Cranial radiotherapy of childhood brain tumours: Growth hormone deficiency and its relation to the biological effective dose of irradiation in a large population based study


Dr Marianne Schmiegelow Department of Growth and Reproduction 5064, Juliane Marie Centre, The National University Hospital, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen, Denmark. Fax: +45 35456054; E-mail:



The study was to determine the incidence of GH deficiency (GHD) following cranial radiotherapy (RT) for a childhood brain tumour in a large population based study and analyse the biological effective dose (BED) to the hypothalamus/pituitary (HP) region as a risk factor.


BED was assessed by use of the linear-quadratic (LQ) model, which gives a means of expressing the biological effect of various treatment schedules in a uniform way. In patients aged ≥ 18 years (n = 53) GH status was assessed by an insulin-tolerance test (ITT) (n = 34), however, in patients with seizure disorders (n = 19), and in 20 children aged < 18 years GH status was assessed by an arginine test. Cut-off levels for GHD, indicating GH substitution, were defined by a peak GH response of < 9 mU/l and < 15 mU/l for patients ≥ 18 and < 18 years, respectively.


Ninety-one children aged < 15 years eligible for the study, diagnosed between 1970 and 1997 in the Eastern part of Denmark, the Faroe Islands and Greenland, with a primary brain tumour not directly involving the HP axis. 84% (n = 76) agreed to participate. Three patients were excluded due to hypothyroidism detected at time of testing.


Serum GH and levels of serum insulin-like growth factor-I (s-IGF-I) and serum insulin-like growth factor binding protein-3 (s-IGFBP-3) were measured. BED was assessed to the HP region.


The median age at the time of RT was 8.7 years (range: 0.8–14.9 years) and the median time of follow-up was 15 years (range: 2–28 years). Fifty-eight patients (80%) had GHD and they had received a median BED of 77.5 Gy to the HP region, whereas the median BED was 54.5 Gy for 15 patients without GHD (P = 0.002). Peak GH and BED were correlated (rs = −0.53, P < 0001). Median IGF-I SDS and IGFBP-3 SDS were −2.5 (−5.2–0.7 SDS) and −1.7 (−5.8–0.9 SDS), respectively, and IGF-I SDS was correlated to peak GH (rs = 0.45, P < 0.001). Peak GH and length of follow-up were related (rs = −0.28, P = 0.018). Stepwise backward multiple linear regression analysis showed that the best-fit model to predict the peak GH release following ITT/arginine stimulation included BED (P < 0.0001) and length of follow-up (P = 0.05).


The data of this study suggest that the majority of long-term survivors of brain tumours develop GH deficiency following radiotherapy in childhood and that the adverse effects of radiotherapy may be directly related to the biologically effective dose. With longer follow-up fewer patients might respond normally to GH stimulation tests.