Many prostate cancer patients and advocates may be less than enthused by data published this week in the New England Journal of Medicine, and based on 11-year follow-up of patients in the European Randomized Study of Screening for Prostate Cancer (ERSPC).
The study managers for the ERSPC have previously reported a 29 percent reduction in prostate cancer-specific mortality among the men in this trial who underwent PSA-based screening for risk of prostate cancer compared to those who had no screening. This most recent study was carried out to assess the degree to which “harms” to quality of life resulting from over-diagnosis and over-treatment can be considered to counterbalance the previously reported benefit.
Heijnsdijk et al. applied a process known as microsimulation screening analysis (MISCAN) to estimate the number of prostate cancers, treatments, deaths, and quality-adjusted life-years (QALYs) gained after the introduction of PSA screening. It should be remembered by readers that this study is based on an exercise of complex modeling skills requiring many assumptions, and so all results are open to considerable question based on assessments of the accuracy of the modeling process.
Having said that, here are the core findings reported by Heijnsdijk and her colleagues:
- For each 1,000 men screened, annual PSA-based screening of men aged between 55 and 69 years of age would result in
- 9 fewer deaths from prostate cancer (a 28 percent reduction in prostate cancer-specific mortality)
- 14 fewer men requiring palliative therapy for prostate cancer (a 35 percent reduction in the need for palliative care)
- A total gain of 73 life-years (i.e., 8.4 years for each prostate cancer-specific death avoided)
- A total gain of 56 quality-adjusted life years (QALYs), with a range of –21 to 97 QALYs
- To prevent a single prostate cancer-specific death
- 98 men would need to be screened
- 5 cancers would need to be detected
- Annual screening of all men aged between 55 and 69 years of age would result in
- An increase in the number of life-years gained (from 73 to 82 per 1,000 men screened)
- No change in the number of QALYs gained (still 56 per 1,000 men screened)
Heijnsdijk and her colleagues conclude that the “benefit of PSA screening was diminished by loss of QALYs owing to postdiagnosis long-term effects.” They are, however, very careful to note their belief that longer follow-up data from both the ERSPC study itself and from appropriate quality-of-life analyses are essential before it would be possible to make any “universal recommendations” about the value of annual PSA-based screening.
The “New” Prostate Cancer InfoLink has not seen the full text of this article, and — even if we had — we do not feel competent to make any reasonable assessment of the validity of the modeling process used by Heijnsdijk et al. We have also not seen the full text of the editorial (by Sox) in the New England Journal of Medicine addressing this article. We will see if we can get copies of the full text of the article and the editorial in the near future. At this point in time we are only able to report what is stated in the abstract of the article.
Other commentaries on this study are already widely available (on HealthDay, on Medscape, on Reuters, on MedPage Today, and on WebMD). Some readers may want to review all of these commentaries. It is apparent from at least some of these commentaries that the editorialist in the New England Journal beleives that these data further justify the importance of discussion between individual patients and their doctors about the risks and benefits of PSA testing before administration of any such test.
Filed under: Diagnosis, Management, Risk, Treatment Tagged: | ERSCP, outcomes, quality of life, screening
Search for new and
ongoing trials on the
CTAG PCa web site
I think you’re misinterpreting the conclusions of this article.
What the article concludes is that for every 1,000 men screened, based on the parameters they use, there is a net gain of 56 “quality-adjusted life years”. If we don’t allow for the harms (e.g., impotence or incontinence in some men treated) then the gain would be 73 life years gained per 1,000 men screened. But because of these harms, the years gained are down-weighted. They do the down-weighting based on average values from surveys of prostate cancer patients.
However, as they point out, this conclusion is sensitive to how one values the relative risk of dying versus the harms of treatment. One can get a range of results, from a loss of 21 QALYs to a gain of 97 QALY.
You should think of QALY as just a unit of measure. They are trying to express all heath-related effects of screening in one measure. To do so, they have to assume how much harms such as incontinence or impotenence reduce life quality. However, this is not something to which there is a right answer for each individual. Different people have different values and preferences.
If you want to make your own calculations, I think the simplest way to do so is to focus on the statistics for life years gained extra prostate cancer treatment. So, what they conclude is that for prostate cancer detected via screening, for every 5 cancers detected, 1 prostate cancer death is averted, which on average increases lifespan by 8.4 years. For those 5 cancers treated, most studies suggest 2 or 3 men would have significant side-effects. So, if you have a cancer detected through screening, you have a 20% chance of treatment saving your life, and if it does, your will live 8.4 years longer. (Or to put it another way, choosing treatment increases your average life span by 1.68 years = 8.4 divided by 5). But you have a 40% or more chance of serious side-effects. Should you choose treatment? It depends.
From a benefit-cost standpoint, what this analysis misses, as they point out, is any consideration of the financial costs of prostate cancer screening or treatment. This is only based on health impacts of the screening and treatment, and ignores any financial costs.
Tim:
I’m not interpreting or misinterpreting anything. I am simply reporting what the abstract said (which is all I have seen, as I carefully point out).
If you want my opinion (which is not what I gave previously) it is that 56 QALYs (and I do fully understand what a QALY is) for 1,000 men who might well live for at least another 20,000 years in total after diagnosis between the ages of 55 and 69 (based on normal life expectancies for men of those ages) is a pretty poor benefit to risk ratio given the risks.
But the 56 QALYs already adjust for the risks. It adjusts for the discomfort and other health hazards of biopsies, the loss in well-being due to knowing you have prostate cancer, and the adverse side effects of definitive treatment due to impotence or incontinence. Now, someone might disagree with the adjustments. They only adjust down from 73 QALY to 56 QALYs. So what they are saying is that the risks reduce about a quarter of the gain due to lives saved.
In addition, that’s the whole point of screening. Only a modest number of men screened have the condition. So by definition it is impossible to have a huge effect per man screened. But since screening is easy and cheap, the small ratio of QALYs gained per man screened is not a big issue. The real issue is how the benefits of screening compare to the harms due to the extra men treated. And their answer is that under plausible parameter values, the former outweigh the latter.
Given that the real personal health care costs of PSA screening are due to the treatment, the real ratio to focus on is the ratio of additional men treated to lives saved, not the ratio of men screened to lives saved. The ratio of men treated to lives saved is what drives the benefit-cost analysis.
Another way to put it is that if we monetize the 56 QALYs at a value of $100,000 per QALY, which would not be an unusual dollar benefit to assign, the 56 QALYs are worth $5.6 million. The value per man screened is $5,600. Obviously the costs of the screening test per man screened will be far less than that. Now, if the definitive treatment costs $30,000 per man treated, and we have 45 extra men treated per 1000 men screened (9 fewer deaths times 5 extra prostate cancers detected per death averted), then the extra costs of definitive treatment due to screening are 45 times $30,000= $1,350,000, or $1,350 per man screened.
So their baseline assumptions result in PSA screening passing a benefit-cost test even if we include the monetary costs of screening and treatment.
Now, if one disagrees with their baseline assumptions, then you get different results. For example, for a man who would rather be dead than impotent, PSA screening clearly makes no sense. For every death averted, we have probably 2 impotent men, so the probability of impotence exceeds the reduced risk of death.
But Tim … PSA testing isn’t even close to being an accurate diagnostic tool.
According to this very paper, 98 men in 1,000 tested with PSA would need a biopsy to prevent a single death from prostate cancer. Many of the “harms” encompassed by this paper are actually associated with the biopsies, not the cancer itself.
I think we are just disagreeing about the value of the two sides of the risk/benefit equation for men who are at no particular known risk for prostate cancer (like me). I have no problem with the idea that men who are at known risk would be wise to get regularly tested. They are the one’s most likely to benefit. My problem comes in the regular testing of all the men (5 in every 6) who are at no particular known risk for clinically significant prostate cancer at all, and who are therefore at higher risk from the harms associated with biopsy (let alone unnecessary treatment) than they are for the benefits of a treatment they are unlikely to ever need or have.
Tim … Your financial argument doesn’t apply to or consider the 55-year-old who dies as a consequence of sepsis resulting from biopsy he never needed to have (let alone his family). You are only including the positive costs, not the negative ones.
I’m afraid you’re arguing with the authors of this article, not with me. All the harms you are mentioning are accounted for in their adjustments that they make to QALYs, from 73 to 56. So what you are really arguing is that in your opinion, the various harms are worth more than 17 QALYs per 1000 men screened. Maybe so. However, the fact is that the baseline scenario in this article, IF you accept their evaluation of harms versus benefits, is extraordinarily favorable to PSA screening.
Not at all. I am saying that average risk is a ridiculous concept because it doesn’t apply equally to those at more or less risk in the beginning. Your argument is only valid when you make the assumption that everyone is at equal risk for prostate cancer, and that quite certainly is not true, which is why the range of the results is from minus 21 to plus 97. You can’t have it both ways! That’s why the authors clearly state (in comments in other reports on this paper) that it is important for men to make individual decisions after discussion with their physicians — a point that I am in entire agreement with.
:O)
THANK HEAVENS — THE RESEARCHERS RECOGNIZE THAT THE ERSPC IS PREMATURE!
Here’s the sentence in Sitemaster’s article giving me modest reassurance the researchers have not joined in the lemming-like rush over the cliff of worthless findings regarding screening for prostate cancer: “… They are, however, very careful to note their belief that longer follow-up data from both the ERSPC study itself and from appropriate quality-of-life analyses are essential before it would be possible to make any “universal recommendations” about the value of annual PSA-based screening….”
May we properly emphasize that word “essential“?
Once again, let us not forget that, at present, the extent of favorable impact of the very loose type of screening done in the ERSPC is impossible to estimate because currently published follow-up is clearly, simply, fundamentally inadequate!!!
This critical point, overlooked in a number of papers citing ERSPC results, rests on these three observations. Firstly, 11-year follow-up is reported from enrollment of apparently healthy men — not diagnosed prostate cancer patients — in the ERSPC trial (actually a set of trials with sundry key features, but Sitemaster has addressed that previously, among other telling flaws). Secondly, therefore, as there was an average lag of at least several years before prostate cancer was diagnosed for those men in the trial who became patients, follow-up from the time of diagnosis in the ERSPC at the “11 year point” was well short of 10 years! Thirdly, as survival of prostate cancer in Europe at the 10-year point is highly likely to be at least not too inferior to survival in the US, and as it has been well established that survival in the US at the 10-year point is virtually 100% for men with low- and intermediaterisk cases, with about 95% survival even for men with high-risk cases, it is extremely unlikely that the ERSPC has entered the point at 11 years where a substantial proportion of men in the trial would have died from prostate cancer!
Here’s a simple (though a tad gruesome) analogy that captures the essence of the problem. Let’s say you wanted to report mortality for laboratory rats falling (with help from the friendly researcher) from the top floor of a building known to be at least 10 stories high, with status at each story on the way down checked with a high-speed camera and other diagnostic tools. Well the ERSPC approach would be to check and report results for rats reaching somewhere between, probably, the sixth through the eighth story down, and naturally the results would be pretty good at that point. A few rats would have suffered unfortunate encounters with protruding air conditioners and perished, but the main body would be doing great. However, if survival at each lower story parallelled survival years after diagnosis for prostate cancer, the vast majority would still be doing fine after descending for at least 15 stories.
Let’s add a thought experiment to the analogy: half the rats are somewhat protected by balloon suits (analogous to men well-screened for prostate cancer), while the rest have no protection. Would we expect any real difference between survival of the groups at, say, the eighth story down? Of course not! Would we expect any difference when they finally hit the ground? Many of us would, though, admittedly, determining the extent of protection is the reason for the experiment. Many of us are convinced we won’t begin to see useful results until at least the 15-year point. Based on current technology for managing and treating the disease, I would not be surprised if it takes well beyond 20 years of follow-up from enrollment before we have confidence that the impact of ERSPC style screening has been captured.
Therefore, any calculations based on the currently profoundly premature ERSPC findings are worthless, except arguably as a way of estimating the inner boundary — the line indicating the lowest benefit of screening, with a vast amount of room for greater benefit to be observed as the trial matures into the zone of meaningfulness.