How badly contaminated were the PLCO trial data?

It is well-understood that a problem with the data from the PLCO screening trial was that many men in the “control” (supposedly non-screened) arm of the trial did in fact get PSA testing and have DREs during the time period of the trial (albeit outside the trial protocol and not at the trial centers). But how bad was that “contamination” of the trial data? A new report by Pinsky et al. suggests that it was probably pretty bad.

Pinsky and colleagues set out to develop a detailed description of the degree of “contamination” of data from patients in the “control” arm of the PLCO study. They sent surveys, inquiring about PSA and DRE use, to a sample of 2,427 men (out of a total of 38,350) in the control arm of the study throughout the screening phase of the trial (i.e., in years 0 to 5 of the trial). A statistical probability model was then used to translate survey results into actual frequency counts of tests.

The results of their study are as follows:

  • Increased rates of contamination were  significantly associated with prior (pre-trial) screening and college education.
  • The estimated mean number of screening PSAs (and DREs) in the control arm was 2.7 ± 1.1 ( as compared to 5.0 ± 3.5 in the screened arm).
  • 1,984 and 2,538 prostate cancers were observed in the control arm and the screened arm, respectively, during the screening phase.
  • In the absence of screening, 960 and 949 would have been expected (based on SEER incidence data from the pre-screening era — between 1985 and 1987)
  • If the men had not been participating in this study, the projected contemporary incidence rates would have been 1,630 and 1,611 (based on data from contemporary SEER data) .

Now it is clear that we will never be able to know the exact level of PSA and DRE use by men in the control arm of the PLCO study. However, Pinsky et al. conclude, based on their data and analysis, that:

  • Use of prostate cancer testing by men in the control arm of the PLCO study was substantial.
  • Testing by men in the control arm was still substantially less than that of men in the screened arm of the trial.

Just to be clear, men in the control arm of the study should not have received a PSA test or a DRE at all unless they showed signs or symptoms suggesting the possibility of a prostate problem that might require treatment. However, the data from this trial suggest that the average (mean) number of PSAs received by men in the control arm over a period of 5 years and who were surveyed was in fact nearer to three than two. On the basis of that degree of contamination, it is hardly surprising that the PLCO study was unable to demonstrate a survival benefit after only 7 years of follow-up.

8 Responses

  1. Aloha,

    I am a member of the PLCO control arm; my wife is a member of the screened arm. In 2000 I participated in a screening by the American Cancer Society and was tested by DRE and PSA. The PSA was only slightly elevated and at the time my health care provider (we) chose no action. I had received a yearly DRE for many years prior to discovery of my prostate cancer. To me these actions were a normal part of the control arm of the PLCO. I would think that any action taken by a member of the control arm would be appropriate for a study such as this. As in my own case, and I suspect most all cases of prostate cancer, the cancer is found because a man presents with urinary problems.

    I don’t think that just because a man in the control arm is given a PSA test, that that action contaminates the PLCO testing. Prostate cancer is so different for each man, and presents in such diverse ways as to be very difficult to fit to a yes/no model. The PLCO data is still valid data. The tested/screened arm was looking specifically for prostate cancer in men. A control arm man could be tested for any problem/present which could be any number of problems, including prostate cancer.


  2. Dear Joe:

    Please don’t get the idea that the PLCO data are not “valid.” The question is only “valid for what?”

    The data from the study referred to above would seem to suggest only that the PLCO study as executed (with regard to PSA testing at least) was not the same as the way it was designed, and that this has profound implications for why it did not show any benefit for “screening” as opposed to “non-screening” at 7 years of follow-up.

    With respect to why most men get diagnosed with prostate cancer, the whole idea behind trials of mass, population-based screening is to see if getting men diagnosed before there are any symptoms of prostate cancer (including urinary tract symptoms) can impact survival. Some would argue that the reason you were diagnosed as a consequence of your symptoms is closely associated with your subsequent clinical outcome, which might have been much better if you had been diagnosed earlier.

  3. “Some would argue that the reason you were diagnosed as a consequence of your symptoms is closely associated with your subsequent clinical outcome, which might have been much better if you had been diagnosed earlier.”

    And “Some” are inevitably correct. The policy issue is, do we value the lives of those folks (who are asymptomatic and are not aware of having risk factors) who would ONLY have their cancers diagnosed (early enough to improve their outcomes) through population-based screenings?

    In today’s reality of unrestrained, skyrocketing health care costs and a health care industry that will do ANYTHING to maintain and grow their profit margins, the subtle, cultural arguments tend towards “No,” these relatively few folks who would be saved are not worth the effort and expense. The rationalization gets couched in talk of the inefficiency and cost of over-testing and over-treatment of those who might have to have a needle stick in their arm or latex gloved-finger in their bum, but really these are pretty trivial costs — both financial and to personal dignity and autonomy — compared to the probability of sparing another human being and his family the pain and permanence of the worst possible outcomes; including permanent disabling side effects and/or premature death.

    I believe we’re at a point in the history of public health policy in which we are ethically “upside down” on this issue. Considering that there’s very good epidemiological evidence that folks who are poorer and/or who belong to minority groups are the least likely to have sustained, consistent access to quality health care, including knowledge of family history and tracking of other important risk factors, by foregoing population-based screening protocols we’re subjecting these folks to the proverbial double whammy: they’re less likely to be screened AND less likely to have access to the most effective (with the least side effects) treatments once they’re diagnosed.

    Folks who fret about “death panels” and end-of-life health care rationing are worried about the wrong side of the equation. The business folks who want to cut health care costs, while keeping all their profit, are clever. They’re not going to blatantly deny folks treatment during phases of illness when treatment is predictably effective in prolonging their lives. Instead they’re simply going to redefine what and when a condition is considered “illness” and what and when we consider an illness “treatable.” My guess is we’ll move toward a system where a lot more folks aren’t diagnosed until the only “humane” solution is to move directly to palliative care (much, much cheaper) for diseases that are too cost-intensive or labor-intensive to be profitable to treat.

    Believe me, somewhere this minute there’s an accountant calculating how much it will cut into his company’s profit for your life to be sustained 5, or 10, or 15 years by a chemotherapy or hormone suppression drug. If the market’s not there, if ROI isn’t profitable at 15 years, the trickle-down effect on public policy and clinical practice will be that you won’t be diagnosed until the point that the treatment can be profitably delivered to the market.

  4. For any of us out here who haven’t a clue what the “PLCO screening trial” was, this is a pretty meaningless article. A paragraph explaining what it was would have made all the difference in the world.

  5. Dear John:

    Your comment is entirely justified. Sorry. The PLCO screening trial was a very big deal a year ago. Please click here to see an original commentary.

  6. I was interested in the PLCO trial until I read halfway through the abstract last year, and then I became angry. While the degree of contamination reported in this excellent piece (thank you) was a major problem, a much greater problem was the totally inadequate number of year of average (median) follow-up. Together, and in conjunction with other significant issues, it should have been clear ther that the interim trial results reported were of no value in assessing the worth of screening. I appreciate all the patients who have participated in the two trials; as the years pass and we reach a meaningful point for follow-up results, then we have a shot at some meaningful data, even if somewhat flawed. Perhaps analytical adjustments will be possible to diminish the impact of the flaws.

    Yet we continue to see favorable references to this trial and to the European trial, references questioning the value of screening. Some of the commentators are people who should no better. One is Dr. Otis Brawley, formerly practicing medical oncology with a concentration in prostate cancer in Emory, but currently medical director of the American Cancer Society. Dr. Brawley initially accepted the trial results uncritically; he then seemed to recognize the shortcomings; however, a few months ago, he referred to the trial results in questioning screening in testimony before a House committee. We need to somehow help Dr. Brawley put this all in perspective and change his negative to neutral stance toward screening!

    The other group whose uncritical acceptance surprised me was the National Comprehensive Cancer Network (NCCN) in its updated guideline on prostate cancer. You will find the reference in its set of talking points for doctors to explain the cons (includes the references) and pros of screening for prostate cancer. I’m hoping they will publish an interim correction that eliminates the misleading references. In fact, as I’ve urged before in this forum, we need to just get on with routine screening without preliminary and confusing talks with the undiagnosed patient, provided the screening is coupled with brief preliminary sound advice on active surveillance, and follow-up extended advice for men diagnosed with low- and very low-risk cases of prostate cancer.

  7. Hi Tracy,

    Your post earlier today contains a questionable statement, though one that fits right in with the conventional wisdom. On July 12, 2010 at 9:22 am you wrote, “In today’s reality of unrestrained, skyrocketing health care costs and a health care ….”

    Actually, regarding the disease we are concerned about — prostate cancer, major areas of cost are plunging and some others are about to or are highly likely to plunge — not skyrocket. I’l give you just a few short instances, but these can be fleshed out with supporting data:

    — Active surveillance has emerged as the go-to approach for low- and very low-risk prostate cancer patients, offering what looks like a better-than-even chance of deferring treatment throughout the patient’s life while ensuring timely, effective treatment if the cancer is a stealthy version that is truly aggressive. With about 190,000 newly diagnosed cases in the US each year, and with the clear majority low risk, the opportunities for cost savings through active surveillance of appropriate cases are huge, let alone the benefits to the patients.

    — The American Urological Association recommended last year that previously routine bone and CT scans not be done in most patients, unless there were unusual circumstances, because they so rarely provided worthwhile information. Again, considering the number of newly diagnosed patients, eliminating these formerly routine scans for most will result in enormous savings, as well as avoidance of unnecessary anxiety, effort, discomfort, and expense for patients and their families.

    — The costs of a number of the major drugs for treating prostate cancer and the side effects of treatment have plunged as generic versions have become/are becoming available. Typically, the initial decrease is about two-thirds, with continuing decreases over time. Consider drugs for hormonal blockade and the side effects of blockade. The heaviest duty component, the LHRH agonist class, has often been Lupron or Zoladex, but the patent for Zoladex is expiring this year and a generic equivalent should soon be available, drastically reducing expenses. The medium duty drug is from the antiandrogen class, and the drug of choice for some time has been Casodex. Casodex, an expensive drug, has now been available as generic bicalutamide for well over a year. The third leg of the triple blockade approach is the 5-alpha reductase inhibitor drug, either Proscar or Avodart. Proscar has been available as generic finasteride for some time now, well over a year or two as I recall it. One of the main side effect concerns of blockade is with bone density, and one of the primary bone density drugs, Fosamax, became available as generic alendronate not long ago. Another important potential side effect is on cardiovascular health, and at least one statin drug, Zocor, has been available as generic simvastatin for a long time and is very low in cost.

    — Finally, prevention of prostate cancer or mitigation of its seriousness is emerging based on lifestyle tactics. While these have some costs (e.g., quality pomegranate supplements or juice, gym memberships), they are likely much less expensive than if drugs or treatment were required.

    On the other side, a very expensive drug, Provenge, was recently approved for men with very advanced cases. While the $93,000 or so per patient is costly, the number of patients is relatively low, compared with the reductions mentioned above. My off-the-cuff estimate is that the net effect of pluses and minuses is going to be much lower average cost per patient treated (or potential patients, factoring in prevention.

    While I have focused on prostate cancer, similar cost reduction developments are affecting other areas of health. Indeed, both the bone density drug and the cholesterol drug I mentioned primarily affect other areas, to say nothing of finasteride, which is primarily used for BPH rather than prostate cancer.

    We need to call to account those political leaders and media commentators and reporters who carelessly talk about “skyrocketing” medical costs.

  8. If anyone measures cost, they must look at cost per course of treatment, including the cost of treating side effects. When we look at the cost of chemotherapy, it is necessary to consider how many treatments a person gets, anti-nausea drugs, transfusions, etc. Not easy. How does that compare to the cost of Provenge?

    I hope we never get to the point where profit is more important than the person.

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: