1452 / 1708
Articles
http://oncology.thelancet.comVol 8 August 2007
701
Median time to progression for the 59 patients who
progressed was 1·6 years (range 0·1–10·2). The 3-year
event-free survival for all 89 patients was 42·7% (95% CI
32·2–52·8) and 5-year 37·5% (27·3–47·7) (figure 3). As
would be anticipated, event-free survival was poorer for
those with metastatic disease (HR 4·1, 95% CI 2·0–8·7,
p<0·0001) with all progressing within 3 years. In the
non-metastatic patients, the 3-year and 5-year event-free
survival was 47·6% (36·2–58·1) and 41·8% (30·7–52·6).
No significant difference was seen in event-free survival
at 5 years between non-metastatic patients with WHO
grades II and III disease: 38·4% (95% CI 26·6–50·0) and
41·7% (15·3–66·5), respectively.
For the 51 patients alive at last follow-up, the median
follow up was 6·0 years (range 1·5–11·3). 38 patients
died: 34 due to tumour, four from post–surgical
complications. The 3-year overall survival for the whole
group was 76·8% (95% CI 66·4–84·4) and the 5-year
overall survival was 60·0% (95% CI 48·4–69·7). As
expected, survival was poorer for those with metastatic
disease (hazard ratio [HR] 3·0 [95% CI 1·2–7·3], p=0·016)
(figure 4). In the non-metastatic patients, the 3-year and
5-year overall survival was 79·3% (95% CI 68·5–86·8)
and 63·4% (51·2–73·4), respectively. Although the
numbers were small, 14 children, those younger than
1 year at diagnosis, seemed to have the poorest survival,
although this was not significant: 44% were alive at
5 years compared with 65% for those aged between 2 and
3 years at diagnosis (HR 1·4 [95% CI 0·9–2·2]; p=0·18;
table 5). Whether the patient was male or female did not
seem to affect outcome. Tumour location in the
supratentorium was associated with a better survival than
for patients with infratentorial tumours (83%
vs
56%),
but this comparison was not significant (HR 3·1 [95% CI
0·8–12·5]; p=0·12; table 5). The 5-year overall survival for
grade II disease was 61·5% (95% CI 48·2–72·4) and
66·7% (33·7–86·0) for grade III.
Patients with a complete resection, based on
neurosurgical review, had better 5-year event-free survival
(48·9% [95% CI 33·8–62·8]), than those with partial
resection or biopsy alone (25·8% [13·7–39·6]).
Importantly, there was a difference in 5-year overall
survival (68·1% [51·2–80·2]
vs
51·8% [35·6–65·8]) for
complete versus incomplete resection, although this is
not firmly established with our study sample size (p=0·07;
figure 5). Overall survival based on the radiological
assessment of residual tumour showed that complete
resection did not confer a better outcome (p=0·28). From
our analysis neurosurgical rather than radiological
assessment of completeness of resection was a more
powerful predictor of outcome (data not shown).
The actual distribution of the SRDChemo achieved
from the combination chemotherapy schedule of table 1
varied according to the total number of cycles received by
each patient. There was a tendency towards lower
SRDChemo in those who had the fewest cycles, whereas
in those who had all seven cycles, the distribution is
skewed towards higher values. The median SRDChemo
achieved was 0·97, very close to the protocol ideal of
unity, but ranged from 0·59 (a patient receiving cycle 1
only) to 1·46 (a patient receiving all seven cycles who was
0
2
6
0·25
0·50
0·75
0·00
Overall survival (%)
1·00
RDIChemo 0·50–0·79 32
28
21
12
8
7
RDIChemo 0·80–0·94 33
31
27
17
15
14
Survival from end of chemotherapy (years)
Numbers at risk
RDIChemo 0·95–1·41 23
21
19
13
10
8
4
RDIChemo <0·79
RDIChemo 0·80–0·94
RDIChemo >0·95
15
19
18
1
3
5
N
HR for death (95% CI) p
Age <1 year at diagnosis
14
1·4 (0·9-2·2)
0·18
Female sex
31
1·1 (0·5-2·1)
0·13
Infratentorial ependymoma
76
3·1 (0·8-12·5)
0·12
WHO III histology
30
1·6 (0·8-3·2)
0·15
Partial resection (judged by neurosurgeon)
45
1·8 (0·9-3·6)
0·07
Partial resection (judged by radiological review)
55
1·5 (0·7-3·0)
0·28
Dose intensity <0·8
32
1·6 (1·0-2·7)
0·04
Table 5:
Risks from different potential prognostic variables
0·25
0·50
0·75
0·00
Overall survival (%)
1·00
Total resection 44
41
29
17
11
6
Partial resection 45
38
21
12
7
3
Time from surgery (years)
Numbers at risk
Less than full total resection
Full total resection
0
2
4
6
8
10
12
··
··
Figure 5:
Overall survival based on neurosurgical assessment of the extent of resection at the end of surgery
Figure 6:
Overall survival from end of chemotherapy by relative dose intensity chemotherapy (RDIChemo)
received