Pituitary Gland Duplication Syndrome: An International Imaging Analysis

DISCUSSION

Fewer than 60 cases of DPG have been described in the literature to date and a complete analysis of the radiologic findings based on a series of patients is lacking.⁸ We now contribute with a detailed analysis of 10 patients harboring a complex, but remarkably repetitive constellation of craniofacial abnormalities. Notably, all our patients presented with thickening of the ventral midline of the hypothalamus, which defines the DPG syndrome and was a criterion for inclusion in this study. DPG should be strongly suspected whenever this feature is noted, prompting assessment of the pituitary region.⁴ This malformation should not be mistaken for hypothalamic hamartoma, which presents in a different clinical context (often with gelastic seizures). In addition, mammillary bodies are often involved in hypothalamic hamartoma, while this is not the case in DPG. Axial images are useful to depict the lateral displacement of the mammillary bodies due to the prominence of the ventral hypothalamic midline in DPG. However, the most important findings that impact patient management are craniofacial dysraphism, the presence of an oropharyngeal mass (ie, teratoma), and associated vascular and vertebral anomalies.¹ In particular, a short basilar artery and other vascular anomalies in association with a limited ventral myeloschisis of the medulla oblongata have not previously been described in such detail.

We noted a strong female preponderance in our cohort (1 boy among 10 patients) but also in the available literature.^6,8,9 Of 52 reported patients, only 7 were boys, which in addition to our series compounds a striking 87.1% female prevalence. The cause of this phenomenon remains elusive, as does the genetic background to the DPG syndrome. One of our patients harbored EFNB1 abnormalities. Significantly, EFNB1 codes for the surface molecule EphrinB1, an important system for morphogenesis of the prechordal plate and notochord in zebrafish,¹⁰ and implicated in an X-linked dominant form of craniofrontonasal syndrome (CFNS) with greater severity in heterozygous females than in hemizygous males.^11,12 In a recently reported series of CFNS, brain abnormalities such as dysplastic corpus callosum were reported, but no mention was made of pituitary gland abnormalities and the images were not available for review.¹³ The remaining genetic variants that were found have unclear clinical significance and do not seem to be associated with craniofacial syndromes.^14⇓–16 We therefore surmise they were not pathogenic.

The radiologic appearances of DPG in our cohort differ from other malformations involving the third ventricle and pituitary region, for example, the persisting embryonal infundibular recess (Supplemental Data).¹⁷ On coronal images, there is apparent duplication of the pituitary stalk due to third ventricle floor extending into the pituitary stalk. Persisting embryonal infundibular recess may be associated with morning glory disc anomaly. Although this malformation is typically associated with a normal hypothalamic ventral midline, a recent report suggests that there may be a continuum.¹⁸ DPG also differs from other syndromic associations with vertebral anomalies, such as VACTERL or the oculo-auriculo-vertebral spectrum.^19,20

Interpreting human malformation syndromes in terms of their embryologic origin is difficult because the phenotype is a result of the causal mechanism but also subsequent events in development. Furthermore, to the best of our knowledge, only 1 case of pituitary duplication has been studied neuropathologically.²¹ All other cases, including ours, have only been assessed by using neuroimaging, and therefore any assumption regarding the ultimate cause of this syndrome can only be retrospective. Still, some considerations can be advanced as working hypotheses awaiting confirmation from further experimental studies.

One of the salient features of these cases is the duplication of the pituitary gland (both, adeno- and neurohypophysis) and pituitary stalk. Furthermore, on coronal sections, the infundibulum is also duplicated.^1,21,22 Thus, this malformation could be interpreted as a duplication of the most rostral ventral midline; in fact, the 2 duplicated lines are not located on a sagittal plane but a parasagittal position on either side of the midline. It has long been demonstrated, first in chickens and then in mammals, that the notochord plays a major role in the induction of the ventral midline of the neural tube.²³ After transplantation of an additional notochord, the neural tube develops 2 floor plates (ie, ventral midline); between these lies a region containing motor neurons (basal plate of the neural tube). At the cranial extremity, the prechordal plate (the most cranial segment of the axial mesoderm) is mandatory for pituitary gland development.²⁴ Furthermore, induction of an additional gland could be elicited by either axial mesoderm or ventral diencephalon.²⁵ Finally, the secreted molecule Sonic Hedgehog (SHH), which is produced by both axial mesoderm and ventral midline of the neural tube, can control forebrain organization.²⁶ All this evidence indicates that the rostral axial mesoderm plays a role similar to that of the notochord for more caudal levels. We therefore propose that duplication of the pituitary gland, pituitary stalk, and infundibulum could be secondary to prechordal plate duplication. Furthermore, such a duplication of the axial mesoderm will lead to an excess of SHH secretion.

The typical appearance of a prominent hypothalamic ventral midline has been variably interpreted as hypothalamic hamartoma,²⁷ or thickened tuber cinereum by tubero-mammillary fusion.^3,6 We disagree with these interpretations and propose an alternative model. By analogy with the results published previously,²³ the midline tissue could correspond to the basal wall of the neural tube (ie, ventral hypothalamus). In 1983, Hori²¹ described the cells that are present in this tissue as magnocellular and medium-sized neurons associated with fibers; these large neurons are normally found in some of the hypothalamic nuclei. Unfortunately, Hori was unable to further characterize these cells since immunohistological techniques were unavailable at that time. However, his results support the theory that the thickened midline tissue is composed of supernumerary hypothalamic nuclei.

The presence of segmentation anomalies of the cervical spine, and particularly the zipperlike appearance in coronal planes, suggests that duplication of the axial mesoderm extends caudally to the cervical notochord. This feature reinforces our interpretation of axial mesoderm duplication. This is also the case for the apparent split spinal cord observed in patient 6 (Supplemental Data), further lending support to the hypothesis of SHH overexpression.

Teratoma (lying either in the oropharyngeal or nasopharyngeal space) was observed in several of our cases (patients 4, 6, and 7) and is also an inconstant finding in the DPG plus syndrome in the literature.² Teratomas are thought to arise from embryonic stem cells first induced by proliferation and then differentiation. Interestingly, SHH is a promoter of embryonic stem cell proliferation.²⁸ Consequently, an excess of SHH could be the trigger for teratoma initiation.

An anomaly of the basilar artery was found in all our cases. This vascular anomaly was first described in association with DPG by Shroff et al⁶ in 2003. In the mouse, Gli2 codes for GLI2, the main activator of SHH pathway. Gli2 -/- mouse lines, for which no GLI2 protein is produced, selectively lack the basilar artery,²⁹ demonstrating the dependence of this vessel on the SHH pathway. Thus, complete, or partial duplication of the basilar artery could be a more likely explanation than a defective fusion. Furthermore, one can postulate that duplication of the basilar artery could result from notochord duplication or increased SHH signaling.

Duplication of the uvula and/or tongue, partial duplication of the mouth, or additional teeth were observed in 3 cases (patients 3, 4, and 7) and have been reported in DPG-plus syndrome.² It is well known that SHH is mandatory for mouth opening and growth in both Xenopus and mouse.³⁰ We propose that duplication of the oral structure could be associated with excess of SHH. The significance of palatal or orofacial clefts is more complex. Indeed, the presence of a tissue mass in the nasopharynx could prevent fusion of the palatal shelves as well as cause tongue duplication. However, it is interesting to note that an excess of SHH does generate such defects in the mouse embryo.^31,32 Additional data are needed to understand the cause (or the causes) of such clefts.

Ciliopathies are associated with SHH defects and may aid in understanding some of the malformations observed in the DPG-plus syndrome. Ciliopathies represent a group of diverse diseases due to defective primary cilia, a cell organelle playing important roles in mechanoreception and in signal transduction. Notably, perturbation of cilia leads to SHH signaling defects with consequences on the ventral neural tube. By contrast, ciliary defects also impair proteasomal activity, relevant for the cleavage of GLI3 into its repressor form (GLI3R). Since GLI3R is a repressor of SHH signaling, the consequence will be an increase of SHH in the dorsal aspect of the neural tube. Defect of GLI3R processing has been proved as the cause of both olfactory bulb and corpus callosum anomalies in ciliopathies,^33,34 and consequently an excess of SHH might also explain olfactory and callosal defects observed in our cases. The significance of the small anterior-to-posterior diameter of the corpus callosum, present in 9 of 10 patients analyzed, remains uncertain. Recently, it was shown that a short corpus callosum on prenatal imaging does not correlate with a specific genetic abnormality if present in isolation.³⁵ However, in our patients, associated callosal dysgenesis was a frequent finding and related to a genetic abnormality.

Another interesting feature linking the DPG-plus syndrome with ciliopathy is the anterior mesencephalic cap dysplasia,³⁶ or interpeduncular heterotopia.³⁷ This abnormal tissue, lying in the peduncular fossa, is formed by aberrant nervous fibers,³⁶ suggesting a defect of axonal guidance, and has been first described in Joubert syndrome, the main ciliopathy affecting the brain. Further anomalies have also been found in ciliopathies, such as absence of decussation of the superior cerebellar peduncles³⁸ and abnormal pontine fibers.³⁶ To the best of our knowledge, however, an abnormal anterior-posterior pontine bundle (Supplemental Data) has not yet been reported.

We also consistently observed a short clivus associated with a tissue stalk, or tethering band, inserted on the ventral surface of the pons or the medulla, which we termed limited ventral myeloschisis because it recalls the typical dorsal tethering bands found in limited dorsal myeloschisis. Such a defect was described by Manjila et al² in 2012. and interpreted as a clival encephalocele, however without any pathologic proof. Significantly, however, already in 1983 Hori²¹ reported the same finding in the sole histopathologic description of a DPG case and found that the stalk was not composed of neural tissue but of connective and muscular tissues. The role of notochord on the development of the clivus ultimately remains obscure and needs further experiments in mammals.

The main limitation of our study is its small sample size, which may underestimate or overestimate the prevalence of the imaging findings described in our report.