Does 120-day mortality after prostate biopsy outweigh any possible benefit of screening?


Back in 2008, Gallina et al. demonstrated that the risk for mortality at 120 days after a prostate biopsy was of the order of 2 deaths per 1,000 prostate biopsies, and historically this is the only population-based study ever to assess this risk.

In a new paper to be presented at the upcoming annual meeting of the American Society of Clinical Oncology (starting this Friday in Chicago), Boniol et al. have used data from the Prostate, Lung, Colorectal and Ovary (PLCO) Study to see if they could obtain a more accurate assessment of the risk of death within 120 days of a prostate biopsy — but based on data from a more carefully monitored population participating in a randomized clinical trial. (However, it does have to be noted that the PCLO trial had structural and other issues that raise questions about its design and impementation.)

Boniol et al. used data from the PLCO study on all men participating who were followed until December 31, 2009. These data included all biopsies performed after initial randomization of the trial participants.

Here is what they say they found:

  • 36 deaths occurred within 120 days after 12,300 prostate biopsies.
    • 17 such deaths occurred after 8,390 biopsies in the “screening” arm of the PLCO.
    • 19 such deaths occurred after 3,910 biopsies in the control group.
  • 17 deaths had “neoplasm” listed as the underlying cause of death.
  • 32 deaths occurred after 9,124 biopsies that were positive for cancer (3.5 deaths per 1,000 biopsies).
  • 4 deaths occurred after 3,176 biopsies that were negative for cancer (1.3 deaths per 1,000 biopsies).
  • Deaths were evident among all age groups from 55-69 to 70-74.
  • There was no difference identified in age between the study arms (p = 0.45) or between those with  and without a prostate cancer diagnosis (p = 0.73).

Boniol et al. conclude that, “The mortality rate at 120 days following prostate biopsy of 1.3 deaths per 1,000 biopsies, in a population free of cancer, is a serious concern” for the computation of risk vs. benefit in the initial diagnosis of prostate cancer.

The authors go on to note that such biopsy-related mortality “would occur earlier than any benefit from a screening program and could reverse any potential gain from screening such as recorded in ERSPC study.” They further note that this mortality rate is “in line with the risk reported by Gallina et al. (2008) and is now based on a properly monitored population.”

While the precise benefit of screening based on the ERSPC trial is still disputed, it does have to be noted that if all those extra biopsies are inducing 1.3 deaths for each 1,000 biopsies carried out, it does make one wonder about whether the benefit is worth the risk for an awful lot of the otherwise healthy men involved. The original (2009) report from the ERSCP showed that 1,400 men needed to be invited to be screened and that 48 men would need to be detected to save a single life from prostate cancer. A later, follow-up report modified those data and stated that 1,055 men would need to be invited to be screened and 37 cases of prostate cancer would need to be detected to save a single life. Even given the more recent data, this means that on a population basis more men are at risk for biopsy-related mortality within 120 days than the number of men whose lives would be saved as a consequence of screening.

We do, on the other hand, feel that it is necessary to point out that these data take no account whatsoever of the risk factors that may give some men (African Americans, Afro-Caribbeans, and men with a significant family history of prostate cancer) a higher potential level of benefit from early detection of prostate cancer than others.

31 Responses

  1. Here are some numbers to help appreciate the very troubling consequence of this observation:

    If 25,000 men get screened and 1 life is saved per 1000 screened then about 25 are saved in 12 years.

    To save those lives, at least 15% get a biopsy (total = 3750). The cancer detection rate is about 4% (1.000) so 11% (2,750) have a negative biopsy.

    If 3.5 deaths occur per 1,000 positive biopsies then 3.5 deaths occurred from the biopsy out of the 25,000 screened and 1,000 found to have cancer. If 1.5 deaths occur/1000 men with a negative biopsy, that means 4.1 (1.5 x 2.75) deaths occurred in those with negative biopsy. Adding to two together totals 7.6 deaths.

    So the net result is: screening 25,000 men saves 25 men in 12 years but causes 7.6 men to die from the biopsy. In addition, about 1.6 die from the surgery (assuming 80% of the 1,000 cancers detected get a radical prostatectomy and the death rate is 0.2%).

    Now the net benefit is 25 lives saved and 7.6 + 1.6 = 9.2 lives lost for a net benefit of 25 – 9.2 = 15.8 lives saved.

    Unfortunately, the analysis cannot end there because the calculation needs to factor in the net impact on life expectancy.

    If we assume that the average age of men getting diagnosed is 62 and the average life expectancy is about 75, then each man dying from the biopsy or treatment loses about 13 years of life expectancy for a total loss of 119.6 years (9.2 x 13). Of the men who were prevented from dying of prostate cancer, some of them did not need to be treated and the entire group may only gain on average a few years of additional life because of the long natural history of the disease. If we say that screening and treatment added on average 4 years of life for the 25 men who avoided dying from prostate cancer, then the net gain in life expectancy is about 100 years (25 men x 4 years each).

    Now the bottom line is: screening added 100 life-years but caused a loss of about 120 life-years for a net loss of life expectancy. If the deaths from the biopsy did not occur, then screening would be increasing life expectancy, but when it is included, the balance now shifts to a net loss.

    No doubt many people will argue about some of the rates being used, but these are very close to the right ones and at the very least, screening does not appear to offer a real benefit to society — but it may help some individuals.

    This analysis excludes the impact of screening on quality adjusted life years, which would tip the scale even further toward a net harm of screening.

    Gerald Chodak, MD

  2. This is an emotional, shoot from the hip response: the prostate biopsies were for screening only? Seriously? How did they get volunteers for this study? In any case, I was happy to see elsewhere in this web site that creative researchers are looking for new ways of screening for prostate cancer.

  3. Dear Dr. Hanline:

    No … The biopsies were carried out only in men with elevated PSA levels in the screening arm of the PLCO trial or in men with clinical signs of risk for prostate cancer in the non-screening arm of the trial.

  4. Thank you for the additional analysis Gerry … very helpful.

  5. In the 2008 paper they actually observed very few deaths in the control group (no biopsies). Maybe I miss something but I cannot understand these statistics because the standard mortality tables show a 1% probability to die yearly for a man in his 60s. This alone explains the 0.3% death rate apparently found for the biopsies (with a cohort of older patients).

  6. PLCO AND ERSPC STUDIES LACK REQUISITE CREDIBILITY

    I appreciate the report of this study and the comments, especially Dr. Chodak’s analysis, and I believe there is information here that will help advance understanding.

    However, I continue to be mystified that any credibility is assigned to the widely reported, updated figure from the ERSPC that 1,000 men need to be screened to save a single life. That figure is not believable for a number of reasons, but the elephant in the room (apparently not visible to all though – advanced cloaking technology evidently) is that the ERSPC bases its calculation on woefully premature follow-up! With a background (though rusty) in demanding statistics and experimental design, having studied both papers, it is painfully obvious to me that the ERSPC follow-up, with a documented, median, updated follow-up of 11 years in the core age group from the date of each participant’s randomization (not from his diagnosis) must involve a follow-up from diagnosis of well under 10 years (unstated). In the USA, survival per SEER data at 10 years is 98%, just slightly below the 5-year survival of virtually 100%, and even 15-year survival is 93%. That’s for all comers, including low-risk patients as well as higher-risk patients. Assuming Europe is not substantially behind the US in diagnosis lead time and treatment effectiveness, it’s evident that very little difference in mortality between the so-called “screening” and control groups would be evident at the so-called 11-year point, especially considering that inter-group differences would be further diminished by flaws such as non-compliance, a sketchy version of screening (every 4 years), and protocol differences.

    Moreover, the 11-year ERSPC follow-up (March 15, NEJM) coupled with the earlier 9 year follow-up practically shout to us that the results are premature:

    — In 2009, at 9 years of follow-up: screenings to save one life = 1,410

    — In 2013, at 11 years of follow-up: screenings to save one life = 1,000

    We should all observe that the number of screenings needed has fallen by 410 with just two more years of follow-up! I will jokingly suggest we should wait for four more years of follow-up at this rate of decline of 205 per year: 1,000 – (4 x 205 = 820) = 180 screenings to save a life. Obviously that is not valid, but neither is the claim so boldly stated, and so out-of-the-context of the earlier claim.

    Is there any reason we should not all join in whacking this mole — the ERSPC assertion of screenings needed to save lives — to its final demise?

  7. I can’t for a moment reconcile Dr. Chodak’s observations and many similar I’ve read from time to time with the fact that there has been a 40 percent decrease in prostate cancer mortality concurrent with the advent of PSA testing.

  8. I hope the paper clearly states that those deaths 120 days after the biopsy are caused by the biopsy itself. Otherwise, Dr. Chodak’s numbers are meaningless…

  9. ROLE OF ALL-CAUSE MORTALITY IN THESE FIGURES

    We need to bear in mind that that the 120 days time frame after the biopsy amounts to one third of a year, and that some men are going to die from unrelated causes during that time.

    Here’s my cut at getting a handle on the number. I’m using the Statistical Abstract of the United States, 128th edition, Table 102, page 75, “Selected Life Table Values 1979-2005, specifically the section for Expected Deaths Per 1,000 Alive at Specified Age, for men aged 65 in 2005. Their mortality was 17.6 per 1,000. Dividing that by 3 to estimate the figure for one third of a year gives us 5.9 deaths.

    There is a slight downward yearly trend amounting to a few tenths per year (e.g., 18.9 annual deaths in 2002). I’m thinking that would result in about 2 fewer deaths by 2013, or an annual rate of about 16, or about 5.3 deaths, somewhat roughly.

    On the other hand, prostate cancer has been compared to the canary in the mine that alerts us to other problems, and its credible to me that men diagnosed with prostate cancer would have somewhat higher mortality risk from other causes than the average man. To a lesser extent, the need for a biopsy is probably also associated with some increased mortality risk.

    Perhaps we can use this perspective to view the results, but, as noted earlier, the elephant in the room is that the ERSPC projection of screenings needed to save a life is simply not credible. To me, at best it is somewhat useful as an indicator of the minimum benefit of screening with a true value expected that will be well above that minimum.

  10. Dear Seymour, Jim, and Ralph:

    (1) Actually it is extremely easy to reconcile the decline in prostate cancer mortality over the past 25 years with the analysis presented by Boniol et al. (which Dr. Chodak has only attempted to clarify). That decline in prostate cancer-specific mortality has utterly ignored the likely increase in biopsy-induced mortality over the same 25-year period! However, no mortality data that I know of have ever listed “prostate biopsy” as a cause of death, even if it was indeed a biopsy that induced infection followed by sepsis followed by death. There are actually several possible reasons why a biopsy might lead to death within 120 days post-biopsy, but the most likely is certainly an infection followed by fever, hospitalization, and death.

    (2) The fact that the follow-up of the ERSPC data is far too short to give us a reliable assessment of the number of men we need to screen to save a life doesn’t affect the study results or Dr. Chodak’s comments on the risks associated with excessive biopsies. (This from the person who pointed out to Jim some time ago now that we might not actually be able to get a reliable result from the ERSPC trial unless we followed all the patients for something like 40 years from their initial randomization.) You can’t blame Dr. Chodak for using the available, accurate data to make his point.

    (3) What is crucial about the analysis by Boniol et al. is the point that Dr. Chodak is making about loss of life expectancy. If you die 120 days after a prostate biopsy because of an effect of that biopsy which found no cancer, you are dead, and you may have lost up to 30 years of life unnecessarily (if, say, you were 55 years of age and men in your family normally live to about 85). By comparison, identification of prostate cancer and effective treatment may only add 5 years to your life, so it would take 6 lives saved to replace the 1 life lost at age 55. This is irrespective of the number of lives actually being saved. Jim has specifically made the point that life expectancy after a diagnosis of localized prostate cancer in the USA today is at least 10 years (even without treatment for most men).

    The critical factor in all of this is not what is right or what is wrong from a statistical point of view. The critical factor is that we just don’t know most of the data we need to know because we keep studying the wrong things or doing those studies badly. We then insist on making decisions based on bad data instead of stating the obvious, which is that we really don’t have the data to be able to tell whether screening is beneficial or not and so in all truth we don’t have a clue whether screening is saving lives today or not (as opposed to whether PSA testing in the 1990s may have saved lives in a previously untested population, which may well have been the case, and which would correlate with the results of the Goteborg screening study, which was again in a previously untested population).

    One of the things that is important about being a good advocate is to point out the truth as opposed to arguing a point of view. The only thing I know today about the utility of prostate cancer screening with the PSA test is that “I don’t know” if it is helpful or harmful for men as a whole. I certainly hope that it is helpful for men known to be at high risk because (at the moment) it’s what we’ve got, but we don’t actually have the data to demonstrate that either! There is absolutely no doubt at all that men are dying as a direct consequence of prostate biopsies, and the truth is that we don’t have any accurate data of how many men that is happening to each year. Bobiol et al. have provided an estimate based on reasonable data. That estimate has implications, but it is only an estimate. If you don’t like the implications, as advocates you need to be pressing for better studies and better data!

  11. Dr. Chodak: Where did you get the number “4 years of life” added.

  12. Mike,

    Do you really think that after millions of biopsies during the last two decades, 3.5 deaths per 1,000 cases from the procedure would go unnoticed? Hard to believe when they claim to need to screen 1,000 to save one …

    Add to that the lack of certainty that the cause of death was actually the biopsy. The Gallina et al study they cite does not provide a clue about the cause of death either. Results:

    1. Overall mortality after 120 days was 1.3% versus 0.3% in the control
    2. Men age 60 or less had a 120-day mortality after a biopsy of 0.2%
    3. Men age 76 to 80 had a 120-day mortality of 2.5%
    4. The number of cumulative biopsies had progressively lower death results.

    I get the impression that results are impacted by:

    1. A higher death rate of older patients.
    2. There is no mention of cause of death in the abstract. Many men could have died of old age, very advanced prostate cancer at diagnosis, etc.
    3. Men with a poor level of health died more often.

    I think that the calculation of loss of life depends much in the actual death rate of the prostate biopsy procedure. In simpler words, if the rate reported is over-stated the calculation is wrong. Recent sources do not support the rate of death reported either. I am sure we will hear more about this. …

  13. Dear Ralph:

    I have been telling people for years that PSA testing is not a benign procedure because it leads to biopsy and that biopsies are related to an increasing rate of infections. (See this report from 2012 as an example.)

    I sincerely do hope we are “going to hear more” about this. I have no better idea that anyone else exactly how many men are dying as a direct consequence of prostate biopsies within 120 days. However, it really wouldn’t be that difficult an issue to track if someone set out to track it. We already know that there has been an increasing rate of identification of infections associated with biopsies, and that those infections have been associated with organisms resistant to standard forms of antibiotic used for prophylaxis pre-biopsy (e.g., specifically including drugs like ciprofloxacin).

    People have an amazing capacity for not seeing what they don’t want to see. It would be nice if some members of the infection control community were to look very specifically and carefully at the medical histories of men in their 50s to 70s admitted to hospital with serious infections to track exactly how many were occurring within (say) 60 days of a prostate biopsy and how many of those men were dying from complications of their infections.

  14. The relevant issue is whether the increase in biopsies due to screening leads to an increase in deaths shortly after biopsy due to possible side-effects of biopsies due to a true causal relationship, rather than correlations due to other factors.

    Judging from the information presented in the abstract, this doesn’t seem to occur in their sample. They report 19 deaths as of 120 days after biopsy out of 3,910 biopsies in the control group, and 17 deaths as of 120 days after biopsy out of 8,390 biopsies in the screening group.

    Therefore, the doubling of biopsies in the intervention group, which can reasonably be attributed to the screening, led to no increase in deaths.

    We might expect the death rate to be lower if the additional biopsies in the screening group were among healthier individuals. But the lack of any increase in deaths whatsoever suggests that a screening-induced increase in biopsies is not associated with extra deaths shortly after biopsy.

    Perhaps the full study presents information that contradicts this conclusion from their abstract. But the abstract by itself tends to suggest that screening-induced biopsies must have very low immediate post-biopsy additional death risk beyond what would have occurred anyway due to a variety of causes.

    As others before have pointed out, because men at this age are dropping dead all the time, it is challenging to use the raw death rate after biopsy as a measure of what death risk is caused by biopsies. And even if we compare a biopsy group to a non-biopsy group, we cannot assume that the biopsy group had the same health status pre-biopsy as the non-biopsy group.

    We need to look at some change in biopsy prevalence that is uncorrelated with other health status variables, and the random assignment in the experiment is one way to do so. As far as I can tell from the abstract, the raw random assignment data does not support a causal interpretation of the raw correlations between biopsies and deaths in the first 120 days.

  15. Mike,

    Ralph’s point about death by biopsy going unnoticed for so long is a realistic question in my eyes. But I reason in my little mind that when death by biopsy occurred in the past that it was chalked up by the medical community as not related to the disease at all but rather a procedural issue. Therefore it was not associated with prostate cancer-related mortality through screening. This being the case, many knew about the issue, they must have, but they dismissed it as an issue with general diagnostics of cancer. It would go a long way for advocates to understand why this crucial information was missed. You thoughts appreciated.

  16. Tony:

    We can speculate as much as we like about what has happened in the past. Your suggestions are certainly one possibility that correlates with other experience in the history of medicine. Tim Bartik’s suggestions are another. However, what we need are accurate data based on really careful study. Speculation can generate hypotheses, but without accurate data that is all we have.

    I would just make one observation. Most men arriving in a hospital emergency room with a fever are probably never asked whether they have had a prostate biopsy in the past month or so. If the hospital doesn’t have that piece of information, how could they possibly connect the dots?

  17. Although I agree that the death rate is surprising and higher than has been reported in other studies, there are several reasons that these findings are likely to be correct:

    (1) It is based on a large, prospective study that makes it easier to capture these results than in retrospective studies.
    (2) Two of the largest published studies on biopsy-related complications that reported much lower rates only measured the complications at 30 days rather than 120 days, which would not capture many of these deaths.
    (3) In the only other report on 120-day mortality that includes a large number of patients, Gallina et al. found that both age and health correlated with the morality after biopsy, suggesting that reasons other than infection may be a contributing factor.
    (4) Infection/hospitalization rates from prostate biopsy have been increasing and resistant organisms are occurring more often.

    As mentioned in my first post in response to this article, applying the lost life expectancy to the full analysis of screening appears to result in a net negative for routine screening. For those questioning my use of an average increase in life expectancy of 4 years for calculating life expectancy in those men whose death was avoided by screening, that number is probably too generous; breast cancer screening only increases average life expectancy by 2.3 years. The negative impact of screening on life expectancy also was reported by Bonioli (BJUI 2012) based on the ERSPC.

    We can expect that if only young, healthy men were getting screened then the mortality rate following prostate biopsy would probably be much lower and the net impact might be more positive, but, older, less healthy men are getting screened. In the VA system, over 30% of men over age 75 get PSA tests. That is why society as a whole is probably worse off because of PSA testing even though some men are definitely better off.

    To coin a line from Spock in Star Trek, “the needs of the many outweigh the needs of the few or the one”. It is time to accept the fact that the overall impact of screening for prostate cancer is doing more harm than good in our society.

    As many others have suggested, men need to get better information about the benefits and harms of screening before they get their PSA measured.

  18. Dear Gerry:

    Could you please check your reference to Bonioli (BJUI 2012)? I can’t identify this paper on the BJUI web site or on PubMed.

    Mike

  19. Contrary to Dr. Chodak, even if the sample is large and has longer-run follow-up, it doesn’t mean anything unless the findings can be given a causal interpretation. The relevant issue is: what is the increased risk of death during the 120 days after a prostate biopsy that is caused by the biopsy. We can have a sample that is huge with great follow-up data, but if all we observe is raw death rates, which can be caused by many things other than the biopsy, we haven’t learned much.

    For a causal interpretation, we need to observe the change in death rates due to some change in biopsy rates that is uncorrelated with unobserved personal attributes that might be associated with the risk of death. In the PLCO study, the most obvious such exogenous change in biopsy rates is the experimental assignment. And as I mentioned below, the data presented in the abstract do not support that this exogenous increase in biopsies led to an increase in deaths during the 120 day follow-up.

  20. Tony,

    The number of deaths when adding those diagnosed and those with negative biopsies in the last 20 years would be something close to 100,000 dead men. Not likely to hide under the rug …

  21. Mike,

    Found two recent references about the topic:

    Carlsson SV, Holmberg E, Moss SM, et al. No excess mortality after prostate biopsy: results from the European Randomized Study of Screening for Prostate Cancer. BJU Int. 2011; 107: 1912–7 … which states that “Only 14/31 (45%) of the screening-positive men who died within 120 days were biopsied and none died as an obvious complication to the biopsy” and concludes that “Prostate biopsy is not associated with excess mortality and fatal complications appear to be very rare.”

    Kakehi Y, Naito S, Japanese Urological A. Complication rates of ultrasound-guided prostate biopsy: a nation-wide survey in Japan. Int J Urol. 2008;15:319–321, which states that “Medical records of 212,065 procedures carried out between 2004 and 2006 were reviewed” and concludes that “Overall, contemporary ultrasound-guided systematic prostate biopsy is safely carried out in Japan. However, minor complications are not infrequent, whereas major ones remain very rare but life-threatening if they occur.”

    I still believe that biopsy mortality in this study is overstated big time.

    Darn, Peter Boyle seems to be on both sides of the issue …

  22. Mike,

    The name is Boniol M, et al. and the reference is:

    Boniol M, Boyle P, Autier P, Ruffion A, Perrin P. Critical role of prostate biopsy mortality in the number of years of life gained and lost within a prostate cancer screening programme. BJU Int. 2012;110:1648-52.

  23. Regardless of the debate about the validity of the study, from the standpoint of an internist whose male patients are predominantly in the older age group, it seems the value of this study is that it gives patients and physicians additional information to help guide decisions about prostate screening or diagnostic studies. The days are over when men whose PSA begins to rise are automatically referred for prostate biopsy.

  24. I have read the Boniol/Boyle study in BJU International (2012) that was referred to above. It is a simulation study that relies on the results of Gallina to project the consequences of a screening program. In their simulations, biopsy deaths are considerably greater than deaths from prostate cancer surgery. They use a figure of 0.2% deaths from prostate cancer biopsy, or 2 in a 1,000. With this number, biopsy deaths lead to a loss of years of life that exceed the years of life gained from the extra and earlier prostate cancer treatment that results from screening.

    You can download the study at Boniol’s website at http://www.mendeley.com/profiles/mathieu-boniol/

    But the key issue is whether this number is right. There are at least two follow-up letters in BJU International that challenge this study. One letter, by Carlsson et al., argues that Boniol et al. have overstated mortality from prostate biopsies by four-hundredfold. This letter goes on to point out that 1 million men annually undergo prostate biopsies in the U.S. The 0.2% mortality figure implies that 2,000 men annually die from prostate biopsies. They then go on to argue that, “On intuitive grounds, we assume that these many deaths would have been noticed …”

    Another letter, by Goldstraw, also argues that the prostate biopsy mortality statistics of Boniol et al. are overstated.

    In looking back at the Gallina study from 2008, which also is available online for free, the key problem is that it is not at all clear that the comparison group is at all comparable to the group getting biopsies. It is clear that the biopsy group is dying at a higher rate per 120 days than the comparison group, but it is not clear that this has anything to do with biopsies. It is interesting that in the Gallina study, the increased death rate during the 120 days after biopsy does not show any sign of being concentrated in the first 30 days or 60 days, which is perhaps consistent with this increased death rate being due to pre-existing differences in health between the two groups.

    I think a fuller verdict on the Boniol et al. 2013 study discussed in the blog post would have to rest on seeing a full copy of the study. However, the statistics presented in the abstract do not support a causal interpretation of post-biopsy mortality for one simple reason: there is no sign that the increased biopsies induced by screening lead to increased all-cause mortality during the 120 day follow-up period.

  25. TAKE HOME THOUGHTS FROM THIS EXCHANGE

    Thanks Tim for your last; thanks Sitemaster, and thanks to all. I have thought about all 24 responses up to this point.

    1. Prostate cancer biopsies that are negative pose virtually no threat to life, at least in the period prior to concerns with MRSA. Contrary to being a serious concern, as described in the study, the information we have reviewed here indicate that a negative biopsy poses virtually no threat to life. As just one of a number of observations indicating that, overall normal mortality appears well poised to account for the mortality noted in the study. MRSA staph infection still appears to be quite rare, with even rarer mortality, but I’m not aware of good information either way about that for prostate cancer biopsies. As a patient advocate/educator, I believe it is best to note the MRSA risk but put it in perspective as an extremely low-level threat, dwarfed, in my opinion, by the value of smart screening that includes screenee appreciation and acceptance of active surveillance as a wise option for appropriately low-risk cases.

    2. We don’t have sound data supporting the lack of benefit of screening. In other words, screening may be as beneficial as many of us believe it is, though we cannot prove that conclusively. As Sitemaster has noted here and previously, the ERSPC and PLCO studies are simply too flawed, including prematurity, to provide any sound indication of the extent of benefit — or lack of benefit — of screening. I accept that those flaws do not pertain to the issue of the harms of biopsies, but they do pertain to the benefit part of the balance related to harms of biopsies, a balance addressed in the paper and in Dr. Chodak’s model. I see no reason to alter a view that the “1,000 screenings needed to save a life,” which was the view stated in the “11 year” update of the ERSPC, is a most invalid and misleading figure that deserves no credence whatsoever. Indeed, that figure needs to be exposed as worthless!

    3. We haven’t thoroughly addresses the biopsy risk portion of 120-day mortality following positive biopsies. Here are three aspects of this:

    First — It appears at first impression that a substantial portion of such mortality is due to advanced disease at the time of biopsy. (That could have been me under the ERSPC protocol of screening every 4 years. Had my initial screen result been normal, but with cancer developing shortly afterward and galloping along with a 3 to 4 months PSADT without another PSA check scheduled for 4 years, the second screening could well have been too late, likely well after development of symptoms.)

    Second — Based on the observations about negative biopsies, it seems most unlikely that the biopsies would have had mortal consequences that are direct medical consequences of the biopsies.

    Third — In contrast to the preceding point, there is some information suggesting that the psychological burden of a positive biopsy could cause death. In his co-authored book Invasion of the Prostate Snatchers, Dr. Mark Scholz cites studies showing that “During the first week after diagnosis, the risk of suicide goes up twenty-two-fold.3 Heart attacks are ten times more likely.4” (Footnote 3 is: Unnur Valdimarsdottir, Completed suicides among newly diagnosed prostate cancer patients. Genitorurinary Cancer Symposium, Abstract 104, February 2008. Footnote 4 is: Fang Fang, Cardiovascular events among newly diagnosed prostate cancer patients. Abstract 104, same source.) I haven’t looked into these studies, but a brief search uncovered the following pertinent doctoral thesis and references that included the word “suicide”:

    Division of Clinical Cancer Epidemiology Department of Oncology — Institute of Clinical Sciences Sahlgrenska Academy at University of Gothenburg, Sweden MAINTAINING QUALITY OF LIFE AFTER PROSTATE CANCER DIAGNOSIS Doctoral Thesis Þórdís Katrín Þorsteinsdóttir Registered Nurse ??/div Gothenburg 2011

    50. Llorente MD, Burke M, Gregory GR, Bosworth HB, Grambow SC, Horner RD, et al. Prostate cancer: a significant risk factor for late-life suicide. Am J Geriatr Psychiatry. 2005;13(3):195-201.

    84. Fall K, Fang F, Mucci LA, Ye W, Andrén O, Johansson J-E, et al. Immediate risk for cardiovascular events and suicide following a prostate cancer diagnosis: prospective cohort study. PLoS Medicine. 2009;6(12):e1000197.

    85. Fang F, Keating NL, Mucci LA, Adami H-O, Stampfer MJ, Valdimarsdottir U, et al. Immediate risk of suicide and cardiovascular death after a prostate cancer diagnosis: cohort study in the United States. J Natl Cancer Inst 2010;102(5):307-14.

    86. Bill-Axelson A, Garmo H, Lambe M, Bratt O, Adolfsson J, Nyberg U, et al. Suicide risk in men with prostate-specific antigen-detected early prostate cancer: a nationwide population-based cohort study from PCBaSe Sweden. Eur Urol. 2010;57(3):390-5.

    The abstracts of these studies suggest to me, at first impression, that there is some risk, though the risk of suicide in the era of widespread PSA testing in the US seems to be negligible (in contrast to two earlier periods). The risk of heart attack could be due to treatment effects, such as clots, as well as stress, it seems to me.

    Dr. Sholz advocates “thorough education and reassurance [of patients] “prior to being biopsied” to “calm men down.” (p. 91) Having first-hand experience with the stunning impact of a high PSA result, that makes a world of sense to me, and I’m convinced that prior to the PSA test there needs to be some brief education about active surveillance and general reassurance about the high success rates with prostate cancer today.

  26. Gentlepersons:

    I am utterly unclear where this discussion went off track, but there is no reason that I am aware of — in theory — to think that a man who has a positive result after a prostate biopsy is at any greater risk of death within 120 days of having a biopsy than a man who has a negative biopsy result. Indeed, the whole point of the original paper and Dr. Chodak’s additional comments are that it is otherwise completely healthy men who are greatest potential risk from undergoing an unnecessary biopsy because they are at absolutely no risk whatsoever of death from prostate cancer at that point in time.

    If having a prostate biopsy really does increase risk for death within 120 days (please note the emphasis on the word if), that would have to be associated first and foremost with events directly caused by the biopsy … of which infection and certain other forms of traumatic consequence would be the most likely candidates. Whether the patient already has cancer or not was never the issue here.

  27. Item number two is often used by proponents to justify the screening process, arguing that it is better to over-test until it is proven to be bad rather than under-test until it is proven to be good. Unfortunately, this is very flawed thinking. The only way we can make progress in medicine is to prove that something is good before recommending it. Following the logic used in the previous comment means that every new drug that comes along should be available to the pubic without proper testing until someone proves it is harmful or not effective. That is not the way to practice good medicine. Far too many people “believed” that something was good for patients only to find out they were wrong.

  28. Mike,

    That is exactly what Boniol reported and you initially wrote: “32 deaths occurred after 9,124 biopsies that were positive for cancer (3.5 deaths per 1,000 biopsies). 4 deaths occurred after 3,176 biopsies that were negative for cancer (1.3 deaths per 1,000 biopsies).”

    That amounts to 4,8 deaths per 1,000 associated with a prostate cancer biopsy. The implication is that prostate cancer biopsies end up killing men. The implication from Boniol is that there is no benefit to screen or diagnose prostate cancer. Why present such data when the cause of death is not clearly defined? I am not supportive of conspiracy theories, but there is something wrong when a bombshell from Lyon is really a small firecracker. …

  29. PLEASE FORGIVE THE ELLIPSIS IN SHIFTING TO POSITIVE BIOPSIES

    Yes, the paper certainly focuses on risks of negative biopsies, and that has been the theme of the discussion.

    The similar plausible non-cancer risks (at least not directly) from positive biopsies are part of a larger pattern relating to the decision of a person to have a biopsy, of course, as there is no way of knowing in advance how the biopsy will turn out. It is reassuring to have gone through this thought process and to have observed that these risks appear to be quite remote. This will boost our confidence in advising sons and others about having prostate biopsies.

  30. Dear Ralph and Jim:

    You are at complete liberty to believe what you like. My only point throughout out all of this discussion has been that I would like much better data to be able to resolve it.

    Boniol and his colleagues have stated an hypothesis that I reported … “warts and all.” I think that hypothesis needs to be either rebutted with real data (which I don’t think we have at this time) or actually tested by collection of accurate data. You seem to think I have some sort of agenda here that I absolutely do not have. What I do know (as Dr. Chodak has just noted) is that the best intentions can commonly have highly unexpected consequences.

  31. “PROOF” THE GOAL — “BEST AVAILABLE INFORMATION” THE REALITY

    Dear Dr. Chodak, Sitemaster, and fellow participants in this discussion,

    The post from Gerald Chodak of June 3 reflects what appears to me to be a dominant perspective in medicine today, at least in academic research circles, that proof is necessary before recommending an approach. The often-used shorthand label for this approach is “evidence-based medicine.” At first impression, that certainly strikes one as reasonable. After all, who would not want evidence behind their medical choices? Moreover, as Dr. Chodak cautions, many of us are aware of drugs and procedures that have looked good on the surface, only to be later revealed as dangerous or at least with some significant risks that are not obvious. Thalidomide in the ’50s (birth defects), DES as androgen deprivation therapy (cardio issues), and Zometa (osteonecrosis of the jaw) come to mind. (I’m still a fan of androgen deprivation as well as thalidomide and Zometa when used wisely.)

    The Phase III, randomized, placebo-controlled, double-blind trial with high compliance of a very large group of patient participants followed for a sufficient number of years with overall survival as the key score is the typical ideal, and that model works very well for many diseases and cancers, especially for those with generally short survival times after diagnosis, such as most pancreatic and lung cancers, but also with mesothelioma, liver, esophageal, and stomach cancers, all of which have 5-year survival rates of 27% or lower per recent SEER data for the US. As a practical matter, it is not that hard to hold together the key members of a research team for, say, 3 to 7 years, allowing completion of research and publication.

    However, is there anyone who will defend the practicality of the foregoing ideal model for prostate cancer, in view of 5-year survival at virtually 100%, 10-year survival at 98%, and 15-year survival at 93%, coupled with the generally elderly age at diagnosis with associated co-morbidities (ACS data)? Indeed, our survival is so long that it is likely there will be substantial advances in technology that affect us during our years of survival, let alone during the shorter periods during which we may be under restrictive rules of trial protocols.

    Regarding the biopsy issue, I too would like to see better information, but I’m convinced we will never have ideal evidence. (This is similar to the screening controversy.) If that is so, doesn’t it leaves us with either silence about biopsies or, on the other hand, using less than ideal evidence, coupled with reason, to assess their worth and make recommendations? For most other issues, such as acceptance of new drugs, I’m with Dr. Chodak in striving for “proof” as often as possible. I don’t want to see new drugs accepted as a standard of care without strong evidence, and I would even like to see a much stronger FDA role regarding supplements.

    But biopsies are so important, for such an important disease as prostate cancer, with an abundance of good but imperfect evidence available, that I’m convinced we should take a stand in favor of biopsies, a stand which not only recognizes but emphasizes the key role of active surveillance and screenee empowerment prior to the procedure, indeed, prior to screening. Regarding the balance of evidence regarding the mortality risk of biopsies, I am especially impressed that the death rate from biopsies within 120 days is well within the death rate that is simply associated with age for men, suggesting negligible additional risk, if any. (The paper suggested 3.5 deaths following positive biopsies in 120 days, versus 1.3 deaths for negative biopsies, and US survival data suggests a substantially higher level of about 5.3 deaths for all US men aged 65 during that period — my post of 5/30 12:52 pm and others.)

    Thanks, Mike, for sticking with this thread.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: