Optimal management remains unknown following prostate-specific antigen (PSA) failure when considering comorbidity and PSA kinetics at recurrence. In order to define randomized controlled trials (RCTs) that can address this issue, this study examined factors associated with the risk of death following PSA failure.
Of 206 men randomized to RT with or without 6 months of androgen suppression therapy (AST), 108 sustained PSA failure and began AST when PSA approached 10 ng/mL and formed the study cohort. Cox regression multivariable analysis was used to determine factors associated with death following PSA failure.
After a median follow-up of 10.3 years of 108 men with PSA failure, 64 (59%) died, with 22 (34%) dying of prostate cancer (PC). Increasing PSA velocity at recurrence was associated with a significant increase in the risk of death (adjusted hazard ratio, 1.21; 95% confidence interval, 1.02-1.45; P = .03). Among men with no/minimal versus moderate/severe comorbidity, PC comprised 42% (20 of 48) versus 12.5% (2 of 16) of all deaths, respectively. Estimates of PC-specific and all-cause death were significantly higher when PSA velocity was greater than as compared with the median or less in men with no/minimal (P < .008) but not moderate/severe comorbidity (P > .15).
Men who have prostate cancer (PC) with unfavorable (intermediate- or high-risk) risk can experience prostate-specific antigen (PSA) recurrence after definitive radiation therapy (RT) despite the addition of androgen suppression therapy (AST).[2, 3] Management of these recurrences generally involves AST if the PSA doubling time (DT) is short, suggesting systemic recurrence, or salvage local therapy or surveillance depending on comorbidity when the PSA DT is long, suggesting a local recurrence. However, management in this setting is not standardized due to the lack of randomized controlled trials (RCTs) comparing these commonly used treatments to surveillance alone.
Given this uncertainty and the known adverse side effects related to AST,[5-8] some physicians have reservations about recommending AST to all men in this setting, particularly those with cardiac comorbidity. Specifically, studies suggest that among men with comorbidities, AST use may be associated with an increased risk of fatal and nonfatal cardiovascular events.[7-9] Therefore, there is a need to appropriately select patients for salvage AST based on comorbidity status and known adverse PC prognostic factors which include a short time to PSA recurrence, Gleason score of 8 to 10, and short PSA DT or rapid PSA velocity.[10, 11]
In 2009, it was shown that a rapid PSA velocity at post-RT PSA recurrence was associated with a significant increase in the risk of all-cause mortality (ACM) among men with no or minimal comorbidity (P < .001) using the 27-item Adult Comorbidity Evaluation-27 (ACE-27) metric. However, among men with moderate to severe comorbidity, this risk was increased and approached but did not reach significance (P = .12). Moreover, following PSA failure, intermittent AST has been recently shown to be noninferior to continuous AST with respect to survival, raising the question of whether intermittent AST or surveillance is best in these men. Therefore, the optimal management of PSA failure following RT with or without AST remains unknown when considering comorbidity and PSA kinetics at recurrence. To define RCTs to address this issue, in this study, we examine factors associated with the risk of death following PSA failure.
MATERIALS AND METHODS
Patient Eligibility and Treatment
Between December 1, 1995, and April 15, 2001, 206 patients were randomly assigned (ClinicalTrials.org identifier NCT00116220) to receive either 70 Gy by 3-dimensional conformal RT alone (n = 104) or in conjunction with 6 months of AST (n = 102), and as of November 20, 2012, 108 men sustained PSA failure and are subject of the current study. These patients were from 6 Harvard-affiliated hospitals and had been diagnosed with T1b to T2b, NX, M0 adenocarcinomas of the prostate (American Joint Committee on Cancer [AJCC] Staging Manual) with at least 1 unfavorable prognostic factor. Unfavorable prognostic factors included PSA of more than 10 ng/mL and ≤ 40 ng/mL; a biopsy Gleason score of 7 or higher; radiographic evidence of extracapsular extension; or seminal vesicle invasion on endorectal coil magnetic resonance imaging. AST included a luteinizing hormone–releasing hormone agonist and the anti-androgen flutamide. The exclusion criteria, registration, randomization, stratification, treatment, and quality assurance guidelines for this study have been described.
Adult Comorbidity Evaluation-27
The ACE-27 score is a validated 27-item comorbidity index for patients with cancer. The senior author (A.V.D.) assigned an ACE-27 comorbidity score to each patient after detailed review of the patients' baseline medical conditions and comorbidities. The decision to use the ACE-27 index was based on its validation of comorbidities, specifically in the setting of newly diagnosed cancer and the selection of clinically significant comorbidities by experts.[15-17] By using the ACE-27 index, grades were assigned to diseases according to the degree of organ system decompensation and its prognostic impact (grade 0 = none, grade 1 = minimal, grade 2 = moderate, grade 3 = severe). Once an individual grade was given, an overall comorbidity score was assigned based on the highest ranked single ailment. If at least 2 moderate ailments were present in different organ systems, the overall comorbidity score was designated as severe. Additional information on the ACE-27 index can be found at http://oto.wustl.edu/clinepi/calc.html.
Patient follow-up consisted of measuring a serum PSA level and performing a digital rectal examination every 3 months for the first 2 years, every 6 months for the next 3 years, and then annually subsequently. PSA failure was defined as 2 consecutive PSA value rises of more than 0.2 ng/mL obtained after a nadir value had been reached, which was the prior ASTRO (American Society for Radiation Oncology) consensus definition of PSA failure at the time this phase 3 RCT was initiated. The time of PSA failure was defined as the time of the initial PSA rise. At time of PSA failure, patients had a routine follow-up assessment, computed tomography or magnetic resonance imaging of the pelvis, and a bone scan for evaluation of metastatic disease. Recommendations regarding PSA measurements and follow-up for patients with PSA failure were left to the discretion of the treating physician. Salvage AST was recommended for patients with PSA levels of approximately 10 ng/mL in both treatment arms. Patients were followed from the date of random assignment until death or last observation through the prespecified analysis date of November 20, 2012.
Description of Study Cohort at PSA Failure Stratified by ACE-27 Comorbidity Level
Descriptive statistics were used to demonstrate patient characteristics at PSA recurrence. Chi-square analysis was performed to compare the distributions of categorical covariates among men with ACE-27 comorbidity scores 2 and 3 (moderate or severe) versus ACE-27 comorbidity scores 0 and 1 (no or minimal). Fisher's exact test was used to analyze the distributions of categorical covariates in the case of a small sample size. The nonparametric Wilcoxon test was performed to compare the continuous covariates of the median PSA velocity and age at PSA failure and distribution of median PSA level at the time of initial treatment stratified by the ACE-27 comorbidity score. Table 1 summarizes these comparisons.
Table 1. Comparison of the Distribution of Patient and Treatment Factors for the 108 Men in the Study Cohort Observed to Experience PSA Failure Stratified by the Degree of Comorbidity Using the ACE-27 Metric
To ensure a normal distribution, PSA levels were log-transformed. PSA velocity at recurrence was defined as the value of PSA velocity obtained using a linear regression[21, 22] of all PSA values after the completion treatment and prior to the time of PSA failure.
Time to ACM Analyses
Cox regression multivariable analysis was used to determine whether PSA velocity at recurrence was associated with the risk of death, adjusting for initial and salvage treatment, ACE-27 comorbidity score, pretreatment PSA level, Gleason score, tumor category, and age at and interval to PSA recurrence. Table 2 illustrates these results. Age, PSA velocity, and interval to PSA failure at PSA recurrence in addition to PSA level at initial treatment were treated as continuous variables. Gleason score (Gleason 7 versus Gleason 6 or less, Gleason 8 to 10 versus Gleason 6 or less), AJCC tumor category (T2 versus T1), treatment arm (RT alone versus RT and AST), and comorbidity ACE-27 score (moderate/severe versus no/minimal) were treated as categorical variables using established clinical cut-points. A significant interaction between comorbidity score and treatment arm was identified in a prior secondary analysis of these patients; thus, an interaction term was included in the multivariable model. To adjust for the use of AST delivered for PSA failure, a time-dependent covariate AST2 was included in the model. Unadjusted and adjusted hazard ratios (AHRs) were calculated for all covariates using the Cox proportional hazards model with associated 95% confidence interval (CI) and 2-sided P values.
Table 2. Unadjusted and Adjusted Hazard Ratios for All-Cause Mortality for Patient and Treatment Factors Following PSA Failure
For the purpose of illustration, the impact of PSA velocity on estimates of prostate cancer–specific mortality (PCSM) and ACM for each comorbidity subgroup (no/minimal versus moderate/severe) is displayed in Figures 1 through 4. PCSM and ACM were calculated from the date of PSA failure. The cumulative incidence method and the Kaplan-Meier method were used to estimate PCSM and ACM, respectively. ACM (%) was defined as 1 minus overall survival (%). The threshold for statistical significance was a 2-sided P value < .05. In estimating ACM stratified by PSA velocity among men stratified by comorbidity levels, a Bonferroni correction was used to adjust for multiple comparisons, and thus, a 2-sided P value < .025 was deemed statistically significant for this analysis. SAS version 9.3 was used for all statistical analyses (SAS Institute, Cary, NC) except for the cumulative incidence analysis, which was performed using R version 2.15.2 9 (R Foundation for Statistical Computing, Vienna, Austria).
Description of Study Cohort at PSA Failure Stratified by ACE-27 Comorbidity Level
As of November 11, 2012, 108 (52%) of 206 men entered on the randomized trial of RT with or without 6 months of AST developed PSA failure, including 85 (79%) men with no or minimal comorbidity and 23 (21%) men with moderate or severe comorbidity. Compared with men who had no or minimal comorbidity, men who had moderate or severe comorbidity were more likely to have PC with Gleason score 8 to 10 and less likely to have PC with Gleason score 7 (P = .01) (Table 1). The rest of the comparisons of the distribution of age at PSA failure (median of 75.62 years versus 74.95 years; P = .91), PSA at initial treatment (median of 12.20 ng/mL versus 12.12 ng/mL; P = .62), AJCC tumor category (P = .61), interval to PSA failure (median of 3.72 years versus 3.66 years; P = .84), PSA velocity (median of 0.65 versus 0.63; P = .63), and AST use for PSA failure (50.6% versus 47.8%; P = .82) were not significantly different in men with no or minimal comorbidity as compared to men with moderate or severe comorbidity.
Time to ACM Analyses
After a median follow-up of 10.3 years (interquartile range = 8.0-11.9), of the 108 men who experienced PSA failure, 64 (59%) men died, 22 (34%) of PC. Increasing PSA velocity at recurrence was associated with a significant increase in the risk of death (AHR = 1.21; 95% CI = 1.02-1.45; P = .03) (Table 2). Other clinical factors that were associated with a significantly increased risk of death include Gleason score of 8 to 10 (AHR = 4.51; 95% CI = 1.79-11.38; P = .001) and Gleason score of 7 (AHR = 3.53; 95% CI = 1.51-8.26; P = .004) as compared to Gleason score of 6 or less. In addition, increasing age at the time of PSA recurrence was also associated with increased risk of death (AHR = 1.06; 95% CI = 1.01-1.12; P = .03) as was an increasing interval to PSA failure (AHR = 1.02; 95% CI = 1.004-1.03; P = .009). As demonstrated in earlier studies, there was a significant interaction between initial treatment arm (RT versus RT and 6 months of AST) and comorbidity level (P < .001). Moreover, within the interaction model, men who underwent RT and AST had a significantly increased risk of death if they had moderate/severe versus no/minimal comorbidity (AHR = 17.71; 95% CI = 5.58-56.2; P < .001).
Estimates of PCSM and ACM
Among men with no or minimal versus moderate to severe comorbidity, PC comprised 42% (20 of 48) versus 12.5% (2 of 16) of all deaths, respectively. Estimates of both PCSM and ACM were significantly higher in men whose PSA velocity was greater than the median or less in men with no or minimal (P ≤ .008; Figs. 1 and 2, respectively) but not moderate or severe comorbidity (P ≥ .15; Figs. 3 and 4, respectively).
Specifically, for men with no or minimal comorbidity, the 5-year estimates of PCSM and ACM were 26.8% (14.33, 41.03) versus 0% (0, 0) and 39.0% (26.03, 55.58) versus 25.0% (13.82, 42.74), respectively, in men whose PSA velocity was greater than the median as compared to the median or less. These respective estimates were 18.2% (2.31, 46.24) versus 0% (0, 0) and 54.5% (29.31, 83.34) versus 54.5% (29.31, 83.34) in men with moderate or severe comorbidity.
In this study, we found that men with moderate or severe comorbidity did not have significantly higher estimates of PCSM or ACM following PSA failure, despite having a rapid as compared with a slow increase in their PSA level during the year prior to PSA failure. Conversely, men with no or minimal comorbidity had significantly higher estimates of both PCSM and ACM following PSA failure when their PSA rise leading to PSA failure was rapid as compared with slow. Therefore, the increase in ACM among men in otherwise good health can be explained by the marked increase in estimates of PCSM among men with a rapid as compared to a slow increase in PSA (Fig. 2). The clinical significance of these findings is 2-fold. First, given the recent RCT finding that intermittent AST was noninferior to continuous AST for men who experienced PSA failure following RT, one could justify the next RCT comparing the impact of intermittent AST as compared to surveillance in men with moderate or severe comorbidity with a prerandomization stratification for rapid versus a slow rise in PSA. It remains unclear whether any AST is beneficial in men with moderate to severe comorbidity, particularly those with a slow rise in their PSA, and therefore such a trial would be justified. Second, given the survival benefit observed in men undergoing either abiraterone or enzalutamide therapy for castration-resistant metastatic PC and the high death rate from PC in men with no or minimal comorbidity and a rapid PSA rise despite continuous AST, one could consider a randomized trial in which these men could be randomized to conventional continuous AST with or without abiraterone and/or enzalutamide.
Several points require further discussion. First, the analyses performed in this study were not prespecified and therefore are hypothesis-generating and require prospective validation. Second, men with no or minimal comorbidity were less likely to have PC with Gleason score of 8 to 10 compared to men who had moderate to severe comorbidity, as shown in Table 1, which likely reflects the practice by primary care physicians to obtain PSA for screening more often in men who are otherwise in good health as compared with those in poor health, allowing healthy men to be diagnosed earlier and therefore with lower grade disease as shown with PSA use in the PSA screening studies.[23-30] Third, a longer interval to PSA failure was associated with an increased risk of death (Table 2). Given that increased time to PSA failure is a favorable prognostic factor, one would not expect this to be due to PC deaths but competing risks. Specifically, among men with a longer interval to PSA failure (ie, > 3 years) versus a shorter interval (ie, 3 years or less, a clinically accepted cut-point) more deaths occurred (36 and 28, respectively). However, the proportion of these deaths due to PC were 5 of 36 (13.9%) patients versus 17 of 28 (60.7%) patients, respectively, indicating that most men with the longer interval to PSA failure died from causes other than PC. Fourth, men in this study received salvage AST at a PSA level of approximately 10 ng/mL, as per protocol guidelines. Whether the results of our study would have differed if salvage AST was administered at a lower PSA level or administered at the time of PSA failure requires further study. It is conceivable that earlier administration of salvage AST in men in good health may have led to lower death rate from PC; however, the significant difference in the estimates of PCSM when stratified by the median value of the PSA velocity would likely remain. Fifth, the use of salvage AST when examined as a time-dependent covariate was not significantly associated with an increased risk of death, as noted in the earlier analysis of this data. This may be explained by limited power in that of the 108 men who experienced PSA failure, 54 (50%) received salvage AST and of these 54 men, 39 died (Table 2) representing only 39 of the 64 (61%) of all observed deaths in the current study. Finally, given the advanced median age of men at PSA failure being approximately 75 and the significant toxicities associated with salvage local therapy for PSA failure, only three men with a slowly rising PSA were medically able to undergo salvage local therapy following PSA failure. In conclusion, despite unfavorable PSA kinetics at recurrence, men with moderate to severe comorbidity may not benefit from AST. RCTs examining the impact on death following intermittent AST versus surveillance are needed in these men. For men in good health and unfavorable PSA kinetics at recurrence, PC death rates are high despite continuous AST, which warrants RCTs to evaluate the impact on death of adding agents that prolong survival in men with metastatic castration-resistant PC to AST.