Albertsen PC, Hanley JA, Fine J. 20-Year Outcomes Following Conservative Management of Clinically Localized Prostate Cancer. JAMA. 2005;293(17):2095-2101. doi:10.1001/jama.293.17.2095
Author Affiliations: Division of Urology, University
of Connecticut Health Center, Farmington (Dr Albertsen and Ms Fine); and Department
of Epidemiology and Biostatistics, McGill University, Montreal, Quebec (Dr
Context The appropriate therapy for men with clinically localized prostate cancer
is uncertain. A recent study suggested an increasing prostate cancer mortality
rate for men who are alive more than 15 years following diagnosis.
Objective To estimate 20-year survival based on a competing risk analysis of men
who were diagnosed with clinically localized prostate cancer and treated with
observation or androgen withdrawal therapy alone, stratified by age at diagnosis
and histological findings.
Design, Setting, and Patients A retrospective population-based cohort study using Connecticut Tumor
Registry data supplemented by hospital record and histology review of 767
men aged 55 to 74 years with clinically localized prostate cancer diagnosed
between January 1, 1971, and December 31, 1984. Patients were treated with
either observation or immediate or delayed androgen withdrawal therapy, with
a median observation of 24 years.
Main Outcome Measures Probability of mortality from prostate cancer or other competing medical
conditions, given a patient’s age at diagnosis and tumor grade.
Results The prostate cancer mortality rate was 33 per 1000 person-years during
the first 15 years of follow-up (95% confidence interval [CI], 28-38) and
18 per 1000 person-years after 15 years of follow-up (95% CI, 10-29). The
mortality rates for these 2 follow-up periods were not statistically different,
after adjusting for differences in tumor histology (rate ratio, 1.1; 95% CI,
0.6-1.9). Men with low-grade prostate cancers have a minimal risk of dying
from prostate cancer during 20 years of follow-up (Gleason score of 2-4, 6
deaths per 1000 person-years; 95% CI, 2-11). Men with high-grade prostate
cancers have a high probability of dying from prostate cancer within 10 years
of diagnosis (Gleason score of 8-10, 121 deaths per 1000 person-years; 95%
CI, 90-156). Men with Gleason score of 5 or 6 tumors have an intermediate
risk of prostate cancer death.
Conclusion The annual mortality rate from prostate cancer appears to remain stable
after 15 years from diagnosis, which does not support aggressive treatment
for localized low-grade prostate cancer.
To determine the need for treatment of localized prostate cancer, patients
and physicians must understand the natural history of this disease. A recent
study by Johansson et al1 documented 20-year
outcomes for a population-based cohort of 223 men diagnosed with localized
prostate cancer between 1977 and 1984. The authors noted a substantial increase
in prostate cancer mortality among the 49 men who were alive more than 15
years following diagnosis.
In 1998, we published a competing risk analysis of 767 men aged 55 to
74 years with clinically localized prostate cancer at diagnosis who were treated
with observation or androgen-withdrawal therapy alone.2 The
purpose of that analysis was to provide an estimate of the natural progression
of prostate cancer if treated conservatively. Because these men have been
followed up continuously by the Connecticut Tumor Registry (CTR), we had an
opportunity to extend our follow-up to 20 years to determine whether prostate
cancer mortality rates declined, remained constant, or increased after 15
Patients followed up in this analysis were the same patients described
in our 1998 study.2 The original study population
consisted of 767 men identifiedfrom the CTR database who were Connecticut
residents when diagnosed with prostate cancer between January 1, 1971, and
December 31, 1984. Of these men, 610 died before March 1, 1997, after a median
follow-up of 15.4 years. Since then, 107 patients have died. The year of last
contact for the remaining 50 men was 1987 (n = 1), 1998 (n = 2),
2001 (n = 1), 2002 (n = 2), 2003 (n = 41), 2004
(n = 3). The final censoring date was October 8, 2004.
The original research was approved in 1990 to 1992 with a waiver of
informed consent by the institutional review boards of the Connecticut Department
of Public Health (DPH), as well as 24 acute care hospitals and the 2 Veterans
Affairs medical centers in Connecticut that had institutional review boards
at the time. Administrative approval waiving informed consent was obtained
in the remaining 11 hospitals. This study was approved in 2004 by the Connecticut
DPH Human Investigation Committee and the institutional review board of the
University of Connecticut Health Center.
Charts were abstracted onsite to confirm the date of diagnosis, metastatic
evaluations completed, method of treatment, and any associated comorbidities.
Patients who had undergone surgery, received either radiation therapy or brachytherapy,
or who were known to have metastatic disease were excluded. In addition, patients
with concomitant cancers and those surviving less than 6 months after diagnosis
were also excluded. Study personnel performing chart abstraction were blinded
to the long-term outcome of the patients as recorded by the CTR. Original
histology slides that were used to secure the patients’ diagnoses were
retrieved from hospital pathology departments and mailed to a referee pathologist
who was also blinded to the long-term outcome.2 Standardized
grading was performed using the Gleason classification system. This system
grades prostate cancers into 1 of 5 morphological patterns according to the
tumors’ glandular differentiation and growth pattern as assessed under
low-power magnification, with 1 indicating well-differentiated disease and
5 indicating poorly differentiated disease. The Gleason score represents the
sum of the pattern numbers of the 2 most common patterns by volume. Scores
range from 2 to 10, with a Gleason score of 10 being the most poorly differentiated
and aggressive tumors.3
Accurate staging information was lacking for many men. Bone scan tests
were performed on only 30% of patients and serum acid phosphatase levels were
confirmed as normal in only 53% of patients. The proportion of patients without
evidence of testing for metastatic disease ranged from 33% for men with a
Gleason score of 2 to 4 disease to 15% for men with a Gleason score of 7 and
8 to 10 disease. No information was available concerning prostate-specific
antigen (PSA) levels at diagnosis because this population had prostate cancer
diagnosed before the clinical application of this test. Approximately 71%
of patients were diagnosed with prostate cancer following transurethral resection
or open prostatectomy, 26% of patients were diagnosed by needle biopsy of
the prostate, and 3% of patients were diagnosed by other or unknown methods.
On March 1, 1997, and again on October 8, 2004, the vital status of
each patient was obtained from the CTR, which is located in the Connecticut
DPH. The CTR is the oldest state cancer registry and has functioned as one
of the sites of the National Cancer Institute’s Surveillance, Epidemiology,
and End Results program since 1973. The CTR uses a variety of sources to obtain
follow-up data for registered patients, including hospital tumor registrars
who rely on hospital records and physician and patient contact, as well as
periodic searches of the DPH Vital Records Section files. The CTR database
is linked annually with files of the Center for Medicare & Medicaid Services,
formerly the Health Care Financing Administration, to ascertain deaths of
men enrolled in Medicare. If a man is not known to be dead, a date of last
contact is assigned that corresponds with the date of the last Center for
Medicare & Medicaid Services/Health Care Financing Administration linkage.
The Connecticut Department of Motor Vehicles files are linked annually and
the National Death Index Plus files are linked every 1 or 2 years to obtain
cause of death for patients known to be dead.
For all men who died since the last analysis, information coded from
death certificates was obtained to classify them as having died from prostate
cancer or another cause. A patient was determined to have died from prostate
cancer if any of the 3 causes listed on part 1 of the death certificate reported
prostate cancer. If prostate cancer did not appear on 1 of these 3 lines,
the patient’s death was attributed to competing medical conditions.
For some patients, only information concerning the date of death was available.
Patients who were not followed up until death were considered alive until
the date of last contact and their subsequent survival time was censored.
The primary outcomes of our study were probability of mortality from
prostate cancer or other competing causes, given a patient’s age at
diagnosis and tumor grade. For the competing risk analysis, we tabulated the
numbers of men with each of the 3 outcomes of interest (alive, deceased from
prostate cancer, and deceased from other causes) for each of the 20 age-histology
combinations. Because of the variable length of follow-up and the small numbers
in some cells, we also performed a second competing risk analysis based on
2 inputs: the rate of mortality from prostate cancer and from other causes,
both fitted as smooth functions of age at diagnosis, Gleason score, and year
of follow-up. These smoothed estimates were derived from regression models
and incorporated the duration of follow-up and the patterns of outcomes in
neighboring cells to allow more stable estimates for all cells.
The regression models used to construct the smoothed competing risk
analysis required information concerning both the date of death and the cause
of death for all patients. For 25 (3.5%) of the 717 men who died, only the
date of death was available. We imputed the cause of death for each of these
25 men separately for each histology score category according to the ratio
of the deaths of known causes for the other men with the same histology scores.2 The rates of mortality from prostate cancer and other
competing medical conditions were estimated respectively using separate Poisson
regression analyses from the 7429 person-years of follow-up by using the Poisson
link in the GENMOD procedure in SAS statistical software version 6.12 (SAS
Institute, Cary, NC), which allowed noninteger numbers of events. P<.05 was considered statistically significant.
To estimate the proportions of men who died from prostate cancer or
other competing medical conditions, or who were still alive 20 years following
diagnosis, we applied the fitted rates of prostate cancer death and other
causes of death to the proportion of men still alive at the beginning of each
successive follow-up interval. This was plotted for each age-histology stratum.
Prostate cancer mortality rates in each of the 5-year intervals after diagnosis
were calculated using the numbers of deaths divided by the number of person-years
of follow-up in the interval; rates were expressed as the number of prostate
cancer deaths per 1000 person-years. The confidence intervals (CIs) for these
mortality rates were calculated by multiple imputation. Because the distribution
of tumor histology among survivors after 15 years was more favorable than
the entire cohort at the time of diagnosis, we also compared the mortality
rates before and after 15 years of follow-up, adjusting for tumor grade and
age at diagnosis using Poisson regression.
The median observation period was 24 years (range, 16-33 years); for
87% of the men, it was longer than 20 years. Of the 107 men who died since
our previous study, we determined the date of death for all 107 men and the
cause of death for 95 men. We were able to obtain the cause of death for an
additional 44 men for whom this information was unavailable at the end of
the first study.2 A description of the original
study population, including information gathered concerning cause of death
during the past 6 years, is shown in Table 1.
The distributions of outcomes at the end of this follow-up are shown
in Table 2. The table stratifies men
by 2 key factors that influence long-term survival: age at diagnosis and the
histology score of the biopsy specimen classified according to the Gleason
system. To standardize the follow-up and provide more stable estimates of
the mortality rates from prostate cancer or other competing conditions, smoothed
estimates are presented as 20-year outcomes as a function of time from diagnosis
(Figure). The results follow the same
trends as noted in 1998. Few men with low-grade tumors identified by prostate
biopsy had progression leading to prostate cancer death within 20 years, although
most men with high-grade tumors died from the cancer regardless of their age
at diagnosis. Among men with a comorbidity score of 0 to 1, 26%, 15%, and
8% survived at least 15, 20, and 25 years, respectively. Among men with a
comorbidity score of more than 1, 11%, 6%, and 3% survived at least 15, 20,
and 25 years, respectively.
Prostate cancer–specific mortality rates were unchanged after
15 years of follow-up (Table 3). The
prostate cancer mortality rate was 33 per 1000 person-years during the first
15 years (95% CI, 28-38) and 18 per 1000 person-years after 15 years of follow-up
(95% CI, 10-29). These rates were not statistically different after adjusting
for the more favorable histology profiles among men who survived more than
15 years from diagnosis (rate ratio, 1.1; 95% CI, 0.6-1.9).
Considerable controversy surrounds the appropriate treatment of newly
diagnosed prostate cancer. Widespread testing for PSA has introduced lead
time and length time effects that further complicate the task of determining
the efficacy of treatment. Although some of the cancers identified by contemporary
biopsy techniques are destined to progress to clinically significant disease,
trends in population-based incidence and mortality rates suggest that a significant
number of prostate cancers identified by PSA testing are unlikely to be clinically
Johansson et al1 recently published 20-year
follow-up data of a population-based cohort of men diagnosed with prostate
cancer between 1977 and 1984, a period of time very similar to our patient
series. They reported an unexpected 3-fold increase in prostate cancer mortality
rates for the 49 patients who were still alive 15 years after diagnosis. A
20-year analysis of our cohort does not support that finding. Two factors
may contribute to this difference: histology classification and cause of death
Despite recent advances in genomics, tumor histology still remains the
most powerful predictor of disease progression. Patients in the study by Johansson
et al1 underwent an aspiration biopsy and the
results were classified according to the World Health Organization grading
system. Our study used the Gleason score grading system, which relies on glandular
architecture to classify specimens. Although both systems show good correlation
between grade and survival, they are based on fundamentally different criteria
and may result in different classifications, especially among men with moderately
Both studies agree that men with well-differentiated tumors rarely die
from their disease and that men with poorly differentiated tumors frequently
die within 5 to 10 years of diagnosis, often despite aggressive interventions.
Men with moderately differentiated tumors have the greatest variation in outcomes.
Based on only 8 prostate cancer deaths from years 15 to 20, Johansson et al1 reported that the prostate cancer mortality rate for
the entire cohort increased from 15 to 44 deaths per 1000 person-years. We
found that men with Gleason scores of 5, 6, and 7 had mortality rates of 12,
30, and 65 deaths per 1000 person-years, respectively, during a 20-year follow-up.
These rates were unchanged from those rates at 15 years. The increased mortality
rate reported by Johansson et al1 may therefore
reflect a classification artifact involving a small number of patients.
A second distinction between these 2 studies is cause of death determination.
Prostate cancer mortality statistics rely on an accurate determination of
whether a patient died of his disease or from a competing medical condition.
This can be difficult to determine when patients have multiple chronic diseases,
such as heart disease or other malignancies (eg, lung cancer or colon cancer).
Johansson et al1 relied on medical record
review to determine cause of death and validated their findings with information
recorded in the Swedish Death Register. We relied on information reported
on part 1 of the death certificate to classify patients as having died from
prostate cancer or a competing medical condition. Although death certificates
are often an unreliable source for identifying a specific cause of death,
we have determined through 2 independent validation studies that prostate
cancer mortality can be classified reliably (concordance with medical record
review of 87%-96%) when cause of death is recorded as a dichotomous variable
(ie, the patient died from prostate cancer or a competing medical condition).6,7 The misclassification of a small number
of patients in either study may explain the different findings at years 15
and 20. Overall, the prostate cancer mortality rates determined in both studies
are remarkably similar.
The clinical implications of both studies should be the same. Men with
well-differentiated disease rarely require treatment, while men with poorly
differentiated disease treated with androgen deprivation alone will usually
die from prostate cancer. Radical prostatectomy may reduce disease-specific
mortality by half.8 However, for those men
who have PSA recurrence following surgery, there is a high probability of
disease progression during a period of 10 to 15 years; this most commonly
occurs in men who have poorly differentiated disease.9
Counseling men who have moderately differentiated disease (Gleason score
of 5-6 and/or World Health Organization grade 2) and a life expectancy of
more than 15 years poses the greatest challenge. A majority of these men will
die from competing medical conditions during a period of 15 to 20 years. Until
better prognostic markers are developed, physicians will continue to recommend
aggressive treatments at the time of diagnosis.
Repeated PSA testing is exacerbating this dilemma by introducing a lead
time of many years. The data from our study and Johansson et al1 are
derived from patients diagnosed with prostate cancer before the advent of
PSA testing. Based on data collected by the Rotterdam section of the European
Randomized Study of Screening for Prostate Cancer, Draisma et al4 have
developed models that estimate the impact of PSA testing on lead time and
the probability of detecting clinically insignificant disease. These depend
on a man’s age at screening. For a single PSA screening test at age
55 years, the estimated lead time was 12.3 years (range, 11.6-14.1 years)
and the likelihood of detecting clinically insignificant disease was 27% (range,
24%-37%). At age 75 years, the estimated lead time was only 6.0 years (range,
5.8-6.3 years), but the likelihood of detecting clinically insignificant disease
increased to 56% (range, 53%-61%). Draisma et al4 estimated
that annual PSA testing from ages 55 to 67 years would yield insignificant
cancers in approximately half of all men diagnosed with localized prostate
cancers and would increase a person’s lifetime risk of being diagnosed
with prostate cancer by 80%.
Thompson et al10 have recently demonstrated
that the pool of subclinical prostate cancer is much larger than prostate
cancer mortality statistics would suggest. After analyzing data from a large
chemoprevention study comparing finasteride with placebo, they found that
the prevalence of prostate cancer was 6.6% among men whose PSA was consistently
less than 0.5 ng/mL and as high as 26.9% among men whose PSA was between 3.1
and 4.0 ng/mL. It is unclear whether these tumors will progress at the same
rate as described by our study and Johansson et al.1 Length
time bias would result in PSA testing preferentially identifying slower growing
tumors. Evidence supporting the impact of length time bias is found in a recent
study by Stamey et al11 who reviewed pathology
specimens from 1317 consecutive men undergoing radical prostatectomy at Stanford
University during the past 20 years. They found that the size of the prostate
cancers identified by PSA testing has decreased steadily since 1983 and that
an increased PSA level in 2004 more commonly reflects benign prostate enlargement
rather than prostate cancer.
A limitation of our study is that many of the patients did not undergo
a complete metastatic evaluation and the actual stage of the disease at diagnosis
may have been worse than our classification. Before the advent of contemporary
transrectal ultrasound–guided biopsies, which often include 10 to 12
cores, 40% of patients with newly diagnosed disease had clinical evidence
of extracapsular disease.12 Catalona et al13 have shown that more than half of all patients presenting
with a serum PSA level of more than 10 mg/mL have pathological evidence of
extracapsular disease. Because none of the patients included in our series
underwent PSA testing, there is a high probability that the series contains
a number of men with extracapsular disease. Therefore, our results most likely
underestimate survival for contemporary patients with localized prostate cancer.
Extended follow-up of our competing risk analysis suggests that prostate
cancer progression rates do not increase after 15 years of follow-up. Men
with low-grade prostate cancer have only a small risk of prostate cancer progression
even after 20 years of management by observation or androgen withdrawal therapy
alone. These results do not support aggressive treatment of localized low-grade
prostate cancer. Men with poorly differentiated disease (Gleason scores of
7 and 8-10) have a high risk of death from prostate cancer; only 3 men were
alive after 20 years. Men with moderate-grade disease (Gleason scores of 5-6)
have an intermediate cumulative risk of prostate cancer progression after
20 years of follow-up. Our data provide what are likely overestimates of prostate
cancer progression when men are treated by observation or androgen withdrawal
therapy alone. Only through randomized controlled trials designed to measure
the efficacy of screening and treatment for prostate cancer can we answer
questions concerning which patients may truly benefit. Such trials are currently
under way in Sweden, England, and the United States.14- 17
Corresponding Author: Peter C. Albertsen,
MD, MS, Division of Urology, University of Connecticut Health Center, 263
Farmington Ave, Farmington, CT 06030-3955 (firstname.lastname@example.org).
Author Contributions: Dr Albertsen had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Albertsen.
Acquisition of data: Albertsen, Fine.
Analysis and interpretation of data: Albertsen,
Drafting of the manuscript: Albertsen, Hanley.
Critical revision of the manuscript for important
intellectual content: Albertsen, Hanley, Fine.
Statistical analysis: Hanley.
Obtained funding: Albertsen, Fine.
Administrative, technical, or material support:
Study supervision: Albertsen.
Financial Disclosures: None reported.
Funding/Support: This study was funded by grant
HS09578 from the Agency for Healthcare Research and Quality.
Role of the Sponsor: The Agency for Healthcare
Research and Quality had no role in the design and conduct of the study; in
the collection, management, analysis, and interpretation of the data; or in
the preparation, review, or approval of the manuscript. Certain data used
in this study were obtained from the Connecticut Tumor Registry in the Connecticut
Department of Public Health. The authors assume full responsibility for analyses
and interpretation of these data.