Further comment on the final results of the PIVOT study


Yesterday we suggested that we would have more to say about this study when we had read the full text of the article by Wilt et al. and the associated editorial by Thompson and Tangen. However, in all truth, others have really done this job for us, and there is little point in reinventing the wheel.

We refer the interested reader to a detailed commentary on the Medscape web site. Yes, you do have to register to read this, but the Medscape site is one we recommend for many reasons.

Everyone is going to put their own spin on the results of the PIVOT study. Yes, it was flawed because Americans couldn’t be successfully encouraged to enroll in a randomized trial of surgery vs. observation. The size of the trial was therefore lower than originally intended. However, there are at least two core pieces of information that come out of this trial to confirm the findings of others:

  • For men with low-risk, early stage, localized prostate cancer who are > 65 years of age and have a life expectancy of not more than 15 years, observation (i.e., active monitoring) is now shown to be every bit as effective as (and a great deal safer than) radical prostatectomy.
  • For men with low-risk, early stage, localized prostate cancer who are > 65 years of age and have a life expectancy of more than 15 years, it is arguable that observation (i.e., active monitoring) may be at least as good an option as radical prostatectomy.

The problem is that the PIVOT data are really limited to these two pieces of wisdom. If one is under 60 years of age, the PIVOT study tells you nothing. And if you you are 65-70 years of age and have intermediate or high-risk disease, it really doesn’t offer you much help either.

We are left waiting again … this time for data from the ProtecT study being carried out in the UK, which (hopefully) is large enough and will offer sufficiently compelling data to be able to give some really strong recommendations regarding the customary use of active monitoring as compared to surgery and radiation as first-line management approaches to early stage prostate cancer in men between 50 and 70 years of age.

10 Responses

  1. If I am following the reviews correctly the low-risk group initially had PSA values of less than or equal to 10 ng/ml. Are there any data as to the progression over time of the PSA and the long-term effect on mortality/thermostatic-icy of that progression?

  2. There is another study, the Goteberg Study, interim results published two years ago in Lancet. I think it is more methodologically robust than the PIVOT trial. It showed that the differences in cause-related survival don’t show up until well into the second decade, when the Kaplan-Meier plots begin to diverge. It showed a benefit to screening: only 12 men needed to be diagnosed via screening to prevent one PC death, and after 14 years, PC-related mortality was reduced by 44%. You can read about it here:

    Everyone agrees that there is no point to screening, let alone treatment, with a life expectancy of 10 years. The ongoing Goteberg study has robust sample sizes (n>7500 in the screened group)and includes men as young as 50 years of age. The USPSTF ought to have given it more weight, IMHO.

  3. Dear Allen:

    (1) The Goteborg study was a study of screening vs. not screening. It was not a study of the comparative efffectiveness of two types of management. As a consequence, the reason that the difference in survival didn’t start to show up until some 15 years after the study started was because they were measuring time from the start of screening, not from time of treatment. In the PIVOT trial they were measuring time from the start of treatment, because men were only enrolled in the trial when they were diagnosed.

    (2) It is absolutely not the case that “everyone” agrees that we shouldn’t be treating (or screening) men with < 10 years of life expectancy. If that was the case we wouldn't be giving PSA tests to 75-year-olds with gay abandon … but we are … millions of them a year … and most of those men are not going to live to 85 … and that will have nothing to do with any risk for prostate cacner.

    (3) There were methodological problems with the Goteborg trial. They weren't as bad as the ones associated with the PLCO and ERSCP trials, but they still existed.

  4. Mike:

    I’m not sure that I understand your question, but yes, the patients’ PSAs were followed over time. However, these data haven’t been analyzed as yet (or if they have these data haven’t been published). Even if they were, I am not sure what you would expect to be discovering because if a patient’s PSA started to rise after surgery or if a patient started to have symptoms of prostate cacner if he never had surgery, he received palliative care (i.e., radiation therapy and/or hormone therapy).

  5. (1) In the Goteberg study, the primary endpoint is cancer-related mortality, which was one of the two endpoints of the PIVOT study. You are right that Goteberg started with screening, while PIVOT picks up men somewhat later in the process — after diagnosis. Because of this, the Goteberg study was able to capture additional information, i.e., that 293 men needed to be screened for 12 men to be diagnosed to prevent 1 cancer death. The point I was trying to make is that 10 years is not enough. Prostate cancer is often a slowly or non-progressing disease, and because we catch often insignificant amounts of prostate cancer so early now, it may take much longer than that to detect the effect of treatment vs non-treatment.

    Of particular interest to many is what these studies say or don’t say about active surveillance (AS). To the extent that “observation” is similar to AS, PIVOT makes the case that AS is a safe alternative for at least 10 years. In the Goteberg study, 40% of the men diagnosed with prostate cancer (12.7% of the screened group) elected AS and 28% remained on it. This gives a sample size of about 460 screened men who elected AS. Their cause-specific survival can be tracked over time and compared to those roughly 560 screened men who chose treatment. Such an analysis of the Goteberg data might give more credible and useful results than the PIVOT study.

    (2) Quite right! I ought to have said that it would not be a reasonable decision to screen or treat men with life expectancy of less than 10 years. That is now part of the NCCN guidelines.

    (3) What were the methodological problems in the Goteberg trial? PIVOT included 20% in the observation group who instead received definitive treatment, and 20% in the treatment group who never received treatment and were only observed. It would be interesting to exclude those (depleting the sample size further) and re-analyze, or alternatively, to retrospectively reassign them (as a purely academic exercise).

  6. I tried to leave comments on the value of this blog but was unable to do so. An excellent source of prostate cancer information. I rate it as a 10 out of 10.

  7. Dear Allen:

    First and foremost, you really can not make the sorts of comparison you are trying to make between the Goteborg study and the PIVOT study. The differences between the two trials are huge and so this sort of comparison simply isn’t scientifically valid. It isn’t even scientifically justifiable to make a direct comparison between the PIVOT study and the Scandinavian trial of surgery vs. watchful waiting (because in the Scandinavian trial the men were inevitably diagnosed with more advanced forms of “localized” disease), but even so, these two trials, “warts and all”, do show an almost identical set of results, with the same recommendation that surgery is inappropriate in men > 65 years of age diagnosed with low-risk, localized disease and a life expectancy of 10 years or less.

    The primary technical problem with the Goteborg trial (as described by Neal in his orginal editorial at the time of publication of the data; click here for discussion) is the fact that the population was completely screening naive at the start of the trial. In other words, one was dealing with a population of men who had never previously been tested for prostate cancer, and which therefore had an inherently higher risk for more more advanced disease that anything that existed in America even by the time we started enrolling men in the PIVOT trial.

  8. I am focusing on the area you are addressing. The outcome of the original grouping has been released at least as an overview. I am wondering what the outcome would have been with a re-staging at 2 years out. I would expect to find a difference in the mortality rate between those who quickly moved on to treatment and those moved slowly. That might not show in the overall study.

    For most, surveillance is kicking the can down the road. Speaking in extremes for clarity … if a significant number of patients left the surveillance pool at 6 months, that is a vastly different outcome than if most were still in the surveillance group at 6 years.

    There is a certain amount of steerage by physicians as each patient is advised to have a specific treatment, surgery, radiation or surveillance. Even if there is a hard to quantify long-term difference in the results of type of treatment in mass, that is probably not true for each individual.

    Part of the art of medicine, knowing who is best served by surgery, who by radiation, and who should choose surveillance takes place before any study of results begins. I believe it takes place most commonly as a patient is sent to a specialist by a generalist. Which specialist receives the referral will often determine the treatment. Real World: I have to believe that most experienced physicians know a high-risk patient when they see one, even if they could not tell you precisely why.

    Real World: the surgery group probably starts with a pool that is truly of higher risk than the surveillance group. Even a study that tries to review and adjust risk factors to compare apples to apples probably does not overcome the prior real world screening.

    There is within the surveillance group a certain amount of “favorable bounce” in the results simply from being studied and closely monitored. The durability of the surveillance group may capture this factor is some rough sense.

    These factors are something I consider when I look at a study. Would I be able to definitively quantify these factors or expect a study to quantify these factors? No.

  9. Dear Mike:

    Real world … The problem is not the high-risk patients, it is the vast number of low-risk men who are treated as if they might be high-risk. The physicans clearly know who the high-risk ones are (Gleason 8 and/or PSA > 20 ng/ml and/or clinical stage T3). The problem is that they treat low-risk men the same way. Active surveillance is a form of individualized clinical study when properly conducted. If you look at the Klotz data (as well as those from other large active surveillance cohorts), you will see that the majority of men who start active surveillance have still not had treatment up to a dozen years later. This is by no means just “kicking the can down the road”. There is still a problem, however. If you define the criteria for active surveillance as being so low-risk as to be “effectively near to no risk at all”, then you will still end up treating an awful lot of men who actually would never need treatment. That is the big difference between the Klotz cohort (which includes a significant number of relatively young men and men with Gleason 7 disease) and the Johns Hopkins cohort (which is restricted to men of at least 60 years of age with Gleason 6, low PSA density, and other risk modifying factors).

    The basic problem is a cultural one … When we hear the word “cancer” most of us still assume it is a deadly disease. However, (as diagnosed today) most low-risk prostate cancers are not ever going to kill most of the men diagnosed with this condition … and especially not most of the men who are diagnosed at age 65 or older. A small subset of those men might need some hormone therapy near the end of their natural life, but hormone therapy for a couple of years when you are in your mid-80s sounds like a better bet to me that 15 years of urinary dribble and/or erectile dysfunction.

    I shall be 65 in 2 years time. I have no intention of getting any form of immediate treatment for low- or even intermediate-risk prostate cancer now. High-risk disease is a very different issue, but the probabilities of a diagnosis of high-risk prostate cancer are relatively low after age 65 if your PSA is still low at that age.

  10. Sitemaster,

    I agree with you. When I say kick the can down the road, I mean that in an entirely positive manner. From the data I have seen, prostate cancer is not a “we must do something quickly disease” (except for a very few). Were that true, PSA screening and cultural changes would have produced a dramatic improvement in outcomes.

    I will kick that can as far as I can and enjoy every day until I reach it.

    My comments were to the study. We have been told the beginning; we have been told the end. Without the middle, the story may be more incomplete than it appears today.

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 )

Google+ photo

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

Connecting to %s