Blastogenesis and Gastrulation (PODs 1 to 13)

During the first 2 weeks of embryogenesis the human embryo undergoes a number of cell divisions and cellular rearrangements (blastogenesis) that ultimately result in the formation of a blastocyst situated eccentrically within a hollow sphere of trophoblast cells (Fig. 175-1A). The blastocyst later forms a two-layered structure with a dorsal layer, the epiblast, adjacent to the amnionic cavity and a ventral layer, the hypoblast, adjacent to the yolk sac.³ A thickening, the prochordal plate, identifies the cranial end of the embryo (Fig. 175-1B).

FIGURE 175-1 Development of the blastocyst—midsagittal illustrations. A, Continued proliferation of cells produces a sphere containing a blastocystic cavity surrounded by an eccentrically located inner cell mass and a surrounding ring of trophoblast cells. B, The inner cell mass develops further into a two-layered structure, the blastodisc, which contains the epiblast adjacent to the amnionic cavity and the hypoblast adjacent to the yolk sac. C, With further development, the blastodisc thickens cranially to form the prochordal plate.

Gastrulation begins with the appearance of a midline structure, the primitive streak (PS), at the caudal end of the embryo; the cranial end of the primitive streak is called Hensen’s node. A midline trough, the primitive groove, runs in the midline within the PS; the cranial end of the primitive groove is the primitive pit. The PS elongates cranially over the next 3 days and forms a midline structure in the caudal half of the embryo (Fig. 175-2A). The PS then begins to regress (shorten) back toward the caudal end of the embryo.³ As gastrulation proceeds, cells of the epiblast migrate toward the PS and invaginate through the primitive groove (Fig. 175-2A). The first cells to ingress (while the PS is still elongating) are endodermal cells, which displace the hypoblast cells laterally and form prospective endoderm (the displaced hypoblast cells form extraembryonic tissues).^6–10 As the PS regresses, mesodermal cells ingress through the PS between the epiblast and newly formed endoderm and become the mesoderm.^9,10 The remaining epiblast cells spread out to replace the cells that have ingressed through the primitive groove and thereby form both the neuroectoderm (the neural tube) and the cutaneous ectoderm (skin). All three germ layers, ectoderm, mesoderm, and endoderm, are derived from the epiblast. Gastrulation transforms the embryo from a two-layered structure to a three-layered structure.¹¹

FIGURE 175-2 Normal human gastrulation. A, Prospective endodermal and mesodermal cells of the epiblast migrate toward the primitive streak and ingress (arrows) through the primitive groove to become the definitive endoderm and mesoderm. B, Prospective notochordal cells in the cranial margin of Hensen’s node will ingress through the primitive pit during regression of the primitive streak to become the notochordal process.

(From Dias MS, Walker ML. The embryogenesis of complex dysraphic malformations: a disorder of gastrulation? J Pediatr Neurosurg. 1992;18:229-253.)

Hensen’s node serves a special role as the “organizer” of the embryo. As the PS elongates, prospective endodermal cells within Hensen’s node migrate through the primitive pit. As the PS regresses, specialized mesodermal cells, prospective notochordal cells, migrate through the primitive pit (Fig. 175-2B) and form the notochordal process in the midline between the neuroectoderm and endoderm.^10,11 As we shall see, the notochord plays an important role in directing subsequent neurulation.

To what extent does the notochordal process extend cranially from Hensen’s node, and to what extent does it elongate caudally by the addition of cells to its caudal end from the regressing Hensen’s node? In the chick, the notochordal process grows largely by the latter mechanism.¹² In mammals, the available evidence suggests that the situation is much more complex and may involve both mechanisms.^13,14 The mechanism of notochord elongation in humans is unknown.

The prospective neuroectoderm is localized to an area of the epiblast that surrounds and flanks Hensen’s node and the cranial half of the PS (Fig. 175-3).¹⁵ Fate-mapping studies in chick embryos suggest that each region of the neuroepithelium contributes to multiple neuraxial levels, with the most cranial neuroectoderm contributing to the forebrain and all caudal levels of the neuraxis and more caudal levels of the neuroectoderm contributing to successively more caudal levels of the neuraxis. The prospective brain is derived specifically from neuroectoderm located cranial and lateral to Hensen’s node.¹⁵

FIGURE 175-3 Location of prospective neuroepithelium in a dorsal view of an avian embryo during gastrulation. The striped region illustrates localization by earlier mapping studies that used carbon particles; the hatched area illustrates additional caudal areas demonstrated by more recent mapping studies using horseradish peroxidase injections and chimeric transplantation.

(From Dias MS, Walker ML. The embryogenesis of complex dysraphic malformations: a disorder of gastrulation? J Pediatr Neurosurg. 1992;18:229-253.)

Formation and Development of the Notochord (PODs 16 to 25)

In both human³ and nonhuman primates,¹⁶ the notochordal process in cross section consists of cells arranged radially about a central lumen called the notochordal canal (Fig. 175-4A).³ The notochordal canal is continuous dorsally with the amnionic cavity through the primitive pit.¹⁷ Between PODs 18 and 20, the notochordal process fuses (intercalates) with the underlying endoderm to form the notochordal plate (Fig. 175-4B). The notochordal plate is therefore incorporated into the roof of the yolk sac, with the notochordal canal becoming continuous with the yolk sac. Intercalation results in a direct communication, the primitive neurenteric canal, that connects the amnionic and yolk sacs at the level of Hensen’s node.³ The neurenteric canal persists for about 3 days, at the end of which the notochordal plate folds dorsoventrally and separates (excalates) from the endoderm (Fig. 175-4C) and the neurenteric canal is obliterated.^3,18,19 Thereafter, the true notochord exists as a solid rod of notochordal cells.¹⁸ The function or functions of the neurenteric canal are unknown.

FIGURE 175-4 Notochordal canalization, intercalation, and excalation. A, The notochordal process contains a central lumen (the notochordal canal) that is continuous with the amnionic cavity through the primitive pit. B, During intercalation, the canalized notochordal process fuses with the underlying endoderm; the communication of the amnion with the yolk sac forms the primitive neurenteric canal. C, During excalation, the notochord rolls up and separates from the endoderm to become the definitive notochord; the primitive neurenteric canal becomes obliterated.

(From Dias MS, Walker ML. The embryogenesis of complex dysraphic malformations: a disorder of gastrulation? J Pediatr Neurosurg. 1992;18:229-253.)

Primary Neurulation (PODs 16 to 25)

Human neuroectoderm is first visible at POD 16 as pseudostratified columnar epithelium overlying the midline notochord (Fig. 175-5A) and contiguous with the surrounding squamous epithelium of the cutaneous ectoderm.²⁰ By POD 17 to 19, a midline neural groove (or median hinge point) develops in the midline of the neural plate immediately above the notochord (Fig. 175-5C).²¹ The neural groove deepens (Fig. 175-5B), and paired neural folds begin to elevate bilaterally.²² Paired dorsolateral grooves (or dorsolateral hinge points) result in further dorsal elevation and medial convergence of the neural folds (Fig. 175-5C). The neural folds separate from the cutaneous ectoderm and fuse (Fig. 175-5D) to form a closed neural tube between POD 21 and 23. Closure generally involves apposition and fusion of first the cutaneous ectoderm and then the neuroectoderm.^3,19 The first part of the human neural tube to close is the region of the caudal rhombencephalon or cranial spinal cord, usually when five pairs of somites are present.¹⁹

FIGURE 175-5 Primary neurulation in chick embryos. A, Thickening of cells to form the neural plate. B, Formation of the midline neural groove and elevation of bilateral neural folds. C, Convergence of neural folds toward the midline around dorsolateral hinge points. D, Fusion of the neural folds to form the closed neural tube and separation of cutaneous ectoderm from neural folds (dysjunction) to form intact dorsal skin.

(From Gilbert SF. Developmental Biology, 7th ed. Sunderland, MA: Sinauer Associates; 2003.)

Closure of the neural tube takes place over a period of 4 to 6 days. Although previously thought to close in linear fashion like a zipper extending cranially and caudally from the point of initial closure, mammalian neurulation instead appears to extend from several initiation sites along the craniocaudal neuraxis.^19,23–25 Cranial neural tube closure may involve the coordinated interaction of as many as four waves of discontinuous neural tube closure.^19,23–26 The spinal cord closes craniocaudally in a linear manner from the point of initial closure to the caudal neuropore.^24,25 The last two closure sites are the cranial and caudal neuropores. The cranial neuropore closes between POD 23 and 25, whereas the caudal neuropore closes between POD 25 and 27.²⁷

Neural crest cells are specialized, multipotential migratory cells that arise from the junction between the neural folds and adjacent surface ectoderm^18,28–30; in humans, a simultaneous origin for some neural crest cells from the surface ectoderm cannot be excluded.¹⁹ Formation of the cranial neural crest begins during elevation of the neural folds before they fuse and continues from the closed neural tube well after the neural folds have fused.^18,19,31 The spinal neural crest arises only after closure of the neural tube.^3,27

Cranial neural crest cells contribute to the branchial arches (Table 175-2) and the arachnoid and pia mater of the cranium (the dura mater appears to derive from the mesoderm). Spinal neural crest cells undergo terminal differentiation into melanocytes of the body wall and limbs, Schwann cells investing the peripheral nerves, spinal cord meninges, dorsal root and autonomic ganglion cells of the spinal nerves, and adrenal medulla.³² Further details regarding control of migration and terminal differentiation of neural crest cells can be found in several reviews.^33–35

TABLE 175-2 Contributions of the Neural Crest to Branchial Clefts

Secondary Neurulation (POD 25)

After caudal neuropore closure at POD 25 to 27, the entire nervous system is covered with cutaneous ectoderm, and more caudal neural development takes place by secondary neurulation.²⁷ This process has been covered in other sources and has no relevance to cranial development; interested readers are referred elsewhere for this information.

Development of Brain Vesicles, Neural Tube Bending, and Formation of Alar and Basal Plate Derivatives (PODs 19 to 40)

During primary neurulation, the cranial neural tube is segmented into three primary brain vesicles: the prosencephalon (forebrain), mesencephalon (midbrain), and rhombencephalon (hindbrain). By POD 21, additional swellings, or neuromeres, become evident—six in the prosencephalon, one in the mesencephalon, and nine in the rhombencephalon. By POD 32, the three primary brain vesicles have further developed into five secondary brain vesicles: the prosencephalon has two divisions, the telencephalon (forming the cerebral hemispheres and basal ganglia) and diencephalon (hypothalamus, thalamus, and epithalamus); the mesencephalon remains undivided and forms the midbrain; and the rhombencephalon has two divisions, the metencephalon (forming the pons and cerebellum) and myelencephalon (forming the medulla oblongata).

Three neural tube flexures develop during this time as well (Fig. 175-6). The earliest to appear, at POD 19 to 21 during primary neurulation, is the dorsally directed mesencephalic flexure. Later, two additional flexures develop—the dorsally directed cervical flexure (POD 28) and the ventrally directed pontine flexure (POD 32). The pontine flexure continues to bend the future brainstem such that the metencephalon (including the cerebellum) comes to lie dorsal to the myelencephalon by the eighth embryonic week.

FIGURE 175-6 Development of primary and secondary vesicles and the formation of flexures. A, At 26 days the brain consists of three primary vesicles (prosencephalon, mesencephalon, and rhombencephalon). The mesencephalic and cervical flexures are just beginning to form. B, By 28 days the flexures are progressing and the primary vesicles are better defined. C, By 35 days the pontine flexure forms between the mesencephalic and cervical flexures; the prosencephalon is now divided into the telencephalon and diencephalon, whereas the rhombencephalon is divided into the metencephalon and myelencephalon. D, At 50 days the telencephalon has further developed into the primitive cerebral hemispheres and the pontine flexure folds the metencephalon back against the myelencephalon; the rhombic lips of the pontine flexure will give rise to the cerebellum.

(From Larsen WJ. Human Embryology, 2nd ed. New York, NY: Churchill Livingstone; 1997.)

The fundamental cytoarchitectural organization of the spinal cord and brainstem (myelencephalon, metencephalon, and mesencephalon) consists of two components: a basal and an alar plate (Fig. 175-7). The basal plate contains paired ventral neuromotor columns that contribute to the primary motor pathways, whereas the alar plate contains paired dorsal neurosensory columns that contribute to the sensory and integrative pathways. As discussed previously, spinal cord neural crest cells form dorsal root ganglia and dorsal sensory nerves, the central processes of which enter the spinal cord and either ascend within the posterior columns to the nucleus gracilis and cuneatus or synapse locally with dorsal sensory neurons within the alar column (Rexed’s laminae I to VII). Alar plate neurons send projections both cranially in the spinothalamic and spinocerebellar tracts and locally that perform integrative functions with other alar or basal plate neurons within the spinal cord. The basal plate motoneurons receive afferent synapses from alar plate neurons (as well as descending projections from cranial portions of the neuraxis) and, in turn, send efferent axonal projections to the periphery via the ventral root.

FIGURE 175-7 Formation of the basal and alar plates. The alar plate lies dorsal to the basal plate; the two are separated by the sulcus limitans. The alar plate will give rise to sensory and integrative neurons, whereas the basal plate will give rise to efferent (motor) neurons. Simultaneously, ependymal (ventricular), mantle, and marginal zones develop.

(From Gilbert SF. Developmental Biology, 7th ed. Sunderland, MA: Sinauer Associates; 2003.)

The brainstem organization follows a similar, albeit much more complicated pattern of alar and basal plate organization. The general dorsoventral pattern of the alar and basal plates in the spinal cord is altered in the brainstem by two processes. First, the brainstem flexures produce an expansion and thinning of the mesencephalic roof that rotate the brainstem such that the alar plate comes to lie dorsolateral to the basal plate (rather than strictly dorsal). Second, some alar plate neurons (particularly those forming the facial nucleus, nucleus ambiguus, and olivary and pontine nuclei) curiously migrate ventrally into the basal plate to form ventral motor nuclei. In particular, migration of the facial nucleus from its original position within the alar plate toward the dorsal midline and then ventrally to lie within the basal plate explains the circuitous route of the facial nerve within the brainstem.

The brainstem nuclei are arranged into seven columns, three motor and four sensory. The three motor columns produce somatic efferent, visceral efferent, and special (branchial) efferent nuclei. The four sensory columns produce somatic afferent, visceral afferent, and two special afferent nuclei—special somatic afferent and special visceral afferent nuclei. Each cranial nerve receives contributions from one or more of these seven columns (Table 175-3).

TABLE 175-3 Derivations of Cranial Nerves

GSA, general somatic afferent; GSE, general somatic efferent; GVA, general visceral afferent; GVE, general visceral efferent; SSA, special somatic afferent; SVA, special visceral afferent; SVE, special visceral efferent.

The cerebellum represents a specific development of the alar plate from the rhombic lips of the metencephalon. During the sixth embryonic week, the paired cerebellar primordia develop and eventually fuse to form the cerebellar anlagen. The caudal region of the cerebellar primordia forms the more primitive flocculonodular lobe, whereas the cranial region produces the much larger (and less primitive) vermis and cerebellar hemispheres.

The organization of the diencephalon and telencephalon is even more complicated by the fact that both lack a basal plate altogether and all structures derived from these two subdivisions therefore arise exclusively from alar plate neurons. The hypothalamus and thalamus are differentiated by a visible sulcus, the hypothalamic sulcus, and the thalamus from the epithalamus by the sulcus dorsalis. In addition to the hypothalamus (including the posterior lobe of the pituitary body), thalamus (including the neural retinae, which extend as outpouchings from the diencephalon), and epithalamus (from which the pineal gland and habenular complex are derived), the diencephalon also forms the globus pallidus. The telencephalon produces the cerebral hemispheres, as well as the corpus striatum (caudate and putamen). Telencephalic cleavage to form the two cerebral hemispheres begins at POD 32. The explosive growth of the telencephalon results in dorsal and posterior expansion of the cerebral hemispheres such that posteriorly the telencephalon ultimately overlies the mesencephalon; the interhemispheric commissural fibers connecting the two hemispheres extend posteriorly to the splenium of the corpus callosum (dorsal to the pineal cistern). The great horizontal fissure (or velum interpositum) is an arachnoid fissure that extends from the foramen of Monro posteriorly to the pineal recess and marks the embryonic plane between the telencephalon and mesencephalon. Within the velum interpositum is the (mesodermally derived) choroid plexus, which extends both ventrally through the pia-ependyma to lie within the roof of the third ventricle and laterally within the choroidal fissure to lie within the lateral ventricle along the caudothalamic groove separating the basal ganglia and thalamus. These embryologic relationships are relevant to understanding the various intraventricular approaches to the lateral and third ventricles.

Neuronal Proliferation, Neuronal Migration, and Formation of Commissures (PODs 24 to 115)

The cellular architecture of the brain and spinal cord is based on a similar unifying theme—embryonic neuroblasts proliferate within a ventricular zone and their postmitotic progeny migrate centripetally through the neural tube to take up locations peripherally to form the mantle layer. Axonal processes from mantle layer neurons extend further to form the marginal zone immediately under the pia (Fig. 175-8). Neuronal proliferation begins on POD 24; neuroblasts are later replaced by glioblasts at the ventricular zone, which then give rise to glial cells (astrocytes and oligodendroglia). The remaining cells of the ventricular zone become ependymal cells.

FIGURE 175-8 Neuronal migration. By 5 weeks’ gestation the embryonic neural tube has three zones: a ventricular (ependymal) zone, intermediate (mantle) zone, and marginal zone. In the spinal cord and brainstem (top), neurons migrate solely from the ventricular zone centripetally toward the intermediate (mantle) zone; their processes extend to the marginal zone. In the cerebellum (middle), neurons migrate along two pathways. Cells migrate centripetally from the ventricular zone to form the Purkinje cell layer (Purkinje, basket, and Golgi cells). Additional cells migrate from the rhombic lip to the mantle zone to make up an additional germinal zone, the external granule cell layer; postmitotic cells from the external granule layer migrate inward to form the internal granule cell layer. In the cerebral cortex (bottom), cells from the ventricular zone migrate to form a second, intermediate germinal zone. Cells from both the ventricular and intermediate zones contribute to the cortex. An additional subventricular zone arises later; these cells migrate to a subplate zone and contribute to the outer cortical laminae as well. The cortex is therefore derived from both the cortical mantle zone and the subplate region and is generated in an inside-out fashion, with superficial layers forming later than deeper layers. The intermediate layer subsequently forms the white matter of the cerebrum.

(From Gilbert SF. Developmental Biology, 7th ed. Sunderland, MA: Sinauer Associates; 2003.)

Spinal cord development follows this basic theme and gives rise to the central location of the spinal cord gray matter with surrounding white matter tracts. However, proliferation and migration at other levels of the neuraxis are altered depending on the specific location—two locations that deserve special discussion are the cerebellum and cerebral cortex. In the cerebellum, the postmitotic neuroblasts migrate centripetally to form two zones: the deep cerebellar nuclear layer (which will form the deep cerebellar nuclei) and the Purkinje cell layer (to which the Purkinje, basket, stellate, and Golgi cells migrate) (Fig. 175-8, middle). The granule cells, in contrast, originate in the rhombic lip, migrate superficially around the surface of the cerebellum, and exist superficial to the Purkinje cell layer as the external granule layer. These cells are still mitotically active and produce numerous granule cell progeny, even as long as 2 years after birth.³⁶ Postmitotic granule cells migrate back toward the ventricular zone, through the Purkinje cell layer, to produce the internal granule layer. This second migration is controlled by specialized glial cells, called Bergman glia, the processes of which serve as guideposts for the migration of granule cells through the Purkinje cell layer.

Neurogenesis and neuronal migration are also modified in the cerebral cortex (Fig. 175-8, bottom). The first neuroblasts send axon that form the marginal zone immediately subjacent to the pia. Postmitotic neurons then migrate away from the ventricular zone to form the intermediate zone between the ventricular and marginal zones. Cells from both the ventricular and intermediate zones migrate further centripetally to form the cortical mantle zone—cells that arrive at the mantle zone first form the deeper cortical layers, whereas those that arrive later form more superficial cortical layers that migrate through the previously established layers. As ventricular zone proliferation ceases, proliferation begins in the subventricular zone, located between the ventricular and intermediate zones. Cells from the subventricular zone migrate through the intermediate zone to form a subplate just beneath the cortical mantle zone. Cells from the subplate zone migrate through the mantle zone to form the most superficial cortical layers. The cortex is therefore derived from both the cortical mantle zone and the subplate region and is generated in an inside-out fashion, with superficial layers forming later than deeper layers. The intermediate layer subsequently forms the white matter of the cerebrum. Neuronal migration in the cerebrum also appears to involve specialized radial glial fibers that span from the ventricular to the marginal zones and along which neurons migrate to their final destinations. Neuronal migration involves the interplay of many different molecular species, including reelin, astrotactin, integrins, neurogulins, and others.³⁶

The striatum and globus pallidus develop from ventricular zone neurons of the telencephalon and diencephalon, respectively, that migrate to deep nuclear zones external to the ventricular zone.

The majority of the commissural fibers of the hemispheres develop from the commissural zone of the lamina terminalis. The lamina terminalis is the site of the cranial neuropore and is located immediately dorsal to the optic chiasm anatomically. Commissural fibers of the anterior commissure, corpus callosum, and hippocampal commissure derive from the dorsal end of the lamina terminalis (the posterior and habenular commissures are formed from the epithalamus of the diencephalon). The earliest commissural fibers, those of the anterior commissure, arise at 54 days, followed by those of the hippocampal commissure during the 11th week and those of the corpus callosum between 84 and 115 days.^37,38 The corpus callosum generally develops in an anterior-to-posterior fashion (genu first, followed by the body and then the splenium); because of the later growth of the frontal lobes, the rostrum is the last portion to form. As a result, incomplete abnormalities of the corpus callosum involve the later developing, more posterior portions (and rostrum) with sparing of the earlier developing anterior portions and lead to selective enlargement of the atrium and occipital horns of the ventricles (colpocephaly). An exception to this rule is holoprosencephaly, in which there is selective agenesis of the posterior portions with sparing of the anterior portions.^38,39

Failure of Neural Tube Closure—Anencephaly and Myelomeningocele

Anencephaly and myelomeningocele are neural tube defects (NTDs), or localized failure of primary neurulation, that can arise through one of two mechanisms. The “nonclosure theory” proposes that NTDs represent primary failure of neural tube closure.⁴⁰ The “overdistention theory,” introduced in 1769 by Morgagni⁴¹ and popularized by Gardner,^42–45 proposes that NTDs arise through overdistention and rupture of a previously closed neural tube. The nonclosure theory is widely accepted for most cases. However, overdistention may contribute to some experimental NTD models, particularly those caused by vitamin A⁴⁶ and the T-curtailed mouse mutant⁴⁷; whether this causes human malformations is unknown. The result is an open, unneurulated segment of neural tube, the nature and severity of which are determined by the location and length of the unneurulated segment. Failure of caudal neurulation produces a myelomeningocele, whereas more cranial failure results in anencephaly.

Because neural tube closure requires the complex interaction of multiple cellular processes, it is not surprising that NTDs may result from a number of embryonic insults. NTDs have been produced experimentally with a number of teratogens (reviewed by Campbell and colleagues²), genetic mutations (reviewed by Copp and associates^47,48), and experimental manipulations (reviewed by Schoenwolf and Smith⁴⁹). Although these processes all suggest a number of potential mechanisms whereby NTDs might arise, the cause of human malformations remains unknown. NTDs are most likely etiologically heterogeneous^1,2,47,48 and represent the end result of a variety of embryonic disorders.

Genetic models of NTDs provide a means of identifying candidate gene or genes involved in neural tube closure^47,48; three such examples illustrate the variety of ways in which genetic mutations can result in an NTD. The splotch (Sp) mouse mutant is identified by a peculiar patch of white fur and exhibits both exencephaly and myelomeningocele. The genetic locus for the Sp mutation is within the Pax-3 gene locus. Although the Pax-3 gene product appears to be involved in apposition and fusion of the neural folds, neither the mechanisms underlying the normal Pax-3 gene product nor the mechanisms whereby the Sp mutation alters Pax-3 function are known with certainty.

A second mouse mutant, curly tail (ct), exhibits posterior NTDs and tail deformities associated with delay in closure of the caudal neuropore.⁴⁸ However, the delay is not the result of faulty neuroepithelial development inasmuch as isolated neuroepithelium from ct mutants undergoes normal neurulation.⁵⁰ Rather, a delay in cell proliferation in the underlying notochord and hindgut endoderm causes an abnormal ventral body axis curvature and impedes posterior neuropore closure.^47,48 If the ventral curvature is corrected by splinting the caudal ct embryo with an eyelash⁵¹ or by retarding neuroepithelial proliferation with retinoic acid,⁵² the posterior neuropore closes normally and the incidence of NTDs is reduced.

A third mouse mutant, T-curtailed (T[c]) produces a lumbosacral myelomeningocele with dorsoventral forking of the caudal neural tube. However, rather than delayed or failed neural tube closure, the T(c) mutant causes rupture of the roof plate and reopening of a previously closed neural tube.⁴⁷

A large number of environmental causes of NTDs have been identified and their underlying mechanisms studied. Recent attention has focused on the role of folate in the embryogenesis of NTDs. Maternal administration of folate antagonists such as aminopterin has long been known to produce NTDs.⁵³ Periconceptional administration of supplemental folate in randomized, placebo-controlled studies reduced both the recurrence rate of NTDs in women with a previously affected pregnancy and the incidence in women who had never had an affected pregnancy.^54,55 However, studies of maternal serum and red blood cell folate levels in mothers of infants with myelomeningocele have produced inconsistent results (reviewed by Wald⁵⁶ and Seller⁵⁷), and folate deficiency does not cause NTDs in mouse or rat embryos (reviewed by Fleming and Copp⁵⁸). These observations suggest that NTDs are not usually the result of an absolute folate deficiency.

More recent attention has focused on the possibility that NTDs are caused by abnormalities involving metabolic pathways (in either the mother or fetus) that require folate⁵⁹; such abnormalities predispose an individual to NTDs and might therefore be overcome by folate supplementation. Folate and its metabolites tetrahydrofolate and 5-methyltetrahydrofolate are important in a variety of mammalian metabolic reactions, including purine and pyrimidine (and therefore DNA) synthesis, and in the transfer of methyl groups during the metabolism of methionine and homocysteine. In particular, the role of folate in methionine and homocysteine metabolism has generated considerable interest.^60–68

The metabolism of homocysteine follows one of two pathways. One involves trans-sulfuration to cystathionine, catalyzed by the enzyme cystathionine synthase. The second is remethylation of homocysteine to methionine, catalyzed by the enzyme methionine synthase and requiring the donation of a methyl group from the folate metabolite 5-methyltetrahydrofolate (which in turn is converted to tetrahydrofolate) by using vitamin B₁₂ as a cofactor. Methionine is activated by adenosine triphosphate to produce S-adenosylmethionine, which is then used to donate a methyl group in a variety of cellular reactions involving protein, lipid, DNA, and RNA metabolism. Tetrahydrofolate is converted back to 5-methyltetrahydrofolate by the enzyme 5,10-methylenetetrahydrofolate reductase.

One hypothesis is that maternal or fetal mutations in either methionine synthase or 5,10-methylenetetrahydrofolate reductase may slow this “methylation cycle” down, drive the conversion of methionine to homocysteine, and lead to methionine deficiency or homocysteine excess, or both. Ingestion of higher doses of folate may overcome this relative deficiency by restoring more normal homocysteine and methionine levels.^59,67,69 A number of observations support this view. For example, elevated maternal serum⁶⁸ and amnionic fluid⁶⁷ homocysteine levels have been identified in women carrying a fetus with an NTD. Disordered methionine metabolism has been demonstrated in nonpregnant women who had previously given birth to a child with an NTD.^67,69 Finally, mutations of the 5,10-methylenetetrahydrofolate reductase gene have been demonstrated in 18% of individuals with NTDs, in 13% of parents of children with NTDs, and in just 6% in controls.^65,70 Abnormalities of cystathionine synthase have also been identified.⁶⁷

A number of teratogens are known to act through a variety of cellular mechanisms to cause NTDs in animals and humans (see Copp and coworkers⁴⁷ for review). One important teratogen, valproic acid (VPA), produces NTDs in both animal models and humans, probably by inhibiting neural fold fusion.⁴⁷ Although the exact mechanism is not known, VPA appears to disrupt folate metabolic pathways,⁷¹ perhaps by interfering with the conversion of tetrahydrofolate to 5-formyltetrahydrofolate.⁶⁷ Maternal folate administration has reduced the incidence of VPA-associated NTDs in some, but not all studies,⁷¹ and mouse strains that are susceptible to valproate-induced NTDs have significantly lower levels of 5,10-methylenetetrahydrofolate after valproate administration than do resistant strains.⁷² In addition, a number of developmental regulatory genes (such as transcription factors and cell cycle checkpoint genes) may be altered in these susceptible strains.⁷³ One hypothesis is that VPA may act by changing folate-dependent methylation of regulatory proteins such as transcription factors.

Incomplete Dysjunction—Dermal Sinus Tracts, Dermoid and Epidermoid Tumors, Meningocele Manqué and Meningoceles

Dermal sinus tracts (DSTs) incorporate a tract of cutaneous ectoderm from the dorsal midline skin that extends for a variable distance into the underlying mesenchymal tissues and, in many instances, penetrates the dura to end within the thecal sac adjacent to or contiguous with the neural tube. Most are spinal in origin; nasofrontal and occipital DSTs are less frequent.

The most widely accepted theory of DST embryogenesis proposes that they arise through faulty separation (dysjunction) of neuroectoderm from the overlying cutaneous ectoderm with a tract of cutaneous ectoderm left sequestered between the skin and neural tube.⁷⁴ The histology of this tract may represent a number of cutaneous ectodermal abnormalities, including epithelial-lined sinuses (DSTs), epidermoid tumors (containing only ectodermal tissue as pseudostratified squamous epithelium), and dermoid tumors (containing both ectodermal and mesodermal tissues such as sebaceous glands and hair follicles) located anywhere between the skin and neural tube. Hoving and colleagues have provided evidence that cranial dysjunction at the anterior neuropore involves apoptosis; misexpression or failure of apoptosis may underlie NTDs, particularly those involving the nasofrontal region.⁷⁵ Another way in which abnormal dysjunction could potentially arise might be through misexpression of cell adhesion molecules or other molecular markers located on cutaneous ectoderm or neuroectoderm, or both.

Dermal sinuses may involve any level of the neuraxis, although there is a predilection for neuropores in both the lumbosacral and frontonasal regions. Frontonasal dermal sinuses arise through incomplete dysjunction at the site of the anterior neuropore and commonly involve the frontobasal skull. The embryogenesis and surgical anatomy of frontonasal dermal sinuses are predicted by the normal development of the anterior cranial base (Fig. 175-9). The prosencephalic primordium, olfactory epithelium, adenohypophysis, and facial cutaneous ectoderm all share a common embryonic origin (at least in avian embryos) from the region of the anterior neuropore.^76,77 The frontal bones develop later from bilaterally paired mesenchymal anlagen and are separated by the metopic suture. Simultaneously, nasal bones develop along the nasal spine and are separated from the frontal bones by a fibrous capsule, the fonticulus nasofrontalis. The nasal spine is separated from the deeper nasal cartilaginous capsule by the prenasal space. During normal embryogenesis, a tongue of dura extends ventrally from the inferior aspect of the anterior cranial fossa and is interposed anteriorly between the frontal and nasal bones at the fonticulus nasofrontalis and inferiorly between the prenasal cartilage and the nasal bones within the prenasal space. During normal development, this dural reflection becomes surrounded by ossifying bone and regresses; remnants of the tract persist as the foramen cecum along the floor of the anterior cranial fossa, between the insertion of the falx cerebri anteriorly and the crista galli posteriorly.⁷⁸

FIGURE 175-9 Normal frontonasal development (A and B) and embryology of frontonasal dermal sinus tracts (C). A, During normal embryogenesis, the fonticulus nasofrontalis forms between the frontal and nasal bones; the prenasal space forms between the nasal bone and nasal cartilage. B, A tongue of dura extends through the foramen cecum toward the midline nasal skin. Later, this tongue of dura is obliterated and the anterior cranial base is formed; the foramen cecum remains as a vestige of this embryonic tract. C, Abnormal dysjunction during closure of the anterior neuropore leaves a tract of cutaneous ectoderm between the commissural plate and the midline nasal skin. Formation of the anterior frontobasal structures results in a tract whose cutaneous opening may be located at the fonticulus nasofrontalis or anywhere along the dorsum of the nose and that extends through the foramen cecum, between the two halves of a bifid crista galli, and along the anterior cranial base; in rare instances, the tract extends all the way to the commissural plate.

(Adapted from Sessions RB. Nasal dermal sinuses: new concepts and explanations. Laryngoscope. 1982;92(suppl 29):1-28.)

Nasal DSTs typically arise along the nasal dorsum or the glabella; 90% end extracranially. When they extend intracranially, they pass through the foramen cecum and typically end within the falx; on rare occasion the tract extends entirely through the falx and travels within the subarachnoid space to end at the lamina terminalis (the site of the anterior neuropore) as predicted embryologically (Fig. 175-9). Associated dermoid tumors may occur anywhere along the DST and involve extracranial or intracranial sites.^78–83 Finally, sequestered neuroepithelium within the prenasal space may develop independently into a mass of neuroglial tissue, the nasal cerebral heterotopia or “nasal glioma.”^84–86

DSTs may also arise in the parietal region, where they fenestrate the superior sagittal sinus, extend intracranially within the falx, pass posterior to the splenium, and end within the subarachnoid space in the pineal recess in close proximity to the posterior and habenular commissures. Intracranial extension of a DST between the frontonasal and parietal locations has never been described, a point that again is explained embryologically. Dermal sinuses arising from the lamina terminalis may have one of two fates. A malformation involving the cranial rim of the neuropore produces a tract that remains anteriorly located and in close association with the anterior cranial structures. Conversely, a malformation involving the caudal rim of the neuropore (i.e., at the caudal margin of the future commissural plate) will be swept posteriorly, behind the splenium, as the telencephalic vesicles and corpus callosum develop.

Finally, dermal sinuses are also common in the occipital region,^87,88 where they arise from either the roof of the fourth ventricle, the cerebellar vermis, or the subdural space posterior to the cerebellum. The tract may extend either beneath the tentorium into the posterior fossa or above the tentorium toward the occipital lobe or branch to involve both compartments. Although some authors have suggested that occipital dermal sinuses arise from the region of the anterior neuropore,⁸⁷ this is inconsistent with the association of occipital dermal sinuses with structures derived from the dorsal aspect of the rhombencephalon (cerebellum and fourth ventricle) well distant from the lamina terminalis. Occipital dermal sinuses more likely arise from the rhombencephalic neural tube during neural tube closure. Why dermal sinuses have a predilection for this region of the neuraxis is unclear. The rhombencephalon in humans is among the first regions of the neural tube to undergo neural fold fusion and may therefore be particularly vulnerable to disorders of neurulation; in addition, the presence of the pontine flexure may lend additional physical stress to the neural tube at this site. Finally, the frequent occurrence of dermal sinuses may simply reflect the relatively large size of this region of the neural tube at the time of neurulation.

Disorders of Gastrulation—Combined Spina Bifida, Split Cord Malformations, and Neurenteric Cysts

A number of seemingly unrelated disorders are thought to share a common embryogenesis, including combined (anterior and posterior) spina bifida (also called the split notochord syndrome); split cord malformations (SCMs); neurenteric cysts; intestinal malrotations, duplications, and fistulas; some cases of the Klippel-Feil malformation sequence; some cases of sacral agenesis and Currarino’s triad; and a number of other unclassified complex congenital craniospinal malformations, all of which exhibit disorders of all three primary germ layers. Variably described as the “split notochord syndrome,”⁸⁹ “endodermal-ectodermal adhesion syndrome,”⁹⁰ “accessory neurenteric canal syndrome,”⁹¹ or “disordered midline integration during gastrulation” syndrome,¹ the embryogenesis of these disorders is fully discussed by Dias and Walker.¹

All these malformations share some type of SCM. All have in common a splitting of the neuroectoderm into two parts over a portion of its length. The two “hemicords” thus formed are typically separated by either a bony or cartilaginous midline extradural spur (type I SCM) or a thinner, fibrous intradural midline band of tissue (SCM type II).⁹² In both cases, atretic and nonfunctional midline dorsal (and in rare instances ventral) rootlets arise from one or both “hemicords” and end on the midline tissue.^93–97 The presence of both lateral and median sets of nerve roots arising from each “hemicord” supports the presence of at least a partial spinal cord duplication in both malformations. Finally, pathologic specimens of both malformations demonstrate neither absolute splitting nor complete duplication of the cord but rather incomplete duplications with relatively well preserved lateral halves and dystrophic medial halves.^93,97–100

SCMs are seen in association with a variety of anomalies, including combined (anterior and posterior) spina bifida; hemimyelomeningoceles; myelomeningoceles (occurring in up to a third of cases); cervical myelomeningoceles; neurenteric cysts; some examples of the Klippel-Feil anomaly, iniencephaly, and caudal agenesis; certain intestinal duplications and diverticula (reviewed by Dias and Walker¹); and Currarino’s triad. All these complex congenital craniospinal malformations have in common both SCMs and anomalies of all three primary germ layers and therefore probably share a common embryonic origin.^1,45 Four theories have attempted to explain the underlying embryopathy.

Beardmore and Wigglesworth¹⁰¹ proposed an adhesion between the epiblast and hypoblast that interferes with notochordal outgrowth. On encountering this “endodermal-ectodermal adhesion,” the notochord might split around the adhesion and produce two “heminotochords,” each inducing a neural “hemicord.” Associated remnants of the adhesion could give rise to endodermal remnants located anywhere between the gut and cutaneous ectoderm. This works only if the notochord extends cranially from Hensen’s node; a modification of this theory⁹⁰ to account for the possibility of notochordal elongation as Hensen’s node regresses postulates an adhesion within the PS caudal to Hensen’s node around which the notochord might be split during node regression.

Bremer noted the similarity between patients with combined anterior and posterior malformations, in which both the involved vertebrae and spinal cord are split midsagittally to form two laterally displaced “hemivertebral” columns and an SCM surrounding a central cleft, and the primitive neurenteric canal of embryogenesis, both seeming to bisect the embryo and connect the amnionic and yolk sacs.⁹¹ This led to the proposal that a more cranial “accessory neurenteric canal” might split the notochord and neuroepithelium and result in a split cord and related malformations.

McLetchie and associates¹⁰² and Saunders¹⁰³ proposed a primary duplication of the notochord that results in a secondary endodermal-ectodermal interaction between the duplicated notochords. Dodds suggested that during normal embryogenesis, bilaterally paired prospective notochordal cell anlagen might be integrated into a single midline structure during regression of the PS.¹⁰⁴ Feller and Sternberg proposed that abnormal rests of undifferentiated cells in Hensen’s node might interfere with proper midline integration and result in paired notochords; subsequent differentiation of these cell rests could give rise to a variety of midline anomalies composed of tissues derived from any of the three primary germ cell layers.¹⁰⁵

Finally, Dias and Walker proposed a disorder involving midline axial integration during gastrulation (Fig. 175-10).¹ During normal development, paired notochordal anlagen within Hensen’s node are integrated to form a single notochordal process; bilaterally paired prospective neuroepithelial cells flanking both sides of Hensen’s node and the PS are similarly integrated to form a single neural plate. According to this theory, complex congenital craniospinal malformations arise when both the paired notochordal anlagen and adjacent prospective neuroepithelial regions remain separate (perhaps because of an abnormally wide Hensen node or PS) during gastrulation. The intervening space between the paired “hemicords” is composed of pluripotent PS cells and could give rise to a variety of tissue types from any of the three primary germ layers (all having been described in association with SCMs).^106–108

FIGURE 175-10 Proposed embryogenesis of split cord malformations and other “complex dysraphic malformations” from disordered midline axial integration during gastrulation. The primitive streak is abnormally wide (perhaps because of maldevelopment of the underlying basement membrane); prospective notochordal cells therefore begin ingressing more laterally than normal. As a result, two notochordal processes are formed. The caudal neuroepithelium flanking the primitive streak also fails to become integrated to form a single neuroepithelial sheet and instead forms two “hemineural plates.”

Although most SCMs involve the thoracic or lumbar spinal cord, rare cranial lesions have been reported.^109,110 Because the cranial notochord reaches only to the clivus (adjacent to the midbrain), one would expect cranial SCMs to involve only the rhombencephalon (medulla and pons) or mesencephalon; SCMs involving the diencephalon or telencephalon have not been described, nor would they be expected based on this embryologic mechanism.

Intracranial Lipomas

Lipomas may arise at any level of the neuraxis, although spinal lipomas are the most common and most frequently involve the lumbosacral spinal cord, conus medullaris, and filum terminale. Spinal lipomas (particularly those that develop cranial to S2) reputedly arise from premature dysjunction during primary neurulation.^111,112 The cutaneous ectoderm, prematurely separated from the neuroepithelium before neural fold fusion, allows surrounding mesenchymal cells access to the central canal of the developing neural tube. These mesenchymal cells, which are normally induced to form meninx when exposed to the outer surface of the neural tube, are instead induced to form fat on exposure to the central canal of the spinal cord.¹¹² More caudal spinal lipomas (those arising caudal to S2) probably involve some (not yet defined) abnormality of secondary neurulation.

Intracranial lipomas, in contrast, most likely represent postneurulation abnormalities arising from an abnormality of the meninx primitiva (the anlagen of the pia and arachnoid derived from the cranial neural crest).¹¹³ A unifying theory by Truwit and Barkovich (based on previous work by Verga, Krainer, and Osaka as referenced in their study) proposes that abnormal or persistent rests of neural crest cells that form the meninx primitiva form intracranial lipomas at predictable sites—the cistern of the lamina terminalis, pericallosal cistern, velum interpositum, and the suprasellar, quadrigeminal, and cerebellopontine angle cisterns. According to this theory, the meninx primitiva forms initially in what is to become the intracranial cisterns. Between approximately PODs 44 and 57 the cisterns of the subarachnoid space develop coincident with regression of the meninx; the cisterns ventral to the brainstem develop earliest and extend lateral and dorsal to the brainstem (41 days), to the telencephalic vesicles and dorsal midbrain (44 days), and finally to the cistern of the lamina terminalis (56 to 57 days). Maldevelopment of the meninx would therefore least frequently affect those cisterns that develop earlier and most commonly affect those that develop later. Both the preferred anatomic locations, the relationship to (and lipomatous extension into) the choroidal fissure, the passage of cranial nerves through the lipomas, and the maldevelopment of surrounding structures such as the corpus callosum are all predicted by this unifying theory.¹¹³

Postneurulation Disorders—Encephaloceles

The embryology of encephaloceles was reviewed by Chapman and colleagues.¹¹⁴ Although thought by von Recklinghausen to arise from failure of closure of the cranial neuropore,¹¹⁵ these malformations are now thought to represent a postneurulation disorder of mesenchyme through which neural tissue herniates.^{114,116–118} Unlike anencephaly (failure of cranial neural tube closure), in which the involved neural tissue is exposed and completely disorganized, encephaloceles are skin-covered malformations containing well-developed neural (unilateral or bilateral basifrontal or occipital cortex, cerebellum, or brainstem) and mesenchymal (choroid plexus, pia/arachnoid) tissues that have undergone considerable histogenesis.^119,120 Marin-Padilla has suggested that mesodermal insufficiency may result in growth impairment of the chondrocranium and a delay in closure of the membranous neurocranium.^116,118 The subsequent explosive growth of the telencephalon may eventually surpass the accommodative capacity of the neurocranium, and neural tissues may then herniate through the mesenchyme to end extracranially. Microcephaly is a common concomitant of encephaloceles; however, it is uncertain whether this is a primary or secondary event. Alternatively, all that may be necessary is a focal area of mesenchymal insufficiency or weakness to allow the rapidly growing telencephalon to herniate; occipital encephaloceles may be produced in chickens simply by incising the occipital mesenchyme overlying the cranial neural tube after neurulation is complete.¹²¹ Chapman and coworkers have suggested that associated hydrocephalus may contribute to the genesis of encephaloceles.¹¹⁴ However, encephaloceles may occur in the absence of hydrocephalus; moreover, associated hydrocephalus may be merely a consequence rather than the cause of the encephalocele.

Although there is greater agreement on the genesis of occipital and parietal encephaloceles as postneurulation disorders, the embryology of nasofrontal encephaloceles and whether they share a common embryogenesis with frontonasal DSTs and nasal gliomas are unresolved issues. Some have described all three as part of a continuum of anomalies sharing a common embryologic mechanism.^78,85 Hoving and colleagues suggest that all three entities arise from abnormal dysjunction involving the cranial neuropore, perhaps through abnormal apoptosis.^75,122 This theory is based on (1) the constant anatomic relationship between these encephaloceles and the foramen cecum and (2) a sustained connection between neuroectoderm and cutaneous ectoderm within the wall of encephaloceles. However, other evidence suggests that nasal encephaloceles do not share a common embryogenetic mechanism with either nasofrontal DSTs or nasal gliomas and arise through a postneurulation disorder. First, nasofrontal encephaloceles take origin from the frontobasal cortex, whereas nasofrontal DSTs and nasal gliomas with intracranial extension take origin from the lamina terminalis and commissural plate, the site of the anterior neuropore.^82,123,124 Second, nasal gliomas contain only disorganized and poorly differentiated neuroglial tissue with almost no cytoarchitectural features,⁸⁵ whereas frontal encephaloceles often contain well-differentiated basifrontal cortex,^119,120 in some cases arising from only one hemisphere, thus suggesting a disorder that follows telencephalic cleavage (at least after POD 32). Finally, frontobasal encephaloceles are common in Southeast Asian populations¹²⁰ and are associated with a number of genetic syndromes,¹²⁵ whereas frontobasal dermal sinuses, dermoids, and nasal gliomas are not.

Disordered Telencephalic Cleavage—Holoprosencephaly

Formation and cleavage of the telencephalon into the two cerebral hemispheres are controlled by the underlying prochordal plate, a mesodermal region anterior to the notochordal process that expresses the gene product sonic hedgehog. Sonic hedgehog performs a number of important functions during development, including induction of the neural groove from the median hinge point cells, as well as directed axonal outgrowth from ventral horn motoneurons. Mouse mutants that misexpress sonic hedgehog resemble humans with holoprosencephaly, and many families with holoprosencephaly have mutations involving the sonic hedgehog gene (reviewed by Gilbert¹²⁶), but the mechanism whereby misexpression results in holoprosencephaly is unknown. Although the molecular cause of holoprosencephaly is probably manifested during the preneurulation stage, holoprosencephaly ultimately represents a postneurulation disorder caused by improper telencephalic cleavage. The three forms of holoprosencephaly in descending order of severity are alobar, semilobar, and lobar. The alobar form is represented by a monoventricle with no apparent telencephalic cleavage; a dorsal interhemispheric cyst represents a dorsal outpouching of the third ventricle. Diencephalic involvement produces midline fusion of the pallidum and thalami with an absent third ventricle, as well as hypotelorism (remember that the neural retina is an outpouching of the diencephalon) or even cyclopia. The semilobar form is characterized by more advanced posterior hemispheric cleavage but abnormal anterior hemisphere cleavage and incomplete formation of the falx cerebri; the thalami are partially fused, the third ventricle is small, and the hypotelorism is less pronounced. The lobar form is characterized by incomplete formation of the anterior falx with unseparated frontal horns and absence of the septum pellucidum. Perhaps considered the mildest form of disordered telencephalic cleavage, septo-optic dysplasia is characterized by partial or complete absence of the septum pellucidum associated variably with optic nerve hypoplasia and abnormal pituitary-hypothalamic function.

Disorders of Commissural Connections—Callosal Agenesis

Abnormalities of corpus callosum development frequently accompany a variety of other disorders. As discussed previously, development of the corpus callosum proceeds in a rostral-to-caudal sequence, with the genu developing first, then the body, and subsequently the splenium. The rostrum, the last portion of the corpus callosum to develop, is the sole exception to this rostrocaudal sequence. Partial disorders of the corpus callosum therefore characteristically involve the more posterior callosum with sparing of the anterior portions. A significant exception to this rule is holoprosencephaly, in which the anterior corpus callosum is absent but the posterior portions are present.³⁸

Predictably, agenesis of the corpus callosum is often accompanied by agenesis of both the anterior and hippocampal commissures because all three are derived from the commissural plate of the lamina terminalis. It is important to understand that commissural axons form but never cross the midline. Instead, these axons are redirected and run along the medial hemispheres as the bundles of Probst.³⁸ Failure of corpus callosum fibers to cross the midline results in eversion of the cingulate gyrus and failure of formation of the cingulate sulcus.³⁸

The prevailing theory explaining the origin of callosal agenesis is failure of apoptosis involving midline glial cells in the commissural plate. These glial cells serve as a passive bed for the passage of axons from the two hemispheres. Subependymal pioneering glial cells from the medial walls of the lateral ventricles migrate toward the commissural plate and serve as a “glial sling” below the interhemispheric fissure; crossing axons use this cellular “bridge” to cross the interhemispheric fissure. Mouse mutants in which the midline glial sling is disrupted exhibit callosal agenesis.^127–129

Disorders of Neuronal Migration—Schizencephaly, Lissencephaly, Polymicrogyria, and Heterotopias

Disorders of neuronal migration occur between the 5th and 16th gestational week, during which neurons migrate from the ventricular zone to their final locations within the cerebrum. These disorders can take several forms. Schizencephaly represents a focal disorder of migration from the periventricular zone during the 5th to 7th embryonic week, perhaps the result of abnormalities of or prenatal injury to radial glial units. Unilateral and bilateral forms exist and are characterized by gray matter–lined clefts (either open or closed lipped) extending from the ventricular ependyma to the pial surface. The gray matter lining the cleft is often polymicrogyric. The origin of the cleft within the ventricle is marked by a small outpouching (or “nipple”). Other neuronal migrational disorders such as cortical heterotopias, lissencephaly/pachygyria, and septal agenesis often coexist (in 80% to 90%).³⁸

Agyria refers to complete absence of sulcation, whereas lissencephaly, or pachygyria, refers to incomplete sulcation, usually with a few broad and flat gyri and shallow sulci. Lissencephaly is thought to represent abnormal migration of neurons during the 12th to 16th week.³⁸ Neurons apparently begin migration but are unable to complete it.¹³⁰ Posterior callosal agenesis and colpocephaly (enlargement of the atria and occipital horns) are common. Two types of lissencephaly have been described: type I largely consists of a four-layered cortex with a molecular layer, a disorganized outer cellular layer, a cell-sparse layer, and an inner cell layer. Histologically, the cortex resembles that of a 12-week-old fetal brain. Type II lissencephaly is characterized by a thickened, disorganized, and unlayered cortex. Lissencephaly has been described in association with chromosomal and genetic conditions; for example, lissencephaly type I is associated with Miller-Dieker syndrome (deletion of 17p13.3), whereas lissencephaly type II is associated with Walker-Warburg syndrome.¹³¹ Lissencephaly has been associated with mutations in six genes, although mutations in the LIS1 and DCS genes, both involved in regulation of microtubule and cytoplasmic dynein function, account for nearly 80%.¹³⁰

Polymicrogyria is a condition characterized by too numerous and abnormally small gyri. Both focal and diffuse forms exist. The cortex appears to be thicker with closely packed sulcal folds. Most commonly, the cortex is four layered, although variations can occur, with different areas or even the same areas containing unlayered cortex or poorly laminated or parallel four-layered cortex coexisting with four-layered cortex.¹³¹ The distinguishing histologic feature of polymicrogyria is cortical laminar necrosis involving cortical layer V. Polymicrogyria appears to represent a postmigrational abnormality, probably ischemic in origin, that occurs later than the 20th gestational week because migration of neurons to the older, outer cortical layers has already taken place before that time.³⁸ However, polymicrogyria is a heterogeneous disorder, and genetic forms have been described as well.¹³⁰

Finally, localized areas of cortical neuronal migration may result in focal heterotopias in which neurons find themselves buried within white matter. A focal abnormality, either misguided or arrested neuronal migration, is the likely culprit and may involve disorders related to the generation or expression of glial precursors or the glial-neuronal interaction that ordinarily results in focused neuronal migration. Nodular and laminar forms are identified. Laminar forms are characterized by the formation of an intermediate, irregular (and often nodular) band of gray matter between the ventricular and cortical zones; in some cases a “double cortex” occurs.¹³¹ Focal nodular heterotopia is characterized by localized nodular heterotopia adjacent to the ventricular zone. Heterotopias are associated with a number of chromosomal and other congenital disorders, including Aicardi’s syndrome, Chiari II malformation (see later), aplasia cutis congenita, and polymicrogyria. Nodular heterotopias may be associated with a defect in filamin-1 located on the long arm of the X chromosome.^130,132

Disordered Rhombic Lip Development—Dandy-Walker Malformation, Rhombencephalosynapsis, and Other Cerebellar Malformations

Dandy-Walker malformation is characterized by the triad of (1) dysgenesis or agenesis of the cerebellar vermis, (2) cystic enlargement of the fourth ventricle, and (3) an enlarged posterior fossa with upward displacement of the tentorium.¹³³ Hydrocephalus is common but is not usually obvious at birth, instead developing over the first 3 months of age.¹³³ Histologically, the membrane of the fourth ventricular cyst is composed of ependymal, neuroglial, and pial layers, thus suggesting agenesis of the cerebellar vermis between the ependymal and pial layers. The fourth ventricular outlets are usually patent, and communication between the ventricular and subarachnoid spaces occurs in more than 80% of cases.¹³⁴ Although the vermis is most severely affected, some degree of hypoplasia involving the cerebellar hemispheres occurs in most cases.¹³⁵ Associated supratentorial abnormalities are present in 68% of cases and include callosal dysgenesis or agenesis (in up to 50%),¹³⁵ schizencephaly, lissencephaly, and cortical heterotopias.¹³⁴ The Dandy-Walker variant describes a similar, although less severe form with a hypoplastic vermis, a less distended fourth ventricle, a less diminutive posterior fossa, and more normal tentorial position. Dandy-Walker malformation appears to arise as a result of an abnormality involving the rhombic lip. Recent studies have identified chromosome 3 deletions in the region of the zinc finger in cerebellum family of transcription factors. Mice mutations involving homologous regions similarly have a Dandy-Walker phenotype.¹³⁶

Rhombencephalosynapsis is a rare example of focal cerebellar dysplasia.^135,137 Its cardinal features include absence of the midline vermis with fusion of the cerebellar hemispheres and, frequently, dentate nuclei.¹³⁵ Abnormalities also involve other midline structures, such as fusion of the colliculi and absence of the septum pellucidum.¹³⁵ Its embryogenesis is unknown; although “excessive midline fusion” of the cerebellar primordia has been proposed,¹³⁸ this fails to account for abnormalities involving other midline structures. The condition is sporadic and not genetically inherited.¹³⁷

Acquired Central Nervous System Anomalies—Chiari II Malformation

The Chiari II malformation, associated almost universally with a myelomeningocele, is an example of a secondarily acquired malformation that to a variable extent encompasses anomalies of virtually the entire neuraxis. Most prominent among these anomalies are caudal displacement of the cerebellar vermis and tonsils into the cervical canal; elongation, kinking, and caudal displacement of the lower brainstem below the foramen magnum; and upward displacement of the superior cerebellum through a dysplastic, low-lying tentorial incisura. A small posterior fossa and lückenschädel of the skull, as well as many associated telencephalic and diencephalic anomalies (Table 175-4), suggest a pancerebral disorder involving much of the cranial neuraxis and chondrocranium.¹³⁹

TABLE 175-4 Chiari-Associated Central Nervous System Malformations