Prostate Cancer: To Screen or Not to Screen?
Prostate Cancer: To Screen or Not to Screen?
In a lively session featured at the 32nd Annual Meeting of the American
Society of Clinical Oncology (ASCO), Jerome P. Richie, MD, Brigham and
Women's Hospital, Boston, and Steven H. Woolf, MD, MPH, Medical College
of Virginia, Fairfax, debated the merits of screening for prostate cancer.
In the January issue, E. David Crawford, MD, University
of Colorado Health Sciences Center, Denver, framed the debate by exploring
the issues surrounding the screening and early detection of prostate cancer.
This month, Dr. Richie presents data supporting the value of screening.
In a subsequent issue, Dr. Woolf will argue against routine screening.
Many different requirements must be met to justify the use of screening
for any disease, Dr. Richie said. "First of all, you have to show
that early treatment is more effective than later treatment. And I think
with prostate cancer we can show that."
Also, it must be shown that the disease is serious from a health standpoint.
The lifetime risk of a man age 50 developing prostate cancer is about 10%,
Dr. Richie noted, and his risk of dying from prostate cancer is about 3%.
Given these "sobering statistics," prostate cancer does qualify
as a serious health problem.
Other requirements for screening must also be met. "You have to
have effective treatments, testing must be safe and inexpensive, and there
should be acceptable sensitivity, specificity, and positive predictive
value," Dr. Richie said.
"There are a variety of biases that are introduced when one screens,
including length-time and lead-time bias, so that you ultimately need to
have a randomized study to prove whether screening will be beneficial or
not," he added. Another important issue is the impact of screening
on patients' quality of life. Most important, however, in the current era
of health-care cost consciousness, is the cost of screening. A critical
question is, how much can we afford to do?
Yet another key issue specific to prostate cancer is its unpredictable
natural history. "We don't really know in an individual patient whether
that cancer will progress and compete as a cause of death. And so we're
lacking the true markers to really know about biologic predictability."
Studies Present Conflicting Views of Natural History
Several papers have dominated the literature on the natural history
of prostate cancer over the last 5 to 10 years, Dr. Richie said. In a study
of 120 patients with prostate cancer published in 1988, George found that
only about 4% of patients died of prostate cancer, whereas about 40% died
of other causes. Similarly, a 1991 paper by Adolffson and colleagues
with a median follow-up of about 8 years showed that only a small percentage
of patients died of prostate cancer.
One of the most widely quoted series in the literature, Dr. Richie noted,
is a study of 223 Swedish prostate cancer patients by Johansson and associates.
This study reported a 5-year cancer-specific survival rate of 94% and a
10-year survival rate of 89%. "The clear implication of this study
was that most of these patients have indolent disease, they don't need
to be treated, and therefore, we are overtreating or trying to diagnose
patients who won't need treatment."
Closer examination of these data reveal, however, that only about 10%
of the patients died of prostate cancer, Dr. Richie noted. The majority
of the patients (148 of 223) had very well-differentiated tumors (grade
1[Swedish system]; approximately equivalent to Gleason score 2-4), whereas
only 66 patients had grade 2 tumors (equivalent to Gleason 5-7) and 9 patients
had grade 3 tumors (equivalent to Gleason 8-10).
Furthermore, in the first 2 years, only patients with very well-differentiated
disease were accrued to the study. "This skews the data and stacks
the odds in favor of patients with very well-behaved disease," Dr.
Richie said. Such patients have low rates of progression and of death due
to cancer. In contrast, most patients detected by screening have grade
2 or 3 disease, which has a higher mortality.
Dr. Richie added that most of the patients in the study by Johansson
et al were elderly. No patient was less than 60 years of age, two-thirds
of the patients were older than 70, and almost 20% were over 80. Prostate
cancer was diagnosed by transurethral resection, and one-third of the patients
had one focus or a very small microscopic focus of well-differentiated
disease. Thus, Dr. Richie asserted, "in any series in any country"
such patients would be followed conservatively. Progression was determined
by rectal or bone scan, and a quarter of the patients were treated with
Many subsequent studies, as well as some of the data analysis by Markov
modeling, rely predominantly on the study by Johansson et al to determine
whether prostate cancers are likely to progress, Dr. Richie noted. "And
in any of these models, if you put in data that are spurious, the answer
that comes out is also spurious."
Another study published in 1994 by Gronberg and colleagues seems
to counter the data of Johansson's group, Dr. Richie said. Gronberg et
al compared close to 7,000 prostate cancer patients in northern Sweden
with age-matched controls. The relative survival of the men with prostate
cancer was about 50% at 10 years, with younger men experiencing an even
greater decrease in survival. "Another way of looking at this is average
loss of life expectancy, and the men with prostate cancer lost an average
of 40% of their expected life," Dr. Richie said.
In 1995, Gann and associates reported on data from the Physician's Health
Study. In this group of 22,000 male physicians who were followed and
whose sera were stored, 2% developed clinical prostate cancer, detected
either on the basis of symptoms or a digital rectal examination. The researchers
then compared each physician with two age-matched controls. This comparison
revealed a very high excess mortality in the study population. Furthermore,
"three quarters of the deaths in men with prostate cancer were caused
directly by prostate cancer. So I think we can't assume that this is an
indolent disease that will not progress," Dr. Richie said.
PSA as a Screening Tool
The beginning of screening for prostate cancer can be dated back to
a 1991 study by Catalona and colleagues at Washington University in St.
Louis, Dr. Richie noted. These researchers performed the then relatively
new prostate-specific antigen (PSA) test in 1,600 men over 50 years of
age. If the patient's PSA was abnormal, a digital rectal examination and
an ultrasound scan were performed, and if either of these was abnormal,
a biopsy was done.
This study showed the importance of PSA as a screening tool. Of the
patients in this series, 6.5% had a PSA between 4 and 10 ng/mL, and about
one-fifth of these men had a positive biopsy for prostate cancer. Approximately
2% of the men had a PSA over 10 ng/mL, and two-thirds of them had prostate
A second large, multicenter study done at Washington University, Brigham
and Women's Hospital, and four other institutions involved over 6,000 asymptomatic
men over age 50. All of these men were screened with both a PSA and
digital rectal examination and underwent an ultrasound scanning and biopsy
if either screening test was abnormal.
In the initial screening, approximately 15% of the men had an abnormal
PSA and a similar percentage had a suspicious digital rectal examination
(Table 1). Of the patients with abnormal
screening tests who were found to have prostate cancer and underwent surgery,
about three-quarters had pathologically organ-confined disease. This is
"extremely discordant" with the rate of organ-confined disease
among men whose prostate cancer was discovered prior to the advent of PSA
screening, Dr. Richie noted. At best, approximately 35% of those men had
An important issue is whether prostate tumors detected by screening
are significant cancers or whether they are indolent cancers that won't
progress. Defining "significant" cancers as moderately or poorly
differentiated tumors or large-volume disease, Catalona et al found that
93% of patients who underwent radical prostatectomy had significant disease.
These data suggest that "we're not detecting the indolent cancers,
but we're detecting cancers that are likely to progress," Dr. Richie