537 692
Data were analysed using SAS 9.3 (SAS Institute, Cary, NC, USA) and
p
<
0.05 was considered nominally statistically significant without
multiplicity adjustment. Adjustments for multiple comparisons were
not made because these were exploratory analyses.
3.
Results
In TERRAIN, 375 patients were randomised (March 22,
2011 to July 11, 2013) to enzalutamide (
n
= 184) or
bicalutamide (
n
= 191) and were included in PRO analyses.
Patients received enzalutamide and bicalutamide for a
median of 11.7 mo and 5.8 mo, respectively
[17]. Groups
were well balanced for baseline demographic and disease
characteristics (Supplementary Table 3)
[10].
3.1.
FACT-P and EQ-5D
Adjusted completion rates, based on patients remaining
on study, exceeded 90% in both arms at all time points for
both questionnaires (Supplementary Table 4). At 61 wk,
84 (46%) enzalutamide patients and 39 (20%) bicaluta-
mide patients were evaluable (Supplementary Fig. 1).
Baseline HRQoL scores were similar between treatments
( Table 2 ).
A decline in scores was observed for all FACT-P domains
for both treatments, except emotional well-being (EWB)
with enzalutamide. The decline was clinically meaningful
(exceeding lower bound of MCID range;
Table 1) in MMRM
analyses for two (Trial Outcome Index [TOI] and prostate
cancer subscale [PCS] pain-related) and eight (physical
well-being, functional well-being, PCS, PCS pain-related,
FACT Advanced Prostate Symptom Index [FAPSI-8], TOI,
FACT–General [FACT-G], and FACT-P total) of 10 domains
with enzalutamide and bicalutamide, respectively
( Fig. 1). A
decline in HRQoL for all FACT-P domains using PMM
analysis was greater than for MMRM analyses. This is
expected since PMM assumes that patients discontinuing
treatment continue to experience a decline in HRQoL
(Supplementary data).
Changes from baseline at 61 wk significantly favoured
enzalutamide (
p
<
0.05) on three FACT-P domains
(EWB, FAPSI-8, and FACT-P) in MMRM analyses and seven
(functional well-being, EWB, PCS, FAPSI-8, TOI, FACT-G, and
FACT-P) in PMM analyses
( Fig. 1 ;Supplementary
Tables 5 and 6). No changes from baseline at 61 wk in
any FACT-P domain favoured bicalutamide over enzaluta-
mide in MMRM or PMM analyses.
EQ-5D visual analogue scale scores were maintained
with no clinically meaningful deterioration at any assess-
ment time point with either treatment in MMRM analyses
( Fig. 2 ). In contrast, EQ-5D utility scores declined over time
with both treatments. However, with enzalutamide, EQ-5D
utility scores were maintained with no clinically meaning-
ful decrease up to 49 wk, whereas with bicalutamide,
clinically meaningful deterioration occurred as early as
13 wk. Between-treatment comparison of changes from
baseline showed no clinically meaningful or statistically
significant differences at 61 wk. Results are confirmed with
PMM analysis
( Fig. 2 ;Supplementary Tables 7 and 8).
The change from baseline for FACT-P and EQ-5D
appeared to favour enzalutamide over bicalutamide in
most subscales in time points prior to 61 wk
( Figs. 1 and 2).
Statistical comparison was only conducted at the pre-
planned 61-wk time point.
Risk of first deterioration in HRQoL outcome (HR, 95% CI)
was significantly lower with enzalutamide for FACT-P total
score (HR: 0.64, 95% CI: 0.46–0.89,
p
= 0.007), FACT-G total
score (HR: 0.70, 95% CI: 0.50–0.98,
p
= 0.04), PCS pain-
related (HR: 0.74, 95% CI: 0.54–1.00,
p
= 0.048), and EQ-5D
index (HR: 0.66, 95% CI: 0.47–0.93,
p
= 0.02) versus
bicalutamide
( Table 3 ). Sensitivity analysis, considering
progression as an event in addition to first deterioration,
showed significantly lower risk on all HRQoL outcomes with
enzalutamide (Supplementary Table 9).
3.2.
BPI-SF
Adjusted completion rates, based on patients remaining on
study, exceeded 90% in both arms at all time points
analysed. BPI-SF baseline values were
<
2 and well balanced
between treatments, although slightly higher with enzalu-
tamide
( Table 2).
Pain at its worst (Item 3) and composite scores of pain
severity and pain interference increased throughout the
study in both groups
( Fig. 3), which reflects increased pain
progression. At 61 wk (Supplementary Tables 10 and 11),
mean increases from baseline were consistently smaller for
enzalutamide versus bicalutamide for all items, but only
significantly smaller for pain interference composite
(0.70 vs 1.62;
p
= 0.01; MMRM analysis), increases in pain
Table 2 – Health-related quality of life (HRQoL) baseline scores
HRQoL baseline scores, mean (standard deviation)
Outcome
Enzalutamide
(
n
= 184)
Bicalutamide
(
n
= 191)
FACT-P
Physical well-being
23.6 (4.13)
23.9 (4.37)
Functional well-being
20.0 (5.20)
20.3 (6.05)
Emotional well-being
18.3 (3.93)
18.8 (3.64)
Social well-being
22.2 (5.10)
22.4 (4.34)
Prostate cancer subscale
32.4 (6.98)
33.0 (6.85)
PCS pain-related score
11.3 (4.09)
11.6 (4.21)
FAPSI-8
23.9 (5.24)
24.3 (5.07)
Trial Outcome Index
76.0 (14.09)
77.1 (15.28)
FACT-G total score
84.0 (13.26)
85.2 (13.38)
FACT-P total score
117 (18.34)
118 (18.75)
EQ-5D
EQ-5D utility index
0.81 (0.20)
0.83 (0.18)
EQ-5D VAS
77.7 (15.48)
76.9 (17.73)
BPI
Item 3: pain at its worst
1.43 (1.66)
1.08 (1.32)
Item 4: pain at its least
0.60 (1.11)
0.56 (1.03)
Item 5: pain on average
1.28 (1.65)
1.15 (1.55)
Item 6: pain now
0.70 (1.20)
0.59 (1.13)
Pain severity score
0.99 (1.22)
0.84 (1.12)
Pain interference score
1.01 (1.56)
0.95 (1.76)
BPI = Brief Pain Inventory; EQ-5D = European Quality of Life 5-Domain
Scale; FACT-G = Functional Assessment of Cancer Therapy–General; FACT-
P = Functional Assessment of Cancer Therapy–Prostate; HRQoL = health-
related quality of life; MCID = minimal clinical important difference;
PCS = prostate cancer subscale; VAS = visual analogue scale
.
E U R O P E A N U R O L O G Y 7 1 ( 2 0 1 7 ) 5 3 4 – 5 4 2
537