ORIGINAL ARTICLE

A retrospective review of pituitary MRI findings in childrenon growth hormone therapy

Sarah L. Tsai & Eoghan Laffan & Sarah Lawrence

Received: 23 June 2011 /Revised: 29 November 2011 /Accepted: 3 January 2012 /Published online: 2 February 2012# Springer-Verlag 2012

AbstractBackground Patients with congenital hypopituitarism mighthave the classic triad of pituitary stalk interruption syn-drome, which consists of: (1) an interrupted or thin pituitarystalk, (2) an absent or ectopic posterior pituitary (EPP), and(3) anterior pituitary hypoplasia or aplasia.Objective To examine the relationship between pituitaryanatomy and the degree of hormonal dysfunction.Materials and methods This study involved a retrospectivereview of MRI findings in all children diagnosed withcongenital growth hormone deficiency from 1988 to 2010at a tertiary-level pediatric hospital.Results Of the 52 MRIs reviewed in 52 children, 26 chil-dren had normal pituitary anatomy and 26 had one or moreelements of the classic triad. Fourteen of fifteen childrenwith multiple pituitary hormone deficiencies had structuralanomalies on MRI. Twelve of 37 children with isolatedgrowth hormone deficiency had an abnormal MRI.Conclusion Children with multiple pituitary hormone defi-ciencies were more likely to have the classic triad thanchildren with isolated growth hormone deficiency. A normalMRI was the most common finding in children with isolatedgrowth hormone deficiency.

Keywords PituitaryMRI . Hypopituitarism . Congenitalgrowth hormone deficiency . Classic triad

Introduction

MRI is a useful modality for delineating pituitary anatomy.Children with congenital hypopituitarism might have theclassic triad of pituitary stalk interruption syndrome, whichconsists of: (1) an interrupted or thin pituitary stalk, (2) anabsent or ectopic posterior pituitary (EPP), and (3) anteriorpituitary hypoplasia or aplasia [1] (Figs. 1, 2, 3 and 4).Children with idiopathic growth hormone deficiency(GHD) have been shown to have a high incidence of pitu-itary abnormalities [2]. MRI findings have a role in identi-fying children who are at higher risk for developingadditional hormone deficiencies, as these patients have morefrequent anatomical abnormalities of the pituitary than thosewith isolated growth hormone deficiency (IGHD) [2–5]. Forthis reason, accurate MRI interpretation is important for theanticipation of endocrine dysfunction.

Idiopathic GHD is a well-recognized endocrinologicalcause of short stature. Growth hormone deficiency can beseen in isolation (IGHD) or in association with multiplepituitary hormone deficiencies (MPHD). GHD can be spo-radic, associated with a familial syndrome, or a geneticdefect.

The purpose of this study was to compare pituitary MRIfindings in children with isolated congenital growth hor-mone deficiency to those with multiple congenital pituitaryhormone deficiencies.

Materials and methods

Children who had been treated with growth hormone from1988 to 2010 were identified using the Children’s Hospitalof Eastern Ontario Division of Endocrinology clinical data-base. Children were excluded if they had secondary GHD

S. L. Tsai (*) : S. LawrenceDivision of Endocrinology, Children’s Hospital of Eastern Ontario,University of Ottawa,401 Smyth Road,Ottawa, Canada K1H 8L1e-mail: s.lenihan.tsai@gmail.com

E. LaffanPediatric Radiology, Children’s University Hospital,Dublin 1, Ireland

Pediatr Radiol (2012) 42:799–804DOI 10.1007/s00247-012-2349-7

caused by intracranial lesion, surgery, radiation or infiltra-tion. Children were also excluded if they were on growthhormone treatment for Turner syndrome, renal failure oridiopathic short stature. This study was approved by theresearch ethics board.

The diagnosis of GHD was based on clinical findings ofshort stature (<2 standard deviations below mean for age andsex), decreased growth velocity (<10th percentile) and poorgrowth hormone response to two stimulation tests (L-arginine,clonidine, or L-dopa/propranolol) defined as a peak GHof <8 μg/L pre-2007 and peak GH of <5.6 μg/L post-

2007 on a two-site immunoenzymatic assay (Unicel® DXL800; Beckman Coulter, Mississauga, ON, Canada). Otheranterior pituitary hormones were assessed by screening ofthyroid-stimulating hormone, free thyroxine, and morningcortisol concentrations. Luteinizing hormone and follicle-stimulating hormone levels were measured in children in thepubertal age range if clinically indicated based on pubertaldelay. Further stimulation testing (i.e. adrenocorticotropichormone, luteinizing hormone-releasing hormone) was under-taken if clinically indicated.

Seventy-two children had congenital GHD, and 52 had abrain MRI available for review. Of these 52 children, 37 hadisolated growth hormone deficiency and 15 had growthhormone deficiency with one or more additional anteriorpituitary hormone deficiencies.

MRI studies were performed with one of two 1.5-teslaunits (GE Healthcare, Milwaukee, WI, USA). Both magnetswere upgraded during the study time period from a SignaHorizon 10.0 and a Signa Excite 12.0 to Signa HDxt 1.5-teslascanners. The imaging protocol included a routine brain pro-tocol and dynamic pituitary imaging protocol. The brain pro-tocol consisted of a sagittal T1-W FLAIR, axial T2-W FSE,axial T2-W FLAIR and coronal T1-W SPGR. Dynamicpituitary imaging consisted of a coronal T1-W FSE pre-contrast scan and repeated five times immediately post-contrast administration (TR0425, minimum TE, ETL04,5 slices per acquisition, slice thickness 2 mm with nogap, matrix 256 × 192, to 512). Sagittal T1-W FSEimaging was also performed once with similar parame-ters to the coronal T1-W FSE imaging. Gadolinium was

Fig. 1 Sagittal T1-W MR image of a 30-day-old boy with growthhormone deficiency. Note the ectopic posterior pituitary bright spot(white arrow) and hypoplastic pituitary (white arrowhead)

Fig. 2 Sagittal T1-W MR image spoiled gradient recalled (SPGR)image of an infant with classic triad—pituitary hypoplasia, absent stalkand ectopic posterior pituitary (white arrow)

Fig. 3 Sagittal T1-W image of an infant with a normal pituitary gland(white arrow)

800 Pediatr Radiol (2012) 42:799–804

administered at 0.1 mmol/kg (Omniscan®, GE HealthcareCanada Inc., Ontario, Canada).

The images were originally read by staff radiologists asper institution protocol. The MRIs were re-reviewed by asingle pediatric neuroradiologist. If there was a disagree-ment with the original report, the film was re-reviewed afteran elapsed period of 6 months by the same neuroradiologist.The reviewing neuroradiologist was not blinded to the pre-vious report or the patient’s hormone status.

A qualitative evaluation was done to examine to location,shape and signal intensity of the anterior and posteriorpituitary gland and adjacent structures. Children were con-sidered to have the classic triad if they had all three criteriadescribed above. The pituitary stalk was described as nor-mal, thin, interrupted or absent. Normal and thin stalks werearbitrarily defined as being readily identifiable or barelydiscernible on sagittal and coronal MR sequences.Pituitary hypoplasia was defined as a pituitary that wasmuch smaller than expected within the sella. The rest ofthe brain was also evaluated for associated malformations,such as Chiari I malformation or optic nerve hypoplasia.Children were described as having a classic triad variant ifthey met only one or two of the criteria. Additional infor-mation such as age, sex, age at diagnosis, date of growthhormone commencement, and dates and quality of MRI(s)were noted for each child.

Results

The films of 52 children with GHD were reviewed, and ofthese 26 (50%) had normal pituitary anatomy (25 withIGHD and 1 with MPHD). Some children in the normalMRI group had subtle abnormalities of unknown signifi-cance, e.g., mild asymmetry to the pituitary gland, or aposterior pituitary bright spot that was subtly abnormal inappearance.

Of the remaining 26 children, 13 had the classic triad (9with MPHD and 4 with IGHD). In children with MPHD, theclassic triad was the most common anatomical abnormality(60%), while the classic triad was found in only 4 of the 37children with IGHD (10.8%). See Table 1 for a summary offindings.

Classic triad variants were seen in 13 children. Of these,five had MPHD and their results are as follows: three hadsepto-optic dysplasia (SOD) and two of the classic triadcriteria, one had EPP and anterior pituitary hypoplasia, andone had isolated EPP. Eight children with IGHD had classictriad variants: four had two of the classic triad criteria (EPPand anterior pituitary hypoplasia), and four had one of theclassic triad criteria (isolated anterior pituitary hypoplasia inthree and isolated ectopic posterior pituitary in one). Of

Fig. 4 Images show characteristics of classic triad. a Coronal T1-Wimage shows ectopic pituitary gland (white arrow). b Axial imageshows a single central incisor (white arrow). c Sagittal T1-W imageof the same child shows ectopic posterior pituitary and absent stalk(black arrow) and anterior pituitary hypoplasia (black arrowhead)

Pediatr Radiol (2012) 42:799–804 801

note, pituitary stalk abnormality was not a feature in anycases of classic triad variant in children with IGHD.

Discussion

MRI has become the single most important imaging modal-ity in the investigation of congenital and structural pituitarygland disorders. Rapid technological advances with strongermagnets, greater field strengths and post-processing softwarehave led to shorter scan times and greatly improved spatialresolution. Sellar structures are very small; therefore, it isimportant to obtain thin-section images [6]. Two- or 3-mmscan thickness using very small or no gap between sectionswas traditionally employed for 1.5-T MRIs [7]. In contrast,modern MRI utilizes 1- or 1.5-mm sections or volumetricscans for even more detailed images of pituitary anatomyand surrounding structures [6].

T1-weighted sequences are mandatory in all patients. Avolumetric sequence should be strongly considered, whereall three planes can be reconstructed later during post-processing. T1 is important as the posterior pituitary isclassically high signal on T1-weighted imaging, known asthe posterior pituitary bright spot and easily discernible fromthe anterior pituitary, which is similar in signal intensity tothe adjacent gray matter. To detect an ectopic posteriorpituitary, the radiologist simply has to find this bright spot,if not in the sella then either along the stalk or at thehypothalamus. Care must be taken when interpreting thepituitary in a neonatal MRI, where the whole pituitary isbright until 2 to 3 months of age.

The size and shape of the pituitary is dependent on ageand gender. The gland is larger in the neonatal period than inchildhood and is generally larger in girls than boys [8]. Inchildhood, the height in the sagittal plane is 2–6 mm [9]. Inadolescence, the gland may swell to a height of 10 mm ingirls and 7–8 mm in boys [10]. However, there is consider-able variability in gland shape, size and signal among nor-mal subjects of identical age and gender [7].

Evaluation of the size of the pediatric gland is complex.The most basic technique involves measuring the maximum

height of the gland [11]. The volume can be calculated bymultiplying the maximal length, width, height and correc-tion factor (0.524 or π/6); however, this is imprecise forglands with an unusual shape, as this formula is based onspherical volume calculation [6]. The product of the sum-mation of the cross-sectional areas of the gland by slice-widthhas been shown to produce robust results in normal prepuber-tal children; however, the direct measures of volume did notcorrelate well with either one-dimensional estimates (height)or 2-dimensional indirectly calculated volumes [12].Numerous studies have shown variability in gland shapeduring childhood, making it difficult to effectively assesswhether a gland is normal in size for age. Ranges of normalvary among investigators, and for this reason definingnormal values for the size of the anterior pituitary orposterior pituitary is very difficult. In our cases, evalu-ation of the pituitary gland was based primarily onradiologist experience. We were unable to find a defi-nition of pituitary hypoplasia in the literature. We arbi-trarily defined it as a gland that appeared much smallerthan expected within the sella. In practice, all radiolog-ists reporting pediatric brain MRIs should familiarizethemselves with the appearance, size and shape of thenormal pituitary gland in all cases and comment on thegland in all reports.

Normal values have been established for each section ofthe pituitary stalk in adults [13]. Unfortunately, similarstandards do not exist in children. Assessment of the stalkgenerally depends on the experience of the radiologist ratherthan age-appropriate reference ranges [6]. In general, a thinstalk is less than 1 mm at any point. [6]. Thin MR slices atthe sella are therefore crucially important. The stalk could

Table 1 MRI findings in children with isolated growth hormonedeficiency and multiple pituitary hormone deficiencies

Isolated growthhormone deficiency

Multiple pituitaryhormone deficiencies

Total

Normal MRI 25 1 26

Classic triad 4 9 13

Classic triadvariant

8 5 13

Total children 37 15 52

Fig. 5 Sagittal T1-W image of child with empty sella syndrome (whitearrow) illustrated by dilated third and lateral ventricles because of theobstruction of the fourth ventricle by a large posterior fossa tumor(white arrowhead)

802 Pediatr Radiol (2012) 42:799–804

also be visualized by the use of contrast enhancement,although in practice, good-quality thin-slice imaging willalmost always suffice.

Of note, pituitary hypoplasia is not equivalent to emptysella syndrome. Empty sella occurs when the superior por-tion of the sella turcica appears to be filled with cerebrospi-nal fluid (CSF) and the pituitary gland is flattened inappearance (Fig. 5). This condition can be caused by a thinor absent sellar diaphragm, which leads to enlargementsecondary to chronic CSF pulsation [8]. Patients generallycontinue to have normal pituitary function and are asymp-tomatic. In some patients, empty sella can be associated withpituitary hormone abnormalities, CSF rhinorrhea or visualfield abnormalities.

The variations in pituitary anatomy identified in thisgroup of patients with growth hormone deficiency are con-sistent with previous studies. One patient with MPHD had anormal MRI, which is consistent with previous studieswhere 0–7% of patients with MPHD had normal MRIs[2–5]. Sixty-seven percent of our patients with IGHD hada normal MRI, which is similar to previous studies wherenormal pituitary anatomy was seen on MRI in 20–70% ofpatients [2–5].

The classic triad was seen in 60% of patients with MPHDand 10.8% of patients with IGHD. Studies have consistentlyshown that this finding is more common in patients withMPHD than in those with IGHD [2–4]. These resultssuggest that close follow-up and long-term screening isimportant with the classic triad and apparent IGHD, asthese patients might be more likely to develop addition-al pituitary hormone deficiencies.

Isolated anterior hypoplasia was seen in three childrenwith IGHD and in no patients with MPHD. Isolated anteriorhypoplasia is very rarely reported in patients with MPHD[3]. A normal MRI or isolated anterior pituitary hypoplasiain patients with growth hormone deficiency is suggestivethat the patient has an isolated deficiency; however, giventhe rare possibility that such patients will develop otherpituitary hormone deficiencies, continued screening shouldbe carried out.

Interrupted pituitary stalk was the least common elementof the classic triad to be identified in classic triad variants.No patients with IGHD and a classic triad variant had anabnormal pituitary stalk. Overall, an interrupted stalk wasfound in 66.6% of the patients with MPHD and in 10.8% ofpatients with IGHD as part of the classic triad.

Several genes, such as LHX4 and HESX1 are importantin early pituitary development [1]. Certain genetic defectsare associated more commonly with particular pituitaryhormone deficiencies and anatomical variations. As veryfew of our patients had genetic testing for these mutations,it was not possible to extrapolate data regarding the associ-ation between pituitary anatomy and genotype.

This study has several limitations. First, the reviewingpediatric neuroradiologist was not blinded to the previousMRI report or the hormone status of the patient. Pituitaryhypoplasia and the thin stalk were qualitatively assessed butnot defined by measurement. The image quality of thestudies available on hard copy might have limited interpre-tation in comparison to those on a PACS workstation. TheMRI software was upgraded significantly during the studyperiod.

Conclusion

MRI of the pituitary gland in patients with growth hormonedeficiency is important as this can guide a clinician withrespect to the likelihood of other hormone deficiencies andthe need for surveillance. Children with normal MRI andisolated anterior pituitary hormone hypoplasia are morelikely to have isolated growth hormone deficiency. Theclassic triad is a more common finding in patients withmultiple pituitary hormone deficiencies than in those withisolated growth hormone deficiency and can guide the inten-sity of ongoing endocrine surveillance.

Acknowledgements This work was presented at the following meet-ings: the Endocrine Society Meeting, June 2010, San Diego, CA, USA(poster); the Canadian Pediatric Endocrine Group Meeting, February2011, Toronto, Canada (oral presentation); and the European Society ofPediatric Radiology Meeting, June 2010, Bordeaux, France (oralpresentation).

We would like to thank Dr. Margaret Lawson for assisting withgeneration of research questions, Ms. Brenda Fraser, RN, for assistancewith the patient database, and Dr. Jeffrey Tsai for review of manuscript.

References

1. Beukhof CM, Lequin MH, Drop ST (2007) The classic triad andvariants. In: Deal C (ed) MRI in congenital hypopituitarism. LillyGenesis References Series, Chicago, pp 69–80

2. Maghnie M, Triulzi F, Larizza D et al (1991) Hypothalamic-pituitary dysfunction in growth hormone-deficient patients withpituitary abnormalities. J Clin Endo Metab 73:79–83

3. Abrahams JJ, Trefelner E, Boulware SD (1991) Idiopathic growthhormone deficiency: MR findings in 35 patients. AJNR 12:155–160

4. Hamilton J, Chitayat D, Blaser S et al (1998) Familial growthhormone deficiency associated with MRI abnormalities. Am JMed Genetics 80:128–132

5. Argyropoulou M, Perignon F, Brauner R et al (1992) Magneticresonance imaging in the diagnosis of growth hormone deficiency.J Pediatr 20:886–891

6. Delman BN (2009) Imaging of pediatric pituitary abnormalities.Endocrinol Metab Clin N Am 38:673–698

7. Delman BN, Fatterpekar GM, Law M et al (2008) Neuroimagingfor the pediatric endocrinologist. Pediatr Endocrinol Rev 5:708–719

8. Ouyang T, Rothfus WE, Ng JM et al (2011) Imaging of thepituitary. Radiol Clin N Am 49:549–571

Pediatr Radiol (2012) 42:799–804 803

9. Hayakawa K, Konishi Y, Matsuda T et al (1989) Development andaging of brain midline structures: assessment with MR imaging.Radiology 172:171–177

10. Elster AD, Chen MY, Williams DW III et al (1990) Pituitary gland:MR imaging of physiologic hypertrophy in adolescence.Radiology 174:681–685

11. Argyropoulou M, Perignon F, Brunelle F et al (1991) Heightof normal pituitary gland as a function of age evaluated by

magnetic resonance imaging in children. Pediatr Radiol21:247–249

12. Fink AM, Vidmar S, Kumbla S et al (2005) Age-related pituitaryvolumes in prepubertal children with normal endocrine function:volumetric magnetic resonance data. J Clin Endocrinol Metab90:3274–3278

13. Rupp D, Molitch M (2008) Pituitary stalk lesions. Curr OpinEndocrinol Diabetes Obes 15:339–345

804 Pediatr Radiol (2012) 42:799–804