ESPE Abstracts

Background: Hypothalamic dysfunction (HD) is life-threatening but precise diagnostic tools are lacking. Normal hypothalamic anatomy is difficult to delineate on MRI. Damage to the area is inferred from a visible lesion, but how widely it disturbs signalling connections or correlates with symptoms is unclear. Furthermore, in congenital/syndromic diseases the hypothalamus appears normal even in cases with clear HD. We aimed to develop novel clinical and radiological tools to quantify the spectrum of HD in populations at risk.

Patients and Methods: 62 children (28F,34M) followed for 8.0±4.0 years for hypothalamo-pituitary (H-P) disorders (34 congenital, 28 acquired), were recruited for study aged 12.5±2.9 years and prospectively screened for features of HD [disordered appetite, thirst, sleep, thermoregulation, and abnormal BMI SDS (<-2 or >+2)]. HD was considered “clinically likely” when there were either ≥3 features or ≥2 and a clear hypothalamic mass/syndrome. 3T high-resolution volumetric MR brain sequences (T1, T2) were prospectively acquired. A protocol to manually segment the hypothalamus was applied to children for the first time to identify and measure regions of interest. In patients with mass lesions, a novel injury score - ranging from 0 to 12 (no to total involvement) based on subsegmentagion of hypothalamic regions - was developed. Imaging data were compared with 25 age- and gender-matched controls (13F,12M, aged 13.6±3.1 years).

Results: The prevalence of “clinically likely HD” was similar in congenital vs acquired disorders (61.8%vs75.0%). 26 patients without visible hypothalamic lesions were subdivided into 13 with (Group1) and 13 without (Group2) “clinically likely HD”. Group1 (6 congenital hypopituitarism, 3 ROHHAD(NET), 1 SRC1 variant, 3 SOD) had smaller hypothalami than Group2 (11 congenital hypopituitarism, 2 SOD) (759.2±130.6 vs 984.0±160.0 mm3, P=0.046) with Group2 being similar to controls (922.5±122.3 mm3). In 36 patients with mass lesions (13 OPGs, 12 craniopharyngiomas, 7 hamartomas, 3 germinomas, 1 arachnoid cyst) the hypothalamic injury score was higher in those with “clinically likely HD” than in those without (6.6±3.3 vs 3.6±2.8, P=0.024).

Conclusions: When prospectively assessed, HD is highly prevalent in both patients with congenital H-P disorders and hypothalamic tumours. We have for the first time to our knowledge manually segmented the hypothalamus to improve identification and quantification of hypothalamic injury on imaging. Our tools developed to assess congenital (reduced volumes) and acquired (hypothalamic score) injury appear discriminatory in defining severity of HD phenotype and may facilitate diagnostic accuracy and development of targeted therapies.