paediatrics Brussels 17

Merchant et al

Table 1. Demographic and Clinical Variables of Patients Categorized According to Age at Irradiation

Age 3 Years (n 40)

Age 3 Years (n 48)

No. of Patients

No. of Patients

Patient Characteristic

%

%

P

Length of follow-up, months Median

34.7

25.6

Range

1.3-60.5

6.1-59.8

Sex

Female

19 21

47.5 52.5

22 26

45.8 54.2

.88

Male

CSF shunting No

31

77.5 22.5

28 20

58 42

.057

Yes

9

Pre-CRT chemotherapy No

35

87.5 12.5

37 11

77 23

.21

Yes

5

Extent of resection Gross-total

34

85

40

83.3

.83

Near-total Subtotal

2 4

5

4 4

8.3 8.3

10

Tumor grade Anaplastic

13 27

32.5 67.5

22 26

45.8 54.2

.20

Differentiated

Tumor location Infratentorial

26 14

65 35

42

87.5 12.5

.012

Supratentorial

6

No. of pre-CRT resections 1

26 14

65 35

30 14

63 29

.45

2 3 4

2 2

4 4

Hydrocephalus No

16 24

46 60

8

17 83

.014

Yes

40

Abbreviation: CRT, conformal radiation therapy.

Disease Control The median length of follow-up was 38.2 months (range, 12.4 to 75.6 months); 20 patients experienced dis- ease progression, and the median time to progression for those patients was 14 months (range, 6 to 26 months). Failures were characterized as local (n 8), local distant (n 4), and distant (n 8). There were no marginal failures. The cumulative incidence of local failure estimate at 3 years was 14.8% 4.0%. The actuarial PFS estimate at 3 years was 74.7% 5.7% (Fig 1). Thirteen of the failures occurred among the 48 children younger than 3 years at the time of irradiation. One patient died whose death was not attributed to radiation therapy. He was censored at the time of death when autopsy showed stable residual tumor. Uni- variate analysis identified statistically significant differences in actuarial 3-year event-free survival estimates based on extent of resection (gross-total resection v near-total resec- tion/subtotal resection; 77.6% 5.8% v 42.9% 16.2%;

whose death was not attributed to disease progression or radiation therapy) as competing risks. The longitudinal trends in neurocog- nitive outcomes were estimated by using linear mixedmodels with random coefficients. 23 SAS software was used for all analyses. 24 Data analyses were performed by the biostatistical coauthors.

RESULTS

Clinical and treatment characteristics of the study patients are presented in Table 1. To identify similarities and differ- ences among the patients, we categorized them according to age (younger than 3 years or 3 years) for comparison. Older patients were more likely to have supratentorial tu- mors ( P .012), and the younger patients were more likely to have hydrocephalus ( P .014) and require CSF shunting ( P .057). Larger proportions of the younger patients received preirradiation chemotherapy and had anaplastic tumors at diagnosis. However, these proportions were not significantly different from those of the older patients.

3158

J OURNAL OF C LINICAL O NCOLOGY