Whatever we may think individually about the recent decision by the U.S. Preventive Services Task Force to recommend against the use of PSA testing as a means of “screening” for risk of prostate cancer, we would be wise to understand that at the heart of this decision is poor science. And that poor science is not poor interpretation of the data available to the task force, it is poor science in the design and conduct of almost every trial carried out with the intent to demonstrate or not demonstrate that PSA-based screening actually either saves lives or improves quality of life of men with prostate cancer.
From the very beginning of attempts to validate the idea that the PSA test might be able to help us know whether early diagnosis would properly lead to early treatment of appropriate patients, there have been inbuilt personal and structural biases that have led to poor clinical trial design and poor clinical trial execution. And this is not some sort of important “new” finding that has been unknown to all those involved.
The initial studies designed to investigate the possibility that we could use the PSA test to screen men of more than 50 years of age for prostate cancer were funded by industry and led by a physician who was already convinced that the answer was “yes.” The data from the study reported by Catalona et al. in the New England Journal of Medicine in 1991 showed something that no one argues with … that if you use a PSA test as well as a DRE and ultrasonography you can increase the likelihood of being able to find men who have pathologically identifiable cancer in their prostates. This is not — and never has been — the critical issue. The issue is whether that finding is clinically meaningful, and if so for which patients.
The data review by Chou et al., which acts as the basis for the recommendation now proposed by the USPSTF, provides a thorough assessment of every significant study that has attempted to demonstrate that screening for prostate cancer with a PSA test (with or without a DRE) is associated with either a prostate cancer-specific or an overall survival benefit. Not one of those studies has ever demonstrated an overall survival benefit. Furthermore, the data provided by the authors in Table 1 of that paper clearly demonstate that not one of the six large studies meets the standards for good evidence-based medicine. This does not, however, mean that they do not provide compelling evidence to “believers” in the value of PSA screening. At least two of them do (the Gõteborg study and the ERSPC study) — but each of these studies was still badly flawed from the point of view of design and implementation.
If these data were ever used in an attempt to have to PSA test approved by the US Food & Drug Administration (FDA) for extending the prostate cancer-specific or overall survival of men with prostate cancer, they would not come close to meeting the necessary standard. But the FDA has never been asked this question. The PSA test was approved by the FDA many years ago for “the measurement of serum PSA in conjunction with digital rectal examination (DRE) as an aid in the detection of prostate cancer in men aged 50 years or older.” And that is exactly what it does. It does not and can not help us to know whether a man needs to be treated unless the PSA level is strongly suggestive of locally advanced or advanced disease.
It has long been unclear to the present writer why the urologic oncology community has been so unwilling for so long to conduct a single, large, randomized study to define — according to the standards of evidence-based medicine — whether PSA testing could, in fact, be proven to extend survival of men diagnosed with prostate cancer. It seems highly likely that if a really well-designed study similar to the Göteborg study — but with improvements — had been initiated in the USA back in the early 1990s (with men carefully stratified by age, by risk factors, etc., with centralized pathology, and with a large enough pool of control patients so that anyone who actually got a PSA test in the control group could have been eliminated from the data set), we would probably know by now — with absolute certainty — that yes, PSA screening could indeed extend the lives of a subset of men diagnosed with at least high-risk prostate cancer at between 50 and 70 years of age. And it may well be that this is the only subset of men in whom we could ever prove such an effect, because the number of men diagnosed with intermediate- and high-risk prostate cancer in their 40s is so small. We have lost the chance to do this study now, and we have lost it because the clinicians who were involved in the early use of the PSA test to screen for prostate cancer were already convinced that the test was compellingly wonderful.
Readers need to understand that the USPSTF is making its recommendation based on the available, prospective, high quality, clinical data required by the fundamental guidelines of evidence-based medicine. They have probably come to a poor decision from the point of view of clinical need. But the fundamental reason for their decision comes back to poor science. The urologic oncology community failed to design and conduct studies that met the highest standards of evidence-based medicine. Many of us who have been around for a while understood just how flawed those studies were from the time they were initiated.
It is easy to pick on the USPSTF if one does not understand exactly what they are charged with doing. They are asked to assess data that meets formal standards for evidence-based medicine and then make a recommendation about whether a particular test or other initiative actually meets those standards in prevention of a specific disorder. We may not like their recommendation. We may not agree with their recommendation. Society may not even have to abide by their recommendation — and it didn’t in the case of the mammography recommendation issued by the USPSTF a while back.
Eliminating access to the PSA test today would be a disaster, but this recommendation from the USPSTF may be a good thing even if it doesn’t get implemented. It may help men and their doctors to think harder about the impications of what they are really doing. It may help to stop thousands of men every year who have an elevation in their PSA at age 75 to not have a biopsy that they don’t really need. It may help thousands of men every year who have a diagnosis of low-risk prostate cance between the ages of 50 and 75 years to decide that active surveillance is wiser than immediate surgery. And it may help thousands of men in their late 70s or 80s to decide that at their age PSA testing is a complete waste of their time because their chances of a prostate cancer-specific death are already near to negligible.
We need to move on. Our focus should be on finding a test that really can discriminate between men who have clinically significant prostate cancer and those who don’t … and when we think we have found such a test, we would be wise to conduct trials that actually prove that it can lead to an overall survival benefit in highly defined groups of men.