A new article published on line late Friday in the Journal of the National Cancer Institute shows “no evidence of a mortality benefit for organized annual screening in the PLCO trial compared with opportunistic screening” for prostate cancer.
This follow-up report, by Andriole et al., on the prostate cancer component of the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial was undertaken to assess the impact of population-based screening (with PSA tests and DREs) on mortality at a maximum follow-up of 13 years after the trial.
As reported in the past, the prostate cancer arm of the PLCO trial included 76,685 men, all aged between 55 and 74 years, who were enrolled at 10 screening centers between November 1993 and July 2001. These men were randomly assigned to the interventional arm (organized screening of annual PSA testing for 6 years and annual DRE for 4 years; 38,340 men) or the control arm (usual care, which sometimes included opportunistic screening; 38,345 men). Screening was completed in October 2006. All cases of incident prostate cancer and deaths from prostate cancer have now been monitored through 13 years of follow-up or up until December 31, 2009.
The updated results of the study are as follows:
- About 92 percent of the participants were followed to 10 years.
- About 57 percent of the participants were followed to 13 years.
- At 13 years of follow-up,
- 4,250/38,340 participants in the intervention had been diagnosed with prostate cancer.
- 3,815/38,345 participants in the control arm had been dioagnosed with prostate cancer.
- The cumulative incidence rate for prostate cancer in the intervention arm was 108.4 per 10,000 person-years.
- The cumulative incidence rate for prostate cancer in the control arm was 97.1 per 10,000 person-years.
- The cumulative mortality rate from prostate cancer in the intervention arm was 3.7 deaths per 10,000 person-years.
- The cumulative mortality rate from prostate cancer in the control arm was 3.4 deaths per 10 000 person-years.
- There was no statistically signifciant difference between the mortality rates in the interventiuon arm and the control arm (relative rate [RR] = 1.09).
Some had hoped that with longer follow-up the PLCO trial might start to show a benefit to mass, population-based screening as compared to “opportunistic” testing of individuals. It is becoming increasingly likely that this will not be the case.
It should be noted that these new, follow-up data, do support the idea that annual, mass, population-based screening for risk of prostate cancer can not be justified by the available data. On the other hand, they do not in any way imply that no one should get PSA tests for risk of prostate cancer. Testing for prostate cancer can still be justified on the basis of individual risk factors, including risk factors defined by baseline PSA levels at age 40 or thereabouts.
Obviously you can find all sorts of other comment on this report all over the recent news media.
Filed under: Diagnosis, Risk | Tagged: mortality, risk, screening, testing |
THE "NEW" PROSTATE CANCER INFOLINK 
Have no access to the full article. I see no comment about screening contamination in the control arm in reading the abstract. I understand these results include patients with the intent to screen that were never screened at all. How these affect results should be explained. After all, this was part of the “good” studies used by the USPSTF to make their recommendation.
BTW, what has happened to the recommendation?
Ralph:
If I am interpreting this correctly, what they are saying is that true screening (everyone gets a PSA test every year) is no better than what happened in the control arm (“normal life” in which some poeople get PSA tests and DREs and others don’t). It is a fact that the control arm was significantly contaminated, so one can’t passible suggest that the true scvreening arm was being compared to an arm in which no one got tested unless they had symptoms of something that might be prostate cancer.
Mike,
The problem is that this study was used as a quality study by the USPSTF to demonstrate that screening is not beneficial in reducing PCa deaths. Now it is used to show how many men were over-treated. This study failed both ways and it doesn’t matter how flawed it is. It has more lives that the Energizer bunny … It keeps coming around.
The PLCO study and USPSTF draft and briefing illustrate the complexity of prostate cancer management and lack of root cause analysis as well as patient outcome considerations from innovation new technology. The recommendation statement by the USPSTF draft statement against PSA screening is critically flawed, and if followed, this will result in increased deaths from prostate cancer.
Screening and management for prostate cancer, from a patient’s perspective is even more complex. Men are receiving conflicting and inaccurate information on the benefit of screening and conflicting deceptive and misinformation for the benefits and risk with each modality.
SEER data show the life-saving benefits of PSA screening.
The 10-year survival rate for prostate cancer improved 33.78% between the 1975-1979 survival rate of 53.5% (before PSA screening) and the 1990-1995 survival rate of 88.28% (PSA screening after 1981 for recurrence).
The 5-year survival rate for prostate cancer improved 30.4% between the 1975-1979 survival rate of 69.2% (before PSA screening) and the 1990-1995 survival rate of 99.6% (PSA screening after 1981 for recurrence).
In my opinion the PLCO article is an excellent example of a poor study that provides no value to patients.
What is needed is a test that can differentiate benign from aggressive prostate cancer, but this doesn’t exist, so why is the task force suggesting that we eliminate the only tumor marker test that is available today?
This makes no logical sense.
Fred
PRE-MATURE PLCO DATA ONCE AGAIN!
Based on the fact that 57% of the men in the study were followed to 13 years, it looks like we have median follow-up of 13 years from the time each man enrolled at a screening center; however, the language describing follow-up is somewhat ambiguous. I’m assuming they mean that 57% of men were followed for 13 years each, or more (and 92% for ten years each, or more). Since we have known for some time from other research that virtually 100% of men with low- and intermediate-risk prostate cancer survive the disease at the 10-year point after diagnosis, and that 95% of even high-risk men survive at 10 years after diagnosis, a study to detect the benefit of screening would need to have median follow-up of a number of years well beyond 10 years after diagnosis in order to get large enough survival numbers so that results would rise above incidental influences. The PLCO study has simply not achieved such follow-up at this point!
Consider this. It is clear that there is a lag between enrollment in a screening-type study like this and diagnosis for those who are diagnosed, so we need to deduct that lag from the follow-up figures to calculate follow-up since diagnosis. We don’t have those figures from the initial 2009 report or from the abstract of the current report. However, the first report of the PLCO results for prostate cancer, the paper published in the NEJM in March 2009, gives us a clue on page 1314 (fifth page under “Mortality”): “At 10 years, the median follow-up time for subjects with prostate cancer was 6.3 years in the screening group and 5.2 years in the control group.” Based on the first page, “10 years” appears to mean total study time from the first patient’s enrollment. If we take 10 years total time minus 6.3 years of follow-up for screening arm men, we get a difference of 3.7 years, and 4.8 years difference for the control group. I’m thinking that none of these figures leads us to follow-up since diagnosis, but they suggest that we can meaningfully adjust total study time downward for those diagnosed. Assuming that is so, then 13 years total study time reported now can be adjusted downward to less than 10 years for those diagnosed (9.3 years for screening arm, 8.2 years for the control arm).
We really need the post-diagnosis follow-up figures, but these figures, which are admittedly shaky, are at least food for thought. It doesn’t take too much digestion to realize that follow-up in the PLCO trial is still much too short to derive any idea of a true impact on survival of assignment to a screening group or to a control group, both of which suffered from compliance with protocol and other flaws, especially in the control arm; it’s clear to me that the results to date are even more premature and shaky for conclusions about the mortality impact of true screening versus non-screening.
Therefore, the conclusion of the study that there was no evidence of a mortality benefit (from organized annual screening versus usual care that included substantial screening) at 13 years of follow-up from start of the study does not add to what we already know: virtually all lower risk patients and the vast majority of even high-risk patients survive prostate cancer beyond the range of follow-up in this study. We drew short straws to be diagnosed with this cancer, but at least we have a cancer with extraordinarily long average survival time and a very high lifetime survival rate.
I’m truly impressed that the research team has been able to sustain this large study to this point. I’m hoping they will be able to get to the 20-year point, at which we would probably see some meaningful results; and, since the cut-off for data was 2 years ago for this report, they are now three quarters of the way toward 20 years. I would like to see substantial numbers of patients reaching 15 years since diagnosis. Consider me as a high-risk patient with extra high-risk features (bPSA 113.6, GS 4 + 3 = 7 Epstein, all cores positive with most 100% cancer, perineural invasion, but CT and BS negative with ProstaScint ambiguous; PSA doubling time unavailable but perhaps in the 3- to 4-month range based on later off-therapy period results). I’m responding well to a third cycle of intermittent triple androgen deprivation therapy (essentially sole therapy) in my 13th year, with my case an illustration of the long survival now achievable for a great many of us higher risk guys. My experience is one of the reasons I’m so aware of the need for truly adequate follow-up in studies related to screening. If I and high-risk friends can do this well, then many other high-risk men with less-challenging cases should also do very well. My experience is also one of the reasons I’m acutely aware that failure to screen can have tragic results; I was a near miss!
I do hope the authors address active surveillance in the discussion section. Based on the study timeframe, very few of the men in the study would have been likely to pursue active surveillance, which we now know should be a preferred strategy for many men with low-risk cases. That strategy seems to me the key to solving the over-treatment problem; the key is not to dispense with screening!
“CONTAMINATION” FIGURES FOR THE PLCO TRIAL
Hi Ralph and Sitemaster. I haven’t seen the complete paper either, but I have a copy of the initial report published in 2009 in the NEJM. It has figures on contamination for the study, and it seems to me they should not have changed as follow-up time increased. I believe the NEJM makes papers available for free that are older than six months or so.
Here’s an addition to my earlier comment.
As Sitemaster has noted previously, the serious flaws in the study that affect statistics hinging on randomization almost certainly have destroyed the prospect of sound inferential statistical analysis of the intent-to-treat results.
However, this study will still give us what should be a valuable look, when mature at around 20 years of follow-up, at mortality for those who got screening (at varying levels of intensity) versus mortality for those who were not screened. Screened men from both the screening and control arms would be lumped together, versus men who were not screened from both arms. While these groups would obviously not have been randomized, precluding use of inferential statistics, we could get some worthwhile comparisons using simple statistics such as percentages and trends.
The lack of controls taint this study no matter the outcome.
Better to spend the research time and money to define which prostate cancers are indolent which are not, in my opinion.
Fred
Hi Jim.
You have addressed the issue of too short a follow-up for the whole cohort at this point in time. I want to address another issue. In this trial, as far as I can tell, one cannot know when after diagnosis both arms of the study were treated. Also, some 380 of the screened patients did not undergo any treatment. The number in the control group not treated was 297. How many of these died of prostate cancer or other causes? What a difference from the Tyrol experiment in which every diagnosed man was treated. Of course, the Tyrol experiment did show a significant benefit to screening and treatment in reducing prostate cancer mortality. Unfortunately, that trial was not considered by the USPSTF …
We can determine which tumors are indolent with a large degree of accuracy. Seriously. Validation study pending (see http://www.Ourlab.net/ov).
Dr.O,
I do not doubt that the present results of biopsies are interpreted better than they were in the past. The question is how are the present biopsy samples representative of the true amount of disease dedifferentiation present in the prostate? I constantly see surgical results upgraded to contain Gleason pattern 4, which has a higher metastatic potential and therefore a higher risk for treatment failure.