Time to rethink the debate on PSA testing

Steven J. Atlas, MD, MPH

For most of us, whether to screen for cancer is a no-brainer. Who wouldn’t want a simple test to prevent cancer or identify it at an earlier, more treatable stage? However, as with many things, the screening decision is more complex than it may appear. For example, the test may not be particularly “simple,” such as undergoing screening colonoscopy. For prostate cancer, even after 30-plus years of using a screening blood test called the prostate specific antigen, or PSA, it still isn’t clear how well it prevents prostate cancer deaths. This has led to conflicting and changing recommendations about whether to screen or not. Earlier this year, I commented on revised recommendations from the US Preventive Services Task Force (USPSTF) — a group that advises primary care physician such as myself about a wide range of topics including cancer screening.

A new look at old studies

The PSA test was used for many years without any clear evidence for or against it, and without the USPSTF providing a specific recommendation. This changed in 2009 after the publication of two large trials, one from the US and the other from Europe. Unfortunately, they showed conflicting results — no benefit in the US trial versus a small benefit in the European one. This led the USPSTF to recommend against PSA screening based on the negative results from the US trial, as well as evidence of potential harms of PSA testing from overdiagnosis and treatment of small, benign-appearing cancers unlikely to spread or lead to death. Earlier this year, the USPSTF proposed changing its recommendations to say that health care providers should now discuss the pros and cons of the test with eligible male patients to help them decide whether to have the test or not, using a process called shared decision making.

If all this wasn’t confusing enough, a new study recently re-examined these two (now old) trials and concluded that they really aren’t as different as we thought. The authors of the new study did new analyses showing that PSA testing decreased prostate cancer deaths similarly in these trials. So, what’s going on, and will the USPSTF again change its recommendation?

In brief, the new study used original data from these two trials and computer modeling to control for important differences. These differences included lower rates of prostate cancer overall and more control patients getting a PSA test done (who shouldn’t have) in the US trial. The European trial performed PSA screening less frequently but called the test result abnormal at a lower value than the US trial. After accounting for all these differences, the new study reported that men in both trials who underwent screening as recommended had lower rates of death due to prostate cancer.

What now?

Though this new study may explain why these two trial results differed, I don’t expect the USPSTF will change its recommendation. Personally, I have long favored shared decision making with my patients, and think the USPSTF finally got it right. Why? First, despite all the fancy modeling in this study, it sought to examine the benefit of the PSA test under ideal circumstances, rather than what actually took place. Moreover, the benefit of PSA testing seems large in relative terms (20-30%), but is small in absolute terms. I’d have to screen over 1,000 men with the PSA test to prevent one prostate cancer death. Moreover, decades of using the test have also uncovered real harms — many men are being diagnosed and treated for prostate cancers that otherwise never would have been detected or caused harm. Finally, prostate cancer therapies for those with serious disease needing treatment have improved a great deal since these trials were done, lowering the value of earlier detection.

I support calls to move on from the overly simplistic debate of PSA test: yes or no. Rather, it is time to figure out how to better inform all eligible men about the benefits and harms of PSA testing. For those who elect to have the test, we need to do a better job figuring out who needs treatment when prostate cancer is found, and how to make sure we avoid harming those without such disease — harm that can be both psychologic in terms of worry and physical in terms of side effects from treatment.


  1. Thomas McPherson

    As a prostate cancer patient, I have a biased opinion.
    My PSA was elevated, sand the MRI showed a suspicious area in my prostate. The biopsy revealed cancer (3+4). I was treated and a followup MRI 6 months later showed nothing suspicious,
    To be sure a biopsy was performed and once more a sample revealed cancer (3*4) in the other side of my prostate. To me this shows that only a biopsy can provide valid results.
    I am being treated at Memorial Sloan Kettering by experts. My last procedure was October 2,2017.

  2. Steve McCarthy

    I think the recommendation of shared decision making between the patient and dictot is poor. Why would the opinion of one doctor be better then a clear consensus of many? My doctor just tells be that his practice recommends PSA testing without any real justification. Men and their need a clear and solif recommendation whether to test or not to test.

  3. David Griffiths

    I believe that two items are consistently overlooked in the debate on the usefulness of the PSA test. One is that the absolute value of PSA (largely dependent on the size of the prostate) is less important than the year-over-year change. For example, my PSA increased 25% in one year. This was followed by very successful (no incontinence or impotence) treatment at RCOG (Radiotherapy Clinics of Georgia) and PSA below detection limits 13 years later. The other is the large drop in cancer deaths after the initiation of PSA testing. CDC numbers indicate 168.6 per 100K in 1999 and 95.5 in 2014. As a (non-medical) scientist, I believe that, properly used, the PSA test is very valuable.

  4. Andrew Goldstein

    “… prostate cancer therapies for those with serious disease needing treatment have improved a great deal since these trials were done, lowering the value of earlier detection.”

    I disagree. Early detection is always beneficial. The doctor and patient can follow up with better non-invasive blood tests like the 4KScore which assesses PCa aggressiveness before moving to MRI or biopsy.

    • Bob

      I wish my PCa was detected earlier before it had escaped the gland. I have had psa tests annually since age 40 but was not aware that finasteride , which I was taking for BPH, lowered test results by half. At age 68 when my psa suddenly doubled from 2.7 to 5.4 , I had a biopsy which found gleason 9 PCa. and I was staged pt1c. However, after radical prostatectomy , my pathology came back as stage pt3b! Since then I’ve had ADT and RT three times due to recurrence.

  5. Chris O'Neill

    a new study recently re-examined these two (now old) trials and concluded that they really aren’t as different as we thought

    Unfortunately such mathematical models of mean lead times largely ignore overdiagnosis bias: http://www.nature.com/bjc/journal/v109/n7/full/bjc2013427a.html?foxtrotcallback=true

    Rather, it is time to figure out how to better inform all eligible men about the benefits and harms of PSA testing.

    Unfortunately, little progress will be made on this while most doctors believe it is more important to PSA screen men without fully informing them than to fully inform them: https://health.usnews.com/health-care/articles/2017-04-11/many-docs-dont-discuss-prostate-cancer-screening-pros-and-cons

Post a Comment to Thomas McPherson Click here to cancel reply.

This blog aims to provide reliable information as well as healthy dialog about the topics covered. We reserve the right to delete comments for any reason, particularly those that do not relate directly to the contents of this post, are commercial in nature, contain objectionable or inappropriate material, or otherwise violate our Privacy Policy. Promotional URLs will be removed from comments. Comments on this blog do not represent the views of our editors or Harvard University, and have not been checked for accuracy. All comments submitted to this site become the non-exclusive property of Harvard University.