The new Memorial Sloan Kettering Cancer Center’s recommendations for prostate cancer screening provide a model for prostate cancer screening that maximizes the benefits, in terms of reduction in prostate cancer-specific mortality, and minimizes harms to men from overdiagnosis and overtreatment.
Andrew J. Vickers, PhD, Attending Research Methodologist, is a co-author on this publication, in press. Dr. Vickers’ research interests focus on models for risk prediction in medicine. One of his findings is that a single measure of prostate specific antigen (PSA) taken in middle age can predict aggressive prostate cancer up to 30 years following. He has also developed a statistical model to predict the result of prostate biopsy based on a panel of markers, and commercially available as the 4kScore.
Here, Dr. Vickers speaks to UroToday about the rationale, deliberations, and conclusions of the authors of the new MSK recommendations to guide prostate cancer screening.
Q: You and your co-authors state that the initiative to review and revise other, older, recommendations for prostate cancer screening was to address three limitations of previous screening guidelines. The first of these you name is an insufficient evidence base. Could you clarify this?
We originally put our guidelines together as an internal document in 2011. At the time, there were certainly some limitations with existing guidelines. This was partly because the randomized trials on prostate cancer screening came out in 2009 and it often takes a bit of time for scientific findings to sink in. What we typically saw in the guidelines at the time was for men to be screened annually, and that the screening should include a DRE and PSA. Yet there was no evidence supporting annual screening, versus less often, such as every 2 years or 4 years. And there was good evidence, even then, that DRE should not be used as a primary screening test.
If a patient has a low PSA, we should not be conducting a DRE. That was the situation at the time, and as we point out in the article, a lot of guidelines have shifted and are now more congruent with best evidence and, as a result, what we said then is what we are saying now.
Q: A second shortcoming of previous screening guidelines the new MSKCC screening recommendations cite is a “…failure to link screening with treatment.” Could you speak to this by an example of what you cite as a “critical deficit”?
Taking a tube of blood and measuring the PSA doesn’t save a patient’s life, but it also doesn’t cause much harm. How we save someone’s life is by treating a cancer when it is localized and preventing it from spreading. How we cause harm is by treating a cancer that never would have affected someone’s life—by giving a man radiotherapy or surgery even though his cancer would never become apparent in the course of that man’s natural life. We can’t think of screening as merely what do we do with the PSA and who we biopsy. We also have to think of whom we treat. We state clearly in the guidelines: prostate cancer is not an indication for treatment. At the time, the guidelines said nothing about treatment: they would say: “Here is how you find prostate cancer.” The NCCN guidelines now say you can’t see this in isolation: you have to link the two together—diagnosing prostate cancer and how you treat prostate cancer – the key thing being to avoid treatment of low risk disease.
Q: Lack of risk stratification, I understand, is a shortcoming of other, or older guidelines. How do the 2016 MSK screening recommendations address this shortcoming?
Lack of risk stratification characterizes both older and even existing guidelines. [According to] guidelines then, and now in many cases, when a man goes in for a PSA test and is told that his PSA is normal, his doctor says, “I’ll see you next year for a repeat test.” What our guidelines say is that, depending on what the PSA is, [clinicians] can tell the patient to come back in 2 years or 4 years or 8 years. That is risk stratification.
There is an important issue in terms of compliance. If I told a man, “You are fine: see me next year,” he may or may not come back. But if I say to him: “It’s nothing to panic about, but you’re a bit high risk, so I really need to see you back in the next couple of years”, he is much likelier to come back. We have shown that nearly half of deaths [from prostate cancer] before the age of 75 occur in men in the top 10% of PSA levels.”
Q: How have you converted that into a recommendation?
We have said that men need to be informed about their risk and that screening should focus on those men at higher risk. You are better off spending your time chasing the man with a PSA of 2.5 ng / mL two years ago and who has not returned, than in getting more blood from a man who had a PSA of 0.5 ng / mL 12 months ago.
MSKCC Risk-Stratified Approach to Early Detection of Localized Prostate CancerPSA test from age 45-70 with possible biopsy for PSA ≥3 ng/mL
PSA test from age 45-70 with possible biopsy for PSA ≥3 ng/mL
Regular screening for PSA ≥1 ng/mL and < 3 ng/mL
Infrequent screening ending at age 60 for PSAs <1 ng/mL
Q: You commented that most men can stop screening at the age of 60. Can you expand on this as a rule of thumb?
Prostate cancer is a very slow-growing disease. If we haven’t seen evidence of it by the time a man is 60, it is not likely to affect him during the course of his natural life.
What we actually showed in a paper in the British Medical Journal2 was exactly that. If a man of 60 has a low PSA, below 1 ng / mL, it is very, very unlikely that PSA testing will save his life. If we continue PSA screening, there is pretty good chance we will find a cancer, and if a cancer is found, it’s likely to be treated. So, a man with a low PSA at the age of 60 gets no benefit from continuing screening for prostate cancer, but can be harmed by over diagnosis and over treatment.
A man with a low PSA at age 60 (<1 – 2 ng/mL) gets no benefit from continued screening for prostate cancer, but can be harmed.
Q: In your Methods section, you and your co-authors make the point that you are addressing “men in good health”, rather than—as other recommendations have phrase it—“men with a projected life expectancy” of, say, 10 years, for example. What was the thinking behind this?
The 10-year life expectancy was never adequately justified. It was not generated by a statistical model showing that if a man’s life expectancy was 10 years or more, then screening is going to help him as opposed to if it is less than 10 years. It was always just a rough rule of thumb—a guess. In the screening trials, there is some evidence of a survival benefit at 8 to 10 years. The problem is that life expectancy at 10 years means that about half of people are going to die before 10 years. So, for prostate cancer screening, that would mean that only about half of men would have any chance of benefit.
So we are more comfortable saying, “Let’s just deal in terms of age and the use of clinical judgment.”
Q: In the text of the 2016 MSK Guideline you describe a decision to depart from the use of a systematic method to search and select evidence for analysis, instead focusing on data from three well-known, high-quality studies, complemented by other data. Could you speak to the reasoning behind this decision?
In guidelines, researchers typically ask questions such as “How should we treat a particular kind of heart problem?” The way to approach this would be to look at all the randomized trials investigating the different treatments for this heart problem and conduct a systematic review of them. The systematic review might show that heparin is better than aspirin, so we are going to recommend that a patient be given heparin rather than aspirin. That makes sense.
How could we do this for prostate cancer screening? We don’t do trials in which, for example, half of patients get screened every 2 years and half get screened every 4 years. We would have to look at the literature. If we do a literature search on whether DRE helps in the diagnosis of prostate cancer, we will end up with thousands of papers, a morass of data with no obvious way of sifting through it. When the USPSTF tried to do this, it came up with nonsense. You are better off restricting a review to a few very good studies.
We also made some pretty uncontroversial statements such as that there is a lot of over-diagnosis. We cite the data for that from outside the three key studies. The question of whether PSA screening works and what should be the intervals for it we based on three high-quality studies. This was a scientific judgment on the best way of getting evidence on how one should screen and what the thresholds should be—from a limited number of high-quality studies.
Q: In the larger sense, the new recommendations, I believe, aim to retain the mortality benefits of aggressive screening, and reduce the harms of over-detection and overtreatment. Are you satisfied that the recommendations could bring about this outcome?
If these guidelines are followed, our best evidence suggests that we would reduce over-diagnosis and overtreatment of prostate cancer and retain most of the mortality benefits of screening.
In the past 10 years the urology community has quite clearly failed, and failed badly, to change practice appropriately. We are still treating men with low-risk disease, even those well into their 70s. We are still biopsying men in the 70s and 80s with moderately elevated PSAs. We’ve failed as a community to communicate to internists and primary care physicians that they should not be doing PSAs on older men.
Urologists have to speak to the primary care physicians and say “You have to stop doing PSAs for men in their 70s and 80s.” They need to be putting 80% or 90% of their low-risk patients on active surveillance.
Here is something that happened to me recently, and it is not an unusual story. A patient presents at MSKCC who is 78, and has morbid obesity and emphysema. He had one microfocus of Gleason 3 disease and was about to go under the knife by a urologist in New Jersey. The family intervened and brought the patient to Sloan Kettering.
This is someone we would not treat: there is no chance that this person will suffer any morbidity from prostate cancer, let alone die from it. This is our routine day-to-day experience at MSK, and it is reflected in national figures.
Q: Has clinical practice not caught up with consensus guidelines or guidelines not caught up with contemporary understanding of the disease’s patterns of progression?
In terms of PSAs, it is not the urologists who are taking the PSAs; it’s the primary care and internists. Our best guess as to what’s going on is that patients are being seen for a heart problem or for their cholesterol to be measured, and [the doctors] run a panel of tests, and while they are at it, run a PSA. Urologists are getting these referrals and they are not proactively going back to the internists and primary care clinicians and saying that they should not be drawing PSAs on these patients.
1. Andrew J. Vickers AJ, James A. Eastham JA, Peter T. Scardino, Hans Lilja. The Memorial Sloan Kettering Cancer Center Recommendations for Prostate Cancer Screening. Urology (2016), http://dx.doi:10.10.1016/j.urology.2015.12.054. Read Abstract
2. Carlsson S, Assel M, Sjoberg D, Ulmert D, Hugosson J, Lilja H, Vickers A. Influence of blood prostate specific antigen levels at age 60 on benefits and harms of prostate cancer screening: population based cohort study. BMJ. 2014 Mar 28;348:g2296.
Written By: Barbara Jones for UroToday