Intermittent Androgen Deprivation for Prostate Cancer
Intermittent Androgen Deprivation for Prostate Cancer
The trial flow is provided in Figure 1. Initial search criteria found 3070 potential reports for inclusion, of which 257 were selected for abstract review and 20 were selected for full text analysis. Most did not fulfil the selection criteria, as they were either (i) not randomised control trials, or (ii) did not directly compare IAD with CAD.
(Enlarge Image)
Figure 1.
Study selection flow diagram. HR, hazard ratio.
Of the 20 reports selected for full text review, seven did not meet the inclusion criteria as four reports were abstract forms of later full published studies, and three did not examine appropriate comparators.
Thirteen studies (5767 patients) fulfilled the criteria for inclusion in the systematic review. Five studies (1099 patients) included in the systematic review did not have published data suitable for inclusion in the pooled analysis. Miller et al., Verhagen et al., and Langenhuijsen et al. were only available as abstracts, and did not publish sufficient data to calculate the HR. Yamanaka et al. only contained interim data with no progression or survival events reported. Hering et al. had a low number of events, and did not publish sufficient data to calculate a HR.
The characteristics of the included studies are summarised in Table 1 . The range of median follow-up was 29–118 months. Study sample size ranged from 31 to 1535 patients. Most studies used a drug combination of gonadotrophin agonists (e.g., goserelin, leuprolide) and oral antiandrogens (e.g., bicalutamide). The study populations were heterogeneous with respect to prostate cancer stage: six studies included locally advanced prostate cancer with and without metastatic disease; two studies included patients with rising PSA after local therapy; four studies included only patients with radiologically confirmed metastatic disease, while one study included any stage of prostate cancer.
All studies except Crook et al. included an induction period of hormone blockade. The most common PSA cutoff point for commencing 'off-treatment' periods in the IAD arm was <4 ng ml, whereas the most common cutoff to start 'on treatment' was >10 ng ml. One study used a fixed treatment interval of 6 months 'off treatment' and 6 months 'on treatment' in the IAD arm. In all studies, the CAD arm hormone treatment regimen was the same as the induction hormone regimen.
The risk of bias summary is included in Supplementary 1. Only one of the included studies provided information regarding random sequence generation for patient randomisation, and no studies reported allocation concealment or blinding of participants and personnel. No study reported blinding of outcome assessor, though in four studies the measured outcomes (OS, PSA progression) were objective and unlikely to be affected by unblinded assessment. Three studies were deemed at risk of attrition bias as they did not report losses to follow-up. All other included studies had acceptable losses to follow-up, and five studies reported intention-to-treat analysis. There were no studies at high risk of selective reporting bias.
There was no difference in all-cause mortality between IAD and CAD (6 studies, 4399 participants, HR 1.01, 95% CI 0.93–1.10, P=0.81) (Figure 2). Pre-specified subgroup analysis did not show any significant difference in all-cause mortality between IAD and CAD in the subgroups of patients with metastatic disease (HR 1.04, 95% CI 0.91–1.19, P=0.58) or in patients with no metastatic disease (HR 1.06, 95% CI 0.91–1.23, P=0.43) (test for subgroup differences χ=0.04, P=0.84).
(Enlarge Image)
Figure 2.
Pooled estimate of hazard ratio (HR) for overall survival. The square on each bar represents the HR for an individual trial, and the bar shows the 95% confidence interval (CI). Analysis performed with inverse variance (IV) weighting. The diamond represents a pooled estimate, with the centre of the diamond giving the HR estimate and the extremes of the diamond representing the 95% CI. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation.
There was no difference in prostate cancer mortality between IAD and CAD (4 studies, 2695 participants, HR 1.03, 95% CI 0.88–1.21, P=0.74) (Figure 3).
(Enlarge Image)
Figure 3.
Pooled estimate of hazard ratio for cancer-specific survival. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation. Analysis performed with inverse variance (IV) weighting.
The criteria used to define progression varied between studies. All studies included PSA in criteria for progression, although cutoffs varied between studies. Other criteria included performance status, metastatic progression, pain, and weight loss due to cancer. Two studes did not report the criteria used for progression. There was no difference in prostate cancer progression between IAD and CAD (7 studies, 3133 participants, HR 0.93, 95% CI 0.84–1.04, P=0.19) (Figure 4). There was moderate heterogeneity (I=57%) between studies, which may be due to variations between studies in the definition of progression of prostate cancer.
(Enlarge Image)
Figure 4.
Pooled estimate of hazard ratio for progression-free survival. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation. Analysis performed with inverse variance (IV) weighting.
There were insufficient data to calculate an overall HR for mortality unrelated to prostate cancer. Five studies published sufficent data to calculate a risk ratio (RR). Although there was a trend for reduced mortality from other causes with IAD (RR 0.90, 95% CI 0.80–1.01) (Figure 5), this result did not attain significance (P=0.07).
(Enlarge Image)
Figure 5.
Pooled estimate of relative risk for mortality unrelated to prostate cancer using Mantel-Haenszel (M-H) method. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation.
Eleven studies published QOL or toxicity outcomes. There were insufficient data for quantititive analysis. The qualitative results are summarised in Table 2 .
Results
Studies
The trial flow is provided in Figure 1. Initial search criteria found 3070 potential reports for inclusion, of which 257 were selected for abstract review and 20 were selected for full text analysis. Most did not fulfil the selection criteria, as they were either (i) not randomised control trials, or (ii) did not directly compare IAD with CAD.
(Enlarge Image)
Figure 1.
Study selection flow diagram. HR, hazard ratio.
Of the 20 reports selected for full text review, seven did not meet the inclusion criteria as four reports were abstract forms of later full published studies, and three did not examine appropriate comparators.
Thirteen studies (5767 patients) fulfilled the criteria for inclusion in the systematic review. Five studies (1099 patients) included in the systematic review did not have published data suitable for inclusion in the pooled analysis. Miller et al., Verhagen et al., and Langenhuijsen et al. were only available as abstracts, and did not publish sufficient data to calculate the HR. Yamanaka et al. only contained interim data with no progression or survival events reported. Hering et al. had a low number of events, and did not publish sufficient data to calculate a HR.
The characteristics of the included studies are summarised in Table 1 . The range of median follow-up was 29–118 months. Study sample size ranged from 31 to 1535 patients. Most studies used a drug combination of gonadotrophin agonists (e.g., goserelin, leuprolide) and oral antiandrogens (e.g., bicalutamide). The study populations were heterogeneous with respect to prostate cancer stage: six studies included locally advanced prostate cancer with and without metastatic disease; two studies included patients with rising PSA after local therapy; four studies included only patients with radiologically confirmed metastatic disease, while one study included any stage of prostate cancer.
All studies except Crook et al. included an induction period of hormone blockade. The most common PSA cutoff point for commencing 'off-treatment' periods in the IAD arm was <4 ng ml, whereas the most common cutoff to start 'on treatment' was >10 ng ml. One study used a fixed treatment interval of 6 months 'off treatment' and 6 months 'on treatment' in the IAD arm. In all studies, the CAD arm hormone treatment regimen was the same as the induction hormone regimen.
Bias Risk
The risk of bias summary is included in Supplementary 1. Only one of the included studies provided information regarding random sequence generation for patient randomisation, and no studies reported allocation concealment or blinding of participants and personnel. No study reported blinding of outcome assessor, though in four studies the measured outcomes (OS, PSA progression) were objective and unlikely to be affected by unblinded assessment. Three studies were deemed at risk of attrition bias as they did not report losses to follow-up. All other included studies had acceptable losses to follow-up, and five studies reported intention-to-treat analysis. There were no studies at high risk of selective reporting bias.
Overall Survival
There was no difference in all-cause mortality between IAD and CAD (6 studies, 4399 participants, HR 1.01, 95% CI 0.93–1.10, P=0.81) (Figure 2). Pre-specified subgroup analysis did not show any significant difference in all-cause mortality between IAD and CAD in the subgroups of patients with metastatic disease (HR 1.04, 95% CI 0.91–1.19, P=0.58) or in patients with no metastatic disease (HR 1.06, 95% CI 0.91–1.23, P=0.43) (test for subgroup differences χ=0.04, P=0.84).
(Enlarge Image)
Figure 2.
Pooled estimate of hazard ratio (HR) for overall survival. The square on each bar represents the HR for an individual trial, and the bar shows the 95% confidence interval (CI). Analysis performed with inverse variance (IV) weighting. The diamond represents a pooled estimate, with the centre of the diamond giving the HR estimate and the extremes of the diamond representing the 95% CI. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation.
Prostate Cancer-specific Survival
There was no difference in prostate cancer mortality between IAD and CAD (4 studies, 2695 participants, HR 1.03, 95% CI 0.88–1.21, P=0.74) (Figure 3).
(Enlarge Image)
Figure 3.
Pooled estimate of hazard ratio for cancer-specific survival. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation. Analysis performed with inverse variance (IV) weighting.
Time to Prostate Cancer Progression
The criteria used to define progression varied between studies. All studies included PSA in criteria for progression, although cutoffs varied between studies. Other criteria included performance status, metastatic progression, pain, and weight loss due to cancer. Two studes did not report the criteria used for progression. There was no difference in prostate cancer progression between IAD and CAD (7 studies, 3133 participants, HR 0.93, 95% CI 0.84–1.04, P=0.19) (Figure 4). There was moderate heterogeneity (I=57%) between studies, which may be due to variations between studies in the definition of progression of prostate cancer.
(Enlarge Image)
Figure 4.
Pooled estimate of hazard ratio for progression-free survival. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation. Analysis performed with inverse variance (IV) weighting.
Mortality Unrelated to Prostate Cancer
There were insufficient data to calculate an overall HR for mortality unrelated to prostate cancer. Five studies published sufficent data to calculate a risk ratio (RR). Although there was a trend for reduced mortality from other causes with IAD (RR 0.90, 95% CI 0.80–1.01) (Figure 5), this result did not attain significance (P=0.07).
(Enlarge Image)
Figure 5.
Pooled estimate of relative risk for mortality unrelated to prostate cancer using Mantel-Haenszel (M-H) method. CAD, continuous androgen deprivation; IAD, intermittent androgen deprivation.
QOL and Toxicity
Eleven studies published QOL or toxicity outcomes. There were insufficient data for quantititive analysis. The qualitative results are summarised in Table 2 .
