Re-analysis of data from the PCLO screening study — what is truth?

A re-analysis of data from the prostate cancer screening arm of the Prostate, Lung, Colon, and Ovarian  (PLCO) cancer screening study has now stated that, “Selective use of PSA screening for men in good health appears to reduce the risk of [prostate cancer-specific mortality] with minimal overtreatment.”

The 76,693 men who participated in this study between 1993 and 2001, did a lot more than just turn up occasionally at one of 10 centers for screening tests. For example, 73,378 of them (96 percent) filled out a questionnaire that asked them about other health conditions (comorbidities) and whether they had had PSA testing before random assignment to PSA testing or non-testing in this trial.

Crawford et al. have now combed through all these data to see if patients with at least one significant comorbidity who were randomized to the screening arm in the trial had a lower probability of prostate cancer-specific mortality [PCSM] than men with no or minimal comorbidity (with suitable adjustments for age and pre-study PSA testing).

The authors report the following results:

  • 9,565 deaths occurred in the screening arm of the trial at 10 years of follow-up.
  • 164/9,565 deaths were directly attributable to prostate cancer.
  • 22 prostate cancer-specific deaths occurred in men with no or minimal comorbidity randomly asigned to intervention versus usual care.
  • 38 prostate cancer-specific deaths occurred in men with one significant comorbidity randomly assigned to intervention versus usual care.
  • The adjusted hazard ratio [AHR] was 0.56, favoring men who had no or minimal comorbidity.
  • In this group of patients, the additional number needed to treat to prevent one prostate cancer-specific death at 10 years was only 5.
  • Among men with at least one significant comorbidity, those randomly assigned to intervention versus usual care did not have a decreased risk of PCSM (62 v 42 deaths; AHR = 1.43).

Now this type of post-study statistical analysis is interesting intellectually, and it may provide ideas for additional studies, but it can not be considered to provide conclusive proof of the idea that selective screening of men in good health does reduce risk for PCSM. Indeed, the authors are extremely careful to use the words “appears to reduce the risk of PCSM” in the conclusion to their abstract.

Quoted in a Reuters report on this newly published paper, one of the study authors (Dr. Anthony D’Amico) says that: “The reason why the original study appeared to be negative is it looked at all patients. What this study suggests is that PSA tests could be used more selectively in men who are in good health,” and that, “For men who aren’t as healthy, it suggests that screening may not make a difference.” (Notice the careful use of words like “appeared,” “could,” and “may not” in this quote.)

In contrast, Ned Calonge, the chairman of the U.S. Preventive Services Task Force has noted that in this study the men who were less healthy and who were screened for prostate cancer seem to have had a higher risk of dying from prostate cancer than the less healthy men who weren’t screened, and it’s hard to explain that particular finding on any logical grounds.

Exactly what the PCLO trial continues to tell us at this time is that it was unable to show a definitive result favoring PSA-based screening of all men of 55 to 75 years of age for prostate cancer according to the trial protocol at 10 years of follow-up. All the analysis in the world isn’t going to be able to change that result. It can only help us to refine a potential hypothesis.

4 Responses

  1. The “take-away” from this study and your comments seems to be that such studies only refine or sharpen questions about how useful prostate cancer screening is, but this quesition has thus far not been resolved.

  2. Rabbi Ed: So long as you recognize that I am only referring to true mass, population-based screening (as opposed to risk- or worry-based early detection), I agree. As yet I see no compelling data that favors true screening. But I cleave to other statements that I have made that anyone who wants to go get tested (for whatever reason) is entirely entitled to do so. It’s not an all or nothing situation.


    I have a lot of respect for Drs. D’Amico and Crawford, but I am still amazed that people are taking the PLCO results at all seriously at this most premature point in follow-up, not to mention other glaring faults. Why can’t everyone see this?

    Here’s a key fact, using the the “128th Library Edition of the Statistical Abstract of the United States, 2009” , as the source. Table 173, page 118, is entitled “Cancer — Estimated New Cases, 2008, and Survival Rates: 1987-1989 to 1996-2004”. It notes, based on SEER data, that 5-year relative disease-specific survival rates for prostate cancer are 99.4% for White men for the period 1996-2004 and 95.9% for Black men. Now it is imperative to keep in mind that participants in a prostate cancer screening study do not have the disease on Day 1 of the study; it is also reasonable to assume that it takes several years to develop. That means that the average disease-free follow-up period for the PLCO study had to be at least several years less than the average follow-up in the study. I’m looking at a copy of the paper, and the average follow-up was 7 years. Therefore it seems highly likely that the average follow-up for men who were diagnosed was under 5 years, in other words under the time span for which disease-specific survival is close to 100%.

    My statistical skills are rusty, but I’m confident in saying that it would take an extremely large number of study participants — surely much larger than the number of participants in this study — to eke out statistically significant results based on differences in survival when virtually all patients are surviving longer than the period of meaningful follow-up. In addition, there are other serious flaws (such as contamination — those being screened who were not supposed to, and vice versa) in the PLCO study.

    As for men with significant co-morbidity, their overall survival would naturally be shorter, due to death from causes other than prostate cancer, than the overall survival of healthier men in the study, resulting in a smaller number available for analysis of prostate cancer mortality. This would further confound the problem of finding significant survival differences. How can you possibly estimate a sound and reasonable need to treat number in view of such profound flaws?

    It gives me pause when thoughtful men like Drs. D’Amico and Crawford think they have found something that looks like nonsense to me, and I would appreciate help from anyone who can resolve the concerns I (and others who are more qualified and authoritative) have raised. (That parallel European study also has very serious flaws, including being premature, though the follow-up was slightly longer.)

    I am convinced that screening, when coupled with active surveillance for low-risk men, is a sound and wise approach. I base that on other evidence and analysis, but recognize there is no adequate clinical trial evidence bearing on the issue.

    I still see that Emperor as buck naked!


    Here are some thoughts about Dr. Calonge’s statement in this article about the paper. The article noted:

    “In contrast, Ned Calonge, the chairman of the U.S. Preventive Services Task Force has noted that in this study the men who were less healthy and who were screened for prostate cancer seem to have had a higher risk of dying from prostate cancer than the less healthy men who weren’t screened, and it’s hard to explain that particular finding on any logical grounds.”

    Here’s my shot at an explanation. In essence, it’s plausible that less-healthy men in the screening arm of the study showed higher prostate cancer mortality than less-healthy men in the non-screened arm because the PLCO study was long enough to capture deaths related to side effects or inadequacy of treatment but not long enough to capture benefits of treatment.

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 )

Google photo

You are commenting using your Google 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: