Presented at the Society for Clinical Oncology/Society of Urologic Oncology/American Society for Radiation Oncology Genitourinary Cancers Symposium, San Francisco, California, March 5-7, 2010.
A study was undertaken to determine the impact of prior coronary revascularization (angioplasty, stent, or coronary artery bypass graft) on the risk of all-cause mortality after neoadjuvant hormonal therapy (HT) for prostate cancer (PC) in men with a history of coronary artery disease (CAD)-induced congestive heart failure (CHF) or myocardial infarction (MI).
Among 7839 men who received radiation with or without a median of 4 months of HT for PC from 1991 to 2006, 495 (6.3%) had CAD-induced CHF or MI and formed the study cohort. Of these men, 250 (50.5%) had been revascularized before treatment for PC. Cox regression was used to determine whether HT increased the risk of all-cause mortality, and whether revascularization altered this risk, after adjusting for known PC prognostic factors and a propensity score for revascularization.
Median follow-up was 4.1 years. Neoadjuvant HT was associated with an increased risk of all-cause mortality (28.9% vs 15.7% at 5 years; adjusted hazard ratio [HR], 1.73; 95% confidence interval [CI], 1.13-2.64; P = .01). Men who received HT without revascularization had the highest risk of all-cause mortality (33.3%; adjusted HR, 1.48; 95% CI, 1.01-2.18; P = .047), whereas men who were revascularized and did not receive HT had the lowest risk of all-cause mortality (9.4%; adjusted HR, 0.51; 95% CI, 0.28-0.93; P = .028). The reference group had an intermediate risk of all-cause mortality (23.4%) and was comprised of men in whom HT use and revascularization were either both given or both withheld.
On the basis of randomized controlled trials showing a survival benefit, neoadjuvant hormonal therapy (HT) combined with radiation is widely used in men with high-risk localized or locally advanced prostate cancer (PC).1-4 However, evidence is mounting that the deleterious effects of HT on cardiovascular health may contribute to excess mortality in susceptible patients.
Specifically, a postrandomization analysis of 1 randomized trial suggested that men with moderate to severe cardiac comorbidity based predominantly on a prior history of a myocardial infarction (MI) did not have improved survival with the addition of HT.2 Moreover, a recent study identified the specific comorbidities that placed men at greatest risk of excess mortality when exposed to HT, and found that men with a history of coronary artery disease (CAD)-induced congestive heart failure (CHF) or MI had a significantly higher risk of all-cause mortality when HT was used in addition to radiation.5
An important unanswered question concerning men with high-risk localized or locally advanced PC and a history of CAD-induced CHF or MI is whether HT should be delivered and if so, what can be done to eliminate the observed excess risk of death associated with HT use. Although coronary revascularization in men with CAD-induced MI or CHF using angioplasty, stent, or coronary artery bypass graft (CABG) is possible, there are no data to suggest that such an approach can eliminate the observed increased risk of all-cause mortality with HT use. Therefore, the purpose of this study is to determine the impact of prior coronary revascularization on the increased risk of all-cause mortality associated with neoadjuvant HT in men with a history of CAD-induced CHF or MI.
MATERIALS AND METHODS
Patient Population and Treatment
From 1991 to 2006, 7839 consecutive men received brachytherapy-based radiation with or without a median of 4 months of neoadjuvant HT for localized or locally advanced PC at 1 of 20 US community-based practices located in Florida, New York, and North Carolina. Among these men, 495 (6.3%) had a history of CAD-induced CHF or MI and a minimum of 1-year follow-up and comprised the study cohort. Among the cohort, 250 (50.5%) had undergone revascularization (angioplasty, CABG, or stenting) before presentation.
Decisions about adjuvant therapies were typically based on the American Brachytherapy Society guidelines, which recommend brachytherapy alone for low-risk disease and suggest the use of supplemental external beam radiotherapy (EBRT) when there is a significant risk of extraprostatic extension.6 Neoadjuvant HT was generally used in the setting of unfavorable-risk disease, and typically only used in the setting of favorable-risk disease if it was needed to downsize the prostate to eliminate pubic arch interference so brachytherapy would be technically feasible.
Neoadjuvant HT consisted of a luteinizing hormone-releasing hormone agonist with or without an antiandrogen and was given before brachytherapy. EBRT was generally delivered using photons in 25 1.8 gray [Gy] fractions to the prostate and seminal vesicles, for a total dose of 45 Gy using computed tomography-based simulation and a 3-dimensional conformal or intensity-modulated radiotherapy technique. The pelvic lymph nodes were not included in the EBRT volume. Brachytherapy was performed using a peripheral loading technique using preloaded iodine-125, palladium-103, or cesium-131 sources and preplanned dosimetry. The prescribed minimum peripheral doses used were consistent with accepted standards within the United States. Specifically, they were 144, 108, and 115 Gy, respectively, for iodine-125, palladium-103, and cesium-131 when used as monotherapy and 108, 90, and 100 Gy, respectively, when used in conjunction with 45 Gy of supplemental EBRT.
Prostate needle biopsy specimens underwent review by a pathologist with expertise in genitourinary pathology at each center. In accordance with federal and institutional guidelines, all research was conducted under an institutional review board-approved protocol permitting collection and analysis of deidentified patient data at baseline and follow-up. The data collection on the study patients was performed by a team of data managers and was overseen by a biostatistician at each facility.
Ascertainment of Comorbidities
For each patient, as part of the initial consultation, the treating physician systematically assessed for the presence or absence of CAD risk factors, including diabetes mellitus, hypercholesterolemia, or hypertension and CAD-induced conditions, including CHF or MI. This information was initially ascertained from medical records provided by referring physicians and then confirmed through verbal communication with each patient at the time of initial consultation. Although no specific diagnostic criteria or threshold laboratory levels were enforced by the study, it was assumed that each patient's primary care physician or cardiologist used up-to-date, evidence-based diagnostic criteria set forth by national, consensus panel statements for each comorbid condition.
Follow-up started on the day of prostate brachytherapy and concluded on the date of death or the date of last follow-up for patients still alive. Patients were generally seen every 3 months for 1 year, every 6 months for an additional 4 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 prostate-specific antigen (PSA) level measurement before the digital rectal examination. At the time of PSA failure, in addition to the routine follow-up assessment, a pelvic computed tomography or magnetic resonance imaging and a bone scan were also obtained. Salvage therapy consisted of a luteinizing hormone-releasing hormone agonist, which was typically initiated within 1 year after PSA recurrence and always before symptomatic or radiographic progression.
All deaths from PC were confirmed using either the National Death Index or attending report.
Comparison of the distribution of baseline patient and tumor characteristics across treatment modalities
Descriptive statistics were used to characterize the clinical characteristics of the study cohort at baseline and are shown in Table 1, stratified by the use of neoadjuvant HT. A Mantel-Haenszel chi-square test was used to compare the distribution of categorical covariates at baseline among men who received HT versus those who did not receive HT. For continuous covariates, including the serum PSA levels, and age at and year of treatment, medians and their distributions were compared across strata using a Wilcoxon rank-sum test. Table 2 uses similar methodology, but the stratification is based on whether men underwent a revascularization procedure.
Table 1. Distribution of Clinical Factors at Baseline Stratified by Whether Men Received Neoadjuvant HT
Cox multivariate regression was used to evaluate whether neoadjuvant HT use was significantly associated with the risk of all-cause mortality, adjusting for known PC prognostic factors (clinical T classification, Gleason score, PSA), whether EBRT was added, and a treatment propensity score for EBRT that incorporated information about PC prognostic factors, age at brachytherapy, and year of brachytherapy.7 Propensity analysis required calculation of a conditional probability for the 2 treatment groups (brachytherapy alone or brachytherapy with supplemental EBRT) using a multivariate logistic regression model.8
A second Cox model assessed whether prior revascularization affected the risk of all-cause mortality because of neoadjuvant HT use, after adjusting for the same covariates as in the prior model, except that the treatment propensity score was constructed to account for the propensity for receiving revascularization. This revascularization propensity score contained information about age at brachytherapy, year of brachytherapy, and whether the patient had a history of diabetes mellitus, or of both CHF and MI versus either alone. The adjusted risk of all-cause mortality was compared within 4 groups defined based on whether patients were revascularized or received HT: 1) revascularization plus no HT, 2) revascularization plus HT, 3) no revascularization plus no HT, and 4) no revascularization plus HT. Given that the adjusted risk of all-cause mortality was not significantly different in the initial adjusted model between groups 2 and 3 (adjusted P = .87), these 2 groups were collapsed in the final model and served as the baseline group to which the other 2 groups (1 and 4) were compared.
For the Cox models, an event was defined as death from any cause. PSA level was log transformed to ensure it followed a normal distribution and was treated as a continuous variable. For the categorical variables of Gleason score (≤6, 7, 8-10) and tumor category (T1c, T2, T3), cutpoints were determined before the analysis on the basis of established clinically relevant strata. Baseline groups for those categorical variables were defined as Gleason score 6 or less and clinical category T1c, respectively. The assumptions of the Cox model were tested, and no evidence that these assumptions were violated was found. Unadjusted and adjusted hazard ratios (HRs) for all-cause mortality with associated 95% confidence intervals (CIs) and P values were calculated for each covariate.
Estimates of all-cause mortality stratified by treatment(s) received
For the purpose of illustration, the method of Kaplan and Meier was used to estimate all-cause mortality (defined as 1 minus the Kaplan-Meier estimate of overall survival) and stratified by whether neoadjuvant HT was used.9 Comparisons of these estimates were also made across the 3 groups as previously defined based on whether a revascularization procedure was performed and whether neoadjuvant HT used. Time zero was defined as the day of prostate brachytherapy. Comparisons of these estimates across cohorts defined by the treatment received were performed using the log-rank test,10 and adjustment for multiple comparisons was made using the Bonferroni method.11 For the revascularization/HT groups, given that there were 3 possible log-rank pairwise comparisons across the 3 groups, a Bonferroni corrected significance level of .05/3 = .017 was used.
SAS version 9.2 (SAS Institute, Cary, NC) was used for all calculations.
Description of Study Cohort
Of the 7839 men, the study cohort was comprised of the 495 (6.3%) patients with a history of CAD-induced CHF or MI. Men with low-risk PC (clinical category T1c/T2a, Gleason ≤6, and PSA <10) accounted for 44% of the cohort, and men with intermediate-risk PC (T2b, PSA 10-20, or Gleason 7) accounted for 39%, whereas men with high-risk PC (T2c or higher, Gleason 8-10, or PSA >20) comprised only 17% of the study cohort.12
Among these 495 men, 250 (50.5%) had been revascularized before presentation, and 315 (63.6%) received neoadjuvant HT. As seen in Table 1, when compared with men who did not receive neoadjuvant HT, the patients who received neoadjuvant HT had more unfavorable PC prognostic factors, including higher PSA (P = .0001), a greater proportion of high-grade disease (P < .001), and higher T category (P = .042), which all contribute to men receiving HT being in higher risk groups (P < .001). As shown in Table 2, the distribution of PSA, Gleason score, T category, and risk group of men who underwent a revascularization procedure did not significantly differ compared with men who had not been revascularized. However, those who were revascularized tended to be slightly younger than those who were not (median age: 72 vs 73 years, P = .01).
HT Use and the Risk of All-Cause Mortality in Men With a History of CHF or MI
After a median follow-up of 4.1 years (interquartile range, 2.7-6.6), 126 (25.4% of 495) men died. As shown in Table 3, neoadjuvant HT was associated with a significantly increased risk of all-cause mortality (adjusted HR, 1.73; 95% CI, 1.13-2.64; P = .01) in men with a history of CHF or MI. No other covariates were significantly associated with the risk of all-cause mortality.
Table 3. HT Use and the Risk of ACM Using Cox Multivariate Regression Analysis in Men With a History of CHF or MI
For illustrative purposes, estimates of all-cause mortality stratified by whether neoadjuvant HT was administered are shown in Figure 1. The 5-year estimate of all-cause mortality for men who received HT was 28.9% (95% CI, 22.8-36.3), and for those who did not receive HT it was 15.7% (95% CI, 9.8-24.6).
Revascularization and HT Use and the Risk of All-Cause Mortality in Men With CHF or MI
As shown in Table 4, men who received HT and were not revascularized had the highest risk of all-cause mortality (adjusted HR, 1.48; 95% CI, 1.01-2.18; P = .047), whereas men who were revascularized and did not receive HT had the lowest risk of all-cause mortality (adjusted HR, 0.51; 95% CI, 0.28-0.93; P = .028). The reference group had an intermediate risk of all-cause mortality, and was comprised of men in whom HT and revascularization were both given or both withheld. Those 2 groups were combined to form the reference group, because their adjusted risk of all-cause mortality did not differ from each other (P = .87).
Table 4. Revascularization and HT Use and the Risk of ACM Using Cox Multivariate Regression Analysis in Men With CHF or MI
Propensity score for revascularization per 1% increase
Figure 2 illustrates estimates of all-cause mortality based on whether a revascularization procedure was performed and whether neoadjuvant HT was used. The 5-year estimate of all-cause mortality for men who received HT and were not revascularized was 33.3% (95% CI, 24.6-44.0). For men in whom revascularization and HT were both given or those in whom neither was given, it was 23.4% (95% CI, 17.1-31.5), and for those who received revascularization but not HT, it was 9.4% (95% CI, 3.8-22.5).
In this study, we found that among men receiving brachytherapy for PC, those with a prior history of CAD-induced CHF or MI (representing 6.3% of this group) were at increased risk for all-cause mortality when treated with neoadjuvant hormonal therapy after adjusting for age, known PC prognostic factors, and a treatment propensity score. This result is concordant with a prior study that also found that men with a history of CAD-induced CHF or MI (5.0% of that population) were at increased risk of all-cause mortality when treated with neoadjuvant HT.5
In addition, we attempted to determine how prior revascularization impacted the increased risk of all-cause mortality attributed to HT use in men with a history of CAD-induced CHF or MI. We found that HT use without revascularization was associated with the highest risk of all-cause mortality. An intermediate risk of all-cause mortality was seen for HT use with revascularization or HT avoidance without revascularization. However, the lowest risk of all-cause mortality was seen for HT avoidance with revascularization. This suggests that prior revascularization reduced but could not completely eliminate the excess risk of mortality associated with HT use in men with a history of CAD-induced CHF or MI.
In current practice, neoadjuvant HT is given to up to 25% to 50% of patients with low-risk PC undergoing brachytherapy in an attempt to shrink the prostate gland to make the procedure more technically feasible, but there is no evidence in the literature that neoadjuvant HT reduces PC-specific mortality in men with low-risk PC.13-17 Therefore, the data from this study suggest that neoadjuvant HT for gland size reduction should not be considered for men with low-risk disease and a history of CAD-induced CHF or MI, even if they have been revascularized.
For patients with unfavorable-risk PC (intermediate or high risk), there are several randomized trials showing that adding HT to radiation improves overall survival.1, 2, 18 However, it is possible that not all subgroups of patients in a randomized trial benefit from an intervention even if an overall effect is seen. Specifically, a postrandomization analysis of 1 of those trials suggested that the benefit of HT appeared to be limited to those with no or mild comorbidity (76% of the study cohort) on the validated ACE-27 scale; those with moderate or severe comorbidity (24% of the study cohort) did not benefit from the addition of HT, and had a nonsignificant decrement in survival when HT was given (8-year overall survival 25% with HT, 54% without HT, P = .08).2, 19 The current study identifies men with a history of CAD-induced CHF or MI as a subgroup that may be harmed by HT and suggests that this harm can be reduced but not completely eliminated by prior revascularization.
Therefore, these data raise the possibility that for men with unfavorable-risk PC, clinicians may need to be selective about which men are offered HT, and will particularly need to individualize care for men with a history of CAD-induced CHF or MI. This study did not have enough men with high-risk PC (17% of the cohort) to perform an adequately powered analysis within that group, and it may be possible that for patients with CAD-induced CHF or MI who have very aggressive PC characteristics, the known benefit of HT on prostate-specific survival may be large enough outweigh the suggested decrement of HT on non–PC-specific survival. This may be particularly true for patients who have been revascularized, as this seemed to reduce the harms caused by HT.
One limitation of this study is that because of its retrospective nature, it is not possible to prove that HT increased the risk of all-cause mortality or that prior revascularization truly modified this risk, but rather we could only show significant associations between these observations. Future randomized studies of PC interventions should be stratified before randomization by cardiac comorbidity and risk factors for cardiovascular comorbidity, and in particular by a history of CHF or MI, to help determine precisely which patients are helped and harmed by additional interventions. In addition, the mechanism by which neoadjuvant HT is increasing the risk of all-cause mortality remains uncertain. The observation that men who had been revascularized had a partial reduction in the excess risk of all-cause mortality suggests that the coronary vessels may play a role in the underlying mechanism, but there are most likely other mechanisms as well that are distinct from the coronary vessels. Further work will be needed to characterize these mechanisms. Finally, the median duration of HT was only 4 months, and so it is unknown from this study what the implications of longer-course HT would be on cardiovascular outcomes.
In summary, in patients with a history of CAD-induced CHF or MI, neoadjuvant HT is associated with an excess risk of all-cause mortality that appears to be reduced but not eliminated by prior revascularization. Therefore, neoadjuvant HT should not be considered for men with low-risk PC and a prior history of CHF or MI regardless of whether they have been revascularized. Decisions about neoadjuvant HT use in men with unfavorable risk PC and a history of CHF or MI should be individualized.