third off-therapy vacation from triple androgen deprivation therapy, my sole therapy.

    As a side note, the Goteborg results are also still premature. Don’t you agree?

    In your second paragraph there is a presumption that surgery is the normal course after diagnosis. A strong case can be made that technology associated with modern radiation is at least as effective at curing and controlling prostate cancer as surgery, and with a competitive, if not favorable, profile of side effects and complications.

  32. Dear Jim:

    So there are a series of very big “ifs” implicit in your comments. They include the ideas that:

    (1) Active surveillance becomes the de facto first-line therapy for all men diagnosed with D’Amico low-risk disease until there is clear evidence of an increase in risk.

    (2) Arguably, active surveillance becomes the de facto treatment of choice for all men with a life expectancy of < 10 years and D'Amico intermediate-risk prostate cancer until there are clear data to show a survival benefit from immediate treatment in such men.

    (3) Regular "screening" becomes risk-based as opposed to annual for everyone. In other words, frequency of PSA and DRE-based screening is based on changes in PSA and DRE over time, and men with a low, stable PSA (up to a value still to be well defined) would need PSA tests at more like every 5 years.

    (4) Frequency of biopsies of men on active surveillance protocols is limited in some way yet to be highly defined.

    I have no problem with your theoretical approach. However, I have grave suspicions about whether even one of the the American Urological Association (AUA), the American Society of Radiation Oncology (ASTRO), the American Society for Clinical Oncology (ASCO), the National Cancer Institute (NCI), or the National Comprehensive Cancer Network (NCCN) could implement (let alone attempt to enforce) a set of guidelines that was based on these principles. Of course one of the large managed care organizations might be willing to try, but only if they were willing to accept the consequent uproar from tens of thousands of patients and clinicians!

    With respect to the Goteborg results, no, I don't think they are premature. In that specific set of highly defined patients there was a very clear prostate cancer-specific survival benefit. What we do with those data is the issue, since they have never been replicated anywhere else.

    The "presumption" that surgery is "the normal course after diagnosis" merely reflects current reality. I don't think anyone is suggesting that reality is necessarily correct.

  33. To all,

    I must now spit out my thoughts before I lose them. Obviously coherency, tact, and finishing a train of thought at one time do suffer.

    Patients have little hard data to base their choices on. Physicians have experience, which probably just increases the awareness of difficulty making medical progress or guiding choices for others. The deviation from the standard treatment I desire for myself is not something I would be able to thrust upon another. I recognize the limited arsenal available to a physician. That said, I obviously have had a preference for an earlier battle.

    I am not as negative as my postings. Frustrated at the lack of access to the treatment course I have chosen, yes. PSA testing is a huge advance even if the stats do not yet capture the results and serve to highlight the difficulty of each forward step.

    Does PSA screening result in unwarranted treatment? I would guess it does. It is probably a pretty close ratio to those men receiving unwarranted treatment without PSA testing. That statement is based upon the stubbornness of the stats to move. It is not PSA testing, it is the overall, extreme difficulty to decipher disease.

    Using untested men as controls does set the bar artificially high. For the reasons Jim details, as well as other reasons, the controls are a bit off. Simply being told you have a life/sexual identity-threatening condition is a mortality risk factor on multiple levels. Depression and distracted driving are two that come to mind. As a control group I probably have more in common with women who have been diagnosed with breast cancer than men who have not been screened for prostate cancer. Who gets screened? Men who have a little nagging worry in their heads. In that sense I am not too concerned that the overall mortality has not dropped. I am more concerned with studies failing to show a clear advantage between current methods. Those results I could use today.

  34. Sitemaster:

    Your comment that the European trial shows no statistically significant effects at the 95% level overall absent the Swedish data is accurate based on the 9-year followup from a few years ago. However, your comment is no longer accurate based on the more recently published followup that is referred to in this post.

    This finding is referred to on p. 986 of the new paper. The estimated reductions in prostate cancer mortality are now statistically significant at the 95% level when any one of the countries is excluded from the data, including Sweden. More details on the country-specific and country-excluded results are in appendix Tables 4A and 6A.

    More specifically, the overall study estimates a reduction in prostate cancer mortality of 21% due to screening, with the 95% confidence interval ranging from a 32% reduction to a 9% reduction. Sweden shows a 44% reduction in prostate cancer mortality, with a 95% confidence interval from a 62% reduction to a 17% reduction. All other countries excluding Sweden show a prostate cancer mortality reduction due to screening of 16%, with a 95% confidence interval from 29% to 2%.

    In the individual country results, the point estimates show prostate cancer mortality reductions of: 29%, Netherlands; 14% Belgium; 44% Sweden; 11% Finland; 14%, Italy, 11% , Switzerland; and prostate cancer mortality was higher in the screening group in Spain, but in Spain, with a small sample size, there was only 1 prostate cancer death in the control arm and 2 prostate cancer deaths in the screening arm, so the Spanish-specific estimate is extremely imprecise. Of these country-specific estimates, only the Netherlands and Sweden are individually statistically significant. However, most of the countries have relatively small samples, so it would be surprising if these individual countries did have statistically significant results. The exception is Finland, which did have a relatively large sample.

    It is unclear what to make of these differences across countries. The authors do not report whether they can reject the hypothesis that “country-specific effects” are jointly insignificant or not. In other words, we would expect some coutnry-specific differences simply due to random chance. The study was designed so that the overall results pooled over all countries could be used to detect a 20% or greater decline in prostate cancer mortality at a statistically significant level. It was not designed with sufficient sample size in each country to detect statistically significant effects in each country. In any event, while I think is is interesting to explore why the results were more positive in Sweden and the Netherlands than in Finland, due to differences in how screening was done and how treatment was done, or differences in the demographics of the countries, there is the very real possibility that the differences across countries are simply due to chance. There is a lot of noise in the individual country estimates.


    Tim Bartik

  35. Tim:

    Thank you for the clarifications about the prostate cancer-specific survival data — nationally and collectively.

    Just out of curiosity, since you clearly have some knowledge of the clinical trials design and analysis process, do you have the same hard time as I do seeing the ERSPC study as one trial? I have no problem with it as a meta-analysis, but if anyone tried to take this study to a regulatory authority such as the FDA or EMEA to use it as the basis for PSA screening, I can’t help thinking they would be told that the trial design was too fundamentally flawed for it to be even considered.

  36. Thank you Tim.

    PSA is the largest but not the only factor leading to earlier date of diagnosis. Unless it is possible to isolate each, it is hard to reject PSA without also rejecting DRE or annual wellness exams which have also lead to earlier diagnosis. Not likely.

    Early detection has separated the date of diagnosis and date of treatment and fed watch and wait.

    Watch and wait will consolidate date of treatment around criteria now being formed. Consolidation will result in a larger pool of narrower criteria. Comparison studies will improve, resulting in progress and fewer debates.

    PSA is a valid scientific tool. Cold but true.

  37. Sitemaster and All,

    My apologies if I have ruffled feathers but I am trying (perhaps clumsily) to provoke a robust discussion on the issues I consider important. I have great respect for the work for men with prostate cancer done on this site and am pleased that the passion I saw in Chicago is still there.

    Sitemaster comments correctly about the desirability of tailoring advice to an individual but my concern is that a large majority (who he is not talking to) are getting very poor general advice, let alone individualized. If we continue waiting for the data and tools to facilitate his desirable approach we may be waiting 1 year, 5 years, 10 years — who knows how long? Men are continuing to have to make important decisions on screening, diagnosis, and treatment based on presented info that I suspect is nowhere near as good as it could be. I don’t know for sure I am right in this but I am trying to find out if I am.

    Tim Bartik has commented on the stats and I think I saw somewhere him saying that if a guy was not treated he had 5% chance of dying (over whatever time period) and if he was treated he had a 30% chance of undesirable side effects from over-treatment. Forgive me if this is not fully correct Tim but I am using the situation to make a point. A patient, given this info, could factor in his own priorities and values and make a decision. If there was any info on the uncertainties in the percentages, that might help a bit more, and if the data could be individualized, even better. Getting a generalized framework seems to me to be the first step however.

    Can we do this now with the info currently available — or not? If we can’t do it fully how close can we come to it? Statistical expertise is obviously needed — maybe Tim could help here or advise other assistance.

    If we can improve in this area, I feel it could be a great step forward for the majority of men who don’t currently have access to the best advice. If we cannot improve for now then at least we have tried.

    Happy to discuss my personal situation but don’t want to have that confuse the important issues above. [Diagnosed age 57 (PSA 5, Gleason 8, unsuccessful radical), now 66 (9 years continuous HT — Lupron then add Casodex then add Avodart then drop Casodex and Avodart and add Nilutamide last PSA 11).]

    I did also make the comment about Gleason being a rough indicator — Sitemaster seemed to support that — do we have any quantification on how rough it is?

    Bit confused about Sitemaster’s comment on my indolent cancer one — but I think I agree with him.

  38. Dear Tony:

    There are two absolute and fundamental problems about the issue you are trying to discuss. They are: (a) the potential accuracy of the application of the available data to the potentially recommendable behaviors of individuals and (b) the fact that enormous numbers of people just don’t care until it is too late.

    There are numerous tools available for helping men to understand and think about their risks of prostate cancer and what they might want to do about those risks. Each of those tools has been developed from pretty large databases. Examples include the tools to help one decide whether one should consider a biopsy based on PSA data and the tools to help consider outcomes of different types of treatment based on known risk factors and the outcomes of other similar men. However, and very importantly, it is the attitudes of individuals to risk of any kind that drives their actual decisions. Some of these tools are highly sophisticated. Others are good and basic. Most of them are freely available over the Web to anyone who wants to use them.

    Consider three men who have exactly the same physical characteristics. They are all 57 years of age; are 6 foot tall; weigh 170 lb; and are in good physical shape. They have each had two PSA tests in the past 5 years. On each occasion their PSA level was < 1.5 ng/ml. How often do they need to get repeat PSA tests?

    The answer is that it depends on their attitude to risk. If they are terrified (for whatever reason) of the apparently small possibility that they will at some point in time get prostate cancer (of whatever level of aggression), they are going to go on going PSA tests regularly (whether that is appropriate or not). If they have no family history of prostate cancer and (accurately) understand that their personal risk is therefore relatively small, they could either ask their doctor to test them (say) every 5 years "just to be on the safe side" or they could decide to ignore the whole thing. If they just don't even think about it, then they just don't think about it unless a doctor tells them they need a PSA test. And then there is a range of possibilities in between.

    We keep trying to produce "correct" guidance for individuals based on data that is uncustomizable to individuals with any real degree of accuracy. We are also making the mistake of thinking that this will help us to "eliminate" prostate cancer, which we can't. Most of the deaths from prostate cancer each year are in men who are beyond cure (for any one of several reasons) at the time they are diagnosed. In particular, those men like you who were diagnosed relatively young with an aggressive form of the disease want to be able to make sure that this doesn't happen to others, but that is like trying to make sure there is never another car crash on the freeway. People don't listen. After 20 years of the availability of PSA testing and public discussion of the risks for prostate cancer, it is still not uncommon to come across men who get diagnosed with metastatic disease in their late 50s or early 60s who have never had a PSA test. These are not necessarily only poor and uneducated men (although the group certainly includes such men). I see this regularly in upstanding middle-class males who ought to know better.

    Men are going to go on getting diagnosed with prostate cancer until we are able to develop some form of simple preventive therapy (a true prostate cancer vaccine, for example) that actually eliminates risk of the disease — if such a thing ever becomes possible. However, take the three men I described above, all of whom now get a new PSA test, which is 3.2 ng/ml. So they go and get a 12-core biopsy. It comes back with two positive biopsy cores of Gleason 3 + 3 = 6, with 20% of each core positive for cancer. Can you tell me what those three men “should” do? I can’t tell you that! I can tell you what I would do, but that’s not the same thing at all. My rough guess is that one of those men will have surgery immediately (the “just get it outta there” approach); a second of the three will agonize for months, get at least three second opinions and finally decide on some type of radiation therapy; the third will shrug and tell the doctor, “Well, let’s just keep an eye on it. There’s no reason to believe this is ever going to kill me, and there’s a very decent chance it is entirely indolent.” Is one of those men “right” and the other two “wrong”? No. Of course not — they just see the problem through very different sets of eyes.

    So … I suppose my real question is, “What are you hoping for?” … Prostate cancer kills about 30,000 men in America every year. The vast majority of those men are in their 80s by the time they die of prostate cancer. By comparison, heart disease kills more like 600,000 men in America each year, and many of those men drop dead suddenly in their 40s and 50s. I was nearly one of them, despite being wildly more educated about health care than the average male, very fit, eating well, being within the recommended “normal” BMI range at the time, and having no family history of heart disease. I look around me, and the vast majority of men I see on the street are overweight, couldn’t run quarter of a mile, and don’t even understand that they are at risk for prostate cancer. (Remember that about half of them still spell it “prostrate” cancer after they have been diagnosed.) I don’t think we have much of a hope of significantly impacting the attitudes of the average male on the street to his health care, let alone to prostate cancer specifically, unless there is some sort of massive, unpredictable, seismic shift in the way humans think and behave. So in the meantime I shall continue to focus on trying to give the best possible guidance I can to the people who find me. I have certainly spent quite enough time around politicians to understand the fact that much of the world can’t even appreciate simple facts when they are explained in simple terms. Are you worried about global warming? Please bear in mind that for one reason or another most of the world is not!

  39. OK, frank discussion it is!

    There is a human problem on both sides. I have had two biopsies and three reports, with a second opinion on the second biopsy. There is no standard format and the second opinion found no infection and no Gleason 7 where the first found both. They disagree on the number of positive cores and neither states if the cancer is on the margin or centered. Are professionals willing to tell their professional counterparts to shape up and fix the obvious need for improvement in their product? Or will the human factor step in.

    From the lab does all treatment and decisions flow.

    Better lab reports do not require medical advancement. Perfect reports yes, better no. A fivefold increase in investment at the first serious step in treatment would improve the entire quality chain.

    I have to believe a needle biopsy contains as much debris as intact cells. I have to believe that in the right culture viscous cancer would soon show itself. I am not talking about research lab protein level stuff, just is it spilling out of the dish like a loaf of bread or not. I would really prefer to find that out before I spend time talking to a physician about options when I know he is looking through a straw and trying to see what type of cancer I have. Or worst yet post op.

    There is good, sound medical advice on the net. There are quacks. The human side of the physician does not dare to risk I will find the quack and there is still a large human side which feels a patient doing research is playing doctor. Not individually maybe, but very real in total. I would be no different; there are not enough hours. If the average physician is going to tell the average patient to make an informed decision then s/he’d better be prepared to spend 10 times the time s/he possibly hase or s/he’d better point out the best source s/he knows.

    Don’t deliver the “informed decision” line when it is only rarely true. Yes there is liability in suggesting a source, there is also liability in not spending time that does not exist. Obviously you and this site are communicating, telling the good and bad. Making up for the time not available to the average physician. I found you after 3 years.

    Trying to market PSA to the masses while researching treatments for the most dire does not fit human nature. You are selling a product with about as much appeal as the problem. No, people do not believe they are going to die in spite of the evidence to the contrary.

    Find a product that extends life before incontinence and impotence begins and survival rates and PSA rates will both increase.

    I do agree the biggest problem with humans are they are human. Within those very large constraints progress must be made. The reluctance to change is only exceeded by the inability to see the need. On both sides of the desk.

  40. Dear Mike:

    I am not at all sure what you are trying to point out here.

    If you are saying that “garbage in/garbage out,” I absolutely agree. If you are saying that there are some things related to the diagnosis of prostate cancer that really could and should be standardized (such as pathology reports), I agree. There is a list of these things as long as your arm. However, that’s a topic you need to take up with the physician communities who control these issues. There is nothing we can do about it! Currently, the physician community can’t even agree whether the PSA test is a useful screening test or not!

    In defense of the physician community, however, I am going to repeat what I have said many time before. Medicine is not a science. It is part science but also part “art.” Even someone with my minimal skill levels can see why two different pathologists can look at the same slide under certain circumstances and interpret what they see in somewhat different ways.

  41. As you said a few days ago, we need to step back and take a fresh look. Your morning post expressed justified frustration. People do always want more while not using what they have. Medicine is far more a social problem than a scientific one. All in all you scientists have been doing remarkably well. And I’ll have a smoke and a drink in your honor.

    What am I trying to say? Just that the easy ground for progress, if there is any, is before a patient reaches the operating room. I see huge efforts after the end of watch and wait, as imprecise as that tipping point is. I see dedicated people struggling with trying to find the best tools and yes “art” they can practice built upon a rather shaky foundation. I do not see the effort being made to keep people from reaching that tipping point. Maybe there is a lot of effort there that I just do not see or find searching the web. By the end of watch and wait it is a catch up game at best.

    No medical training, spent my time estimating and negotiating. Got pretty good seeing where the weak link in the process was. Almost always in the hand off from one step to the next. Researchers and cutting edge facilities are striving to try to do it a little better, a little differently. I just do not see the same start as finish. If the underpinning of research, treatment and my life depends on art, that’s not acceptable. Change the method to bring the “art” to the absolute minimum or fail trying to change it. Art should be and always will be the fill in for scientific technique, not an excuse for not progressing science. I am good with that. Not good with that’s the way it is.

    Garbage stops when it is sent back not when garbage is discussed. Medicine is art. It is also a culture. Good and bad. Not trying to go there, trying not to. Tell me people don’t die because physicians and patients have decorum and I will never post again. (I will still read, have learned more here than past several years of hunting.)

    I understand how long the list is and I understand the power of “physician communities”. I understand much of what physicians do and don’t do and patients complain about is just the other side of the same circle of responsibility and liability. But that is just as big a barrier to lowering prostate cancer death as people not having PSA tests.

    Why are lab reports not as precise as they might be? Patients sue anybody they can. Also because it is rude to say to another professional, “That may be your area but it really is not what I need.”

    “Physician communities who control these issues”. Ouch. Those issues are controlled by those communities because other physician communities allow it. What is the current estimate, 18, 20 seconds to get a physician’s attention. Do not think any patient will have that much time with a physician community who controls any issue.

    “Nothing we can do about it.” That is just your frustration showing. PSA testing may not be universally accepted. It is getting there. I do not hear of anyone making a living traveling the world saying global warming is a fraud anymore. Still not acting upon it much, true. Progress is slow.

    Just my view from the bottom.


    Hi Sitemaster,

    I’m admiring your endurance in covering this topic. I have some comments on your thoughtful response of 4/9 1:11 pm (if this keeps up I’ll soon have to add the year) to my immediately preceding response of 4/9 12:01 pm. I’ll split them up in separate responses.

    In my response I had expressed the view that the Goteborg results, with a median follow-up of 14 years in the 2010 Lancet Oncology paper, were still premature. I asked if you did not agree, and you replied that you did not think the results were premature. I looked again at the abstract of the study (can’t put my hands on the complete paper at the moment), and I agree that the follow-up is long enough, considering all the aspects of the study, to determine that there was a survival advantage to screening, which I think is what you meant, with the understanding that the finding was in the context of a particular study that has not [yet] been replicated.

    Where we may disagree is whether the follow-up is long enough to give a good indication of the benefit. The paper includes an observation that 293 men would need to be invited for screening and 12 to be diagnosed to prevent one prostate cancer death. I find that observation misleading — underestimating the value of screening. While the finding is acceptable as a cross-section for a screening study (not since diagnosis) with median 14-year follow-up, it just has so little weight for me against the context of near universal survival for low- and intermediate-risk men, and about 95% survival of even high-risk men at the 10-year point since diagnosis, at least in the US. If memory serves, there is an indication of the gap from start of follow-up (at invitation to screening?) to diagnosis in the Goteborg study, but general experience suggests a gap of at least several years. That would mean the follow-up time from diagnosis to 12/31/2008 — the check point reported, would be in the neighborhood of 10 years, the period during which virtually all patients are alive except for those with extremely high-risk cases. In future reports I’ll bet we are going to see the “number needing invitations to prevent a death” declining substantially from the 293 reported. As years of follow-up from diagnosis tick by beyond the 10-year mark, I’ll bet that number continues to decline, with an ever-increasing indication of the value of screening.

    The reality today is that we are just living much longer with prostate cancer, even if we have cases with some challenging features, provided we get sound treatment. I’ve mentioned my own case, now at the 13th year point. Our group just lost a friend who was diagnosed with a PSA of 86, but he survived for 23 years, doing very well until recently. Another friend is just taking a vacation from ketoconazole after first-line (limited) androgen deprivation was insufficient. He has been on ketoconazole for 7 years. I realize that is anecdotal, but my impression is that research also indicates long survival for many of us with challenging cases. If that perspective is a good one, then the follow-up reported to date for the PLCO and ERSPC trials (and subsets) does not seem sufficiently mature to give us a good idea of the real bang for the buck with screening. I suspect that a doctor will not be helping a patient when he quotes that “293 needed to screen to save a 1 life” to a patient.

  43. Sitemaster,

    What I am hoping for is quite modest. I am not trying to reform the slackness and stupidity tendencies in the human race. I am not expecting to find a cure for cancer. I simply want the best advice readily available to men re the prostate cancer issue at the times they most need it — when they are deciding what to do or not do about screening, diagnosis, and treatment.

    To take a simple example regarding treatment … A man is told he has prostate cancer with whatever Gleason — If he is also told at that point that he has an x% chance of dying with no treatment and a y% chance of unnecessary side effects (honestly defined) if he is treated, then he has the capacity to make an informed decision on what he does next, based on this information and his own priorities and lifestyle.

    In my experience, information of this clarity is not presented to men at any time, let alone at points of decision. You mention numerous aids on the Internet. None of the ones I have seen achieve the standard set in my example above. If you can indicate any that do I would be most pleased to get that advice. In my experience, also, individualised advice usually comes as just that – comments on factors peculiar to the individual but disconnected or divorced from any generalised risk framework for the disease. Savvy people like us often assume that everybody is up to speed like us but forget how much of the info we have was obtained retrospectively. Some men are not on the Internet, and I suspect for many others their Internet search skills may be downgraded at moments of traumatic decision.

    I have mentioned before that my priority is the risk of premature death from prostate cancer ahead of risks of over-treatment. This line of thought can be easily promoted by hinting death, death, death to vulnerable people. I do also, however, acknowledge that over-treatment is real and important and must be minimised. A balance between these two competing issues is essential — decisions by informed patients in consultation with their doctors.

    I don’t know whether the quality of advice I seek is fully or partly achievable. I am asking for help to try and establish that.

  44. Tony:

    I have to tell you that I do not believe that the quality of guidance you seek is either fully or partially achievable at this time — for the simple reason that prostate cancer is not one disease. It is a spectrum of diseases that are not yet well characterized, let alone easily differentiable.

    Here are just some of the things we would need to be able to give any specific patient that level of guidance with accuracy at the time of diagnosis:

    — Near to absolute certainty about whether any individual patient has micrometastatic disease at the time of diagnosis. This is currently impossible (but we do know that there are men who get diagnosed with apparently low-risk disease who turn out to have incurable prostate cancer).
    — The ability to differentiate with near to absolute certainty between the 25-40% of men with indolent disease (who are never going to have even symptomatic prostate cancer, let alone die of it) and the remaining men with clinically significant disease (which includes a lot of men who will also never die of prostate cancer). This is also currently impossible.
    — The ability to categorize the men with potentially clinically significant disease into really meaningful risk groups that are way more accurate than the D’Amico risk groups. There is no way to do this yet.
    — Precise knowledge of how men accurately and specifically categorized according to the above (unknown) criteria actually respond to different types of treatment, which is utterly unknowable at this time.

    There are, of course, some things we can tell you with certainty:

    — A man diagnosed with a Gleason score of 8 to 10 is at very high risk for progressive disease and potentially for prostate cancer-specific death at some point in the future if he does not get treated relatively soon. He may be completely curable … but then again, he may not.
    — A man diagnosed with prostate cancer who has a PSA of 50 ng/ml or higher is at high risk for metastatic disease at the time of diagnosis (even if he has negative bone, CT, MRI, and even PET scans) … but he may still be able to live with his disease for 20+ years, just as Fred Gersh did and others are doing.
    — A man diagnosed with prostate cancer in his 40s may be able to live for 20+ years even if he just “watches and waits” and takes active step to manage his general health well for most of the first 15 of those years. Terry Herbert is living proof of this, and he may well still live for another 10-15 years. However, other men diagnosed with prostate cancer in their 40s may be dead within 24 months despite the very best treatment in the world.

    I could go on, but I hope you get the point. The problem is that the vast majority of men being diagnosed today are being diagnosed with clinical stage T1c, Gleason 6 disease, and a PSA of < 10 ng/ml. Most of these men could go onto active surveillance for at least a while until we knew whether they really needed treatment, but most of of them won't even consider this option for socio-cultural reasons, and many physicians are still not comfortable with not treating these patients immediately because of the risk that even one of them will have progressive disease that might have been curable if treated at the time of diagnosis (because they have been trained to try to "cure" this type of cancer by eliminating the cancer cells and because here in the US there are legal consequences for failing to do so if it might have been possible).

    The bottom line is that we are still treating "prostate cancer" as if it was one disease that we can divide into four groups: "not risky because you are going to be dead from something else within 5 years," "risky, but not too risky," "more risky", and "very risky". However, it is really something like 25 different diseases between which we are not yet capable of differentiating at a prospective clinical level at the time of diagnosis, so we don't even know how best to approach treatment.

    Look at the progress that has been made in some other forms of cancer where we have been able to subcategorize the patients with great accuracy at the time of diagnosis (based on the presence or absence of specific clinical markers) and then treat them based on that subcategorization: most patients with chronic myelogenous leukemias can be and are now be managed on drugs like Gleevec for decades; a small subset of lung cancers are now almost completely curable with Xalkori; women with breast cancer with the HER2 mutation are treated very differently and very effectively with Herceptin compared to those without that mutation. We don't have any options like this for prostate cancer yet.

    If I was a urologist today (which I am of course not), all I could do is use the available nomograms to advise an individual patient of the possible (and sometimes but not always the probable) and give him my personal perspective on how he should consider the available options. The very best doctors do exactly that. Others don’t; that I grant you. (There are bad doctors and there are also bad patients.) But there is no agreement whatsoever within the medical community about how to accurately diagnose and manage a patient at risk for prostate cancer today, so you can hardly expect that anyone is going to come up with a universally acceptable model of how to advise men (a) whether or not they need a PSA test; (b) whether to have a biopsy (or an MRI) based on the results of that PSA test; or (c) what to do if that biopsy is positive. There isn’t even consensus among the patient advocacy community about how that advice could or should be delivered.

    While I praise your idealism, I am grounded very much in the real world, where I help those I can help. I would also note that if I really held your level of idealism, I would be working with future generations to change the socio-cultural model, not the current generations. It is extremely difficult to modify adult behaviors and mindsets. Children, on the other hand, can and do want to learn.

  45. Jim:

    My interpretation of the Goteborg study is that it has shown a clear, prostate cancer-specific survival benefit to screening in this specific cohort of men based on this particular study design. No other study has ever shown an effect of this magnitude, so how to further interpret this result is certainly open to question. I don’t think this study was ever powered to be able to show an overall survival benefit. On the other hand, there was not the slightest suggestion in the data from this trial that there ever would be an overall survival benefit.

    Your own “challenging case” is, unfortunately, not representative of the situation most men are trying to deal with. Most men get diagnosed with D’Amico low-risk prostate cancer. You very clearly needed early and aggressive treatment (as did Tony Maxwell). Men being diagnosed with D’Amico low-risk cancer have no way to know whether they really need immediate treatment or not. Every single one of them certainly could go onto active surveillance for at least a while. Our current socio-cultural and medical approaches to the management of cancer are unlikely to encourage such an approach, however.

  46. JIm ,Tony, Sitemaster:

    Four excellent posts. Tony, I am not sure Sitemaster answered your question for you, I am not you, but your question was also a question of my mine which Sitemaster answered in a rare degree of clarity.

    I have underestimated to a large degree the social pressure to just do something, anything, at a very low degree of medical risk. That would greatly diminish PSA’s $cost/lifecost ratio.

    A model requiring no medical advancement and not much social change;

    Example: PSA screening leads to a couple of biopsies on a moderate slope to eventual 6-7 radiation pellets, more biopsies, and eventual surgery for a percentage.

    Alternate: First biopsy showing small tumor; 1 or 2 pellets merely delaying surgery for a time, a percentage of biopsies become PSA tests; some increased percentage simply ages out prior to surgery.

    True financial model is probably based in local custom as much as medicine.

    The social benefit is a reduction in those who simply do not want to know (until it is too late). The $cost/lifecost ratio would probably be greatest in those without much immediate demonstrable medical need.


    Hi again Sitemaster,

    Thank you again for your thoughtful comments.

    It is clear to me too that active surveillance is not yet widely accepted by men; the figure I recall from talks by Dr. Klotz is 10% of eligible patients are now choosing active surveillance (AS), with the other 90% turning to immediate treatment instead. However, while I would not place a large bet, I believe AS will be chosen by a far larger portion of eligible, low-risk patients before the decade closes. Here are some reasons for optimism.

    Until 2003, we had no research reporting outcomes for AS, even with as little as 5 years of follow-up. Now we have at least a half dozen major centers reporting on their AS series, and their reports have favorably addressed major concerns about risks of AS. Not until 2007 was there a major conference of leading experts on AS (led by Dr. Peter Carroll). Until just a few years ago, no major medical organization had endorsed AS; but now the AUA and the NCCN have strongly endorsed it, and you mentioned other organizations in one of your responses that I thought indicated other endorsements. It was only a few months ago in December that the NIH held its major consensus development conference on AS.

    I’m thinking that these developments are very recent in terms of the time and information flow needed for medical practice and patient acceptance to evolve. But now that the key information is essentially in place and credible, I believe practice will change and patient acceptance too. We are already seeing that locally, both from patients who are more willing to consider AS and who hear about it from veterans at support groups, and from doctors who present to us who now say they are recommending it as a prime choice for certain patients, a change from just a few years ago.

    To me the advantages of AS for the right patient are compelling, if the patient will listen, think, and recover his poise after diagnosis: with very high likelihood, AS will allow him to enjoy additional years of life — often for the rest of his life — free of the burdens of treatment, but AS will also catch any stealthily aggressive cancer in time for treatment virtually as effective as immediate treatment. Those are powerful points for any man concerned about minimizing disease risks as well as potential complications and side effects, and now these points can be communicated with confidence. For the first time the pieces are in place for men to decide for themselves that AS makes good sense for them and their families.

    I recognize that you have been at this far longer and intensively than I have, and I hope I am not being overly optimistic.

  48. “I have underestimated to a large degree the social pressure to just do something, anything, at a very low degree of medical risk. That would greatly diminish PSA’s $cost/lifecost ratio.”

    This is a core problem for morbidity and cost reasons:

    (1) The current data show a very high morbidity to mortality benefit. The benefit pertains mainly to Swedes.

    (2) People are generally risk (anxiety) averse. Thus those assigned to the watchful waiting group may irrationally (in a statistical sense) elect to proceed to early RP.

    (3) The fiscal cost of the aggregate of statistically irrational decisions may lead to a blow-out in the cost of population screening.

    (4) In countries where health dollars are scarce and screening general (e.g., New Zealand), this would lead to long waiting lists for radical prostatectomies and pressure on the public service to perform repeated biopsies on a majority (95%) who will never die from prostate cancer. The endgame being over-servicing by the private sector in order to compensate. Given the limited supply of urologists, this would drive up the cost of urology services to all New Zealanders.

    (5) My country cannot afford this. We need to place the cost of concerns re prostate cancer alongside many other — and more “cost effective” — demands on voter health.

    I often advise patients of the statistical probability of serious illness associated with a particular symptom complex. Be it one in ten or one in a million, it is irrelevant if that patient has the illness associated with my analysis — because for that person it is 100%. I also advise them of this fact.

    There is no certainty where there is no certainty.

    What is certain is that all countries have limited health resources.

  49. Hi again Sitemaster,

    (Looks like we’re headed for 50 exchanges. Whew!)

    Earlier I had questioned the maturity of the Goteborg results (4/11 9:11 PM and you had kindly replied at 4/12 7:59 AM). I had glossed over the point I was aiming for in mentioning several cases and research involving challenging cases and long survival. (I suppose I was overly indirect as the line of speculation potentially involves my own mortality as a patient with a challenging case, not that comfortable a topic for me.)

    What I meant was this: beyond the extremely high-risk cases, the next large group of deaths in the screening trials will probably come from high-risk patients, men with challenging cases; however, if these patients are surviving for a timespan from diagnosis that is far longer than survival spans we can infer from current median survival being reported in the screening trials, then we will likely have to wait a good number of years before we begin to see whether there is a difference in survival between the groups. (As I’ve mentioned before, I would really like to see results reported that grouped those actually screened versus those not screened as opposed to the intent-to-screen or intent not-to-screen breakdown.)

  50. Sitemaster,

    I dont know why it is this hard but maybe we are gradually moving towards a meeting of minds — I hope so. My confidence in my ability to explain what I want is taking a battering but I guess I need to try harder.

    From my perspective you take my (perhaps ill-explained) position, raise the bar 1000% (with lots of accuracy added) then announce that it is all impossible. Most people would say it is impossible, I would also say it is impossible. That is not the point. That is not what I am proposing. I am not proposing an idealistic nirvana land diagnosis machine with heaps of accuracy and very little uncertainty. I am trying to establish whether or not we can assemble existing information and offer better practical advice to men at diagnosis on the general parameters of the disease than we currently do. If we can do better I am sure substantial uncertainty would still be part of the mix but hopefully a bit less than we currently offer. My impression is that there is currently virtually no general model covering the 25 or whatever number of PC variants in aggregate and we offer individual advice based on high Gleason, high PSA or whatever in the absence of any general framework. I think we can do better.

    In randomised screening trials we attempt to detect prostate cancer-specific mortality differences related to screening decisions. There is a big gap between the screening decision and eventual death and contaminations in randomisations and the disciplines of the trial create frustrations in the clarity of results. The gap between diagnosis and death is smaller and while still fraught with complications is probably a better target on which to build initial general models.
    Diagnosis (incidence) and death also happen to be the main reference points in national and regional cancer stats (SEER in USA and AIHW in Australia). While having some extra detail (splits into age groups or a time series over a number of years) these are essentially snapshots of the whole population at particular points of time. Medical discriminations and the ability to follow particular cohorts in the statistics are areas that I am less certain of. Cancer registries hold a lot more detailed info (I think) and may allow segregation into medical subgroups of various kinds and the tracking of those subgroups. The medical subgroups may include diagnostic and/or treatment variations.

    I presume the Internet tools you mentioned previously would derive their data from cancer registries of various kinds. Could you please forward me a list of such tools which I can then check out and try to come to grips with what they offer and their sources of data.

    This whole thing may still come to nothing but I am determined to follow it to the end — to the best of my ability.

  51. Jim:

    This type of speculation is just that. Speculation.

    We have no real idea at all how the way someone like you has been treated has affected mortality compared to “standard practice” in Goteborg for a man diagnosed in the mid to late 1990s. As a consequence, for all I know, your chance of 20-year survival from time of diagnosis may be 10 times as high as that for such a Swede … or exactly the same. Trying to turn that lack of knowledge into a set of meaningful population statistics is just not meaningful.

  52. Dear Tony:

    I don’t think you appreciate the scope of the problem.

    The reason we can’t do what you are proposing is because the data you are looking for on outcomes doesn’t exist (quite apart from the data on diagnosis). There are two major missing factors — the individual information about the outcomes from treatment by the specific treating physician and the comparative information on the outcomes after different types of treatment.

    We know that there is MASSIVE variation in outcome depending on the skill and expertise of individual treating physicians and we know that there MAY be significant differences in outcomes based on types of treatment. But we have no data on either of these. (We don’t have the latter because none of the physician groups really wants to know the answer and — at least in America — few patients are willing to be randomized to clinical trials to test the questions that need testing. The currently ongoing ProtecT study in the UK may help us to answer two of them. We should have initial, meaningful data from that trial in about another 5 years.)

    The really big databases like SEER don’t contain any really relevant information because it is all de-identified and they collect basic information that is standardized (when, where, how patient treated, did he live or die). They do not always collect critical clinical information (PSA level, clinical and pathological stage, Gleason score, treatment type, etc.) They sometimes do, but it isn’t actually required. Sometimes one can use data from SEER and correlate that to data from the Medicare database to make retrospective estimates, but I have to tell you that these estimates are not particularly good because so much key information is missing.

    The tools on outcome projection that are available are based on the patient series from selected, large volume, academic surgeons and radiation oncologists. For example, the Partin tables are based on the series of patients treated by Walsh and a few others at Johns Hopkins; the Kattan surgical nomograms are based on the series of patients treated by Scardino at Baylor and later at Memorial Sloan-Kettering. We can use these with considerable accuracy to project individual responses to surgery in men with specific charactistics based on the assumption that they are treated by a surgeon who is at least as good as Walsh or Scardino was when they were doing these surgeries. But that is all we can do.

    The large databases compiled as part of the screening trials were never designed to produce long-term, detailed follow-up data on individual patients. All anyone ever did in those studies was determine whether the patients did or didn’t die and whether or not it was from prostate cancer. What happened in detail to these patients between the time that they were diagnosed and the time that they died is utterly unknown.

    I am not trying to produce a perfect world at all. What I am trying to do is show you that anything we tried to do today is going to havecome with a very high risk for being misleading on an individual basis because (a) it is based on extremely limited information and (b) it massively discounts the potential benefit of active surveillance.

    As Jim has pointed out earlier in this conversation, the Klotz series strongly suggests that every single patient diagnosed with low-risk prostate cancer should not, in fact, receive immediate treatment at all. He should be monitored until it is apparent that treatment is necessary. This radical re-think of they way we should approach the management of low-risk prostate cancer would necessitate a complete re-evaluation of all the available outcome data from series of patients like those of Scardino and Walsh because many (if not most) of the men in those series were, in fact, low-risk patients, so the outcome data are meaningless under the revised model. It is my entirely personal opinion that the entire prostate cacner advocacy community should get much more strongly behind that approach than anything I have seen to date.

    I entirely understand what you are trying to do. My skepticism is based on two associated issues: (1) the fact that I don’t have a clue how to do what you want to do with any degree of projectable accuracy for the individual patient (because the data are just not available) and (2) the fact that even if it was doable I don’t think many people would use it anyway because socio-culturally they do not react to a diagnosis of cancer with logic. That is something that most people come to only later — after they have been diagnosed and treated and they start to understand how few data are available that could perhaps be available to help people make much better decisions. I can say this with some confidence because (to my knowledge) I am the only man in the world outside some sort of research center who became a prostate cancer advocate with absolutely no personal skin in the game (i.e., no peronsl diagnosis, no family member, no personal risk factors, etc.). For me it was initially an educational exercise — in teaching physicians about advances in prostate cancer management — and then, more importantly, an exercise in learning what could be done through the Internet to improve patient knowledge about all disease, but using prostate cacner as a paradigm model.

    I have spent 20 years watching people like you and Jim and Terry Herbert and even Michael Milken struggle first with their diagnosis and treatment and then with what we “should” do about the problem. If we had set up one (or more) real national prostate cancer registries that collected accurate data on tens of thousands of patients starting 20 years ago (which I had certainly suggested to people as long ago as 1993-4), we might have been able to answer some of your questions by now. We didn’t … and we still haven’t … anywhere in the world. Why? Because (a) it would cost a small fortune; (b) the different physician groups would have a really hard time all agreeing on the data that needed to be included over time; and (c) if we tried to do it for prostate cancer there would be immediate howls that it need to be done for several other types of cancer too (and then you would be talking real money).

    Michael Kattan (now at the Cleveland Clinic) has built an entire career out of trying to develop projection models that physicians and patients can use to estimate their outcomes after treatment for prostate and other forms of cancer. I am pretty confident that if he had been able to find data to support the development of more accurate models, he would have done so. Dozens of people have created sophisticated educational tools to try to advise patients how to make good decisions about how to be treated. They work very well … but newly diagnosed patients generally don’t use them. Why? Probably because we are humans. We tend to trust our doctors more than we trust educational tools. (In the case of prostate cancer, at least, such trust may contain a large dose of naivety.)

    I honestly wish you well in what you are trying to do, but I don’t think it can be done, and I think that — as a committed prostate cancer advocate — you would do much better and help many more patients if you were to focus on three much more simple and achievable goals:

    (1) Making sure that every patient diagnosed with low-risk prostate cancer goes on active surveillance until it is clear that he really needs treatment (with the full recognition that, at present, some men really will need treatment despite their low-risk disease because they just can’t deal with the anxiety they suffer on active surveillance.)

    (2) Making sure that every patient diagnosed with intermediate- and high-risk prostate cancer gets evaluated and advised by someone who really knows what the heck they are doing and talks to the patients about the risks and benefits of all available managementy options (including active monitoring in men who have limited life expectancies for other reasons) in an unbiased manner. Only then should they see someone who is actually going to treat them. In other words, we need to clearly separate the diagnostic and initial informational process from the treatment process in order to eliminate all the biases. To give you a clear example, if I was diagnosed tomorrow, I really wouldn’t give a **** whether a surgeon thought I needed surgery. What I would want would be an unbiased opinion on the best way to manage my diagnosis and then, once I had decide on that “best practice” a referral to someone who was truly expert at doing it.

    (3) Raising money and advocating for research into better diagnostic and prognostic tests that will help us to discriminate with high accuracy between clinically signifciant and clinically insignificant disease so that we can stop over-treating men who don’t need treatment and work harder to cure the men who really have clinically significant but curable disease.

  53. Thank you for your excellent discussion and the summary on the issues associated with the work which needs to be done to help select the best option for the persons diagnosed with prostate cancer.

    However, speaking as one who got diagnosed with prostate cancer about 16 months back and had to deal with it, I would like to say that unless I had confidence that the diagnostic tools are good enough to “discriminate with high accuracy between clinically significant disease and cliniclly insignificant disease,” as you express in your wish list no 3, I personally would be hesitant to go on active surveillance unless I was sure that I am not increasing my chance of dying from prostate cancer by waiting for a treatment until my symptoms became worse. Thus, it becomes a personal trade between the chances I am willing to take between the possible side effects of the various treatments and dealing with prostate cancer later at a harder-to-treat stage.

    If, however, I was sure that waiting for a while before treatment could not possibly make my situation worse (or at least significantly worse), the choice to go with active surveillance would be much easier.

  54. Sitemaster,

    Your three goals are worthy candidates for further action if this all comes to nothing — or maybe if it does not. I am gradually realising the difficulties in my proposal but humour me a little longer while I exhaust all reasonable options.

    (1) Have had a look at the Partin/Han and Klotz stuff. Partin/Han is interesting – most particularly if you are a Johns Hopkins patient but probably much wider as well. Klotz is interesting on active surveillance. When I Google Partin some other similar tools seem to come up. Are there others that I should be looking at?

    (2) Is it at all possible with SEER-type databases to track cohorts (based on incidence and mortality only — no medical stuff) rather than the whole population snapshot approach? So identify a cohort diagnosed in a particular year (perhaps split by age groups) then track survivorship of that cohort over succeeding years? SEER may not publish data in this format but do they have the capacity to produce info in this form from their base data — or not? Any medical additions to this would be lovely but cohort survivorship data would be a start. I will make enquiries here in Australia on this as well — Aust Inst Health Welfare (AIHW) nationally and NSW Cancer Institute regionally.

    (3) Can patients build their own databases to generate information. Patients Like Me is an example in the US but it has hardly any prostate cancer members. This sort of data will obviously be biased but a database of substantial size (even if biased) may be a lot better than no information at all. Privacy would need to be ensured as necessary but once the thing was set up the cost of the information would be very low. People would need to be attracted to give data and I guess contamination could be a problem.

  55. Sitemaster

    What about Yananow (The Palpable Prostate?) as site(s) for patient database. Yana seems more descriptive than data centred.

  56. Tony:

    Re your question (1) … See this page on this web site and look under the “Tools” heading. There are literally hundreds of different types of tool available today, but the accuracy of many of these is debatable. For the most recent evolution of the Kattan nomograms for prostate cancer, see this link and then look under the prostate cancer heading. I make no attempt to list or even keep upo with all the available nomograms. I tend to focus on the ones that are actually believed to work well.

    Re your question (2) … You would need to ask an expert on the use of the SEER and associated SEER-Medicare databases to know exactly what might be possible. My guess is that you could probably identify cohorts of some type … but so what? I don’t see what that gets you that we don’t already know.

    Re your question (3) … Sure patients could build their own databases … but the vast majority of patients get facts wrong all the time … and so do even really good researchers. Without a highly structured system to be able to validate the accuracy of all data entered, these databases are of very limited value at anything more than a “gross” level … and we know what to do at a “gross” level already. Just as an example, if we wanted to really know men’s levels of impotence after different types of treatment, it would have to be done according to a highly structured and validated scoring system. I have grave doubts as to whether this could be done well without experienced research oversight and management. The CaPSURE database here in the USA, which is based on data provided by a significant number of urology practices over about the past 15 years now, can produce some quite valuable data, but even in that database, which costs a couple of million dollars a year to run, there are limits to the consistency and accuracy of the data — and it only deals with some men who are treated by some community urology practices.

    Re Yananow and other possible “places” to build such patient-centric registries … I don’t actually think building the registry system is that difficult. There are many patient organizations that have developed patient registries (particularly in the rare disease space where we are only starting the learn the natural history of some of these disorders). As indicated above, the problem is validation of accuracy of the data. An associated problem in prostate cancer is the fact that patients are doing (or not doing) all sorts of other things that may be affecting their outcomes — from diet and exercise and taking or not taking statins to the exact range of supplements that many men seem to be obsessed with. This is a VERY complex data collection problem. If anyone was going to do it, my bet would be that it would be the Prostate Cancer Foundation … they’re the one with the money and the research-centric focus. And then there is the whole issue of getting men to actually participate and whether the system would inherently be biased by such things as internet access and other sociological factors.

  57. Dear akt1000 … I think that Dr. Klotz and Dr. Bal Carter at Johns Hopkins would tell you they already are good enough … for specific men who meet highly specific low-risk criteria. The disagreements start over the minimum age at which active surveillance should be recommended and exactly what different research groups mean by “low risk”. Dr. Klotz appears to be more willing to include younger men with slightly higher-risk “low risk” disease than the Johns Hopkins group. When I use the term “low risk” I am simply using the D’Amico definition (PSA < 10 ng/ml, Gleason 3 + 3 = 6, clinical stage T1c or T2a).

  58. THANK YOU !

    Dear Sitemaster,

    On 4/14 at 9:08 am your comment included this: “… I am the only man in the world outside some sort of research center who became a prostate cancer advocate with absolutely no personal skin in the game (i.e., no personal diagnosis, no family member, no personal risk factors, etc.) ….”

    Words cannot express my gratitude that you have taken this on and continue to pursue it with such dedication, energy, insight and effectiveness. I am amazed at what you are accomplishing, and I offer my profound thanks.


    It is usually difficult for us human beings to change our ways of thinking, and our prostate cancer terminology is no exception. Unfortunately, many men and doctors still do not draw the important distinction between “active surveillance” (AS) and “watchful waiting” (WW).

    WW is essentially waiting after diagnosis until symptoms become evident. In sharp contrast, AS is essentially deferring therapy only as long as key case indicators continue to indicate a non-aggressive case. Therefore, with well done AS, patients who are determined to need treatment get it at a very early stage, long before symptoms develop, and published research suggests excellent outcomes virtually on par with immediate treatment. Again in sharp contrast, WW patients who develop symptoms typically have advanced disease. At that point the opportunity for cure is usually gone, treatment options are limited, and available treatments, such as androgen deprivation therapy, are often not as effective as they would have been if implemented before the cancer had built momentum. I am a big fan of AS, but, with a few exceptions (revolving around short life expectancy), I think WW is nuts! It’s waiting for the cancer to clobber you before you begin to act.

    Dr. Montomery and akt1000 — It appears that you are not distinguishing AS from WW. Dr. Montgomery, in your comments of 4/14 at 8:43 pm you wrote in part: “(2)People are generally risk (anxiety) averse. Thus those assigned to the watchful waiting group may irrationally (in a statistical sense) elect to proceed to early RP ….” Would it make a difference to you if those patients were assigned to an AS group rather than a WW group? To me it makes a vital (sometimes literally) difference.

    akt1000, your comment of 4/14 at 8:43 pm included this statement: “However, speaking as one who got diagnosed with prostate cancer about 16 months back and had to deal with it, I would like to say that unless I had confidence that the diagnostic tools are good enough to “discriminate with high accuracy between clinically significant disease and cliniclly insignificant disease,” as you express in your wish list no 3, I personally would be hesitant to go on active surveillance unless I was sure that I am not increasing my chance of dying from prostate cancer by waiting for a treatment until my symptoms became worse.” Do you understand that you have confused WW with AS? Under AS you do not wait until symptoms become worse. Rather, case indicators that are almost sure to be below the patient’s level of awareness (therefore not symptoms) are key with AS, including such indicators as an increase in the extent or grade of cancer observed in follow-up biopsies, PSA pattern over time, and DRE findings. These indicators allow early and timely treatment intervention if needed.

    A decade ago we did not know whether deferring treatment under AS would work. Now we do; it does! Moreover, AS technology is steadily improving, promising ever better results than the impressive success already achieved. In effect, AS becomes part of the diagnostic process that separates truly indolent from truly aggressive prostate cancer.

    Do you see what I’m getting at here?

  60. I did not understand there was a difference between WW and AS. The terminology is confusing — but thus far has not been relevant in NZ.

    My role is to spend as much time as is required in order that the patients understand the complexities in so far as it is possible. I will request as many PSAs as people want. I will also advise of national recommendations re screening. Currently we don’t have any.

    I do know that there is enormous variation with respect to risk. Practising medicine is like driving — the landscape is constantly changing and only the irrational will choose to drive over a cliff.

  61. Dear Dr. Montgomery:

    If the distinction between WW and AS is not yet widely appreciated in New Zealand, I would respectfully suggest that that is indicative of a real medical problem from a patient perspective. Understanding of and appeciation of the potential benefits of AS compared to WW (most particularly for those men diagnosed early with low- and very low-risk prostate cancer and a life expectancy of 10 or more years) is critical to the risk/benefit decision process that relates the benefits of immediate aggressive treatment for localized disease to the potential for loss of quality of life.

    For details I would refer you, particularly, to the publications by Laurence Klotz and his colleagues in Toronto and to those of H. Ballentine Carter and his colleagues at Johns Hopkins in Baltimore. There have also been several good review articles on this topic in journals like the Journal of Urology, the European Journal of Urology, and BJU International. The major area still to be resolved in active surveillance is the frequency of repeat biopsy. Some major research centers now suggest a follow-up 12-core biopsy to confirm low-risk disease at the time of initiation of active surveillance and an annual endorectal MRI (along with careful monitoring of PSA, PSA velocity, PSA density, and other clinical tests, such as regular DRE) as triggers for further repeat biopsies. Other centers are still using annual biopsies as a monitoring process, although that seems excessive to me unless specific patients are at elevated risk for progressive disease). The normal trigger for treatment is a Gleason score of 7. Obviously, there are some men whose anxiety level makes active surveillance an impossible management strategy for them as individuals, just as there are some older patients with more advanced disease who are not good candidates for WW because they feel compelled to just “do something.”

    When talking to patients, I usually try to make the key distinction that WW is a passive management strategy designed to minimize the risk for unnecessary side effects of things like hormone therapy whereas AS is an active management strategy designed to ensure curative treatment — if and only if it really does become necessary.

  62. Thank you for your response.

    The salient measure will — I am sure most agree — be the number needed to undergo RP or radio (brachy or external beam) treatment in order to prevent one death from prostate cancer-specific mortality.

    Do you have any current data or projections wrt this?

  63. Dr. Montgomery:

    This is a much-debated statistic, and in all honesty I don’t think anyone really knows the correct answer to your question.

    The most recent analysis of data from the ERSPC study has suggested that 37 men need to be diagnosed with prostate cancer to avert one case of prostate cancer-specific death (see article above). How many of those men actually need to be actively treated was not specified. The data from the Goteborg trial suggested that it could actually be as low as 12 men needing diagnosis to save a single prostate cancer-specific death (at least in the Swedish population of Goteborg, based on the standard treatment protocols in place at the time of the trial). However, we also know from the Goteborg data that 82.8% of the men treated sugically in that trial, and who claimed to be fully potent prior to their surgery, were either impotent or sexually inactive at 18 months post-surgery. That’s quite a cost in terms of quality of life if 11 of 12 men really didn’t benefit in terms of their overall survival.

    I think it is a reasonable argument to state that well-coordinated active surveillance would actually minimize the number of men who really needed to be actively treated over time … so long as we are sensible about active surveillance and don’t start giving every patient diagnosed with low-risk prostate cancer excessive numbers of PSA tests, follow-up physicals, high-tech imaging scans, and biopsies that they don’t really need. This is likely to be an easier process to manage in countries like Australia, New Zealand, and most of the European countries with national health care systems than it will be in America, where access to “high tech medicine” is viewed as something approaching a “right”, whether it is really necessary and helpful or not.

    As an eternal optimist, I am hopeful that the ProtecT trial currently ongoing in the UK will actually be able to give us much better insight into this question … although I also expect there to be much debate over how to interpret the results when they become available.

  64. Good thread on WW versus AS. Another indication that a smaller, closely defined study may have more insight than a larger study contaminated with the perception of authors of believing they have a common definition when they actually do not.

    On a “marketing” approach, Dr. Montgomery’s phrase “active management” sounds sounds so more more proactive than “active surveillance.” Much more appeal to the “do something, anything” emotion.

    Question: Is there any evidence that early, light treatment increases the percentage of patients who never reach the critical care stage?

    Catch it early, kill it or wound it severely, never reaching or delaying the mega life/cost impact stage. … Or am I still day-dreaming.

  65. Mike:

    There is no really compelling evidence that I am aware of that allows us to state categorically that any form of early, “light” treatment will stop a progressive form of the disease. However, there is clear, anecdotal evidence that lifestyle changes and dietary changes can affect the risk for progression in at least some men. The problem, of course, is that the men who those lifestyle changes may seem to be helping may be the men who have indolent or very low-risk disease anyway.

    I can tell you that if I was diagnosed tomorrow with low-risk disease at my current age of 64 years, I would certainly go on to some form of active surveillance and I would make greater efforts than I currently do to improve my diat and exercise better. (Of course if I was really smart I would be doing that already … but my wife is a gourmet cook and I spent 45 years exercising like a maniac and it has become very boring!)

  66. On Dr. Montgomery’s question, on how many need to be treated to avert one death, I provided info from the European trial on this issue at the beginning of the thread. The European trial implies that as of an average followup of 11 years after screening was begun, around 21 men received some form of definitive treatment (surgery, radiation, chemo) for every life saved from prostate cancer mortality. This is lower than the ratio of diagnoses to averted prostate cancer deaths, as a higher percentage of the screening group versus the control group received no treatment other than AS/WW.

    So in my view, the tradeoff under current treatment patterns is between a 5% reduction in the 11-risk of prostate cancer death vs. a much larger (50%?) probability of serious side-effects of the surgery. So, on the one hand, the probability of serious side-effects of treatment exceed the reduction in prostate cancer deaths from the treatment. On the other hand, I think most men would view death as a more serious consequence than the side-effects, although some men may take the other view.

    I also suspect that with more development of AS as a treatment approach, the numbers in the European study could be improved upon. It appears that many more cases in the screening group in the European study were plausible candidates for AS, but ended up choosing definitive treatment. If more men could be persuaded to pursue AS, side effects could probably be reduced with relatively little impact on death risks.

  67. In response to Jim Waldenfels comment on April 15, 2012 at 4:57 pm:

    Maybe I should have used the word “cancer” instead of “symptoms” in my statement “I personally would be hesitant to go on active surveillance unless I was sure that I am not increasing my chance of dying from prostate cancer by waiting for a treatment until my symptoms became worse.”

    Then it would have read, “I personally would be hesitant to go on active surveillance unless I was sure that I am not increasing my chance of dying from prostate cancer by waiting for a treatment until my cancer became worse.”

    Do I still feel that way after reading all the discussion? Yes, because the idea of active surveillance is essentially to wait until the cancer becomes worse. Maybe the probability of survival has not changed significantly. But we don’t seem to know how big a change that is.

    Also, I would hate to depend on a biopsy to tell me the change in cancer over a year’s period, because it seems to be a hit or miss proposition. When I had a negative biopsy, the first thing my urologist warned me about was that it could be false negative. When I did have cancer show up in the biopsy, he had no idea about what part of the left prostate gland the cancer was in. Unless we have imaging techniques capable of detecting cancer in prostate, which are good enough to directly compare the extent of cancer change from, say, year to year, I would continue to feel uncomfortable with the active surveillance approach.


    Hi again Sitemaster. Your response to me of 4/12 7:59 am got me thinking about the differences between Swedish treatment of prostate cancer in the ’90s and subsequent outcomes versus recent treatment in the US and outcomes. My bottom line is that I’m realizing that the Goteborg results, considering the regional and temporal context, are more mature than I had been thinking, though still, I think, less mature than needed for useful, need-to-treat to prevent death figures for patients and doctors making decisions today. I’ve come up with a couple pieces of evidence, and maybe you or someone else can take them further than I have.

    The major evidence is from the well-known Holmberg/Bill-Axelson study of surgery versus watchful waiting in some Scandinavian countries; as the study was led by Swedish researchers with a majority of Swedish institutions participating, I’m thinking most of the patients were probably from Sweden. The time frame for study entry was 1989-1999, virtually matching the frame of the ERSPC and also spanning the December 1994 launch of the Goteborg study. I’m thinking the Holmberg/Bill-Axelson study should give useful insight for the treatment and outcome situation in Goteborg at the time of the Goteberg study.

    I hit on this after trying to come up with a way to address the point you made in your 4/14 7:34 am post that outcomes stemming from the Goteborg study environment could be quite different from the outcomes I was describing for a later US environment. While, unfortunately for the current maturity issue, the “15 year” update published in the NEJM in May 2011 does not give us diagnosis-to-death time span statistics (and I could not tease it out from the complete paper), the update does give enough information to suggest that substantial numbers of prostate cancer deaths were occurring. The diagnosis-to-death statistics obviously exist for the Homberg/Bill-Axelson study, but they do not appear to have been published in this or other related studies, based on my quick look at titles using PubMed.

    The second evidence is from SEER data of 5-year relative survival rates (US). It shows projected survival probability at the 5-year point for those diagnosed in 1987-1989 (85.4%), 1990-1992 (95.3%), 1993-1995 (96.2%) and 1996-2004 (99.4%) for White men, with parallel figures for Black men. While survival has been virtually 100% in the US recently, it is clear that survival was substantially lower for patients in the US SEER locations who were diagnosed around 1994 when the Goteborg study was launched.

    Food for thought!

  69. All,

    Looks like my thoughts are coming to naught — will puddle on a bit longer to tie up any loose ends and notify when done.

    On some other matters … This site has more than 140 posts so far in 2012. About 30-40 have no comments to date, this one has 66, a very small number have 10-20 comments and the rest have 1-9 comments, so I guess that makes the PSA testing controversy the top of the pops. But the most extraordinary thing is the effort to get those posts up and then to subsequently manage them and the debates involved. I think the culprit is obvious (I saw the passion in Chicago) and I want to thank you, mate, for the effort and commitment on this.

    Fred Gersh gave me an origami bird in Chicago. You are giving us all origami birds with this work.

  70. Dear akt1000 … This is exactly why some research centers now carry out annual, specialized, endorectal MRI scans as opposed to annual biopsies on their active surveillance patients, and base the need to do biopsies on the results of those scans. Is it as good a method as we need? No. Is it better than just doing a biopsy every year? Almost certainly.

  71. Jim:

    I don’t think I can speculate meaningfully on the relative importance of these very different types of data.

  72. Tony:

    It’s not that your “thoughts are coming to naught” at all. Your thoughts are part of the continuing need for motivated patient advocates to understand what will constitute good and valuable scientific evidence that will change clinical practice for the better. It is not enough to have a passion. One has to be able to use that passion to encourage meaningful change over time. Get yourself integrated into one of the prostate cancer clinical research teams at a major academic center. Become their patient representative as they discuss and consider new trials and new ideas. Keep on learning … and share what you learn with other patients. Try to help to stop so much money getting wasted on studies that will never tell us anything useful.

    With regard to the PSA issue being “top of the pops” … On its own, actually it isn’t (in my opinion). The “top of the pops” is the combination of the value of screening and the subsequent “what do you do with that information?” problem.

    We are coming from a place in which the hypothetical assumption was that earlier diagnosis would necessarily lead to not just superior prostate cancer-specific but also superior overall survival. What we appear to have demonstrated (so far) — from an epidemiological point of view — is that earlier diagnosis does indeed lower prostate cancer-specific mortality, but it does so at a massive cost in terms of complications, side effects, and impact on quality of life, and it may have no significant impact on overall survival at all. Conversely, the entire screening and outcome controversy has “re-booted” the whole issue of whether careful active monitoring over time is a necessary and sufficient initial method for managing many thousands of men diagnosed each year with low-risk disease. This is a critical question from the point of view of the individual patient. I would argue that the currently available data signals a strong “Yes” in answer to that question. I just hope to heck that the ProtecT study will be able to confirm this.

    If we can generate data that can convince akt1000 that > 95% of men with low-risk disease can be effectively managed over time with active surveillance and curative therapy for those men on active surveillance when it becomes clear that they need it, then we may be able to convince an awful lot of other men (and their doctors) that immediate treatment is not even a very ethical response to a diagnosis of low-risk prostate cancer.

    Last but not least … Thank you for the kind words, but I am just providing a forum for sharing of knowledge. It is what others will do with that knowledge over time that is really important!

  73. Dear Sitemaster:

    I have made some comments previously on this blog that were subsequently interpreted in a manner I did not intend.

    I have read very widely re this subject.

    My current preoccupation is understanding the statistics from the ERSPC. I have searched the net but have been unable to locate the specifics from the Finnish arm. This showed a small benefit I believe. There is also a claim that a subset from Sweden (20,000 studied) displayed an NNT of 12 to 1. I want to know the NNT data from these arms as they comprise near 50% of the pooled data.

    In addition, with regard to to active surveillance (AS) — I have some questions.

    As I understand it, AS requires serial biopsies. Prostate biopsy places men at risk of sepsis. (5% as I understand it.) In addition there is a cumulative psychological and financial cost. Have these costs been quantified?

    I would suggest that there would be a “tipping point” at which many men would say “B…..r this (NZ expression), I’m having it out.” What “counselling” strategies exist to deal with this contingency.

    I have read that 10% of men experience depression post-radical prostatectomy. This may well be correlated with morbidity. Depression may be an unfamiliar psychological state to many — but if it is predicated upon an irreversible physical state which is core to male function, then it may assume a significance as debilitating as death from an awful disease.

    Life is not just about how long we live — but how we live.

    I do not know — and do not want to know — my PSA. I am 56 years old.

  74. akt1000,

    “Not knowing what part of the left prostate the cancer was in”. I agree, that should be a standard finding. That information was at the time of biopsy just as readily available as knowing it was a left core. It probably is available today, depending on preservation of samples and working biopsy notes. It is not reported, because even those physician recommending AS have not reached the rather obvious (to you and I) significance of the question.

    Many physicians have not yet stepped beyond the mindset that the actual treatment will be the complete removal of the prostate, the only questions are will you die first another way, which method and when to remove, and how much collateral damage.


    I am assuming that there is also a linkage between type of treatment (surgery, radiation, chemo) and the viewed risk of each cancer. Is the chemo post surgery/radiation or a stand-alone treatment in Europe?

  75. Dear Dr. Montgomery:

    There appear to be four very different issues that you are raising, so let me see if I can deal with them one at a time.

    (1) The data from the ERSCP trial: I can’t help you further with this issue. You need to go to the source and talk to the investigators if you want that level of clarity. The e-mail addresses for the lead authors of the various studies are usually available on the abstracts of the published data in the original journals on line (but not always on PubMed abstracts). I may be able to help you find e-mail addresses for specific individuals if they are not easily found. My experience has been that many of these investigators are very willing to be helpful by providing complete texts of published data or posters from meetings.

    (2) Issues associated with active surveillance: (a) There is considerable controversy associated with the need for and frequency of repeat biopsies in men undergoing active surveillance. This has been addressed in at least two of the prior comments I have made in the past couple of days. There is no “standard practice” today, and I think it is important to understand that what may be being done under a research protocol at Johns Hopkins (for example) may not be necessary in day-to-day clinical practice. (b) Biopsies do place men at risk for infection, but the rate of sepsis is not anywhere close to 5% that I am aware of. Most of this risk appears to be avoidable with the use of appropriate prophylactic antibiotic therapy based on culture of bacteria from rectal swabs prior to the biopsy. (c) There are all sorts of psychological and financial costs associated with the management of prostate cancer — by active surveillance and by immediate intervention. I have no idea how one could accurately quantify these because they are much too dependent on the predispositions of the individual patients. I would also note that many of the same concerns come up with the management of breast cancer. We don’t deal with them well in that setting either. (d) There is most certainly a tipping point for some men at which they do indeed say “Bugger this, I’m having it out.” I would only point out that this tipping point seems to affect a relatively small percentage of men who have committed to AS in the well documented clinical trials. The centers that have been involved in these trials have always allowed their patients to “opt out” of AS and have active treatment if that is what the patients want to do. It’s the patient’s cancer and the patient’s body after all. As I understand it, the degree of counseling has always been tailored to the patient. The goal has not to be to keep the patient on the trial but to allow him to discuss his concerns and come to his own decision about how he wants to act.

    (3) Issues related to post-treatment depression: Many men suffer depression before and/or after treatment. So do many of their wives and partners. This issue comes up all the time. It is one of the very best arguments for active surveillance as a form of management. It is also one of the very best arguments for a much greater need for selectivity in the application of screening programs and biopsies. The risk for depression is predicated not only on the occurrence of “loss of manhood”, however. It also depends on a host of other factors. I know of men who, diagnosed with aggressive, high-risk stage T3 prostate cancer, have been treated with surgery, radiation, adjuvant hormone therapy, and recovered to resume their very positive attitudes to life (in some cases with apparently complete return of sexual function). I know of other men who have been paralyzed with fear at the idea that they needed a biopsy … for years. The range of human response to stress is astonishing. I don’t think we should over-simplify this issue. However, you are quite correct when you suggest that — at least for some men — post-treatment depression “may assume a significance as debilitating as death from an awful disease.”

    (4) Your own specific attitude to your own specific situation: One of the things that most people completely fail to appreciate is the distinction between medical policies and guidelines and the actual practice of medicine. I have no personal problem at all with your decision about not knowing and not wanting to know your PSA level. I have never believed in mass screening for prostate cancer. Indeed, I have been taking flak for this from most of the prostate cancer community for 20 years. My belief has always been that a man needs to come to his own conclusions (in conversation with whoever he wants to talk to about this) whether he needs to get PSA tests. We can clearly define some groups who appear to be at higher risk, and (as I said before in this long discussion), if I was a Black American of sub-Sarahan origins, I suspect that I would be much more likely to think that regular PSA tests might be a good idea. However, I am a boringly white Caucasian with no known risk factors and a 200-year family history of almost no cancer at all. … The only reason I have a PSA test each year is that it makes my very competent primary care physician happy and my life insurance requires it. I will add that even if my PSA suddenly came in at 3.00 ng/ml (as opposed to the usual 0.5 or 0.6 ng/ml), I would most certainly want at least two repeat PSA tests before I decided I needed a biopsy. I am in complete agreement with your comment that “Life is not just about how long we live — but how we live.” However, you need to appreciate that others may not have quite the same attitude as you and I do. In saying all of the above, I would just add that if your father, one grandfather, and one brother all died of metastatic prostate cancer and you aren’t getting PSA tests on a regular basis, then you may have lost your grasp on reality! :O)

    A someone suggested early on in this discussion … people always want simple answers to complex problems. It is human nature. One of the reasons that I was willing to do what I am doing (all over again) is because prostate cancer offers a paradigm for an increasingly complex problem: “How do we, as a society, develop systems that offer the highest reasonable standard of health care to the largest possible number of people?” A key component of that, for me, is “How do we ensure that patients are given useful, unbiased information about the health care problems they may be facing, unadulterated by the prejudices of specific interest groups?” At a very deep level, I understand that our current social and medical systems are not really addressing this issue. The patient community is largely excluded from most of the discussions of medical policy and development of meaningful guidelines. Are we making any progress? I hope so. Is it fast enough? No. Can I tell you and other people what is “right” or “wrong” for them as individuals? Of course not. That’s not the idea at all. Indeed, arguably, the idea that we even could is a fundamental problem that many in the medical community ignore to the detriment of their patients.

  76. Dr. Montgomery,

    I agree about not knowing your PSA if that is your choice. A choice which only you and each individual man must make for themselves. (Well, those with access to healthcare.) If you actively desire to not know, you have probably already made your decision about treatment, intentionally or not. That is different from having you head in the sand.

    Some urinary symptoms led to my first PSA about 6 years before biopsy. Symptom-free, I would not have had more than the first PSA, and maybe not even the first. No annual PSA until ramp-up began. Three years between biopsies and still borderline tipping point.

    Panic PSA and biopsies, no thanks. None at all, no thanks.

    Knowledge is neither good nor bad. Only in usage is it one or the other, usually both at the same time.

  77. I agree that if specialized, endorectal MRI scans provide good tracking of the cancer progress (with biopsy required only for confirmation), then that will make active surveillance a lot more acceptable. Hopefully such scans’ usefulness will be proven and then accepted by the medical community soon, so that the facilities to have such scans done are available at more places.

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

%d bloggers like this: