1624 / 1708
0.10-0.76,
P
Z
.011)
( Fig. 2,
Table 2), without a difference be-
tween treatment arms in weight decrement
( Table 2). The mean
group decrement from mid-parental height
z
-score (ie, genetic
target) was also significantly greater after HFRT, by 0.55
( Fig. 2 ).
Impact of demographic characteristics and clinical
events
After adjustment in a regression model for age, sex, and the
presence of cerebellar mutism (or, alternatively, all perioperative
complications), the association between HFRT and lower (ie,
better) executive function
z
-scores in participants of all ages was
unchanged (adjusted mean intergroup difference 0.48, 95% CI
0.15-0.80,
P
Z
.005). In a sensitivity analysis, exclusion of the 6
participants with premorbid developmental impairment did not
materially alter the effect size of treatment allocation on executive
function (mean intergroup
z
-score difference 0.44, 95% CI
0.11-0.77,
P
Z
.01).
We looked for an interaction between age category (see Pa-
tients and Methods) and the effect of treatment allocation on the
principal outcomes. These interactions, which the study was not
powered to detect, fell short of statistical significance but were
substantial for the outcomes in which a main effect of treatment
allocation was found (interaction estimates 0.62, 95% CI 0.07 to
1.30,
P
Z
.077 for executive function
z
-score; 0.48, 95% CI 0.20
to 1.16,
P
Z
.16 for height decrement
z
-score). When younger and
older participants were analyzed separately because of these in-
teractions, the effects of treatment allocation on executive function
and height decrement
z
-scores were 3-fold larger in the younger
group (
<
8.0 years at diagnosis)
( Table 3). As with the unstratified
analysis, mean intergroup
z
-score differences were not appreciably
altered by adjustment for sex and cerebellar mutism or by
exclusion of the 6 patients with premorbid developmental
impairment. The differences in executive function
z
-score were
not, however, supported by any significant differences between
treatment groups at any age with respect to scores on cognition-
related subscale measures of health status (HUI3 cognition) or
quality of life (PedsQL school functioning, QLQ-C30 cognitive
functioning) (not shown).
Hormone and other therapies, ototoxicity, and
adult social and employment outcomes
Approximately half of participants in each arm had received
growth hormone (GH) and thyroxine replacement therapies
( Table 4). In both treatment arms, state benefits were being
claimed in one-third, special educational support was required in
more than half, and the majority used therapy services and
reported problems with their appearance
( Table 4 ). Compared
with those receiving STRT, use of hearing aids was reported in a
significantly higher percentage in the HFRT group (10% and 23%,
respectively;
Table 4). This difference was, again, clearer in the
group aged
<
8 years at diagnosis (6 of 40 [15%] after STRT, 10 of
30 [33%] after HFRT) than in those older at diagnosis (4 of 51
[7.8%] after STRT, 6 of 40 [15%] after HFRT). However, neither
the HUI3 hearing attribute (mean rank single attribute function
scores 54.4 and 51.7 in the HFRT and STRT groups, respectively)
nor the previously reported audiogram data from this study (see
Table 1
Demographic characteristics and postoperative neurology in participants and nonparticipants by treatment
Characteristic
Participants
Nonparticipants
n1, n2
(HFRT, STRT)
HFRT
STRT
n1, n2
(HFRT, STRT)
HFRT
STRT
Demographic characteristics
(74, 77)
(43, 50)
Median [range] age at
diagnosis (y)
74, 77
8.7 [3.2-20.8] 9.7 [3.3-20.4]
43, 50
9.0 [4.2-17.6]
8.5 [5.0-17.8]
Median [range] age at
assessment (y)
74, 77
14.9 [7.5-29.9] 15.9 [8.6-29.6]
43, 50
15.1 [9.1-26.4
] *14.5 [10.6-23.6
] *Median [range] interval
from diagnosis (y)
74, 77
5.7 [4.2-9.9]
5.8 [4.1-9.8]
43, 50
7.1 [4.2-9.9
] *6.5 [4.2-9.9
] *Males, n (%)
74, 77
51 (69)
46 (60)
43, 50
25 (58)
29 (58)
Midparental height
z
-score (SD)
68, 71
0.09 (0.80)
-0.10 (0.90)
4, 10
NT
NT
Premorbid developmental
impairment, n (%)
47, 49
1 (1)
5 (6)
11, 18
2 (5)
0
Postoperative neurology,
n (%)
y
(74, 77)
(43, 50)
Impaired consciousness
72, 74
1 (1)
2 (3)
41, 50
2 (5)
2 (4)
Impaired nerves III, IV, VI
70, 71
32 (46)
22 (29)
37, 45
16 (37)
13 (26)
Ataxia
64, 68
36 (49)
37 (48)
29, 39
19 (44)
23 (46)
Cerebellar mutism
74, 77
6 (8)
4 (5)
43, 50
6 (14)
3 (6)
Abbreviations:
HFRT
Z
hyperfractionated radiation therapy (see Patients and Methods); NT
Z
not tabulated as insufficient data; STRT
Z
standard
radiation therapy (see Methods and Materials).
* For nonparticipants, median age at assessment and interval from diagnosis was estimated using January 1, 2011, the midpoint of the assessment
period, as the notional assessment date.
y
Percentages are expressed as a percentage of total number of participants or nonparticipants, using conservative assumption that the feature was not
present in cases not reported.
Volume 88 Number 2 2014
Quality of survival in the PNET4 study
295