A 13-year update from the ERSPC screening trial

So one of the presentations given at the recent EAU meeting in Stockholm was an update, by Dr. Jonas Hugosson, at a median 13 years of follow-up, from the European Randomized Study of Screening for Prostate Cancer (ERSPC)  — and there is some good news, but maybe not enough to get carried away by … yet.

The most recent previous report (at 11 years of follow-up) was published in the New England Journal of Medicine in 2012.

According to the report on the presentation given in Stockholm, now available on the Medscape Oncology web site, the 13-year follow-up data has shown the following:

  • The relative risk for a diagnosis of prostate cancer between the screening and the control groups at 9 years of follow-up was 1.91, at 11 years of follow-up was 1.66, and at a median of 13 years of follow-up was 1.57, but, according to Hugosson, …
  • “[W]e still have a 57 percent increased incidence of prostate cancer in the screening group, compared with the control group.” (In other words, getting screened for prostate cancer increases the risk for diagnosis of clinically insignificant disease and the risk for consequent over-treatment.)
  • The absolute difference in prostate cancer mortality per 1,000 patient-years was 0.31 vs 0.37 at 9 years, 0.35 vs 0.46 at 11 years and is now 0.43 vs 0.54 at 13 years.
  • The difference in the relative risk for mortality was 0.85 at 9 years, 0.78 at 11 years, and is now 0.79 at 13 years; in other words, the relative risk for prostate cancer-specific mortality appears to have stabilized.
  • The number of men who need to be invited for screening to prevent 1 prostate cancer death has decreased from 979 at 9 years to 781 at 13 years.
  • The number of men who need to be diagnosed and treated to prevent 1 death has decreased from 35 at 9 years to 27 at 13 years.

However, Hugosson also noted that there continue to be truly massive differences between the data sets from the different countries involved in this study:

  • In Finland (the country that provided the most patients for the study), there has been a 9 percent reduction in mortality.
  • In Sweden (where all the patients came from a single highly defined area), there has been a 38 percent reduction in mortality.
  • In Spain, there has been a 46 percent reduction in mortality.
  • In Switzerland, almost bizarrely, there has been a 14 percent increase in mortality among the screened patients.

As The “New” Prostate Cancer InfoLink has noted several time previously, it is truly unreasonable to look at the ERSPC as a single trial. It is actually a meta-analysis of data from several concurrent trials in several countries, and each of those countries used different screening protocols (e.g., frequency of PSA tests, rates of biopsies, median PSA levels required to trigger biopsies in invited patients, and duration of screening). In addition, the characteristics of the actual population being screened varied from country to country too. The men screened in Sweden, for example, were all from the Göteborg area, and almost none of these men had ever been given a PSA test prior to the initiation of this study in that area — so they were a truly unscreened population, and we know, from prior experience in the USA and elsewhere, that one is more likely to find prostate cancer in such a population.

What are we to really make of these data?

There is absolutely no doubt that, among the Swedish patients from the Göteborg area, there was a very real benefit from initial screening in a truly unscreened population. It is reasonable to conclude, based on these data and on early data from the use of the PSA test here in America, that initial use of PSA testing in such truly unscreened populations can lead to a survival benefit. The same may well be true for the Spanish subset of patients in this trial (but data on this patient set is much less rigorous and based on fewer patients to begin with).

Unfortunately, what this trial can never tell us is whether regular, long-term PSA testing of any type (i.e., true “screening” over a period of say 30 years, during a time frame when a man ages from his early 40s to his early 70s) is going to be any more beneficial than (for example) a single baseline PSA test when a man is in his early 40s, with risk-based testing for higher risk men based on that initial PSA value as compared to a repeat PSA at (say) age 50 and again at age 60 for men who appear to be at much lower risk based on their baseline PSA.

It is certainly gratifying to see that the number needed to treat to prevent a single prostate cancer death has fallen from 35 to 27 at 13 years of follow-up in this study. However, that does still mean that 26/27 men (96 percent) being treated as a consequence of this study did not gain any survival benefit from treatment, but they almost certainly all had some degree of complications and side effects of their treatment!

6 Responses


    It is good to see this 13-year update of the whole ERSPC, which is generally consistent with the 12.8-year update from Rotterdam and the previous update from the Swedish group. The trends in incidence and mortality are favorable to screening (should be smart screening with active surveillance as a key option), as we expected.

    Perhaps Dr. Timothy Wilt, MD, who is a former key member of the USPSTF and general practitioner — neither a urologist nor an oncologist, and current members of the USPSTF will reflect on the clear prematurity of the data in the trials at the time they so confidently issued their profoundly flawed recommendation that was essentially against screening. I have some small hope that they may appreciate the trends in incidence and mortality and come to see that they need to revise their USPSTF guideline.

    With modern improvements in therapy that are leading to longer survival, other results in the trials become more prominent, especially the proportion of patients in each arm diagnosed with metastatic disease.

    Thanks for providing this important report!

  2. Dear Jim:

    If I was a member of the USPSTF (and thank goodness I am not), what I would tell you is that these data still do not demonstrate (let alone confirm) a value for “screening” (i.e., annual or less frequent, mass, population-based PSA testing of all men aged between 50 and 74). What they do suggest is that there is a very real value to appropriate use of the PSA test to evaluate risk, especially in previously untested men … but that’s not the same as “screening” at all. The precision of what is meant by the term “screening” is a truly critical factor to what is being implied in this debate.

  3. I fully agree with your response.

    I used the term “smart screening” and mentioned active surveillance as shorthand for what you said. However your statement that, “What they do suggest is that there is a very real value to appropriate use of the PSA test to evaluate risk, especially in previously untested men …” is sharply at odds with statements by Dr. Wilt at the conference on prostate cancer in Los Angeles last September, comments that are staunchly opposed to screening (okay, “PSA testing” also works for me), though he will order the test if requested, of course. I have taken very detailed notes from the DVD of his talks and can provide almost verbatim accounts with minute and second cites to the DVD.

    I am going to be at a health and fitness exposition within a month, participating in staffing a booth about prostate cancer. We have materials regarding PSA testing, and they are definitely NOT advocating mass screening. They do prominently highlight active surveillance, as well as noting that a biopsy is at times not the best response to an elevated PSA (such as: “Comprehensive prostate cancer screening involves yearly PSA blood tests and a prostate exam called a DRE. If both PSA and DRE appear troublesome, a biopsy of the prostate gland is the next step …”, with the rest of that particular pamphlet, “PSA Results — What They Mean”, providing key context, including these topics The PSA Test, Who Should Be Tested, The Screening Controversy, PSA Density, PSA Velocity, Percent-Free PSA, PSA Doubling Time, What a High PSA Means, and It’s Not Necessarily Cancer.

    Regarding a PSA timing protocol tailored to circumstances, which I recognize is still undetermined, our materials do provide some guidance, which makes sense to me as the alternative at this time is no guidance, with the assumption likely for many envisioned readers that an annual test is best. Our materials state: “All men should get a first time PSA test at 40 to be repeated every other year. Begin routine yearly testing at 50. Men at higher risk for prostate cancer such as African American men and those with a family history of prostate cancer should begin yearly testing at 45.” That statement was written by Dr. Charles “Snuffy” Myers, MD. (Dr. Myers wants to see that baseline PSA at 40 because he has had a number of patients come to him in their 30s who already had metastatic prostate cancer. It’s rare, but it does happen, and a PSA at 40 will catch many of these men before their cancer spreads widely.) I’m going to have more detailed material for those interested that shows some research support for a one time PSA with an indication of much reduced risk for PSA results of 1.0 or lower (as reported on The “New” Prostate Cancer InfoLink) with the suggestion that a more liberal testing schedule would be reasonable, depending on the person’s view of risk.

    Of course the ideal starting age for a PSA is also undetermined, but the same thinking about giving some advice applies as for timing.

  4. Dear Jim:

    So I have a very distinct problem with the statement that “Comprehensive prostate cancer screening involves yearly PSA blood tests and a prostate exam called a DRE” unless one also points out very clearly that for the vast majority of individual men such comprehensive screening is associated with a far higher likelihood of no survival benefit (96% in the ERSPC at 13 years of follow-up) than any actual benefit.

    In addition, while I have the greatest respect for Snuffy Meyers, I cannot agree with his recommendation about all men getting PSA tests every other year starting at age 40. There are no data to support this — at all, and while one might well pick up very rare cases of prostate cancer among some men in their 40s (particularly among men at theoretically higher risk levels), one is also opening the door to massive over-diagnosis and over-treatment. It is hard enough today to persuade a man in the 70s to go onto active surveillance when he is diagnosed with low- or very low-risk disease. How much harder is it going to be to make that recommendation (however reasonable and well supported by good data) to a man of 45 once he is told he has “cancer”?

    About 2 men in every 100 now die of prostate cancer in the USA. However, about 50 men in every 100 in their 50s will be found to have some prostate cancer if one looks hard enough. Where is the obligation to actually protect these 48% of American men from risk of unnecessary diagnosis and the side effects involved, associated unnecessary testing, not to mention unnecessary treatment? What happened to “first do no harm”? A prostate biopsy is a much higher-risk procedure than most people are willing to acknowledge.

  5. You should note that the ratio of 27 to 1 is really not the ratio of men treated with surgery and radiation to the lives saved, but rather, it is the ratio of extra men diagnosed to the extra lives saved. A much higher percentage of the screening group, compared to the control group, received no treatment except for some form of watchful waiting/active surveillance. Therefore, it is not the case that 26 out of 27 men received treatment that would lead to a risk of serious side effects. If one looks at ratios of additional men receiving surgery or radiation to the lives saved, one would get much lower ratios. The latest data should be looked at in detail to see what that lower ratio is.

  6. Tim is correct … Mea culpa.

    In the 2012 article in the NEJM, at 11 years of follow-up, the number needed to be diagnosed (NND) to save a life was 37. In the current study it appears to have been 27, so it has fallen.

    Table 9A in the Appendix to the 2012 NEJM paper provides a breakdown of the numbers of patients in each group who received various types of treatment. According to that table, 2,984 (35.8%) and 1,549 (18.6%) of men in the screening arm (which totaled 8,345 men altogether) received surgery and radiation therapy. By comparison, in the non-screening arm, 1,813/6,621 (27.4%) and 908/6,621 (13.7%) of men received surgery and radiation therapy, respectively. A better estimate of the numbers of men who received treatment that had no survival benefit in the screening arm is therefore something like 15 out of each 27 diagnosed in the screening arm, implying that about 15 patients who received treatment would have suffered from the side effects of treatment without any evidence of a survival benefit. To my thinking that is still not a very good ratio.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: