DR. HOFFMAN: Prostate-specific antigen (PSA) testing was introduced in the 1980s as a tumor marker for cancer surveillance. However, it was rapidly and widely adopted for prostate cancer screening. From the beginning, PSA screening has been controversial. Based largely on diagnostic performance data, The American Urological Association (AUA)1 and the American Cancer Society (ACS)2 began recommending routine PSA screening and digital rectal examination (DRE) in the early 1990s. In contrast, the US Preventive Services Task Force (USPSTF) did not endorse screening because they believed that the available evidence was insufficient to either recommend or oppose screening. Large randomized screening trials were initiated in Europe (the European Randomized Study of Screening for Prostate Cancer [ERSPC]3) and in America (the Prostate, Lung, Colorectal, and Ovarian [PLCO] Cancer Screening Trial4).
The introduction of PSA screening was associated with a decreased incidence of advanced stage disease and a slight decrease in prostate cancer mortality. However, screening often detected indolent localized cancers that did not require treatment. Observational studies also suggested that treating localized cancers could adversely affect urinary, bowel, and sexual function.5 Consequently, while awaiting results from the randomized trials, the guidelines began acknowledging the complexity of screening decisions and encouraging clinicians to help patients make informed decisions.
In 2009, the publication of the long-awaited mortality results from 2 large randomized prostate cancer screening trials did not provide sufficient data to resolve controversies. The ERSPC showed a small benefit from screening after 9 years, but the American PLCO Cancer Screening Trial reported only negative findings. Recent updates with longer follow-ups confirmed the original findings.6,7 Revised guidelines have actually heightened the controversies.8 The ACS and AUA still recommend helping men make informed decisions regarding screening, though the 2009 AUA guideline expanded the age range for screening as well as criteria for biopsy referral.9 Meanwhile, the USPSTF recently recommended against prostate screening under any circumstances.10 Regardless of this, after more than 20 years of use, PSA testing is not going to be suddenly discontinued. Patients and physicians will still face the challenging decisions of whether and how to screen for prostate cancer. We’ve assembled a panel of experts to discuss clinical trial results, guideline controversies, and strategies for supporting informed decision making for prostate cancer screening.
DR. HOFFMAN: I’m Dr. Richard Hoffman, Professor of Medicine at the University of New Mexico and a staff physician at the Albuquerque VA Medical Center. I’ll be moderating this discussion.
DR. PENSON: I am David Penson; I’m Professor of Urologic Surgery at Vanderbilt University. I direct our Center for Surgical Quality and Outcomes Research.
DR. VOLK: I’m Bob Volk. I’m a professor in the Department of General Internal Medicine at the MD Anderson Cancer Center. I’m a decision scientist and a developer of patient decision aids.
DR. WOLF: I’m Andy Wolf. I am an Associate Professor of Medicine at the University of Virginia School of Medicine. I was the first author of the ACS Guideline for the Early Detection of Prostate Cancer that came out in 2010.
DR. HOFFMAN: I want to start briefly by talking about the epidemiology of prostate cancer with regard to PSA screening. We’ve seen a large and persistent increase in cancer incidence rates. Cancer mortality rates initially increased slightly after the introduction of PSA testing followed by a decrease in mortality rates. Dr. Penson, can you comment on this?
DR. PENSON: When we assess prostate cancer mortality rates in the United States, interestingly, we see a spike in the early 1990s, which we have alluded to, that’s consistent with the introduction of PSA testing. We also see a spike for incidence, and as we examine that spike, we see that incidence remains higher than it was before the introduction of PSA testing.
Certainly, that’s what you’d expect with a new screening test, but what’s interesting is that when we follow-up the patients into the late 90s and the early 21st century, we see a decline in prostate cancer mortality when we look at the Surveillance Epidemiology and End Results (SEER) data.11 This gives rise to the question of why we are seeing this decline in mortality.
Proponents of screening will tell you this is proof that screening works and that we’re catching cancers earlier and treating them earlier. However, although it certainly makes those of us who believe that screening is a positive thing feel optimistic, it isn’t conclusive evidence.
The problem is that there are other possible influences that may reduce mortality. The first is the change in the way we treat prostate cancer. There was an increase in the use of hormone therapy in the late 1990s and early 2000s, and some people have said that this may be the cause of the decline in mortality.
Some models have tried to account for changing patterns of treatment and for the potential benefit of hormone therapy. Most of the work has been done by a group called the Cancer Intervention and Surveillance Modeling Network (CISNET), which is a group of biostatisticians and modelers who are funded by the National Cancer Institute (NCI).
I’ve actually participated in that group, and there was a paper in Cancer Causes and Control12 approximately 3 or 4 years ago that is often quoted. We were able to show that even after accounting for these changes in treatment, over time, you’re going to see that PSA screening or prostate cancer screening is responsible for anywhere from 40% to 70% of the observed drop in mortality.
Again, this is a model. It’s not a randomized clinical trial, so it is looked at quite skeptically. That being said, we are definitely seeing a decline in mortality. There is some preliminary evidence showing that this may be due to screening, but it’s not conclusive.13
DR. WOLF: Dr. Penson, interestingly, when you review the European Randomized trial data, which shows a potential reduction in mortality due to screening, it seemed that the benefit was observed approximately 10 years after the onset of screening.14 And yet, the initial decline in prostate cancer mortality seemed to begin only 2 or 3 years following the advent of PSA screening. The critics of screening say that the mortality decline occurred too soon after the advent of PSA for it to be attributed to PSA screening. I’m sure you’ve thought about that, but is that a legitimate criticism of PSA screening?
DR. PENSON: I think it is a legitimate criticism to some degree. If you only observed a decline in the first few years after screening was introduced, which then dissipated, it would be a very valid criticism, but here, you continue to see the decline. A recently published study from Austria observed a decline within 1 to 2 years of prostate cancer screening.15 No one gives that result credit because of exactly what you said: It occurs too early. When you look at these models that Ruth Etzioni and Alex Tsodikov have developed,16 you notice that some of the decline in mortality, but not all of it, is attributed to screening. The former change may be related directly to treatment that was stopped after the first 2 or 3 years as patients were receiving hormonal treatments earlier. I think that’s what accounts for that initial drop, but the fact is, you see a prolonged drop over time.
When I look at that decline in mortality—and again, this is an opinion because it has not been shown in a randomized clinical trial—my opinion is that prostate cancer screening has an effect, but we can’t say that the entire decline in mortality is due to screening. It’s probably due to other reasons as well, and that’s why the models make more sense to me when you say, “Somewhere between 40% and 70% of the decline can be attributed to screening.”
DR. HOFFMAN: Dr. Wolf, that’s actually a good lead in, because clearly, the epidemiologic data are not going to be conclusive with regard to whether screening accounts for a decline in mortality. We awaited publication of the results from the ERSPC and PLCO randomized screening trials for many years. Can you tell us about the ERSPC study design and results and how you have interpreted the findings?
DR. WOLF: The European trial was more of a meta-analysis of multiple trials going on simultaneously because it involved 7 European countries. It started in 1991 and included randomized 162,000 men who were between the ages of 55 and 69 years, to PSA screening or routine care.
However, this trial was being conducted in 7 different European countries, and there were some real differences in the methodologies between the countries. In addition, there were differences in the PSA level cutoffs that prompted further evaluation and also differences in screening intervals. Therefore, it wasn’t a pure randomized trial; it was more of a meta-analysis. Nonetheless, it provided a very large cohort of men who were randomized. They were followed up for a mean period of approximately 11 years at the most recent update, which was published in 2012.
In that study, there was a statistically significant difference in prostate cancer mortality in the men who underwent PSA screening compared with those who received routine care. There was a 21% relative risk reduction in prostate cancer death in the screened group as compared to the unscreened group at 11 years of follow-up. Thus, approximately 1000 men needed to be screened to save 1 life and approximately 37 men needed to be diagnosed with prostate cancer to save 1 life over 11 years. These findings highlight an important limitation of prostate cancer screening: overdiagnosis and overtreatment. Although screening may reduce cancer mortality overall, many men who undergo screening will be diagnosed with and treated for prostate cancer, who never would have developed clinically apparent disease if they had not been screened. In other words, they would have lived a normal life and died of an unrelated cause before their prostate cancer became apparent. This is an issue with many conditions that are screened for, but it is a particularly prominent issue for prostate cancer screening, because of the often long latency between cancer onset and the development of clinical illness.
Of note, the European trial did not show any impact on overall all-cause mortality, although this is often the case in most cancer screening trials. They’re aimed at assessing cancer-specific mortality reduction. Therefore, this study did show a statistically significant decline in prostate cancer-specific mortality.
The European study has been criticized for a number of reasons. Perhaps the most serious criticism is that there was some concern that the men in whom prostate cancer was diagnosed in the screening arm were more likely to receive treatment in university settings than those in whom prostate cancer was diagnosed in the routine care arm or control arm, who may have been more likely to receive care in community hospitals. That difference, if big enough, may have affected the outcome.
Another criticism is that the study is, in fact, a meta-analysis and that the methodologies differed significantly between the countries. In fact, one of the concerns of the USPSTF was that only 2 of the countries—Sweden and the Netherlands—showed a statistically significant mortality reduction that was not seen in the other 5 countries. This raised the question of whether the reduction in mortality was a generalizable finding.
The contamination rate, referring to the number of patients who were in the control arm and who were actually screened, was relatively low; it was only 10% compared to an approximately 40% to 50% contamination rate in the American study, which we will discuss further on.
DR. HOFFMAN: Dr. Penson, the American PLCO Cancer Screening Trial is essentially negative. Can you comment on that study?
DR. PENSON: It’s really nice to follow that great discussion of the ERSPC trial. At the same time that ERSPC was being conducted, there was a similar trial in the United States—the PLCO Cancer Screening Trial—involving prostate, lung, colorectal, and ovarian cancer screening.
In the prostate arm of that trial, the researchers randomized just over 76,000 men for either screening or routine care. This was accomplished at 10 centers from 1993 through 2001, and they originally published their results in The New England Journal of Medicine.17 They have published a longer-term follow-up earlier this year in the Journal of the National Cancer Institute.18 This study was essentially a negative study. With 13 years of follow-up, there was no difference between the intervention arm and the control arm.
However, I don’t view this study as a screening versus no-screening one. This is a study of a little bit of screening versus a lot of screening. In the intervention arm, there were patients undergoing annual screening as would be suggested if a patient decided to undergo screening via various guidelines. However, there is high contamination in the control arm because at 3 years of the study, just over 40% of the men in the control arm underwent a PSA screening test, and by the end of the study, more than half the men—I believe it was 52%—had a PSA screening test. So, it’s not really a fair comparison of screening versus no screening, but the study does provide us with some important information.
Screening is not to be performed for everyone. For example, in an older patient who has more comorbid diseases, screening is probably not worth it because, as was discussed earlier, the life expectancy should be adequate in order to see a benefit (if one exists). You have to live long enough to get that benefit—I would say, at least 10 years.
There was another analysis of PLCO Cancer Screening Trial that was published in The Journal of Clinical Oncology,, with David Crawford as the lead author, that looked at the subgroup of patients in the PLCO Cancer Screening Trial who were younger and healthier.19 In that subgroup, they found prostate cancer screening to have a benefit. We should keep in mind that this study was a subgroup analysis, but I think it starts influencing our decision making because it helps us understand which patients are going to benefit from the screening.