Endocrine Deficiency As a Function of Radiation Dose to the Hypothalamus and Pituitary in Pediatric and Young Adult Patients With Brain Tumors

Abstract

Purpose: There are sparse data defining the dose response of radiation therapy (RT) to the hypothalamus and pituitary in pediatric and young adult patients with brain tumors. We examined the correlation between RT dose to these structures and development of endocrine dysfunction in this population.

Materials and methods: Dosimetric and clinical data were collected from children and young adults (< 26 years of age) with brain tumors treated with proton RT on three prospective studies (2003 to 2016). Deficiencies of growth hormone (GH), thyroid hormone, adrenocorticotropic hormone, and gonadotropins were determined clinically and serologically. Incidence of deficiency was estimated using the Kaplan-Meier method. Multivariate models were constructed accounting for radiation dose and age.

Results: Of 222 patients in the study, 189 were evaluable by actuarial analysis, with a median follow-up of 4.4 years (range, 0.1 to 13.3 years), with 31 patients (14%) excluded from actuarial analysis for having baseline hormone deficiency and two patients (0.9%) because of lack of follow-up. One hundred thirty patients (68.8%) with medulloblastoma were treated with craniospinal irradiation (CSI) and boost; most of the remaining patients (n = 56) received involved field RT, most commonly for ependymoma (13.8%; n = 26) and low-grade glioma (7.4%; n = 14). The 4-year actuarial rate of any hormone deficiency, growth hormone, thyroid hormone, adrenocorticotropic hormone, and gonadotropin deficiencies were 48.8%, 37.4%, 20.5%, 6.9%, and 4.1%, respectively. Age at start of RT, time interval since treatment, and median dose to the combined hypothalamus and pituitary were correlated with increased incidence of deficiency.

Conclusion: Median hypothalamic and pituitary radiation dose, younger age, and longer follow-up time were associated with increased rates of endocrinopathy in children and young adults treated with radiotherapy for brain tumors.

Trial registration: ClinicalTrials.gov NCT00105560 NCT01063114 NCT01288235.

Figures

Fig 1.

(A) Cumulative incidence curve for the development of any hormone deficiency with 95% CIs. The model prediction is overlaid as a blue line. (B) Cumulative incidence curve for the development of any hormone deficiency stratified by hypothalamic dose ≤ 20 GyRBE, 20 to 40 GyRBE, and ≥ 40 GyRBE, with 95% CIs (shaded). The model prediction is overlaid as dashed lines. (C) Cumulative incidence curve for the development of any hormone deficiency stratified by age at time of treatment < 6 years old, 6 to 10 years old, and > 10 years old, with the model prediction overlaid as dashed lines.

Fig 2.

Cumulative incidence of hormone deficiency stratified by mean D50 (median dose) for the hypothalamus and pituitary for (A) growth hormone (GH), (B) thyroid hormone, (C) adrenocorticotropic hormone (ACTH), or (D) gonadotropin.

Fig 3.

Cumulative incidence of hormone deficiency stratified by age at time of radiation treatment of (A) growth hormone (GH), (B) thyroid hormone, (C) adrenocorticotropic hormone (ACTH), or (D) gonadotropin.

Fig 4.

Models relating mean D50 (median dose) to the hypothalamus and pituitary with risk for hormone deficiency at 5 years, stratified by age at time of radiotherapy (RT) for (A) any hormone, (B) growth hormone (GH), (C) thyroid hormone, (D) adrenocorticotropic hormone (ACTH), and (E) gonadotropin. Solid lines represent the model prediction and dashed lines represent the 95% CIs for select groups.

Fig A1.

(A) Plot of median hypothalamic dose versus median pituitary dose for each patient showing the correlation between doses to the two structures. (B) Plot of mean D50 (median dose) for hypothalamus and pituitary versus median hypothalamus dose, and (C) median pituitary dose for each patient showing the correlation between dose to the composite variable and these two structures. Development of any hormone deficiency is noted by gold triangles, and no deficiency is noted by blue circles. The gray line represents a perfect 1:1 correlation, and the blue line represents the least squares linear regression.