RT indicates radiation therapy; AST, androgen suppression therapy.
Kaplan-Meier estimates of overall survival are based on the method of Kaplan and Meier19
and cumulative incidence estimates are based on Gaynor et al.20 Error bars at x=8 years are 95% confidence intervals.
RT indicates radiation therapy; AST, androgen suppression therapy.
Kaplan-Meier estimates are based on the method of Kaplan and Meier.19
Error bars at x=8 years are 95% confidence intervals. P<.025 (.05/2) was considered significant.22
Customize your JAMA Network experience by selecting one or more topics from the list below.
D’Amico AV, Chen M, Renshaw AA, Loffredo M, Kantoff PW. Androgen Suppression and Radiation vs Radiation Alone for Prostate Cancer: A Randomized Trial. JAMA. 2008;299(3):289–295. doi:10.1001/jama.299.3.289
Author Affiliations: Departments of Radiation Oncology (Dr D’Amico and Ms Loffredo), Pathology (Dr Renshaw), and Medical Oncology (Dr Kantoff), Brigham and Women's Hospital and Dana Farber Cancer Institute, Boston, Massachusetts; and Department of Statistics, University of Connecticut, Storrs (Dr Chen).
Context Comorbidities may increase the negative effects of specific anticancer treatments such as androgen suppression therapy (AST).
Objectives To compare 6 months of AST and radiation therapy (RT) to RT alone and to assess the interaction between level of comorbidity and all-cause mortality.
Design, Setting, and Patients
At academic and community-based medical centers in Massachusetts,
between December 1, 1995, and April 15, 2001, 206 men with localized but unfavorable-risk prostate cancer were randomized to receive RT alone or RT and AST combined. All-cause mortality estimates stratified by randomized treatment group and further stratified in a postrandomization analysis by the Adult Comorbidity Evaluation 27 comorbidity score were compared using a log-rank test.
Main Outcome Measure Time to all-cause mortality.
As of January 15, 2007, with a median follow-up of 7.6 (range,
0.5-11.0) years, 74 deaths have occurred. A significant increase in the risk of all-cause mortality (44 vs 30 deaths; hazard ratio [HR],
1.8; 95% confidence interval [CI], 1.1-2.9; P = .01)
was observed in men randomized to RT compared with RT and AST. However,
the increased risk in all-cause mortality appeared to apply only to men randomized to RT with no or minimal comorbidity (31 vs 11 deaths;
HR, 4.2; 95% CI, 2.1-8.5; P < .001).
Among men with moderate or severe comorbidity, those randomized to RT alone vs RT and AST did not have an increased risk of all-cause mortality (13 vs 19 deaths; HR, 0.54; 95% CI, 0.27-1.10; P = .08).
Conclusions The addition of 6 months of AST to RT resulted in increased overall survival in men with localized but unfavorable-risk prostate cancer. This result may pertain only to men without moderate or severe comorbidity, but this requires further assessment in a clinical trial specifically designed to assess this interaction.
clinicaltrials.gov Identifier: NCT00116220
Several randomized studies1-5 have documented a prolongation in overall survival, prostate cancer–specific survival, or both when androgen suppression therapy (AST) is combined with external beam radiation therapy (RT) compared with RT alone in the management of unfavorable localized and locally advanced prostate cancer. Therefore, in men with these stages of disease, RT and AST have become a standard of care.
However, evidence from pooled analyses of randomized studies6
as well as large patient cohort studies7,8
suggests that AST administration is associated with an increased risk of fatal6
and nonfatal7,8 cardiovascular events in men of advanced age. A possible explanation for this association is that as men age, they often acquire comorbid illnesses and these comorbidities may increase the negative effects of specific anticancer treatments such as AST. As a result, it is possible that the survival benefit observed when AST is added to RT may vary among specific subsets of patients defined by their comorbid illness profile.
In this study, we first report the results of the long-term follow-up of a randomized study of 6 months of AST and RT vs RT alone.
The results of shorter-term follow-up, ie, at 4.5 years, have been previously reported.5 In addition,
we performed an analysis of overall survival in subgroups defined by their level of comorbidity at the time of randomization and evaluated whether an interaction existed between the level of comorbidity and treatment group with respect to time to all-cause mortality.
At academic (Dana-Farber Cancer Institute, Brigham and Women's Hospital, and Beth Israel Deaconess Medical Center) and community-based (St Anne's Hospital, Metrowest Medical Center, and Suburban Oncology Center) medical centers in Massachusetts, between December 1, 1995,
and April 15, 2001, 206 men (median age, 72.5 years; range, 49-82
years) with 1992 American Joint Commission on Cancer (AJCC) Clinical Stage9 T1b to T2bN0M0 adenocarcinoma of the prostate and at least 1 unfavorable prognostic factor were randomized to receive RT alone or in combination with 6 months of AST. Radiation therapy alone consisted of 3-dimensional conformal RT. Androgen suppression therapy consisted of a luteinizing hormone–releasing hormone agonist and the antiandrogen flutamide.
Unfavorable prognostic factors included a prostate-specific antigen (PSA) level of more than 10 ng/mL (maximum, 40 ng/mL); a biopsy Gleason score of 7 to 10; cancer or radiographic evidence of extracapsular extension; and/or seminal vesicle invasion using endorectal magnetic resonance imaging (MRI). A complete compilation of preexisting medical conditions and comorbidities was ascertained by the treating physician and recorded at baseline before randomization to satisfy the inclusion criterion that the patient have at least a 10-year life expectancy excluding death from prostate cancer and an Eastern Cooperative Oncology Group perfomance status of 0 or 1, where 0 indicates fully active, able to carry on all predisease performance without restriction, and 1 indicates restricted in physically strenuous activity but ambulatory and able to perform work of a light or sedentary nature (eg, light housework, office work).10 Patients with a history of a prior malignancy except for nonmelanoma skin cancer or prior pelvic RT or AST were excluded. Prostate needle biopsy specimens underwent central review by a pathologist (A.A.R.) with expertise in genitourinary pathology. Before study entry, all men signed an institutional review board–approved, protocol-specific informed consent form in accordance with federal and institutional guidelines.
Randomization was centralized at the Quality Assurance Center of the Dana Farber/Harvard Cancer Center, Boston, Massachusetts. A permuted blocks randomization algorithm was used with a block size of 4. Stratification factors were based on the baseline PSA level, centrally reviewed Gleason score, or endorectal MRI findings as follows:
for group 1, a PSA level of more than 20 to 40 ng/mL; for group 2,
a Gleason score of 7 to 10; for group 3, a PSA level of more than 10 to 20 ng/mL and a Gleason score of 6 or less; and for group 4, endorectal MRI evidence of extracapsular extension or seminal vesicle invasion and low risk (defined as PSA <10 ng/mL and a Gleason score ≤6 and AJCC tumor category T1c or T2a).
Using detailed information on preexisting medical conditions and comorbidities collected at baseline before randomization, a comorbidity score was assigned by the principal investigator (A.V.D.) using the Adult Comorbidity Evaluation 27 (ACE-27), a 27-item validated comorbidity index for use in patients with cancer.11-13
instrument was selected because the clinically relevant comorbid ailments used to obtain the comorbidity score were selected based on prior research by experts.14-16 In addition, the ailments were validated specifically for the case of the newly diagnosed patient with cancer. The index was used to assign grades to diseases of specific conditions into 1 of 4 levels of comorbidity (grade 0 [none], grade 1 [minimal], grade 2 [moderate], or grade 3 [severe]) according to the severity of the individual organ system decompensation and prognostic impact.
Once a man's individual comorbid conditions were classified,
an overall comorbidity score was assigned based on the highest ranked single ailment. For the case in which 2 or more moderate ailments occur in different organ systems, the overall comorbidity score was designated as severe. An example of the scoring in the case of the cardiovascular system is that a prior history of myocardial infarction (MI) within 6 months, more than 6 months, or an old MI by electrocardiogram only (age undetermined) would be scored as severe, moderate, and minimal comorbidity, respectively. The instrument can be found
Follow-up started on the day of randomization and concluded on the date the patient was last observed or the date of death through the prespecified analysis date of January 15, 2007, whichever came first; no patient was lost to follow-up. The median for the date of last follow-up in living men was September 8, 2006 (range, November 2, 1998-January 15, 2007).
The attending oncologist who followed the patient until death determined the cause of death. To record a death as being due to prostate cancer, there had to be documented hormone refractory metastatic prostate cancer and evidence that the PSA level was increasing at the time of the last follow-up visit despite the use of second-line hormonal maneuvers and cytotoxic chemotherapy before death. To record a death from cardiovascular disease, an acute MI needed to be determined as the immediate cause of death.
Patients were observed every 3 months for 2 years, every 6 months for an additional 3 years, and then annually thereafter. At each follow-up, a history and physical examination including a digital rectal examination was performed in addition to a serum PSA level before the digital rectal examination. At the time of PSA failure in addition to the routine follow-up assessment, computed tomography or MRI of the pelvis and a bone scan were also obtained. Androgen suppression therapy administration for PSA failure was recommended when the PSA level reached 10 ng/mL.
There were 30 occurrences of AST use for PSA failure in the 104 men randomized to RT.
Sample Size Calculation and End Points.
The study had 80% power and allowance for a type I error of 5% to detect a difference in time to PSA recurrence assuming the true median time to PSA failure of 2.7 years and 4.8 years among patients treated with RT or RT and AST, respectively. Full power was projected to occur after 2.7 years of accrual at 100 patients per year and an additional 2 years of follow-up.
The primary end point of the original study was time to PSA recurrence. Prostate-specific antigen recurrence was defined as a PSA level of more than 1.0 ng/mL and increasing by more than 0.2 ng/mL at 2 consecutive visits following treatment. However, before the first planned interim analysis, which was to be performed 3 years after completion of accrual, the follow-up was extended to evaluate the prespecified secondary end points of overall survival and prostate cancer–specific mortality (PCSM) because results17
became available that revealed a higher than expected (2-fold) reduction in death when AST was added to RT in men with locally advanced prostate cancer. Overall survival stratified by randomized treatment group and further stratified in a postrandomization analysis by the ACE-27 comorbidity score11-13 was also analyzed.
Overall Survival and PCSM. Descriptive statistics were used to characterize the patients at study entry. A χ2 metric18
was used to compare the distribution of comorbidity scores between men in the 2 treatment groups. The methods of Kaplan and Meier19
and cumulative incidence20
were used to estimate and characterize overall survival and PCSM, respectively. Comparisons of these estimates were performed by using a log-rank test and k-sample test,21
respectively. Overall survival was measured from the date of randomization to the date of death or the date of last follow-up. For the purpose of illustration estimates of overall survival stratified by treatment group in a prespecified analysis and also further stratified by the ACE-2711-13
defined comorbidity score in a postrandomization analysis were graphically displayed. Two-sided P<.05 was considered statistically significant, with adjustments made for multiple comparisons using a Bonferroni correction.22 All results reported are from an intention-to-treat analysis.
Assessment for Interaction Between Comorbidity Score and Treatment Group. A multivariate Cox regression postrandomization analysis23
was used to evaluate whether a significant interaction existed between the ACE-27 comorbidity score11-13
and treatment group, adjusting for known prognostic factors. Prognostic factors examined included the PSA level, highest biopsy Gleason score and AJCC tumor category, and age at randomization in addition to the comorbidity score as determined using the ACE-27 instrument. To assess whether an interaction between comorbidity score and treatment existed, an interaction term (treatment × comorbidity score) was included in the multivariate model in addition to the categorical covariates of treatment and comorbidity score. To adjust for the use of AST delivered for PSA failure, a time-dependent covariate24
named AST2 (t) was included in the multivariate model. For categorical variables, cut points were determined before the analysis based on established clinically relevant strata.25 Age and PSA level were treated as continuous variables and the PSA level was log transformed to ensure that the values were normally distributed, whereas treatment, Gleason score,
comorbidity score, and AJCC tumor category were considered categorical variables. Baseline groups for categorical variables were defined as comorbidity score 0 (none) or 1 (minimal), treatment with RT and AST, Gleason score of 6 or less, and clinical category T1.
For all regression analyses, the assumptions of the Cox proportional hazards regression model were tested and no evidence that these assumptions were violated were found. Unadjusted and adjusted hazard ratios (HRs)
for all-cause mortality23
with associated 95% confidence intervals (CIs) and P values were calculated for each covariate from the Cox proportional hazards regression23
and Fine and Gray regression26 models, respectively. The Fine and Gray regression was used to compute the HR for PCSM to use a proportional hazards model that accounted for competing causes of mortality. R version 2.1.1 (R Foundation for Statistical Computing, Vienna, Austria) was used for all calculations pertaining to Gray's k-mean test and Fine and Gray regression. SAS version 9.1.3 (SAS Institute, Cary, North Carolina) was used for all remaining statistical analyses.
As of January 15, 2007, with a median follow-up of 7.6 (range,
0.5-11.0) years, there were 74 deaths among the 206 men (44 deaths among men randomized to RT and 30 deaths among men randomized to RT and AST) (Table 1). Of the 74
all-cause mortality deaths, 18 were attributed to prostate cancer and, of these, 14 and 4 were observed to occur in men randomized to RT vs RT and AST, respectively. Estimates of overall survival were significantly higher (log-rank P = .01)
for men who were randomized to RT and AST compared with RT, with Kaplan-Meier 8-year survival estimates of 74% (95% CI, 64%-82%) and 61% (95% CI,
49%-71%), respectively (Figure 1). Figure 1 also illustrates that cumulative incidence estimates of PCSM significantly favored (k-sample P = .007) the RT and AST group. Specifically,
there was an increased risk of PCSM (14 vs 4 deaths; HR, 4.1; 95%
CI, 1.4-12.1; P = .01) that translated into an increased risk of all-cause mortality (44 vs 30 deaths; HR,
1.8; 95% CI, 1.1-2.9; P = .01)
in men randomized to RT compared with RT and AST (Table 2).
No significant differences (P = .99)
were observed in the distribution of comorbidity scores for the 104
and 102 men randomized to RT vs RT and AST, respectively (Table 3). Specifically, 68 men (65%),
11 men (11%), 22 men (21%), and 3 men (3%), and 67 men (66%), 11 men (11%), 21 men (21%), and 3 men (3%) randomized to receive RT or RT and AST had comorbidity scores of 0, 1, 2, and 3, respectively. All 6 men with severe baseline comorbidity (score = 3) attained these scores because of moderate comorbidity (score = 2)
in 2 different organ systems. A history of MI more than 6 months before randomization was the most common reason to assign a moderate comorbidity score being noted in 5 of the 6 men (83%) with a severe comorbidity score and in 11 of 22 men (50%) and 13 of 21 men (62%) with moderate comorbidity who were randomized to RT or RT and AST, respectively.
Of the 157 and 49 men with no to minimal or moderate to severe comorbidity, there were 42 (27%) and 32 (64%) deaths, respectively.
Of these deaths, 17 (40%) and 1 (3%) were due to prostate cancer,
meaning that prostate cancer was not a major contributor to all-cause mortality in men with moderate to severe comorbidity. After adjusting for AST use for PSA failure, age, comorbidity and Gleason score, PSA level, and AJCC tumor category, men randomized to RT compared with RT and AST remained at increased risk for all-cause mortality (44
vs 30 deaths; adjusted HR, 3.0; 95% CI, 1.5-6.4; P = .003)
A significant interaction was noted between comorbidity score and treatment (adjusted HR, 0.14; 95% CI, 0.05-0.39; P < .001)
(Table 2 and Figure 2). Specifically, for the 157
men with no or minimal comorbidity scores treatment with RT and AST compared with RT was associated with a significantly higher survival (31 vs 11 deaths; HR, 4.2; 95% CI, 2.1-8.5; P < .001),
with Kaplan-Meier 8-year survival estimates of 90% (95% CI, 79%-95%)
and 64% (95% CI, 49%-75%), respectively (Figure 2). Among the 49 men with moderate or severe comorbidity,
those randomized to RT alone vs RT and AST did not have an increased risk of all-cause mortality (13 vs 19 deaths; HR, 0.54; 95% CI, 0.27-1.10; P = .08). For these men randomized to RT and AST compared with RT, the Kaplan-Meier 8-year survival estimates were 25% (95% CI, 9%-44%) and 54% (95% CI, 32%-72%), respectively (Figure 2).
Our results provide evidence to support the continuation of the previously reported5
overall survival benefit when men with localized but unfavorable-risk prostate cancer of median age 72.5 years received RT and 6 months of AST compared with RT alone. Prior investigators12,27
have noted that comorbidity is an independent prognostic factor for men diagnosed and treated for localized or locally advanced prostate cancer. Still others have noted that life expectancy tables28
can overestimate survival in men treated with radiotherapy for prostate cancer because of competing causes of mortality. Therefore, in the current study, a hypothesis-generating postrandomization subgroup analysis was performed evaluating overall survival in men randomized to receive RT and 6 months of AST or RT stratified using a comorbidity index11-13 that has been validated in individuals with cancer.
For the overall study, there was a 1.8-fold increased risk of death observed; whereas, this risk was 4-fold in men with no or minimal comorbid illness randomized to receive RT compared with RT and AST.
This risk translated into a 13% and 26% absolute improvement in overall survival at 8 years for all men or specifically those with no or minimal comorbidity, respectively, who were randomized to receive RT and AST compared with RT alone. However, men with moderate to severe comorbidity at randomization who were randomized to the combination therapy were not observed to have a prolonged survival when compared with men randomized to receive RT. A possible explanation for this lack of a survival benefit was the presence of a significant interaction between comorbidity score and treatment. Prior studies6-8 have suggested that AST administration can shorten the time to MI in men older than 65 years, and the results of our study are consistent with these prior studies.
The clinical significance of this finding is that preexisting comorbid illness may increase the negative effects of specific anticancer treatments such as AST. Therefore, future randomized studies evaluating the impact on survival of adding novel therapies to the current standards of practice in men with clinically localized or locally advanced nonmetastatic prostate cancer should consider a prerandomization stratification by comorbidity score. Performing this stratification in a study with at least 80% power to observe a difference would permit an assessment of whether a survival benefit was achieved in all men or just those with minimal or no comorbidity.
Two points require further discussion. First, future studies will be needed to discern the specific types of comorbid illnesses that may shorten life expectancy in men who are undergoing AST. Our study suggested that the relevant underlying comorbidity may have been a prior history of MI more than 6 months before randomization but numbers were too small to exclude other comorbidities such as diabetes. Second, in addition to life expectancy, health-related quality of life outcomes require further study. Specifically, it is possible that while life expectancy may not be altered, health-related quality of life may be affected more in men with certain underlying comorbidities when AST is administered. Therefore, the impact of specific comorbidities such as preexisting cardiovascular disease on both life expectancy and health-related quality of life should be studied prospectively in future randomized studies in men with prostate cancer.
In conclusion, the addition of 6 months of AST to RT resulted in increased overall survival in men with localized but unfavorable-risk prostate cancer. This result may pertain only to men without moderate or severe comorbidity, but this requires further assessment in a clinical trial specifically designed to assess this interaction.
Corresponding Author: Anthony V.
D’Amico, MD, PhD, Brigham and Women's Hospital, Department of Radiation Oncology, 75 Francis St, L-2 Level, Boston, MA 02115 (email@example.com).
Author Contributions: Dr D’Amico 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: D’Amico.
Acquisition of data: Renshaw, Loffredo.
Analysis and interpretation of data:
D’Amico, Chen, Kantoff.
Drafting of the manuscript: D’Amico,
Chen, Loffredo, Kantoff.
Critical revision of the manuscript for important intellectual content: D’Amico, Chen, Renshaw, Loffredo,
Statistical analysis: Chen.
Administrative, technical, or material support:
D’Amico, Loffredo, Kantoff.
Study supervision: D’Amico.
Financial Disclosures: None reported.