Central diabetes insipidus (CDI) is a rare entity in the paediatric age group, with an annual incidence of 4 patients per 100000 individuals.1,2 It is due to the deficient secretion or release of antidiuretic hormone (ADH), also known as vasopressin, and polyuria and polydipsia are the most frequent presenting symptoms at the time of diagnosis.3 It needs to be differentiated from nephrogenic diabetes insipidus, due to resistance of the kidneys to ADH, and from primary polyuria due to excessive fluid intake.

The aetiology of CDI in the paediatric population is heterogeneous and includes forms secondary to central nervous system tumours (germinoma or craniopharyngioma, among others), cranial malformations, neurosurgery, head trauma, autoimmune hypophysitis or diseases that cause infiltration in the hypothalamic-pituitary region (neoplasia, sarcoidosis or histiocytosis). Genetic aetiologies are infrequent (genetic defects in ADH synthesis—mutation in the AVP gene—that may have an autosomal recessive, autosomal dominant or X-linked pattern of inheritance). Nephrogenic diabetes insipidus is a disorder with X-linked inheritance caused by mutations in the arginine vasopressin receptor 2 gene (AVPR2) or with autosomal or recessive inheritance cause by mutations in the aquaporin 2 gene (AQP2).1

All patients given a diagnosis of CDI should undergo a magnetic resonance imaging (MRI) examination to confirm the diagnosis (absence of the bright spot corresponding to the posterior pituitary in T1-weighted images) and perhaps even determine its aetiology.4 The presence of a thickened pituitary stalk at diagnosis of diabetes insipidus or during the follow-up is suggestive, although no pathognomonic, of pituitary infiltration associated with a tumour. At any rate, a variable percentage of CDI cases, which depending on the case series varies between 20% and 50%, with or without pituitary stalk thickening, are considered idiopathic (ICDI).5,6

The aim of our study was to analyse the clinical characteristics and outcomes of a group of patients given a diagnosis of CDI during childhood or adolescence, and to evaluate the association between pituitary stalk thickening and a neoplastic aetiology and the presence of other hormone deficiencies.

• Sample. We performed a retrospective review of the cases of 15 patients given a diagnosis of CDI between 2003 and 2016 managed at a children's hospital. The diagnosis of CDI was based on a positive personal history of polyuria and polydipsia and the results of the water deprivation test.7 To determine the central aetiology of the diabetes, patients underwent a desmopressin test (DDAVP; 0.1IU/kg delivered intravenously), and CDI was considered confirmed in case of a 50% increase in urine osmolality relative to baseline. In patients what presented with a plasma osmolality greater than 295mOsm/kg and a plasma sodium level greater than 145mEq/L, the water deprivation test was not performed and was replaced by direct administration of vasopressin.

• Clinical and auxological evaluation. We retrieved data on the symptoms, anthropometric measurements and Tanner stage of pubertal development recorded at diagnosis and every 6 months during the follow-up.8,9

• Pituitary function. Every year, serum levels of insulin-like growth factor-1 (IGF-1) and insulin-like growth factor binding protein-3 (IGFBP-3) were measured as indirect markers of growth hormone (GH) deficiency. Also, serum levels of thyroid-stimulating hormone (TSH), free T4 and cortisol and the serum testosterone/estradiol ratio were measured every 6 months. When considered necessary to diagnose a potential deficiency, patients underwent a growth hormone stimulation test (clonidine stimulation test and insulin-induced hypoglycaemia test), gonadotropin stimulation test (luteinising hormone-releasing hormone [LHRH] test) and adrenocorticotropin (ACTH) stimulation test.

- Thyroid-stimulating hormone deficiency was defined as a decreased free T4 level (<0.65ng/dL) accompanied by decreased levels of TSH or a normal TSH value inconsistent with the level of free T4.

- Adrenocorticotropin deficiency was diagnosed in patients with basal serum cortisol levels (8 am) of less than 8μg/dL and a peak cortisol level of less than 18pg/mL in the ACTH stimulation test.

- Gonadotropin deficiency was suspected in male patients with a testicular volume of less than 4mL at age 14 years and female patients that had not exhibited thelarche by age 13 years. When serum levels of sex steroids were in the prepubertal range in these patients, a LHRH stimulation test was performed to confirm the diagnosis of gonadotropin deficiency due to an absent or insufficient response of luteinising hormone (LH) and follicle-stimulating hormone (FSH).

• Neuroimaging. The neuroimaging evaluation consisted of a cranial MRI scan following the routine protocol (T1-weighted sagittal view, FLAIR axial view, T2-weighted coronal view and diffusion-weighted imaging) and a specific study of the pituitary region (T1- and T2-weighted sagittal views, T1-weighted coronal view and, following administration of intravenous paramagnetic contrast, post-contrast T1-weighted sagittal and coronal views). The same protocol was applied when MRI was required during the follow-up. Two neuroradiologists measured the thickness of the proximal pituitary stalk independently, with diameters of less than 3mm considered normal. Pituitary stalk thickening was classified as mild (3–3.9mm), moderate (4–6.5mm) or severe (>6.5mm).2 In patients in whom it had not been possible to establish the aetiology of the diabetes in the initial evaluation, the follow-up included performance of a cranial MRI scan every 6–12 months. Patients that received a diagnosis of neoplasm or histiocytosis underwent follow-up MRI examinations in adherence to the oncology protocol applicable to each case.

We made a retrospective review of the characteristics and clinical outcomes of the 15 patients given a diagnosis of CDI in the past 13 years in our hospital. The study was particularly focused on the presence at diagnosis or subsequent development of a thickened pituitary stalk and its association with the presence or development of other pituitary hormone deficiencies.

In our sample, the time elapsed from onset of symptoms and the aetiological diagnosis of CDI was typically long, of more than 2 years, especially in patients with mild or no abnormal imaging findings at the time of diagnosis of CDI, which can be explained by the challenges in diagnosis a disease with a slow course and localised in a hard-to-access region that has multiple functions, such as the hypothalamic-pituitary region.

Forty percent of the patients in the sample (6–15) received an initial diagnosis of idiopathic CDI; a percentage that dropped in a relatively short period of time (between 1 and 2.2 years) by 50% to only 20% of the patients. This reduction over time in the proportion of cases of CDI considered idiopathic was greater compared to similar studies, which have reported decreases of approximately 20% (18%–22%), with a median duration of follow-up of 2.3–10 years.6,10,11

The final proportion of ICDI (20% in our sample) varies widely among recently published case series,11,12 ranging between the 5.7% reported by Hunter et al.13 and the 52% reported by Maghnie et al.14 This variability is mostly due to the heterogeneity of the follow-up protocols applied to these patients, especially in relation to the interval at which follow-up MRI examinations are performed, as shorter time intervals, especially in the first few years after diagnosis (in our hospital, ever 6–12 months for at least the first 2–3 years) allow the early detection of pituitary stalk thickening, thus manifesting the presence of an underlying disease, frequently of a neoplastic nature and potentially severe.15,16 However, the detection of pituitary stalk thickening, especially if mild, does not necessarily correspond to the presence of a tumour. Two of our patients with suspected ICDI developed mild thickening of the stalk during the follow-up that eventually resolved spontaneously. These transient thickening suggests the possibility of hypophysitis of an inflammatory and/or autoimmune aetiology.17 There have been occasional reports of the detection of antibodies against vasopressin and other antibodies in this context whose measurement could be useful in the diagnosis of some patients, but which was not performed in our sample. This clinical picture is associated with a low risk of developing histiocytosis or germinoma in the future.18

In our sample, the most frequent aetiology was neoplastic, identified in 60% at the time of diagnosis and in 80% by the end of follow-up.19 Early detection of thickening is essential, as our case series demonstrates that the presence of a thickened pituitary stalk is associated with an increased risk of underlying tumours. In the case series presented by Werny et al.,6 the presence of a thickened pituitary stalk at diagnosis was significantly associated with a neoplastic aetiology.

Germinomas amount to approximately 5% of all central nervous system tumours.20 In our study, it was the most frequent cause of CDI, accounting for 60% of cases, although other studies with larger samples have reported lower percentages (10%-38%).21,22 Thickening of part or all of the pituitary stalk may be the sole presenting feature of small germinomas.10

When it came to the patients with LCH, two ouf of three had a thickened pituitary stalk at the time of diagnosis of CDI; Grois et al.23 found this feature in the initial evaluation in 71% of cases of CDI secondary to LCH. This difference could be due to the small number of patients given an initial diagnosis of histiocytosis. We did not find any cases of familial neurogenic diabetes insipidus or CDI of a genetic aetiology, which was consistent with previous studies11,13; in larger studies, monogenic forms account for less than 10% of the total cases of CDI.6,10,14

In our study, the hyperintense signal characteristic of the posterior pituitary was absent in the initial MRI scan of 100% patients with CDI, but this was a higher percentage compared to studies with larger samples (81%6–94%14). Thus, the presence of this bright spot does not rule out CDI.23

The hormone evaluation revealed more than 1 deficiency in 67% of our patients, which was consistent with previous studies that have reported percentages of patients with more than 1 hormone deficiency ranging from 53.8% (Liu et al.)12 to 81.4% (Di Iorgi et al.).10 The most frequent deficiency concerned GH, which was also consistent with the literature.

Another test that should be performed in patients with CDI of suspected neoplastic aetiology is the measurement of AFP and β-HCG levels in plasma and CSF, which are tumour markers.24 As observed in other studies, most patients with germinoma had normal serum levels of AFP and β-HCG with elevation in CSF. The measurement of these markers in plasma show low sensivity sensitive as in CSH. Thus, they should be measured routinely in CSF, especially considering that their elevation is highly suggestive of a germ cell tumour.25

Based on a detailed analysis of the data in our study and a review of the literature, we may conclude the following:

• •

  All patients with CDI require monitoring of auxologic parameters and hormone levels at least every 6 months to identify pituitary deficiencies that can develop gradually over time, especially in the first 2 years following diagnosis of CDI, with particular emphasis on GH deficiency on account of its high frequency.

• •

  The follow-up of patients with ICDI should include a MRI scan every 6–12 months based on the detected hormonal abnormalities, at least for the first 2–3 years after diagnosis, as 50% of the patients in our sample received a diagnosis of germinoma and histiocytosis in this time interval.

• •

  The presence and degree of pituitary stalk thickening in patients with CDI could be directly correlated to the presence or future development of other hormone deficiencies.

The authors have no conflicts of interest to declare.