Relationship between prostate cancer mortality and number of unfavourable risk factors in men treated with definitive brachytherapy

Authors


Gregory S. Merrick, Schiffler Cancer Center, Wheeling Hospital, 1 Medical Park, Wheeling WV 26003, USA.
e-mail: gmerrick@urologicresearchinstitute.org

Abstract

Study Type – Therapy (case series)
Level of Evidence 4

OBJECTIVE

To explore whether the number of unfavourable pretreatment risk factors predicts cause-specific mortality in men treated with prostate brachytherapy.

PATIENTS AND METHODS

Between April 1995 and March 2006, 739 patients were treated who had at least one of the following adverse risk factors: pretreatment prostate-specific antigen (PSA) level of >10 ng/mL, a Gleason score of ≥7, clinical stage ≥T2b, or a PSA velocity (PSAV) of >2 ng/mL/year. Supplemental external beam radiotherapy (EBRT) was delivered to 464 (62.8%) men and 301 (40.7%) received androgen deprivation therapy (ADT). Of men with more than two risk factors, 87% received EBRT and 62% received ADT.

RESULTS

The biochemical progression-free survival (bPFS), cause-specific survival (CSS) and overall survival for all patients were 95.0%, 97.9% and 70.0% at 12 years. Men with three or four risk factors had a prostate cancer-specific mortality (PCSM) at 12 years of 5.3%, vs 1.7% for men with one or two risk factors (P= 0.006). When ‘percentage of positive biopsy cores >50%’ replaced PSAV as a risk factor, men with two or more risk factors had a PCSM of 8.9%, vs 1.0% for men with one or two risk factors (P= 0.001). There was no difference in all-cause mortality between the groups in either analysis.

CONCLUSION

Multimodal brachytherapy results in high rates of bPFS and CSS, even for men with several unfavourable risk factors. Men with two or more unfavourable risk factors had a slightly greater risk of PCSM and no difference in all-cause mortality. The presence of three or four unfavourable intermediate-risk factors does not appear to clearly identify a group that requires further treatment intensification, although the percentage of positive cores might be more predictive than PSAV.

Abbreviations
CSS

cause-specific survival

bPFS

biochemical progression-free survival

OS

overall survival

PSAV

PSA velocity

PCSM

prostate cancer-specific mortality

ACM

all-cause mortality

EBRT

external beam radiotherapy

ADT

androgen-deprivation therapy

%+BC

percentage of positive biopsy cores.

INTRODUCTION

Clinical tumour stage, Gleason score and pretreatment PSA level have been identified as the primary risk factors for disease recurrence [1,2] and cause-specific survival (CSS) [3] in men treated with definitive therapy for prostate cancer. Men with low-risk disease based on these factors have excellent long-term outcomes [4–6]. Men with intermediate-risk disease, based on the presence of one of these risk factors, tend to have very good outcomes, although not as good as those in low risk individuals [4,6–8]. Another unfavourable risk factor, an elevated PSA velocity (PSAV), has also been found to significantly predict an increased prostate cancer-specific mortality (PCSM) [9,10]. Men with several unfavourable risk factors, including elevated PSAV, have been found to be at significantly greater risk of treatment failure [11–14].

Recent studies [11,15] suggest that men with three or four of the above-mentioned risk factors and treated with external beam radiotherapy (EBRT) or prostatectomy are at particularly high risk of death from prostate cancer. Nguyen et al.[15] found that prostate cancer deaths accounted for 50–80% of total deaths in this group of men. The implication is that these men might require novel therapies, beyond existing standard treatments, to improve the potentially poor prognosis.

In the present study we evaluated the relationship between prostate cancer mortality rate and number of unfavourable risk factors in men treated with definitive brachytherapy. Our goal was to assess whether the presence of three or four unfavourable risk factors in men treated with primary brachytherapy has the same prognostic value as in those treated with EBRT or surgery.

PATIENTS AND METHODS

Between April 1995 and March 2006, 739 patients were treated with permanent prostate brachytherapy and had at least one of the following adverse risk factors: a pretreatment PSA level of >10 ng/mL, a Gleason score of ≥7, clinical stage ≥T2b, or a PSAV of >2 ng/mL/year. Of the 739 patients, 464 (62.8%) received supplemental EBRT. In general, EBRT consisted of 45 Gy delivered in 1.8-Gy fractions with 15–18 MV photons delivered via a conformal technique that used four fields (opposed laterals and anterior–posterior and posterior–anterior) with custom treatment devices to spare as much normal tissue as possible. For patients with a ≤10% risk of pelvic lymph node involvement [16], the target volume consisted of the prostate gland and seminal vesicles. For patients with a >10% risk of pelvic lymph node involvement, the pelvic lymph nodes were also included in the target volume. In all cases, supplemental EBRT was delivered before brachytherapy.

Of the 739 patients, 301 (40.7%) received androgen-deprivation therapy (ADT); 199 received ≤6 months of ADT, whereas 102 received >6 months of ADT. In patients receiving ADT, the ADT was initiated 3 months before seed implantation and consisted of a LHRH agonist and an antiandrogen.

The brachytherapy target volume consisted of the prostate gland with periprostatic treatment margins including the proximal 1.0 cm of the seminal vesicles [17,18]. The minimum peripheral dose was prescribed to the target area with a margin. Of the 739 patients, 646 (87.4%) had implantation with 103Pd and 93 (12.6%) with 125I. At implantation, the prostate gland, periprostatic region and base of the seminal vesicles were implanted [17,18].

Patients were monitored by physical examination, including a DRE, and serum PSA determination at 3- and 6-month intervals. The endpoint of the analysis was cumulative PCSM, biochemical progression-free survival (bPFS) and overall survival (OS). Several clinical, treatment and dosimetric variables were evaluated for their effect on survival. The cause of death was determined for each patient who died. Patients with metastatic prostate cancer or hormone-refractory disease with no obvious metastases, and who died of any cause, were classified as dead from prostate cancer; all other deaths were attributed to the immediate cause of death. Biochemical failure was determined using a threshold definition of a PSA level of ≥0.4 ng/mL after the nadir. Patients who had a nadir of >0.4 ng/mL or those whose PSA increased to >0.4 ng/mL after a nadir below that level were considered to have had a biochemical failure.

Kaplan-Meier analyses were used to compare cumulative PCSM across risk categories. Univariate Cox regression analysis was used to determine univariate predictors of bPFS, CSS and OS. Variables examined included pre-implantation PSA level, Gleason score, clinical stage, pretreatment PSAV, number of adverse risk factors, age at implantation, duration of follow-up, percentage of positive biopsy cores (%+BC), body mass index, prostate volume, the percentage of the target volume receiving 100% of the prescription dose (V100), the minimum dose as a percentage of the prescription dose, which covered at least 90% of the target volume, D90), perineural invasion, use of supplemental EBRT, use of ADT, hypertension, diabetes, and tobacco use. Variables with P < 0.10 were then included in a multivariate model of a Cox regression as a means of identifying multiple predictors. For all tests, P≤ 0.05 was considered to indicate significance.

RESULTS

Table 1 summarizes the clinical and treatment characteristics of the 739 men with one or more of the following unfavourable risk factors: PSA level >10 ng/mL, Gleason score ≥7, clinical stage ≥T2b, or a PSAV of >2 ng/mL/year. The median age was 67 years and the median follow-up was 6.7 years. Supplemental EBRT was used in 63% of the men and ADT in 41%. The median D90 after implantation at day 0 was 120% of the prescribed dose.

Table 1. 
Continuous clinical, treatment, and dosimetric variables in all 739 patients, and the distribution of unfavourable risk factors
VariableValue
  • *

    Unavailable for four patients.

Continuous; median, mean (sd)
Age at implant 67.0, 66.6 (7.0)
Follow-up, years  6.7, 6.9 (2.9)
Pre-treatment PSA level, ng/mL  6.9, 8.3 (5.4)
Gleason score  7.0, 6.9 (0.8)
%+BC 33.3, 39.6 (24.8)
Body mass index, kg/m2 27.7, 28.5 (4.6)
Prostate volume, mL 33.5, 33.9 (9.5)
V100, % 98.0, 96.6 (4.7)
V150, % 71.3, 68.0 (14.0)
V200, % 40.3, 38.5 (12.6)
D90, %120.0, 119.5 (13.7)
Most recent PSA level, ng/mL  0.02, 0.04 (0.08)
Categorical, n (%) 
 Clinical stage; <T2b630 (85.3)
  ≥T2b109 (14.7)
 Pretreatment PSA level, ng/mL; ≤10586 (79.3)
  >10153 (20.7)
 Pretreatment PSAV, ng/mL/year; ≤2389 (52.6)
  >2350 (47.4)
 Gleason score; <7183 (24.8)
  ≥7556 (75.2)
 % positive biopsies; ≤50561 (76.3)
  >50174 (23.7)
 Isotope; 125I 93 (12.6)
  103Pd646 (87.4)
 EBRT; No275 (37.2)
  Yes464 (62.8)
 ADT, months; 0438 (59.3)
  ≤6199 (26.9)
  >6102 (13.8)
 Risk; intermediate495 (67.0)
  high244 (33.0)
 Perineural invasion; No486 (65.9)
  Yes251 (34.1)
 Hypertension; No377 (46.5)
  Yes395 (53.5)
 Diabetes; No632 (85.6)
  Yes106 (14.1)
 Tobacco use; Never273 (36.9)
  Former364 (49.3)
  Current102 (13.8)
Unfavourable risk factors 
 Clinical stage; <T2b630 (85.3)
  ≥T2b109 (14.7)
 Pretreatment PSA, ng/mL; ≤10586 (79.3)
  >10153 (20.7)
 Pre-PSA velocity; ≤2389 (52.6)
  >2350 (47.4)
 Gleason score; <7183 (24.8)
  ≥7556 (75.2)
 %+BC*; ≤50561 (76.3)
  >50174 (23.7)

Table 1 also shows the distribution of unfavourable risk factors; a PSAV of ≥2.0 ng/mL/year and Gleason score >7 were the two most prevalent risk factors. Table 2 stratifies the patients by the number of unfavourable risk factors and details the percentage of men receiving EBRT and ADT by group. Men with three or four risk factors were more likely to receive supplemental EBRT and/or ADT than those with only one or two risk factors. Table 2 also shows the distribution of risk factors stratified by year of treatment.

Table 2.  The prevalence of ADT and EBRT with number of risk factors, and number of risk factors with year of implant
n (%) variableRisk factorsP *
1234
  • *

    Chi-square;

  • Fisher’s exact test, P= 0.004.

No. of patients4272069511
ADT140 (32.8) 95 (46.1)61 (64.2) 5/11<0.001
EBRT220 (51.5)152 (73.8)82 (86.3)10/11<0.001
Year of implant     
1995–2000125 (50.6) 79 (32.0)35 (14.2) 8 (3.2) 
2001–2006302 (61.4)127 (25.8)60 (12.2) 3 (0.6) 

Figure 1 shows the Kaplan-Meier curves for bPFS, CSS and OS for all the patients, with rates of 95.0%, 97.9% and 70.0%, respectively, at 12 years. The Gleason score (as an ordinal rather than binary variable) and %+BC were the strongest predictors of bPFS (Table 3). More unfavourable risk factors (two or fewer vs more than two) was not associated with increased risk of biochemical failure. None of the variables examined predicted the CSS rate. Age and tobacco consumption were predictors of all-cause mortality (ACM, Table 3). None of the other variables examined (listed in the Methods) predicted the ACM. Although Gleason score and %+BC were predictors of biochemical recurrence, on Kaplan-Meier analysis they had no significant effect on ACM or PCSM.

Figure 1.

Kaplan-Meier curves for CSS, bPFS and OS; percentages provided in parentheses are 5-year values, with 523 patients at 5 years.

Table 3.  Predictors for biochemical failure and ACM (Cox regression analysis)
VariablesUnivariate PHazard ratio (95% CI)Multivariate PHazard ratio (95% CI)
Biochemical failure
Pre-implant PSA level (≤10 vs >10)0.8483.324 (1.012–10.920)0.917 
Gleason score (<7 vs ≥7)0.0482.931 (1.410–6.091)0.149 
Stage (<T2b vs ≥T2b)0.004   
Pre-implant PSAV (≤ 2 vs >2)0.525 0.795 
Risk factors (≤2 vs >2)0.0892.003 (1.401–2.865)0.0031.768 (1.215–2.573)
Gleason score<0.0011.025 (1.012–1.037)0.0031.019 (1.006–1.032)
%+BC<0.0012.525 (1.039–6.136)0.461 
EBRT (yes vs no)0.0410.399 (0.189–0.842)0.0140.393 (0.186–0.831)
Hypertension (yes vs no)0.016   
ACM
Pre-implant PSA (≤10 vs >10)0.194   
Gleason score (<7 vs ≥7)0.978   
Stage (<T2b vs ≥T2b)0.156   
Pre-implant PSAV (≤2 vs >2)0.517   
Risk factors (≤2 vs >2)0.7521.099 (1.067–1.133)<0.0011.110 (1.076–1.144)
Age at implant<0.001 <0.001 
Tobacco (never vs former vs current)0.0060.476 (0.293–0.774)<0.0010.372 (0.228–0.607)
Current (0) vs never (1)0.0030.542 (0.350–0.840)<0.0010.387 (0.248–0.604)
Current (0) vs former (1)0.006   

Figure 2 shows the cumulative PCSM stratified by the number of unfavourable pretreatment risk factors. Men with three or four risk factors had a PCSM rate of 5.3% at 12 years, compared to 1.7% for men with one or two risk factors (P= 0.006). The number of risk factors had no effect on cumulative ACM (Fig. 3).

Figure 2.

Cumulative hazard function for PCSM, stratified by the number of risk factors. Risk factors are PSA level >10 ng/mL, Gleason score ≥7, clinical stage ≥T2b, and %+BC >50. Percentages in parentheses are the 5-year values.

Figure 3.

Cumulative hazard function for ACM stratified by the number of risk factors (details as in Fig. 2).

Because, on multivariate analysis, the %+BC seemed to be more predictive of treatment outcome than PSAV, we conducted an additional analysis replacing PSAV with %+BC as an adverse risk factor (keeping pretreatment PSA level, Gleason score and clinical stage as the other three risk factors). Figure 4 shows the cumulative PCSM stratified by number of the revised pretreatment unfavourable risk factors, with %+BC included. Men with three or four of these risk factors had a PCSM rate of 8.9% at 12 years, compared to 1.0% for men with one or two risk factors (P= 0.001). The number of risk factors had no effect on the cumulative ACM in this additional analysis (Fig. 5).

Figure 4.

Cumulative hazard function for PCSM stratified by number of risk factors (details as in Fig. 2).

Figure 5.

Cumulative hazard function for overall mortality stratified by number of risk factors (details as in Fig. 2).

DISCUSSION

Over the last decade, significant progress has been made in identifying men with localized prostate cancer who benefit from intensification of their initial definitive therapy. Three large randomized trials [19–21] reported notably superior outcomes when ADT was combined with definitive EBRT in higher-risk patients. Although there are no similar randomized trials in higher-risk brachytherapy patients, good results from retrospective studies [22–24] led to intensification of therapy for many of these men, with the addition of supplemental EBRT and/or ADT to their primary brachytherapy.

Despite these advances, prostate cancer continues to be a leading cause of cancer death in men in the USA [25]. Although chemotherapy has historically been limited to the metastatic setting for prostate cancer, investigators recently began introducing chemotherapy into the definitive setting in an effort to further reduce mortality among high-risk men. The Radiation Therapy Oncology Group trial 99–02 examined adding paclitaxel, estramustine and etoposide to definitive EBRT and ADT in high-risk patients [26], although this trial was closed early due to increased treatment toxicity of the regimen. Radiation Therapy Oncology Group trial 0521 is currently addressing a similar issue, with a change in the chemotherapy regimen to docetaxel and prednisone, in an effort to reduce toxicity.

The primary objective of the present study was to assess whether the number of unfavourable risk factors (pretreatment PSA level, Gleason score, clinical stage, and PSAV) could be useful for identifying men treated with primary brachytherapy who would be candidates for further treatment intensification. There was a relatively small absolute difference (3.6%) in 12-year PCSM for men with one or two of our original unfavourable risk factors compared to men with three or four of these risk factors. Although the difference was statistically significant, the absolute rate of PCSM at 12 years for men with three or four risk factors was still relatively low, at 5.3%.

Our results differ noticeably from those published by Nguyen et al.[15] for patients treated with prostatectomy and EBRT. In that study there was a large absolute difference in PCSM between men with one or two unfavourable risk factors and those with three or four risk factors. The 8-year PCSM was <10% for men with only one or two risk factors compared to >40% for men with more than three or four (P < 0.001). This led to the conclusion that men with more than two unfavourable risk factors ‘should be considered for clinical trials designed to assess whether survival is prolonged with the addition of novel agents to current standards of practice’.

The reason for the seemingly discordant findings in these two studies might be that the patients who had EBRT in the study of Nguyen et al. received what currently would be considered a substandard dose, particularly for high-risk patients, of 70.2 Gy. When the subset of EBRT patients in that study who received ADT as a radiation sensitizer in addition to 70.2 Gy were evaluated, the absolute PCSM and the differences between men with more or fewer than two risk factors was much smaller (0% vs ≈10% PSCM at 8 years, P= 0.04). With higher radiation doses, it is possible that the overall PSCM might decrease even more and the predictive effect of more risk factors be potentially further diminished.

Recent reports have raised questions about the predictive value of pretreatment PSAV as an adverse risk factor before initial definitive therapy [27,28]. Our primary analysis, which included PSAV as a risk factor, did not identify a large absolute difference in PCSM between the groups. In our subsequent analysis we found that replacing PSAV with %+BC increased the discriminatory power of the algorithm. The 12-year PCSM was 8.9% for men with three or four risk factors, compared to 1.0% for men with only one or two (P < 0.001). However, this did not translate into a difference in ACM.

High-risk prostate cancer is often defined as a PSA leve of >20 ng/mL, a Gleason score of ≥8, clinical stage >T2b, or more than one unfavourable intermediate-risk factor (PSA level of >10–<20 ng/mL, Gleason score = 7, or clinical T2b). The present study and others [22,23] showed that high-risk men, so defined, treated with high-quality brachytherapy implants, have excellent outcomes and a low risk of PCSM. Based on our analysis, men undergoing primary brachytherapy with three or four unfavourable intermediate-risk factors (PSA level >10 ng/mL, Gleason score ≥7, clinical stage ≥T2b, or PSAV >2 ng/mL/year) are at only a slightly greater risk of PCSM than men with fewer unfavourable risk factors. It is not clear that setting a threshold of fewer than two intermediate-risk factors is the best way to identify the subset of patients having primary brachytherapy who are at greatest risk of PCSM and would benefit from treatment intensification with additional systemic agents. Because of the toxicity associated with many of the chemotherapeutic agents, a risk stratification that more narrowly identified a set of patients with a higher predicted mortality rate would be useful. Including the %+BC of >50 as an adverse risk factor might be of benefit.

Because only a few men in the present cohort had all four unfavourable risk factors, we were unable to further stratify outcomes to assess whether there was a prognostic difference between three or fewer risk factors and all four factors. Of men in the study <2% had all four unfavourable factors, so it would probably require a very large sample to discern any relationship if there is one. In addition, there might be other variables which we did not consider that in combination with the variables examined, or perhaps replacing one of the variables included, would help to better identify a subset of patients with a much higher risk of CSM. However, the variables that we included are those that have been most consistently validated in previous reports having prognostic significance.

As described in Table 2, men with three or four risk factors were much more likely to receive intensified treatment which included ADT and EBRT. In the absence of this treatment intensification, it is possible that these men would have had notably higher CSM rates. The relatively modest differences in CSM between men with one or two and those with three or four factors probably reflects the efficacy of existing treatment intensification strategies more than an underlying lack of risk of prostate cancer death in men with three or four risk factors.

In conclusion, men with higher-risk prostate cancer treated with quality primary brachytherapy are at a relatively low risk of PCSM. The presence of three or four unfavourable intermediate-risk factors does not appear to clearly identify a group who will benefit from further treatment intensification. Additional research might be warranted to identify the minority of men who are at highest risk of failure with the current multimodal brachytherapy treatment regimens.

CONFLICT OF INTEREST

None declared.

Ancillary