Last week we reported data from a study by Tangen et al. suggesting that men diagnosed with metastatic prostate cancer are living longer today than they were in the 1980s and early 1990s. Frankly, we don’t find this finding particularly surprising. Why this is happening is a whole different issue.
According to a commentary on Reuters.com, published last Friday, Dr. Brawley (the Chief Medical Officer of the American Cancer Society) appears to have gone out of his way to suggest not only that we cannot attribute this increase in survival exclusively to the application of PSA testing (which we entirely agree with, and which was clearly stated by the authors of the paper) but also to imply that this survival benefit may not even be real (which we find to be an extraordinary suggestion).
Now, to be fair to Dr. Brawley, it is not entirely clear exactly what he did or did not say when interviewed by Reuters. Only some of what he supposedly said is actually placed in quotation marks and therefore should be assumed to be what he actually stated.
In some places he is very clear that you cannot attribute this survival benefit to PSA testing alone and appears to be suggesting that there may be other good reasons for the survival benefit. This is a precise quote from Dr. Brawley as reported by Reuters:
You cannot say that PSA testing is the cause for this improvement, but when you eliminate all the other potential explanations there is something else there that is playing a role.
However, Dr. Brawley also seems to have indicated that data from the management of other cancers would suggest that the perceived survival benefit is questionable for any reason. The following information from the Reuters commentary includes a direct quotation from Dr. Brawley, but was not all actually stated by him verbatim:
… other trials done by the same research group but involving non-Hodgkin lymphoma, a group of blood cancers, also show patients live longer over time, according to Dr. Otis Brawley. … And that’s despite the facts that doctors don’t screen for the disease and that patients got the same treatment for their disease. … “There is a cohort effect that we have seen numerous times in medicine,” said Brawley, …. He added that the reasons for the improved survival over time are unclear, but could be related to advances in medical care and infection control.
Of course medical care and infection control have improved over the past 30 years. Thank goodness for that! In addition, it is reasonable to consider that we do in fact monitor people for risk of blood cancers like non-Hodgkin’s lymphoma (a group of several related types of blood cancer). Risk for non-Hodgkin’s lymphoma can be identified by swollen glands in the neck, armpit, and groin, by the presence infections (particularly of certain specific types of infection), and by the presence of excessive numbers of white blood cells in your blood. Thus, anyone who gets a regular annual check up is being monitored for the possible risk of lymphoma. Do we have a test to screen specifically for non-Hodgkin’s lymphoma? No we don’t. And we don’t have a test to screen specifically for prostate cancer either, because the PSA test is not cancer-specific.
Dr. Brawley has long expressed a strong personal belief that widespread use of the PSA test (i.e., mass, population-based “screening”) is not associated with any significant survival benefit in the management of prostate cancer. Others agree with him about this. However, unless Reuters has misinterpreted Dr. Brawley’s remarks, he is now at least hinting that any appearance that men diagnosed with metastatic disease may be living longer today than they did 30 or so years ago may also be “bunk”.
What the authors of the original paper were very, very careful to note was that while this increase in the survival benefit of men diagnosed with metastatic disease occurred during the “PSA era” it was not possible to know exactly why it had occurred.
Dr. Ian Thompson, the senior author of the current study and the current co-chair of the prostate cancer committee of the Southwest Oncology Group (SWOG), apparently told Reuters that,
I guess I have a difficult time explaining the difference [in survival].
Dr. Thompson is a highly regarded and careful scientific investigator who is also well known for his care and tact in his public remarks to the media. His comment is entirely in line with the conclusions of the original paper. He is correct. Explaining in detail why this survival benefit is occurring is extraordinarily difficult and may well involve a multitude of factors (inclusive of the advent of PSA testing). However, Dr. Thompson appears to have made no suggestion that this survival benefit is not “real.”
What The “New” Prostate Cancer InfoLink would like to understand is whether Dr. Brawley’s remarks reflect the official opinion of the American Cancer Society — which he presumably represents, as its Chief Medical Officer — or his personal opinions, which are most certainly at least “colored” by his views on PSA testing. Dr. Brawley is entirely at liberty to express his personal opinions (as are the rest of us). However, once again he seems to have failed to distinguish between those personal opinions and the official position of the American Cancer Society.
Filed under: Living with Prostate Cancer, Management, Treatment Tagged: | Brawley, metastasis, survival
Search for new and
ongoing trials on the
CTAG PCa web site
It seems unworthwhile to speculate about what might have been said or thought that was not reported.
However, once people’s reactions are sorted out, I will be interested to see whether anyone bothers to perform proper matching of apples to apples, quinces to quinces and oranges to oranges.
Apples: Men whose first medical encounter occurred when they already had metastatic-by-1985-standards prostate cancer, and whose eventual cause of death was determined by a single agreed-upon set of criteria.
Quinces: Men whose prostate cancer was diagnosed using 1985 screening criteria before it had reached metastatic-by-1985-standards disease, a subset of whom went on to develop metastasis, and whose eventual cause of death was determined by a single agreed-upon set of criteria.
Oranges: Men whose prostate cancer was diagnosed by 2005 screening criteria before it had reached metastatic-by-1985-standards disease, a subset of whom went on to develop metastasis, and whose eventual cause of death was determined by a single agreed-upon set of criteria.
Fruit salad: Metastatic prostate cancer patients who die (but whose cohort is unknown, whose background population is unknown, whose reason for being biopsied is unknown, whose metastasis-determination ranges across a gamut of different meanings from “M1-by-1985 standards” to “M1-by-2012-standards”, and whose eventual cause of death is determined by an unknown standard).
My father provides a case in point. In 1986, his death certificate listed “prostate cancer” as a cause, even though his proximate cause of death was an infection acquired during non-RP prostate surgery that found prostate cancer as an incidental local finding.
In 2012, I hope, a 73-year-old with no previous history of prostate cancer and no metastasis would not have “prostate cancer” listed as a cause of death. (However, a small voice within me wonders whether this mislabeling still occurs, and maybe preferentially in patients whose families are absent, distant, ignorant, or litigious.)
If, in the quarter-century since then, a seemingly small administrative/labeling change of this kind has occurred in thousands of men (as I believe and hope it has), then this non-medical reason could perhaps explain a large fraction of the “progress in fighting prostate cancer” that has occurred since 1985.
In other words: In 1985, it might be reasonable to expect a large fraction of men in their 60s and 70s with indolent local prostate cancer to have “cause of death: prostate cancer” appear within a few years. In 2010, only a very small fraction of exactly the same men receiving exactly the same care and meeting exactly the same outcome would be assigned the same cause of death.
If, in 1985, many men whose cause of death was then “prostate cancer” would in 2010 have been assigned a cause of death of “hospital-acquired infection” instead, then the unintelligent data-crunching of epidemiology would correctly report fewer and fewer men dying from prostate cancer — even if no medical progress whatsoever had occurred. Also, in this circumstance, if hospitals from 1985 to 2010 got better at controlling infections, then a decrease in deadly hospital-acquired infection over 25 years would be classified as a decrease in prostate cancer deaths. All because of the labeling.
I am not imputing such analysis or opinions to Dr. Brawley, and I do not even know whether the effect of changing the label/meaning relationship is enormous or merely large.
Takeaway message: The problem of relabeling vs label-repurposing is rife, poorly understood, poorly measured, and largely ignored. As a result, we may not really know at all what we think we know and/or what we have the ability to find out. A highly significant correlation of 0.836 between X1 and Y vs a correlation of 0.872 between X2 and Y can be measured with impressive p values by mathematically correct statistics. But it is meaningless if the unquestioned starting assumption that “Y = Y” is false.
I am really surprised by the comment What The “New” Prostate Cancer InfoLink would like to understand is whether Dr. Brawley’s remarks reflect the official opinion of the American Cancer Society … or his personal opinions, and … he seems to have failed to distinguish between those personal opinions and the official position of the American Cancer Society. Surely, Mike, you of all people are fully aware of how lax media reports are in establishing “the truth” as opposed to what might make a good headline?
PaulC’s comments are excellent in my opinion and I believe my personal experience is a case in point. I was diagnosed in 1996 as a result of an elevated PSA, almost certainly due to prostatitis. It is doubtful that I would have been diagnosed as having prostate cancer in terms of the current protocols had they applied at that time, since one of the pathologists gave me a Gleason score of 5. But we can probably assume that the diagnosis would have been one typified as “low risk” or “insignificant” now. In 1996 there were no such distinctions.
Choosing what was then termed Watchful Waiting, part of my watching was to have bone scans at appropriate intervals. After one of these scans, in December 2006, the radiologist identified a lesion as “suspicious for metastasis” and a leading oncologist concluded that this scan in conjunction with a rising PSA was sufficient grounds to commence ADT (androgen deprivation therapy).
So my current survival time exceeds the SWOG 8494 49 month median quite comfortably. But should my results indicate a longer survival period thanks to PSA testing or merely a very early diagnosis of a form of the disease that would not have killed me up to now but which might well do so at some time in the future? Without PSA testing, I might not have been diagnosed until a very late stage and my survival on ADT would then be very much shorter, but as it is, it seems I am a Quince or an Orange and if I was included in a study I might be used to demonstrate longer term ADT survival linked to PSA testing when marked against an Apple.
Terry:
(1) The studies that Dr. Brawley (and I) are referring to are about survival time from diagnosis with M1 prostate cancer. They do not include time from earlier diagnosis, so the extended survival of someone like you is not reflected in the data from these studies.
(2) I think that people in Dr. Brawley’s position are ethically required to be very clear to distinguish between their personal opinions and the formal positions of the organizations that they represent. The ACS has already released formal statements on other prostate cancer-related issues that diverge significantly from Dr. Brawley’s publicly stated views, so this is not a “new” problem.
(3) The need to distinguish clearly between apples and oranges and quinces is most certainly important, and studies started back in the 1980s and early 1990s did not do this as well as we would now like … but then back in the 1980s and early 1990s we did not know what we know now, and it can be hard to raise the money to fund collection of all the data that truly assiduous researchers would like to be able to collect while doing large studies like this. (Dr. Hussain has already stated, publicly, that the National Cancer Institute was unable to provide the funds to collect some of the data initially recommended for S9346 — just as an example.)