No survival benefit to screening in Spanish arm of ERSPC trial

A newly published article has concluded that, among the 4,000+ men enrolled in the Spanish cohort of the European Randomized Study of Screening for Prostate Cancer (ERSPC), there were no differences in overall or prostate cancer-specific mortality at 15 years of follow-up.

As a reminder, the ERSCP trial — data from which was first published 3 years ago in the New England Journal of Medicine — accumulated data from several different countries, and each country used slightly different screening principles. To date, only one well-defined subset of all the patients included in the ERSPC study has clearly suggested a prostate cancer-specific mortality benefit from screening (the patients in the Göteborg group of the Swedish cohort). It has previously been shown that there was no overall or prostate cancer-specific benefit in the Rotterdam cohort from the Netherlands, and even the patients in the Göteborg group showed no overall survival benefit.

According to the new article by Luján et al., the Spanish cohort of the ERSPC enrolled and randomized 4,278 men aged between 45 and 70 years of age between 1996 and 1999. If they were in the screening arm of the study, these men were given a PSA test every 4 years and a prostate biopsy when their PSA was ≥ 3 ng/ml. The men in the control arm of the study were given no tests. Relevant data were collected on all enrollees on an annual basis.

Here are the results reported by Luján et al.:

  • 2,416 men were recruited and randomized to the screening arm.
  • 1,862 men were recruited and randomized to the control arm.
  • The average (mean) age of the patients was 57.8 years.
  • The median follow-up was 13.3 years.
  • 427 deaths occurred in total and just nine of these were from prostate cancer.
    • Prostate cancer accounted for 2.1 percent of all deaths (9/427).
    • 9/4,278 men (0.2 percent) died of prostate cancer.
    • 6/2,416 men in the screening group (0.25 percent) died of prostate cancer.
    • 3/1,862 men in the control group (0.16 percent) died of prostate cancer.
  • Most relevant causes of death were malignant tumors (52.9 percent), cardiovascular disease (17.3 percent), and respiratory disorders (8.9 percent).

The authors conclude that the Spanish arm of ERSPC failed to reproduce the long-term results shown in the whole study and that these results suggest limited value of regular, mass screening for risk of prostate cancer screening in Spanish men.

These data are beginning to confirm prior suggestions that only the data from the Göteborg cohort of the ERSPC actually demonstrate a prostate cancer-specific survival benefit from mass screening for risk of prostate cancer.

6 Responses

  1. Is there any information on how the men in the screening arm were dealt with?

    Did some receive radical treatment? Were some allocated to active surveillance?

    Do the results means anything at all without this information? Surely the benefits of screening, if any, depend very heavily on how men were treated when positive biopsy results (for example) were found?

  2. This article is a bit confusing. It states that the study was conducted by giving a PSA test every 4 years. The conclusion is that annual screening would be of limited value. On the face of it, that doesn’t seem to follow. Maybe I’m missing something?

  3. Richard: I have only seen the abstract of this paper. However, all of the European countries were pretty consistent that in this study treatment was appropriately offered to all men diagnosed with prostate cancer. What that treatment was would have been based on the individual characteristics of the patients at the time of diagnosis. My bet is that in Spain at that time it would have included primarily surgery, external beam radiation, brachytherapy, immediate hormonal therapy, and watchful waiting. (Active surveillance would have been unusual in Spain or anywhere else at that time.) If you can get a copy of the full paper, I am sure it would include the relevant information.

    Doug: Mea culpa. I have adjusted the text of the article above appropriately. I had meant to do that before I hit the “publish” button … and then the phone rang! Thanks for pointing this out. :O)

  4. Any information about “contamination” of the control group? For the PLCO this was significant. Also, was the tumor stage of the patients in the control arm indicated. If they were cT1 or cT2 this would represent PSA and DRE for them, otherwise not allowed.

  5. Again … I can’t answer these questions. I haven’t seen the full text of the paper.

  6. This study simply doesn’t have sufficient sample size to detect plausible effects on prostate cancer specific mortality. You can intuitively see this because there are only three prostate cancer-specific deaths in the control group, so there’s not much room to reduce prostate cancer mortality given the sample sizes.

    The overall European study estimated that as of 12 years after the study’s beginning (the closest we can get to this study’s 13 years), we would expect to have prostate cancer mortality of 0.43% in the screened group vs. 0.64% in the control group, that is the reduction in mortality is expected to be 0.21%, or about 2 less deaths per 1000 men. This is very hard to detect with a screening sample size of 2,416 vs. a control arm of 1,862.

    In fact, if you do a “power analysis”, you find that with their sample sizes, the probability of detecting a statistically significant difference (at the 5% level) of this size in prostate cancer mortality would be only 0.1198, or less than 12%. To get a power of 0.80 (a conventional power level, which means a 80% chance of detecting a statistically significant effect at a significance level of 5%), you would need, with equal sample sizes, about 47,000 each in the screening arm and the control arm.

    It is even harder to detect overall mortality effects. If overall baseline mortality is 10%, as it was in this study, it is very hard to tell if prostate cancer drove the probability down by 0.21%, that is to 9.79%. To get a power of 0.80 to detect such an effect, you would need a sample size of 319,000 each in the treatment arms and control arms.

    At the average follow-up of 9 years in the European study, the differential in mortality is less: 0.284% in the screening group vs. 0.355% in the control group, a difference of 0.071%, about one-third the prostate cancer mortality difference at 12 years. To have a power of 0.80 to detect this difference in prostate cancer mortality, you need a sample size of 100,000. The overall European study had close to that, which is why they were able to detect a statistically significant effect. On the other hand, to detect a decline in overall mortality from 10% to 9.929%, at a power of 0.80, would require a sample size of 2.8 million each in the screening and control arms. Good luck on ever getting the funding to conduct such a study!

    In other words, much of this discussion about overall mortality effects of prostate cancer screening is a waste of time with current studies, as they simply are nowhere near large enough to detect such effects, even if we do a meta-analysis that tries to pool all these samples. Furthermore, it is problematic to try to make too much of differentials even in prostate cancer mortality results in small segments of the European study, as you run into problems with sample sizes.

Leave a Reply

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

You are commenting using your 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 )

Google+ photo

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

Connecting to %s


Get every new post delivered to your Inbox.

Join 1,074 other followers