High Prevalence of Early Endocrine Disorders After Childhood Brain Tumors in a Large Cohort

Abstract

Brain tumors are the most common type of solid cancer in the pediatric population (1, 2), with an incidence rate of 5.42 per 100 000 in the United States (1) and 2.99 per 100 000 in Europe (3). Endocrine complications, including anterior or posterior pituitary deficiencies, due to the tumor or its treatment, account for an important number of long-term complications in these patients (4-12). Oncology patients, and in particular children treated for a brain tumor, may have a decreased life expectancy and quality of life (QoL) (13, 14) linked to tumor relapse, neurological sequelae, need for long-term medical care and, in many cases, need for hormonal substitution.

Teams from the Childhood Cancer Survivors Study and the St Jude Lifetime Cohort in the United States (11, 15, 16) have addressed long-term pituitary deficiencies in adults, but there is less available data from European countries. Information from these existing adult patient cohorts consists mainly of data from questionnaires, with the limitation of self-reported information, and in some cases of clinical evaluation using basal hormone screening (without dynamic testing), which may underestimate hormonal deficiencies, especially growth hormone deficiency (GHD).

For tumors that are localized far from the pituitary region, radiotherapy (RT) may cause secondary pituitary deficiencies (15-19), through hypothalamus and pituitary radiation exposure. Detailed description of data concerning endocrine disorders related to tumor type and location is scarce. Finally, treatment of cancer in general is continuously evolving, and so it is important to provide new data assessing the impact of more focalized radiotherapy and preservation of cerebral tissue when possible.

Our team had previously published data from a cohort of children treated for a brain tumor, specifically focusing on sellar and suprasellar tumors, and hydro-electrolyte disturbances, which showed that an important proportion of these patients may display pituitary deficiencies at diagnosis and after treatment (90.3% GHD, 68.9% thyrotropin deficiency [TSHD], and 66.2% adrenocorticotropic hormone deficiency [ACTHD] at last visit) (20). In this new study, we aim to bring new data, with a detailed endocrine description of a large monocentric cohort of patients with a history of pediatric primary brain tumor, including basal and dynamic biologic assessment and follow-up.

The secondary objectives were to describe tumor type, localization, and the type of treatment including radiotherapy, hormonal dysfunction according to the tumor profile and hormonal replacement therapy, including GH treatment modality, in order to propose recommendations for current practice.

Methods

A noninterventional monocentric study was conducted in patients with a history of pediatric primary brain tumor, who had been followed in the Pediatric Endocrinology Unit at Hôpital Universitaire Necker – Enfants Malades between 2010 and 2015.

All patients with a diagnosis of primary brain tumor before the age of 18 years, who visited the outpatient consultation at least once between January 1, 2010 and December 31, 2015, in our Pediatric Endocrinology Unit, at Hôpital Universitaire Necker—Enfants Malades, in Paris, and who did not refuse to participate in the study were included. Exclusion criteria were insufficient data, pituitary adenomas, untreated tumors, or refusal to participate in the study.

Oncology data, including the type of treatment (surgical interventions, radiotherapy including radiation dose received on the pituitary area and the thyroid, and chemotherapy, classified as high, moderate, or low risk for gonadotoxicity (21), Table 1 shows the detail of chemotherapy classification) were collected from medical records, and included data from follow-up until June 15, 2020.

Patients were followed in consultation every 6 to 12 months, with clinical evaluation, basal hormone tests performed once a year (or more often if needed).

Anthropometric Data

• - Height, weight, and adult height were reported using the French growth charts (Groupe Français d’Auxologie—Pr Michel Sempé, revised in 2012). Mean adult height for these curves is 163 cm for women and 174.5 cm for men.

• - Overweight was defined as a body mass index (BMI) z-score of ≥ +2 SD, and malnutrition or underweight as a BMI z-score of ≤ −2 SD.

Hormone Level Assessment

At the first endocrine visit in our unit, all patients had a clinical evaluation. A basal blood sample was performed (including hormonal status: insulin-like growth factor-1 [IGF1], thyrotropin [thyroid stimulating hormone; TSH], free thyroxine [FT4], 8:00 am cortisol level, luteinizing hormone [LH], follicle-stimulating hormone [FSH], testosterone, anti-Mullerian hormone [AMH]/inhibin B levels), and according to clinical and laboratory findings dynamic tests were programmed when necessary (GH stimulation test, LHRH test, ACTH test). A first thyroid ultrasound to screen for nodules was generally performed 5 years after spinal radiotherapy, or earlier if there was an abnormal thyroid palpation.

IGF1 levels were measured by IDS-iSYS IGF1 assay (22) since August 1, 2013, and before this date by Cis Bio International IRMA assay. Reference values for IGF1 were given according to age and pubertal status. FSH, LH, FT4, TSH, and cortisol were measured using the Beckman Coulter Access assays. Testosterone levels were quantified by RIA with extraction “Testo-RIA-CT”, also with the Cis Bio International assay. AMH was quantified using ELISA assays (DSL until 2011, Immunotech/Beckman Coulter until 2015), and after 2015 with the Beckman Coulter Access 2 assay. Inhibin B was quantified using EIA type sandwich assay.

• - GHD was defined by 2 stimulation tests with a GH peak < 20 mUI/L, or one abnormal test when there was a history of radiotherapy dose to the pituitary area ≥18 Gy. Tests used in these patients were glucagon test or arginine-insulin test. Normal IGF1 values (> -2 SD for age and pubertal stage) did not exclude GHD, since IGF1 levels may be normal and thus misleading for the diagnosis of growth hormone deficiency in patients who have received radiotherapy (23). During transition (after having attained adult height) GHD was defined as a peak GH level after insulin tolerance test of < 15 mUI/L. Severe persistent GHD at transition was defined by a low GH peak < 15 mUI/L after insulin tolerance test with a low IGF1 level. Patients who had multiple (≥ 3) pituitary hormone deficiencies at transition were considered to have persistent GHD (24).

• - Thyrotropin (TSH) deficiency, or TSHD, was defined by low free thyroxine levels with inappropriately low, normal, or slightly elevated TSH levels.

• - Primary hypothyroidism was defined by a TSH level above normal ranges with normal free thyroxine levels.

• - Corticotropin (ACTH) deficiency, or ACTHD was defined by a low cortisol level at 8:00 am (< 6 µg/dL or 165 nmol/L), or failure to attain a cortisol peak ≥ 18 µg/dL (500 nmol/L) after a tetracosactide (Synacthen) test.

• - Precocious puberty was diagnosed when pubertal signs (breast development in girls, testicular enlargement) or detectable levels of testosterone > 0.07 ng/mL or 0.24 nmol/L in boys appeared before the age of 8 years in girls, or 9 years in boys (25, 26). Early puberty was diagnosed if these signs appeared between 8 and 9 years in girls or between 9 and 11 years in boys.

• - Delayed puberty was defined by absence of pubertal signs (breast development, enlargement of testicular volume) by the age of 13 in girls or 14 in boys, or lack of appropriate progression of puberty, with more than 4 years between the first pubertal signs and menarche in girls or the onset and completion of testicular growth in boys (26). Hypogonadotropic hypogonadism was defined as delayed puberty that needed treatment by steroids, in absence of gonadal toxicity by the criteria described below.

• - Gonadal toxicity was defined by basal LH or FSH levels above the upper limit of the reference range and/or low AMH levels in girls or low inhibin B levels in boys.

• - Diabetes insipidus (DI) was diagnosed in patients who had hypernatremia (>145 mmol/L), polyuria (> 3 mL/kg/h), polydipsia, and an adequate response to treatment with desmopressin.

Classification of Tumors

For finer analysis of data concerning pituitary deficiencies, we separated our entire population into 2 distinct groups: suprasellar tumors (SS), when the tumor involved the sellar and/or suprasellar region, the hypothalamus or optic pathways, and non-suprasellar tumors (NSS) which included all the other tumor locations. The hypothesis was that SS tumors may have pituitary deficiencies associated with the tumor, probably present since diagnosis, and that NSS tumors have endocrine disorders as late effects of treatments, that occur later in time.

The SS group included patients with craniopharyngiomas, hypothalamic or optic tract gliomas, germ cell tumors and chordomas, and the NSS group included patients with medulloblastomas, other gliomas, ependymomas, rhabdoid tumors, pineal tumors including germ cell tumors, and other tumor types.

For analysis of postradiotherapy pituitary deficiencies, only NSS patients who had received cranial radiotherapy (CRT) or craniospinal radiotherapy (CSRT) were included, and they were divided into 3 groups by dose of radiation received to the pituitary area, as recommended by the Endocrine Society for endocrine follow-up (¹⁶): <18 Gy, 18 to 30 Gy, and >30 Gy. Incidence of pituitary deficiencies and time lapse between radiotherapy and onset of deficiencies were compared among these 3 groups.

Statistical Analysis

Demographic and clinical data are presented as means ± SD or medians (range) or percentages. Parametric and nonparametric tests were utilized according to data distributions. Statistical analyses included t tests, ANOVA, Chi-square, Wilcoxon rank sum, and Kruskal-Wallis tests. Logistic regression model was used to assess the association between baseline biomedical factors and the development of endocrine disorders. Multiple linear regression models examined the associations of baseline biomedical factors with adult height SD and BMI SD at last visit. Analyses were performed using STATA software (version 14.0). An alpha level of ≤ 0.05 was used to determine statistical significance.

Results

A total of 221 patients were included (Fig. 1), 48.9% females, with a median age at diagnosis of 6.7 years (range, 0 to 15.9 years) and a median postdiagnosis follow-up time of 6.7 years (range, 0.3 to 26.6 years). The median age at last visit was 15.6 years (range, 3.5 to 30.9 years), and 30% of all patients had at least 18 years of age.

Figure 1.

Population included in the study.

Open in new tabDownload slide

Of this cohort, 12 patients died (8 boys and 4 girls). Eleven patients were still undergoing oncological treatment at last follow-up visit.

Adult height was available for 113 patients: 58 female and 55 male patients.

The oncology profile of the cohort is depicted in Table 2. In Table 3 we show the frequency of hormone disorders in the cohort, at last follow-up visit in our unit.

A total of 174 patients were treated by radiotherapy, 128 patients received photon beam therapy and 46 received proton therapy. In 41/46 patients receiving proton therapy, the tumor was in the suprasellar area, so there was a high dose to the pituitary area. For the patients with NSS tumors who received proton therapy, 3 had a high cumulated radiation dose to the pituitary area ≥30 Gy (2 of them because of the need for subsequent CSRT) and 2 had low pituitary doses.

Tumor Type and Endocrine Disorders

Medulloblastoma

Patients with a medulloblastoma (n = 83) accounted for 37.6% of the cohort and represented the most frequent tumor type encountered. Age at diagnosis was 7.5 ± 3.5 years, and there was a sex ratio in favor of males (57%). Most of the patients, 62 cases (74.7%), were treated by a combined protocol comprising surgery, radiotherapy, and chemotherapy. In 73/82 cases radiotherapy administered CSRT and in 9/82 cases it included only the posterior fossa (CRT), with no spinal irradiation. Pituitary radiation dose was in median 38 Gy (range, 2.5-68). Chemotherapy used was high-risk for gonadal toxicity in 61/63 patients treated by chemotherapy.

GHD was diagnosed in 71 patients (85.5%), 70 of whom had received cranial RT (CRT or CSRT), with a median dose to the pituitary of 38 Gy (range, 3-54). Hypothyroidism was diagnosed in 32 patients (38.6%), 13 with primary hypothyroidism (12/13 having received thyroid irradiation) and 19 with TSHD (all of which also had postradiotherapy GHD). Only 5 patients in this group (6.0%) had ACTHD; all of them had GHD, and 3/5 also had TSHD. Precocious or early puberty was diagnosed and treated in 9 patients (10.8%), all after radiotherapy. Gonadal toxicity was diagnosed in 28 patients of this group (33.7%), all of whom had received gonadotoxic chemotherapy (high-risk chemotherapy in 27/28, moderate-risk chemotherapy in 1 patient).

Only 3/83 patients with medulloblastoma had ≥ 3 pituitary hormone deficiencies; nonpituitary hormone disorders were mainly related to high-dose chemotherapy (gonadal toxicity) or thyroid irradiation (primary hypothyroidism and thyroid nodules).

Craniopharyngioma

Craniopharyngioma represents the second largest group of our cohort, with 64 patients (29%), and was the group that had the highest frequency of endocrine disorders, especially pituitary. The mean age was 8.2 ± 3.7 years, with a sex ratio in favor of males (56%). Surgery was the main initial treatment in 63/64 patients, and radiotherapy completed the treatment in 47 cases (74.6%). Pituitary radiation dose was in median 53 Gy (range, 27-56 Gy). One patient received as the only treatment an intracystic injection with bleomycin.

GHD was diagnosed in 61 patients (95.3%), TSHD in 58 patients (90.6%, all but 1 also had GHD), ACTHD in 57 patients (89.1%, 56/57 also with GHD, and 55/57 also with TSHD). Hypogonadotropic hypogonadism was diagnosed in 34/43 patients (79.1% of girls ≥ 13 years old and boys ≥ 14 years old). A large proportion of patients of this group, 56/64 (87.5%) had ≥ 3 pituitary hormone deficiencies.

DI was diagnosed in 48 patients from this group (75%); in 12 patients at the time of tumor diagnosis, and in 36 patients DI appeared shortly after surgery (median time, 1 day after surgery; range, 0-68 days).

Glioma

In this cohort, 45 patients had a glioma (20.4%), 33 in the SS group and 12 in the NSS group. These patients were on average younger than those with the other tumor types (mean age, 4.1 ± 3.2 years; P < 0.0001 vs craniopharyngiomas and vs medulloblastomas), with a predominance of female patients (64%). Thirty-nine were treated by chemotherapy, 36/45 by surgery and 17/45 by radiotherapy (only 2 cases with CSRT). Pituitary radiation dose was in median 39 Gy (range, 4-54 Gy). GHD was diagnosed in 24 patients (53.3%), 18 with a SS tumor and 6 with a NSS tumor who had received pituitary irradiation ≥ 11 Gy. TSHD was diagnosed in 19 patients (42.2%), precocious or early puberty in 17 patients (37.7%, 15/17 in the SS group), and ACTHD in 14 patients (31.1%, all with SS tumors). Only 6/45 patients had ≥ 3 pituitary hormone deficiencies. DI was present in 13 patients, all of whom had previous surgery.

Tumor anatomical location and endocrine disorders: SS vs NSS

We divided the entire cohort into 2 groups: SS tumors (48%) and NSS tumors (52%). A comparative table (Table 3) shows differences in patients according to the tumor location.

The first Endocrinology visit took place earlier for patients with SS tumors, in a median 0.2 years after diagnosis (range, −1.0 to 6.7) than for those with NSS tumors: median 1.6 years (range, 0.1-10.1), P < 0.0001. The first endocrine evaluation was performed in most cases after the patient had already started treatment, so we had no basal pretreatment hormonal status.

Height at first endocrine visit of patients was similar between both groups (SS: −0.3 ± 1.6 SD vs NSS: −0.1 ± 1.2 SD), but the BMI was already significantly higher in SS (+0.8 ± 2.1 SD, 21% obese) than in NSS (−0.2 ± 1.3 SD, 5.2% obese, P < 0.0001).

A comparative figure for endocrine disorders between the 2 groups at last visit is shown in Fig. 2.

Figure 2.

Proportion of endocrine disorders according to tumor location. Abbreviations: SS, suprasellar; NSS, non-suprasellar; GHD, growth hormone deficiency; ACTHD, ACTH deficiency.

Open in new tabDownload slide

GHD incidence was very similar in both groups (83.0% SS; 76.5% NSS; P = 0.338) and was diagnosed a median 1.8 years after tumor diagnosis (range, 0-12.4 years). Diagnosis of GHD was made earlier in the SS group, at a median of 1.1 years after diagnosis of tumor (range, −0.8 to 12.2), compared with the NSS group, who were diagnosed with GHD at a median of 2.3 years after tumor diagnosis (range, 0.5 to 12.4), P = 0.0082.

Postradiotherapy GHD was diagnosed after a median of 1.6 years from end of RT (range, 0.2-10.7). When separating patients according to pituitary radiation dose, the low dose group (< 18 Gy, n = 12), was diagnosed in a median of 3.0 years after end of RT (range, 1.6-10.7), which was significantly later than for patients in the intermediate group (18-30 Gy, n = 18), diagnosed in a median of 2.1 years after RT (range, 0.2-6.4), P < 0.005. Patients who received a high pituitary radiation dose (>30 Gy, n = 54) were diagnosed with GHD after a median of 1.4 years (0.4-8.2), significantly earlier than the low dose group (< 18 Gy, P < 0.001), but similar to the intermediate group (18-30 Gy, P = 0.54).

A summary table concerning the impact of radiotherapy on endocrine disorders, comparing this study and previous publications, is shown in Table 4. This table includes only patients with NSS tumors having received ≥ 18 Gy to the pituitary area.

Treatment by GH was initiated at a mean age of 10.5 ± 3.0 years, in an average of 0.7 ± 0.7 years after diagnosis of GHD. There was no difference in age of GH initiation between SS and NSS patients (P = 0.7090).

Hypothyroidism was very common in SS patients (76.4%, all with TSH deficiency), appearing in most cases shortly after diagnosis, after a median time of 0.1 years (range, −0.4 to 12.5). Postradiotherapy TSHD in these patients concerned only 9/81 patients, after a median of 2.0 years (range, 0.3-4.2). In NSS patients, hypothyroidism was diagnosed in 39/110 patients (33.9%): in 16/39 as primary hypothyroidism and in 23/39 as TSHD. In 38/39 patients it was diagnosed after radiotherapy, after a median of 2.3 years (range, 0.5-9.4).

ACTHD was very frequent in our SS patients (69.8%), and as expected was rarer in NSS patients (6.1%). In NSS patients, ACTHD was diagnosed only in patients who had received a high-dose pituitary radiation (median, 37.2 Gy; range, 25-39 Gy), at a median time of 2.6 years after RT (range, 0-9.2 years).

Precocious or early puberty was noted in 16% of SS and in 12.2% of NSS (all but 1 of the NSS group after radiotherapy). Hypogonadotropic hypogonadism was predominant in SS (63.1% vs 1.3% NSS), P < 0.001, and gonadal toxicity in NSS (29.6% vs 2.8% SS), P < 0.001. All patients with gonadal toxicity had received moderate or high-risk chemotherapy.

Diabetes insipidus was almost exclusively present in the SS group, with an incidence of 63.2%. DI was most often diagnosed before the tumor diagnosis or shortly after surgery, at a median time of 0.0 years after tumor diagnosis (range, −1.1 to 12.0). Only 1 patient in the NSS group (0.9%) was diagnosed with DI, in the context of a very large hemispheric astrocytoma treated by surgery during the neonatal period.

Patients who had received spinal or cervical radiotherapy were considered “high-risk” for developing thyroid nodules (86/221). Of the 86 patients in this high-risk group, 45 had performed at least 1 thyroid ultrasound (Fig. 3). Thyroid nodules were found in 13/45 patients after a median of 8.0 years (range, 4.1-12.8), those measuring ≥10 mm had biopsy by fine needle aspiration. Among these 13 patients there were 4 cases of thyroid cancer: 3 papillary carcinomas, and 1 follicular carcinoma (after 4.8, 5.6, 8.3 and 11.5 years of RT). Only 1 of these 4 patients had a cancer predisposing condition: a TP53 mutation (Li-Fraumeni syndrome) in a girl diagnosed with thyroid cancer 8.3 years after RT, at the age of 13. Interestingly, three-fourths of patients diagnosed with thyroid cancer had normal thyroid function tests. In addition, 7 patients in the “low-risk” group (no history of cervical irradiation) underwent a cervical ultrasound, and 2 of them had micronodules (≤ 5 mm) currently under surveillance.

Figure 3.

Description of patients with thyroid nodules. Abbreviations: US, ultrasound; CSRT, craniospinal radiotherapy.

Open in new tabDownload slide

Adult height

Adult height according to tumor site, depicted in Fig. 4, was significantly different between the SS group: −0.2 ± 1.4 SD and the NSS group: −0.8 ± 1.4 SD (P = 0.013).

Figure 4.

Difference between initial and adult height in standard deviations according to tumor location. Abbreviations: SS, suprasellar tumors; NSS, non-suprasellar tumors.

Open in new tabDownload slide

Patients with SS tumors had a significantly lower adult height than their mid-parental target height: −0.2 (F: 162 cm; M: 173.5 cm) vs +0.2 SD (F: 164 cm; M: 175.5 cm), P = 0.04.

Patients with NSS tumors had an even more significant difference in adult height compared to their mid-parental target height: −0.8 ± 1.4 SD (F: 158.5 cm; M: 170 cm) vs +0.2 ± 0.9 SD (F: 164 cm, M: 175.5 cm), which gives a loss of 1.0 SD compared with their parents, P < 0.0001).

When separating patients by tumor type, adult height in patients with craniopharyngiomas (+0.3 ± 1.1 SD) was similar to their target height (+0.6 ± 1.2 SD). Patients with gliomas were significantly shorter (−0.9 ± 1.4 SD) than their target height (0 ± 1.2 SD), P = 0.02. Patients with medulloblastomas also had a significantly lower adult height (−1 ± 1.4 SD) when compared with their parents (+0.2 ± 1.0 SD), P < 0.0001, despite GH treatment starting at a similar age as patients with craniopharyngiomas.

When we compared patients treated with GH vs untreated, we did not see a significant difference in adult height (treated: −0.4 ± 1.4 SD vs untreated: −0.7 ± 1.5 SD; P = 0.4740). All patients treated had a diagnosis of GHD, and most of the untreated patients with GHD were diagnosed after attaining adult height. Only 3 children were untreated after diagnosis of GHD: a boy aged 12.1 years, with a diagnosis of medulloblastoma treated by CSRT, due to parental refusal (adult height: −2.8 SD), a girl aged 12.6 years, with a diagnosis of craniopharyngioma, not treated by RT, due to regular growth velocity (adult height: −0.5 SD, within her parental target height) and a girl aged 7.4 years, with a diagnosis of glioma, not treated by RT, due to regular growth velocity (adult height: +0.5 SD, also within her target height).

The impact of spinal radiotherapy on adult height was also assessed, independently of tumor type or site: patients having received spinal radiation (with CSRT) had a lower adult height compared to their parental target height (−0.8 ± 1.4 SD vs +0.2 ± 1.0; P < 0.0001), with loss of 1.0 SD. Patients who did not receive spinal radiation were also shorter than their parents (−0.3 ± 1.4 SD vs +0.2 ± 1.1; P = 0.0001), with a loss of 0.5 SD; there was no significant difference between both groups (P = 0.0644). When we compare height in the NSS group, there is no difference in adult height between those having received spinal RT vs those that did not receive spinal RT (P = 0.9573), but the number of patients in the second group is small (14 patients).

Change in body mass index

BMI increased significantly in both groups from initial endocrine visit to last follow-up visit, both in SS (initial: +0.8 ± 2.1 SD, final: +2.0 ± 1.8 SD) and NSS patients (initial: −0.2 ± 1.3 SD, final: +0.5 ± 1.4 SD), P < 0.001, with obesity noted in 46.2% in SS and 17.4% in NSS (Fig. 5). When separated by tumor type, there was a significant difference between medulloblastomas and craniopharyngiomas (P < 0.001), and between medulloblastomas and gliomas (P = 0.003), but patients with craniopharyngiomas and gliomas had a similar BMI at the initial visit BMI (P = 0.626) and also at last visit (P = 0.115). It is important to note that in the SS group there was no significant difference in initial or final BMI for patients with craniopharyngiomas when compared with other tumors in this group.

Figure 5:

Change in BMI between first endocrine visit and last follow-up visit, in SD, according to tumor location. Abbreviations: SS, suprasellar tumors; NSS, non-suprasellar tumors.

Open in new tabDownload slide

Analysis of risk factors associated with endocrine disorders

A logistic regression model was established to screen the risk factors associated with GHD. Presence/absence of GHD was used as a dependent variable. Age, gender, and radiotherapy were used as independent variables. Results showed male gender and radiotherapy were closely associated with GHD (P < 0.05). Age at diagnosis was not significantly associated with the development of GHD. The odds ratio of radiotherapy was of 5.7 (95% CI, 2.5-13.0), and was statistically significant (P < 0.001). The odds ratio of gender (male) was 2.1 (95% CI, 1.0-4.6, P = 0.045).

The type of radiotherapy (CRT vs CSRT) was also screened as possible predictor of the development of GHD in a logistic regression model; however, no significant results were found.

For the other endocrine disorders, including TSHD, ACTHD, precocious puberty (PP), and hypogonadotropic hypogonadism no significant associations with demographic and biomedical factors were found in the exploratory analyses.

Factors impacting adult height and BMI at last follow-up visit

A multiple regression model controlled for height at the first visit and target height was run to predict adult height. Gender, age at diagnosis, tumor site, and radiotherapy were used as predictors. A younger age at diagnosis and tumor site (NSS tumors), statistically significantly predicted a lower adult height, F(7, 95) = 14.05, P < 0.0005, R² = 0.508.

A multiple regression model controlled for BMI z-score at the first visit was run to predict final BMI z-score. Gender, age at diagnosis, tumor site, radiotherapy, and chemotherapy were used as predictors. Only tumor site (SS group) significantly predicted a higher BMI z-score at last visit (P < 0.05), F(6, 213) = 27.95, P < 0.0005, R² = 0.440.

Discussion

This is one of the largest European-based cohort studies including young patients with a history of pediatric primary brain tumor. Almost all patients included had a high risk for developing pituitary deficiencies, as most of them had either a SS tumor or a NSS tumor treated by cranial or craniospinal radiotherapy.

The division into 2 different categories, SS and NSS, highlighted differences in endocrine effects between the groups. Both groups had a similar incidence of GHD, but the SS group had a higher incidence of other pituitary deficiencies and obesity, whereas the NSS group showed a higher incidence of gonadotoxicity linked to high-dose chemotherapy, primary hypothyroidism and thyroid nodules/cancer related to cervical radiotherapy. This division seemed therefore useful to guide the endocrine follow-up.

GHD was diagnosed earlier in the SS group in part because of the earlier referral, but in some cases, these patients already had a diagnosis of GHD at tumor diagnosis or before starting oncology treatment. In the NSS group, some patients had postradiotherapy GHD diagnosed at first visit, but others developed GHD later on, over the following years.

Patients with NSS who received ≥ 18 Gy to the pituitary area developed postradiotherapy GHD at a median of 1.6 years after RT, which is quite early when compared with previous studies (4, 11). This may be explained by an early first endocrine consultation after the end of oncological treatments, but also by the predominance in our cohort of patients having received a high pituitary dose (> 30 Gy). When compared with previous studies (Table 4), we noticed a similar high incidence of childhood onset GHD, as reported by Laughton (19) in 2008, in children that were also treated with high pituitary doses. This incidence contrasts with the lower incidence observed in other cohorts, such as that observed by Shalitin (10) and the different studies analyzed by Mulder (17). It is important to note that we performed GH testing quite largely, in patients that started to decrease their growth velocity or that showed no growth acceleration during puberty, even if they were not yet below −2 SD.

Persistent GHD during transition, after attaining adult height, was also very high, in almost all (92.4%) patients that had received RT ≥ 18 Gy. Severe persistent GHD in our cohort (44.4%) had a similar incidence when compared to Chemaitilly’s cohort (15) (46.5%), as expected, since it was based on low IGF1 levels.

Despite the timely hormonal substitution, many of these patients have short adult height. In this study we observe loss of 1.0 SD (5.5 cm) in adult height of patients with NSS tumors compared with their parental target height, independently associated with young age at diagnosis. This decrease in adult height can be explained by the large predominance in this group of patients with medulloblastoma, who received in most cases spinal irradiation. This association has already been described in the literature (27-29). Children with NSS who were growing on a higher height SD curve at first visit attain a lower SD curve at adult height, despite adequate GH replacement treatment (Fig. 4). Patients in this group who had no spinal irradiation also had a low adult height compared with their target height. This may be explained by other factors, such as tumor relapse, with temporary withdrawal or delay for GH treatment initiation, or the presence of associated syndromes, such as type 1 neurofibromatosis (in patients with gliomas), which can be on its own an independent factor for short adult height. It is therefore essential that these children receive an appropriate evaluation of their GH secretion after a year of remission, in order to start replacement therapy when needed, as soon as possible, even if the child has not yet fallen below the −2 SD growth curve.

We unexpectedly found short adult height also in patients from the SS group, especially those with a diagnosis of glioma. Potential hypotheses include a delay in diagnosis and treatment of GHD in children from this group (see Table 3), possibly after a long period of treatment by chemotherapy. We cannot rule out the influence of other factors in adult height, such as syndromes associated with short stature (e.g., type 1 neurofibromatosis), or influence of chemotherapy on bone metabolism and growth, especially in very young children.

We conclude that an adequate estimation of response to GH treatment and adult height must take into account the tumor type (craniopharyngiomas respond better than other tumor types), the age of diagnosis (young age at diagnosis is associated with a lower adult height) and the treatment received (spinal irradiation worsens the adult height prognosis).

TSHD had a similar incidence when compared with pediatric populations, and a higher incidence when compared to the adult cohort of Chemaitilly. We hypothesize that this higher incidence in children is due to unmasking of TSHD during GH replacement treatment, because of an increase in peripheral conversion from FT4 to FT3 (30, 31). Adults with untreated GHD will have higher basal FT4 levels (32), preventing diagnosis of milder forms of TSHD.

The early diagnosis of postradiotherapy ACTHD, noted in few patients (6.1% of NSS), was comparable to the literature in terms of frequency (8, 16), but after a median time of only 2.6 years (range, 0-9.2 years). These patients had received a pituitary dose between 23 and 39 Gy and had an age at diagnosis and treatment that was comparable to the other patients in this group. High RT dose may partially explain this finding. This supports the need for annual screening, before the recommended time of 10 years after RT (¹⁶), in all patients with post-RT GHD, and even in the absence of overt clinical signs. We suggest an annual basal 8:00 am cortisol level and performing ACTH (Synacthen) test when basal cortisol level is in the lower normal ranges or if symptoms suggesting adrenal insufficiency appear, especially when there is already a diagnosis of GHD and/or TSHD.

In patients with gonadal toxicity who had received a high pituitary radiation dose it was very difficult to exclude the possibility of partial hypogonadotropic hypogonadism. We decided to give preference to the diagnosis of gonadal toxicity when criteria were present (low AMH in girls or low inhibin B in boys, with normal or high gonadotropins; we did not find any patient with low gonadotropins), these patients were thus classified as having gonadal toxicity, but an impaired pituitary—gonadal axis is not completely excluded in these patients. However, no patient in the “non-chemotherapy” group developed hypogonadotropic hypogonadism, despite the radiotherapy received. Longer follow-up is needed to assess if these patients will develop hypogonadotropic hypogonadism in the following years.

Obesity was already present in a fifth of the patients in the SS group at first visit, attaining nearly one-half of patients in this group at their last follow-up evaluation. Patients in the NSS group were less concerned by overweight at first visit, but increased their BMI throughout follow-up, with an obesity incidence of one-fifth of the patients at last follow-up visit. We conclude that nutritional evaluation and follow-up are of paramount importance in all patients with a brain tumor, in order to prevent or limit excessive weight gain, even if their initial nutritional status is normal. A special focus on children in the SS group (not only in patients with craniopharyngiomas) is mandatory, since inclusion in this group was the strongest predictor for developing obesity. Continuous efforts are currently being made by the oncology and surgery teams, in order to limit the endocrine late effects of treatments; hypothalamus-sparing surgery for treatment of craniopharyngiomas has shown an important impact on preventing or limiting severe obesity (33, 34). New treatments by GLP-1 agonists may provide additional help in patients in whom weight gain control is unsuccessful despite adequate nutritional and surgical measures (35). Routine vitamin D supplements and adequate calcium intake (through dietary intake or medication) are recommended to optimize bone mass formation (36).

Childhood cancer survivors may have a decreased QoL in adulthood, associated with growth and endocrine disorders (12, 37). Replacement treatment, especially with thyroid hormone and testosterone (in males), but also by GH in children, has been reported to have a positive impact on QoL (37, 38). Substitution of most endocrine disorders is possible even during oncological treatment, and a standard waiting period of 1 year after tumor therapy completion is considered as reasonable before starting GH replacement treatment (5, 22). It is very important to establish a multidisciplinary follow-up to detect and treat endocrine, metabolic, and other complications in a timely manner. The transition period into adult care in this population is also of the utmost importance, and good communication between the pediatric and the adult team is key to success in follow-up of these patients (39), in whom late effects may appear at a distance from the end of treatment.

This cohort shows an early incidence of endocrine secondary effects after brain tumor treatment when compared to previous studies (4, 8, 9, 11, 16, 25), including pituitary deficiencies and obesity. Our data strongly suggests the need for a first endocrine consultation shortly after diagnosis in children with a SS tumor, and no later then 1 year after end of radiotherapy in children with NSS tumors, especially when the dose delivered to the pituitary area is ≥ 18 Gy.

One of the strengths of this study is its homogeneous population, which includes only patients with a history of brain tumor, followed by a single pediatric Endocrinology team, and therefore with similar practice guidelines. Fine clinical and laboratory data description is available, so endocrine and metabolic status is objective. Most of these patients are referred to the adult Endocrinology department at La Pitié-Salpêtrière University Hospital, so medium- and long-term follow-up data concerning transition and young adulthood is available through their files. The most important weakness is that the study is partially retrospective, and some patients included are currently lost to follow-up in our team.

It is important to note that these so called “late endocrine effects” of treatment for a brain tumor can appear very shortly after oncological treatment, in the first few years of follow-up, so we must encourage an early referral to the endocrine clinic.

Abbreviations

Abbreviations
 
• ACTH

  adrenocorticotropic hormone

 
• ACTHD

  adrenocorticotropic hormone deficiency

 
• AMH

  anti-Müllerian hormone

 
• BMI

  body mass index

 
• CSRT

  craniospinal radiotherapy

 
• CRT

  cranial radiotherapy

 
• DI

  diabetes insipidus

 
• FSH

  follicle-stimulating hormone

 
• FT4

  free thyroxine

 
• GH

  growth hormone

 
• GHD

  growth hormone deficiency

 
• IGF1

  insulin-like growth factor-1

 
• LH

  luteinizing hormone

 
• NSS

  non-suprasellar tumors

 
• QoL

  quality of life

 
• RT

  radiotherapy

 
• SS

  suprasellar tumors

 
• TSH

  thyrotropin (thyroid stimulating hormone)

 
• TSHD

  thyrotropin deficiency

Acknowledgments

The authors are grateful to Magali Viaud for her help and assessment in ethical regulations concerning this study, and to the medical and nursing staff at Hôpital Necker—Enfants Malades. We thank also Dr Delphine Jaquet for her careful and thorough reading of the manuscript. We would like to acknowledge Prof. Michel Zerah, who contributed largely to our study through the care of many of the patients included, during his many years working in the Neurosurgery Unit at Hôpital Necker, and who passed away in September of 2021.

Author Contributions

Laura G. González Briceño and Michel Polak wrote the manuscript. All other authors performed follow-up of included patients and revised and approved the manuscript.

Financial Support

L.G.G.B. received partial funding for this study from Novo-Nordisk France.

Disclosures

D.K., D.S.B., E.G., J.B., S.B., B.F., S.P., C.S.R., C.A., G.P., M.L.P., S.B., S.A., I.A., K.B., M.B., T.R., T.B., F.D.R., C.T., C.P., C.R., S.J., K.B., A.S., F.B., L.L., L.G.R., D.O., P.T., C.D., J.G., and M.P. have nothing to declare. F.D. has received fees for advisory board roles from Bayer, BMS, Roche, Servier; travel expenses from Bayer, BMS, Roche; and consultancy roles from Servier.

Data Availability

Some dataset generated during the current study are not publicly available but are available from the corresponding author on reasonable request.

References

Author notes