A tapered and unusually slender dural sac is frequently seen near the lower end of the spinal canal in caudal agenesis, particularly in the part below the cord terminus. This tapering is most obvious and the sac itself ends higher in cases of high bony agenesis, so that the dural sac of total sacral agenesis ends in an abrupt cone at L1 or L2. The cauda equina is mostly not constricted; rather, the short tapered dural sac and correspondingly narrow bony canal give the appearance of a caudal atretic process.

Rarely, however, the caudal dural sac is extremely narrow and resembles a spindle on magnetic resonance imaging (MRI) or myelography, sometimes with scalloping at multiple levels. The nerve roots of the cauda equina are tightly bunched together in a single solid bundle without the normal spread out pattern. These patients typically have severe radicular pain and sensorimotor paralysis that respond only to decompression duraplasty. These rare examples of symptomatic dural stenosis have a striking myelographic picture not seen in any other condition besides caudal agenesis.

The Caudal Spinal Cord

An extreme form of the stretched conus is occasionally seen in which the spinal cord literally ends opposite the tip of the bony spinal column and its caudal segment is distended by a large hydromyelic cavity reminiscent of the ventriculus terminalis of the secondary neural tube. There is no discernible filum. and the elongated cord is attached directly to the dural cul de sac.

Other associated conus lesions include terminal and transitional spinal cord lipomas, intradural dermoids, skin covered meningoceles and LDM. Split cord malformations, either type I with its osseocartilaginous spur and double dural tubes or type II with a fibrous midline septum and a single dural tube have also been associated with caudal agenesis. The most dramatic and complex caudal cord lesion is the terminal myelocystocele in which the terminal cord balloons out into a large hydromyelic cyst. The cyst protrudes through a posterior spinal defect into an expanded caudal meningocele sac and fuses partly with the dorsal subcutaneous tissues and partly with a distal intraspinal lipoma opposite the last sacral piece.

Correlation between Neurological and Neuroimaging Findings

Correlation between the Severity of Bony Aplasia and the Neurological Level

The severity of the bony aplasia correlates well with the motor level but not the sensory level. Thus, the patients with the highest motor levels (L3, L4) also have high "last normal vertebra" (LNV), often with exact corresponding levels; and patients with mild (low) motor deficits also tend to have low sacral defects. In contrast, sensation is either totally or near totally preserved in 30 patients regardless of the level of the bony aplasia. Some patients do not follow the above pattern and have motor levels much higher than the LNV; as in patients with terminal myelocystoceles and transitional lipomas. Presumably, these additional caudal cord malformations "super­imposed" neural disorganization on the already dysplastic conus. (Their sensory levels are also much higher than the LNV.)  For the cases of hemisacrum, the side of the hemidefect always has more profound motor weakness and is also invariably the side of the unilateral renal anomaly.

Correlation between the Level of Bony Aplasia and Caudal Cord Lesions

For reasons relevant to the embryogenesis of caudal agenesis, aplastic lesions with no vertebra below S1 are considered high bony malformations, and those with preserved S2 or lower sacral pieces are considered low bony malformations. When the level of the last vertebra present (LVP) is correlated with the level of the conus, a definite pattern emerges. Patients with conus ending above the lower body of L1 (group 1 conuses), almost all have high bony malformations. In contrast, of the patients whose conuses end below L1 (group 2 conuses), most have low bony malformations, mainly of the S2 or S4 subtotal type. This group also includes all cases of hemisacrum, in which the lower sacral pieces are present on at least one side. Thus, most of the stretched­out conuses with thickened filums or other additional malformations such as myelocystoceles, lipomeningoceles, and extreme elongation of the conus are associated with low sacral malformations.

Embryology

The aforementioned clinical, pathologic, and neuroimaging data can now be integrated into a theory to explain the classic features of caudal vertebral aplasia and its embryogenetic relationship with various non-neural and neural malformations.

Embryology of the Caudal Region

Since imperforate anus and genitourinary anomalies form part of the caudal agenesis triad, embryogenetic mechanisms purporting to explain caudal spine and spinal cord dysplasia must also account for the spectrum of hindgut and genitourinary malformations. Primordia of these three organ systems in the caudal region of the embryo originate from a common source, the caudal eminence or tailbud. This is a mass of pluripotential tissue first distinguished in the stage 9 embryo [O'Rahilly staging; postovulatory day (POD) 20] as a continuation of the primitive streak. By stage 12 (POD 26), primary neurulation has ended with closure of the caudal neuropore. and the primitive streak almost completely regresses. The correspondingly more prominent caudal eminence now stretches from the shrinking Hensen's node to the rostral end of the cloacal membrane anlage, and subsequently takes the place of the primitive streak to provide structures comparable to those derived more rostrally from the three germ layers. Thus, the caudal eminence will give rise to all the tissues of the caudal embryo including the neural cord (the caudal continuation of the neural tube), caudal neural crest cells, caudal notochord, somites caudal to somite 30, caudal mesenchyme, and hindgut.

While primary neurulation of the neural plate accounts for formation of the spinal cord down to the lumbosacral junction, neural material for the caudal embryo (sacral and coccygeal segments of the spinal cord) is laid down not as a superficial (neuroectodermal) plate but as a solid neural cord in a process called secondary neurulation. In the chick embryo, secondary neurulation results in the formation of the medullary cord (equivalent to the neural cord in humans) which is composed of a dense outer cell layer and a loose inner cell layer. Cavitation begins cranially between these two lay­ers and produces multiple lumina within the medullary cord. The inner cells eventually become resorbed, and the multiple small lumina, lined only by the outer cells, coalesce to form a single, large central lumen. This secondary neural tube is initially not continuous with the primary neural tube, but lies slightly dorsal to it in an overlap zone. Fusion between these two tubes takes place in a final step.

Unlike the chick, the secondary neural tube in the mouse is always in direct continuity with the primary neural tube. It extends caudally from the caudal neuropore by accruing cells from the neural cord to form a medullary rosette. Moreover, the lumen of the secondary neural tube in the mouse is single rather than multiple, and is always continuous with the central canal of the primary neural tube. There is therefore no overlap zone or fusion site. Caudal growth of the secondary neural tube in the mouse occurs by additional cavitation of the medullary rosette and recruitment of additional cells from the caudal eminence.

The characteristics of secondary neurulation and the integration of the neural cord with the primary neural tube are incompletely understood in the human. According to Mϋller and O'Rahilly, and Schoenwolf, the human situation resembles that of the mouse. Their specimens described a single lumen (ventriculus terminalis) within the secondary neural tube continuous with the central canal of the primary neural tube without an overlap zone. In contrast, human secondary neurulation as described by Lemire et al. and Bolli more closely resembles that of the chick, with its multiple lumina transition stage. Whichever view ultimately prevails, the final stage of caudal spinal cord formation is known to involve a retrogressive differentiation of structures formed during canalization, resulting in the regression and complete disappearance of the embryonic tail. Atrophy of the caudal neural tube is responsible for formation of the filum terminale, which connects the future conus (containing, for a while, the ventriculus terminalis) with the coccygeal medullary vestige, a small ependymal cell rest stuck to the coccyx.

According to Mϋller and O'Rahilly, somite 31 and all subsequent somites in the human develop from the caudal eminence. Secondary neurulation thus begins opposite somite 31. Also, the boundary between the "primary" somites (originating from the primitive streak) and caudal somites (originating from the caudal eminence) is thought to be at the S1 and S2 vertebral junction. Thus, the material for the sacral vertebrae 1 to 5 being derived from the mesenchyme for somites 30-34 is mainly laid down during stage 12, at a time when primitive streak activity is winding down and is subsequently replaced by that of the caudal eminence. Consequently, formation of the first sacral vertebra is linked to end-stage primary neurulation, whereas the S2 to S5 pieces are linked to the developmental fate of the caudal eminence. It is unclear whether the caudal notochord exerts inductive influences over the caudal neural cord during secondary neurulation in the same manner that the more cranial notochord regulates primary neurulation, or whether the caudal notochord is also involved in the subsequent retrogressive differentiation. At least in the chick, the caudal notochord appears to be important in the successful formation of the caudal neural tube.

Prior to caudal neuropore closure at stage 12, the bilaminar (ectoderm and endoderm) cloacal membrane lies in a dorsal position just caudal to the primitive streak . With subsequent enlargement of the caudal eminence during secondary development, the cloacal membrane is gradually pushed in a curving arc from a dorsal to a ventral position as the tail (caudal) fold of the embryo takes shape. During folding, part of the yolk sac is incorporated as the hindgut, the terminal part of which soon dilates slightly to form the cloaca. The cloacal membrane now forms the ventral wall of the cloaca. During the fourth to seventh weeks of development, a urorectal septum arises from the caudal mesenchyme in the angle between the allantois and hindgut and grows caudad until it fuses with the cloacal membrane, dividing this into a posterior anal membrane and an anterior urogenital membrane. The cloaca is correspondingly divided into a posterior anorectal canal and an anterior primitive urogenital sinus.

With the emergence of the metanephric blastema from the intermediate mesoderm and its apposition with the ureteric bud from the mesonephric duct, the future permanent kidney gains connection with the upper (vesical) part of the urogenital sinus. the future urinary bladder. The lower pelvic and phallic portions of the urogenital sinus become the future urethra. During closure of the anterior hypogastric wall of the embryo, ventral migration of caudal mesenchyme from the somatopleure of the lateral mesodermal plate forms a pair of elevated cloacal folds, which fuse with each other in front of the cloacal membrane.

With subsequent division of the cloacal membrane into the urogenital and anal membranes, the cloacal folds are likewise segregated into the genitourethral fold anteriorly and the anal folds posteriorly. In the male, the genital tubercle, the prominent fusion site of the genital folds in the cranial end of the urogenital membrane, elongates to become the phallus, while the two flanking urethral folds close over the slit-like urogenital membrane to form the penile urethra. A secondary pair of scrotal swellings at the base of the urethral folds enlarges to form the scrotum. In the female, the genital duct is formed from the mesenchyme parallel to the mesonephric system, called the paramesonephric or mϋllerian ducts. These fuse in the midline behind the vesical urogenital sinus (future bladder) and ultimately open into the vestibule behind the urethra as the vagina, while their cranial parts expand into the uterus and fallopian tubes.

Embryogenesis of Caudal Anomalies

As the pluripotential source for most of the tissues in the caudal embryo, the caudal eminence has been imputed as the teratogenic target for malformations concomitantly involving the caudal spinal cord, the anorectal complex, and the genitourinary tract. Duhamel thought that a complete absence of the caudal eminence leads to aplasia of all three primordia and results in the most severe form of caudal agenesis, the rare human sirenoid monster that consists of (1) agenesis of the lumbosacral spine; (2) fusion of the lower limbs (symmelia): (3) anorectal agenesis; (4) renal, urethral, vesical and ureteral agenesis and (5) agenesis of all mϋllerian duct derivatives with sparing of the gonads. The symmeli a is thought to be due to approximation and fusion of the two lower limb buds when the pelvis anlagen are drawn to the midline by the void of the missing caudal tissues, while the sparing of the gonads is explained by the fact that the primordial germ cells forming the gonads come from the yolk sac endoderm near the allantois and not from the caudal eminence. Indeed, Wolff had produced the sirenoid monster in chick embryos by destruction of the caudal eminence.

For caudal malformations less severe and complex than the siren, Duhamel and others postulated less extreme and more localized deficiencies of the caudal eminence. Since S1 is probably derived from more cranial somites (somites 29 and 30) that have already been formed before the onset of caudal eminence activity, an isolated failure in the caudal eminence-derived somites (from somite 31 downward) would result in subtotal sacral defects that spare S1. This was observed in Renshaw's patients and is certainly the case in the Pang's experience, in which 29 of 34 patients had completely preserved S1, two patients had preservation of at least half of S1, and only three patients had no S1 at all. The few instances of lumbar agenesis remain difficult to explain on the basis of isolated caudal eminence failure. It is possible that severe teratogenic insults that affect the caudal eminence actually cause regression of already formed somitic mesoderm in more cranial locations, or, alternatively, that teratogenesis occurs at the time of transition between primary and secondary development of the embryo.

A secondary effect of caudal somitic agenesis may well be that the hypoplastic caudal axial structures also fail to provide adequate impetus for the caudal-ventral push of the caudal fold and cloacal membrane, causing malpositioning of the cloacal membrane and changes in the primitive cloaca. Such perturbation of the caudal mesoderm and cloacal membrane may not inhibit downgrowth of the urorectal septum, but may be sufficient to forestall its final fusion with the cloacal membrane so that the cloaca is incompletely segregated into the urogenital and anorectal portions. The terminal segment of the anal canal could no longer be completed, nor could the canal perforate the anal membrane, thus giving rise to anorectal atresia and imperforate anus, often associated with a rectourethral or rectovaginal fistula. If the urorectal septum does reach the cloacal membrane but deviates dorsally before fusing with it, the anal canal would be open but stenotic. The same insult that inhibits the somitic (axial) mesoderm may also adversely affect the intermediate mesoderm (the nephrotome) that normally forms the metanephric blastema and the ureteric buds. and cause malformations ranging from agenesis to duplication or abnormal fusion of the kidneys and ureters. Similar insults to the adjacent Mullerian ducts in females result in agenesis or dysgenesis of the salpingouterovaginal tract. Abnormal location or orientation of the urogenital membrane in the male affects its eventual invasion by the caudal mesenchyme so that the genital tubercle and urethral folds ultimately fail to complete ventral fusion, giving rise to hypospadias and bifid scrotum. Anomalous condensation of the genital tubercle at the caudal rather than the rostral end of the urogenital membrane will explain the lack of dorsal fusion of the phallus in epispadias.

Other abnormalities of the caudal eminence may result in an overdeveloped and nonregressing cloacal membrane that acts as a mechanical barrier against the ventral migration of the lateral somatopleure mesoderm, which is vital to the formation of the infraumbilical abdominal wall (the "wedge effect" proposed by Marshall). Alternatively. CO2, laser injury to the portion of the caudal eminence destined to supply the ventral mesoderm has been shown to cause mesodermal arrest and likewise an abdominal wall defect. Subsequent rupture of the cloacal membrane prior to completion of the urorectal septum allows the dorsal wall of the cloaca to eventrate through the abdominal wall opening. turning its future bladder and bowel fields inside out. This process produces an exstrophied central bowel field surrounded by two hemibladder fields; formation of an omphalocele above the cloacal exstrophy completes the OEIS complex. If the anterior part of the cloacal membrane ruptures after completion of the urorectal septum, the resultant exstrophy will consist only of the midline bladder field (classic bladder exstrophy).

Relationship between Sacral Aplasia and Development of the Caudal Spinal Cord

Major deficiencies in the caudal notochord and axial mesoderm can be expected to interfere with secondary neurulation. The simplistic way to explain spinal cord dysplasia in severe (high) sacral agenesis would be the loss of the inductive influence from the missing caudal notochord and paraxial mesoderm on the corresponding segments of the secondary neural cord during the condensation or early canalization stage. The affected caudal cord segments either simply do not form at all, or form badly.

There is evidence to suggest that "simple" cord aplasia, at least of the ventral motor portion, does occur in severe (high) sacral agenesis. First, reports show that the motor level corresponds well with the level of the vertebral defect. Second, high sacral defects are correlated with short conuses that are blunted and have lost their usual taper as if the caudal portion is missing (not formed). Third, neuropathologic studies on patients with lumbar or high sacral agenesis have confirmed severe distal cord dysplasia. Rusnak and Driscoll found few anterior horn cells in the caudal cord in a patient with L3 agenesis and Price et al. also showed in a patient with L4 agenesis that the degree of myelodysplasia increased toward the termination of the cord in that more anterior horn cells dropped out and the normal cytoarchitecture became increasingly unidentifiable. Others have also reported atrophic nerve roots caudal to the last vertebra in high sacral malformations. Finally, patients with a short conus and high sacral defect also have a narrow and short caudal dural sac as if the meninges are also affected by the same atretic process. Thus, severe insults to the caudal eminence appear to cause complete inhibition of the caudal somites and simultaneous arrest or profound disturbance in neural cord condensation, resulting in severe motor deficits and incompletely formed conus. Such an atretic process also affects the dural sac since the latter depends on induction from the neural tube.

A low sacral malformation logically must be associated with a lesser degree of caudal cord dysplasia, and a conus less club shaped and ending nearer to L1. A correlative study between the level of bony aplasia and conus level suggests otherwise. Low sacral malformations are highly associated not with near-normal conuses but ones that are extremely elongated to well below L1 and variously tethered by thick filums or distended by large terminal hydromyelia. These conus shapes suggest defective retrogressive differentiation rather than incomplete neural cord formation. It follows that milder or later insults to the caudal eminence occurring after the mesodermal deposition for the midsacral somites must affect secondary neurulation not during condensation of the neural cord or its canalization, but during the subsequent retrogressive differentiation. Thus, premature termination of the retrogressive process would result in a low conus and thickened filum. whereas complete lack of retrogression and anomalous persistence of the ventriculus terminalis would produce the extremely elongated hydromyelic conus. A variation of this theme, perhaps further complicated by abnormal ventral migration of caudal mesoderm and traction on the caudal notochord as in cloacal exstrophy, may explain the trumpeted and grossly hydromyelic caudal cord in terminal myelocystocele. In short, milder insults causing lesser disturbance of the caudal axial somites (as in low sacral agenesis) are more likely to affect secondary neurulation in ways that do not cause abrupt cessation of neural tube formation but instead give rise to a low-lying, malformed and "tethered" conus.

Regardless of the bony level, conus length and the severity of the motor deficits, the dorsal (sensory) portion of the caudal cord in sacral agenesis always seems to develop better than the ventral (motor) portion. Published reports consistently describe sensory levels several segments below the motor levels. The dorsal cord develops better than not only the ventral cord, but also the axial mesoderm. In addition. the wedge shape of the conus in some cases also suggests a better developed dorsal half. Lastly, histopathologic studies have verified that dorsal sacral roots and dorsal root ganglia (DRG) in patients with sacral agenesis are better preserved than the corresponding ventral roots and the dorsal conus is also less disorganized than the corresponding ventral half.

Two mechanisms have been postulated for this relative dorsal sparing. First, studies on human and pig embryos have shown that the notochord and ventral regions of the neural tube receive their first blood supply earlier than the dorsal region - that is, shortly after the newly developed embryonic circulation begins spreading from its primary location in the walls of the yolk sac. Given the temporary preferential circulation to the ventral cord, delivery of the teratogenic substances at this specified period would affect the notochord and the ventral neural tube to a greater extent than the dorsal half. Second, the peripheral sensory neurons are dependent on the central nervous system (CNS) for development only during a restricted critical period. Kalcheim and LeDourain showed that even though the early differentiation and survival of neural crest cells destined to become DRGs are absolutely dependent on signals from the neural tube, this dependency changes from being total to only partial at a slightly later stage of development. Thus, segregation of migrated neural crest cells from the influence of the neural tube using an implanted silastic membrane in the 30-somite chick embryo at day 2 resulted in complete disappearance of DRGs, but similar silastic implantation performed on day 4 to 4.5 of development, after the formation of dorsal root connections, did not prevent the development of a DRG on the somitic side of the implant. It is thought that with the establishment of connections with peripheral targets. the DRG cells become increasingly less dependent on CNS signals but instead rely more on trophic support from the target organs. Therefore, if the teratogenic insult in sacral agenesis occurs after the establishment of this peripheral dependency, the DRGs will survive at levels where the vertebra and/or neural tube segments do not. In humans with caudal agenesis. DRGs with full central and peripheral connections have in fact been found at several levels caudal to the last vertebra and the last ventral root. Also, neural crest cells that are autonomic precursors behave like a distinct lineage from DRG precursors and are not significantly influenced by the neural tube even during early development. This explains the intact autonomic nerve fibers and ganglion cells found in rectal biopsies from patients with total absence of the lumbosacral spine,

Association with Other Syndromes

It is understandable how other complex multiorgan syndromes comprising of all or part of the triad of lumbosacral aplasia, imperforate anus and genitourinary anomalies may be linked to caudal agenesis. These syndromes include VATER, OEIS and the unusual entity called "cloacal malformation." all having many overlapping anomalies in addition to the central (unifying) feature of sacral aplasia. The same types of conus malformations are also found in all of these syndromes. Even though strict boundaries between these syndromes are difficult to impose for the sake of recognizing regional defects and properly assigning treatment priorities,  these syndromes probably should be individually designated with the proviso that transitional forms will sometimes defy rigid classification. Thus. the keystone of VATER is the tracheo-esophageal fistula, whereas in caudal agenesis, tracheo-esophageal abnormalities, radial aplasia and cardiac defects are rare. In OEIS the lower abdominal wall is always defective, and the ileocecal junction eventrates through this defect in between two hemibladder fields below an omphalocele, whereas in caudal agenesis the abdominal wall is closed and the genitourinary anomalies are hidden, except for hypospadias. Also, while sacral aplasia is the sine qua non of the caudal agenesis syndrome, vertebral abnormality is not an absolute requirement for OElS, in the series of OEIS of Hurwitz et al. only 48 percent of patients had vertebral anomalies and in the series by Ziegler et al. 86 percent had vertebral malformations but only 53 percent had sacral aplasia.

Similarly. even though sacral aplasia is seen in 40 percent of patients with cloacal malformation. in which the urinary, genital, and intestinal tracts converge into a common outflow chamber, the cloaca (Latin for sewer), true cloacal malformation is found exclusively in phenotypic females, and it generally lacks the miscellaneous osseous malformations in caudal agenesis such as those involving the pelvis, legs, ribs and extrapelvic vertebrae.

On the standpoint of embryogenesis, however, the distinctions are far less clear between these syndromes that affect overlapping but different subsets of organs. It recently has become known that morphoregulation of large segments of the embryo is dependent on concentration gradients of morphogens controlled by the expression of homeobox-containing genes. It has been postulated that multiorgan malformations involving large "developmental fields" are due to disorders of the homeobox gene expression. Different but related multiorgan syndrome complexes (OEIS, VATER, caudal agenesis. etc.) may simply be part of the same homeobox defect with involvement in dissimilar yet also overlapping developmental fields.

Etiology of Caudal Agenesis

Evidence suggests that both extrinsic and genetic factors contribute to caudal agenesis. Karyotype screening of patients with this syndrome consistently turned up no abnormality. However. Frye et a!. described sacral aplasia in a group of Swiss laboratory mice and defined the transmission mode as autosomal recessive. Familial occurrences of caudal agenesis in humans are rare but have been reported. Welch and Aterman delineated three distinct forms of familial sacral defects, and suggested that all three were autosomal recessive traits.

Except for the rare familial forms, all reported sacral agenesis cases are sporadic. Although familial cases do not show other linkage traits, sporadic caudal agenesis is strongly associated with maternal diabetes mellitus. Pooled data showed that 16 percent of infants with caudal agenesis have diabetic mothers. but only 1 percent of infants born to diabetic mothers have caudal agenesis. In addition, maternal prediabetic and latent diabetic states are also associated with caudal agenesis. It was initially thought that some subtle action of the diabetogenic genes might be a critical etiologic factor. However, if this were the sole explanation, an association between paternal diabetes and sacral aplasia would also be expected, which has not been observed. Furthermore, an incidence higher than 1 percent would be expected among all pregnancies of diabetics. A more likely explanation is that the maternal diabetic or prediabetic state provides a specific maternal milieu, perhaps hyperglycaemia, that provokes an abnormal fetal response in the presence of a specific (but unidentified) fetal genetic constitution which renders it susceptible to the maternal diabetic milieu. This would explain the low frequency of caudal agenesis in pregnant diabetics because only certain fetuses will react to this milieu by developing spinal aplasia. The fetal genetic constitution is therefore also a critical factor. Welch and Aterman postulated that this susceptibility may be associated with a relatively unusual human leukocyte antigen (HLA) haplotype.

Besides maternal diabetes, other extrinsic factors may contribute to the teratogenic environment for caudal agenesis. Minute embryonal trauma producing longitudinal kinking of the long embryonic axis has been known to cause caudal suppression. Drastic temperature changes also produced frog embryos with caudal aplasia, which led to the implication of maternal pyrexia as a risk factor. Maternal diets deficient in certain vitamins have also been suspected to produce piglets without hind limbs. In addition, a long list of putative chemicals haw been named. Because many diabetic mothers of children with sacral defects were insulin users, insulin itself has been implicated. Landauer produced rumpless chickens by injecting insulin and other sulfur­containing molecules such as cysteine, cystine, and thioglycolic acid into the yolk sac of chick embryos. Lithium salts have been known to suppress the development of the notochord. Trypan blue injected intraperitoneally into pregnant rats produced offspring with vertebral and neural aplasia. Shenefelt also reported a wide spectrum of malformations, including lumbosacral agenesis, after giving a single dose of retinoic acid (a carboxylic derivative of vitamin A1 alcohol) to golden hamsters at different stages of pregnancy.

The exact mechanism by which these noxious agents exert their deleterious effects are unknown. Recently, Alles and Sulik demonstrated increased cell death in the primitive streak and caudal eminence of mouse embryos given retinoic acid and suggested that deficiencies in the gastrulating mesoderm may be responsible.

Trypan blue is known to uncouple oxidative phosphorylation;  its teratogenic action may rest on inducing anoxia in tissues undergoing active differentiation with a high metabolic demand. such as the caudal eminence. How these deleterious effects are translated into dysmorphogenesis is also unknown. Gardner and Nelson invoked a disorder of the axial mesodermal "developmental field" responsible for orchestrating migration and determination of prospective caudal eminence cells during gastrulation. Others have proposed failure of a caudal "organizer." At the cellular level, the "developmental field" or "organizer" is thought to involve cell-cell interactions mediated by cell surface morphogenesis proteins.

Progressive Neurological Deficits in Caudal Agenesis

Given the combined handicap imposed by the skeletal and neurological lesions in caudal agenesis, it is important to determine whether early diagnosis and treatment can improve the neurological outcome and thus the total functional level of these patients. While a certain neurological "baseline" is fixed in each patient, Pang and Hoffman found in 1980 that this baseline can get worse in some cases. Other reports have since confirmed that the progressive deficits in these patients are mainly caused by lesions tethering or compressing the conus. Early detection and treatment of these lesions not only can prevent further neurological deterioration but also can help regain lost function in some cases. Neuroimaging studies are obtained on all caudal agenesis patients to classify lesions according to whether they primarily involve the bony canal, the dural sac or the caudal cord itself.

Abnormalities of the Bony Canal

Instances of a tight bony canal causing compression of the conus or cauda equina have been reported. Rarely, this occurs when the canal is smoothly funnel-shaped and diffusely stenotic, but more likely the thecal sac is indented by bony excrescences projecting inward from the vertebral body and neural arches near the terminal bony column, where anomalous osteogenesis is common. Dense fibrous bands stretching between bifid neural arches have also been shown to compress the cauda equina.

Abnormalities of the Dural Sac

A tapered and narrow caudal dural sac near the lower end of the spinal canal is common in caudal agenesis but this seldom causes neurological problems. Rarely, however, the caudal sac is extremely narrow and tightly constricts the lumbosacral nerve roots. The myelographic picture of true dural stenosis is unique to caudal agenesis, but the symptoms are similar to the neurogenic claudication caused by spinal stenosis and may likewise be due to cauda equina ischemia. Improvement is dramatic with decompression duraplasty.

Abnormalities of the Caudal Spinal Cord

Except for the few cases of bony and dural stenosis, neurological symptoms and deterioration in caudal agenesis occur exclusively in patients with a low-lying and tethered conus. The description "low-lying" requires clarification here because the "normal" position of the nontethered conus in caudal agenesis might actually be one or two levels higher than L1 since all the club- or wedge­shaped conuses end opposite T11 or T12. It is in fact uncommon to find the prototype "normal" conus in caudal aplasia, i.e. one with a gentle taper ending at the mid- to lower level of the L1 vertebral body. The norm may well be the short, blunt conus seen so often, which is actually a favourable finding since neurological deterioration is not associated with it.

It is far more likely to find a tethered conus in patients with hemisacrum and low type III agenesis (preserved S2) than high type III agenesis (absent S2). Thus, the suspicion for tethering should paradoxically be heightened when a patient is found to have a "mild" case of sacral agenesis. Total sacral agenesis (types I and II) is rare; in the three cases included in Pang D. study, two were associated with tethering lesions.

The list of caudal neural tube lesions in caudal agenesis likely to cause progressive deficits runs the gamut of occult dysraphic lesions known to tether the conus. These lesions are all associated with intact lumbosacral skin, which supports the hypothesis that the teratogenic insult in caudal agenesis affects the caudal eminence after primary neurulation and closure of the caudal neuropore. The most common tethering lesion in caudal aplasia, as in occult spinal dysraphism, is the thickened filum terminale. Only two cases of the extremely elongated conus with terminal hydromyelia have been reported in caudal agenesis: in these cases, the postulated complete lack of retrogressive differentiation of the secondary neural tube predicts absence of the filum. and the conus is directly anchored to the dural cul-de-sac near the end of the bony column. Spinal cord lipomas have been associated with caudal agenesis. The few reported cases of skin-covered "meningoceles", probably represent limited dorsal myeloschisis. in which a stalk consisting of glial tissue, nerve roots, fibrous bands and blood vessels arises from the dorsal aspect of the cord and extends into the base of the sac through a narrow dural fistula; the cord is thereby tethered to the subcutaneous base of the meningocele by this fibroneural stalk. Both types I and II split cord malformations have been reported with sacral agenesis. In such cases, the cord is not only tethered at the site where the hemicords are transfixed by the midline septum. but usually also by a thick filum distally. The most complex lesion seen in sacral agenesis is the terminal myelocystocele, in which the trumpeted hydromyelic cord is tethered to the subcutaneous fat and usually also to an intraspinal lipoma. It is frequently associated with cloacal exstrophy although at least two cases have been reported with uncomplicated sacral aplasia without additional defects.

Diagnosis of Caudal Cord Lesions

Subsequent prospective screening with MRI of all patients with caudal agenesis, with or without symptoms, revealed a much higher incidence of tethering and compressive lesions than previously assumed. The incidence reported in 1993 is 70 percent among all known cases of lumbosacral agenesis. This figure is almost certainly spuriously high since many caudal agenesis patients with mild stable deficits and no conus lesions no doubt still remain undiagnosed, but it clearly underscores the need to utilize a more inclusive screening protocol among high risk candidates. especially when we now know that functions lost due to cord tethering are often not reclaimable if treatment is rendered late.

MRI should be used for screening for the bony as well as the dural and intradural abnormalities. If a complex lesion such as a myelocystocele, LDM, transitional lipoma or split cord malformation is identified, its details can be further delineated by computed tomographic myelography, which also serves as a surgical road map. The following groups of patients are recommended for MRI screening:

1. Infants having the classic external features of caudal agenesis such as flat buttocks, low intergluteal cleft, tapered legs, talipes deformities or bilateral congenital dislocation of the hips.

2. Patients with imperforate anus (sacral aplasia is found in 13 to 54 percent of patients with imperforate anus).

3. Male infants with genital duct anomalies such as hypospadias, epispadias, bifid scrotum or ambiguous genitalia.

4. Female infants with genital duct anomalies such as salpingouterovaginal atresia or bicornuate or didelphic uterus.

5. Infants with urinary tract anomalies such as hydroureter, caliectasis, double collecting system, pelvic or horseshoe kidney and unilateral renal agenesis.

6. Neonates with unexplained recurrent upper or lower urinary  tract infections.

7. Patients with known caudal agenesis, with or without neurological deficits.

8. Patients with multiorgan syndromes involving the caudal eminence derivatives, such as cloacal exstrophy (OEIS complex), VATER association, the cloacal malformation and bladder exstrophy.

Management

With the rare exception of an associated open myelomeningocele, neurosurgical intervention is usually delayed until other life-threatening anomalies (as in OEIS and VATER) are dealt with. For infants with OEIS, the initial series of operations include closure of the omphalocele, internalization of the open bowel field and construction of a colostomy and ileostomy, repair of the open hemibladders, vesicostomy, and gender reassignment with genioplasty. In the VATER infants, the tracheo-esophageal fistula and imperforate anus are repaired during the first few days of life. For the nonsyndromic cases of sacral agenesis, the isolated imperforate anus, if present, is always primarily repaired at birth either by colostomy or a pull-through procedure.

After the first few weeks of life, the anomalies in caudal agenesis likely to cause serious health problems in infancy are still not neurosurgical but those that threaten renal function. The three main goals of urologic management are (1) preservation of renal function. (2) prevention of infection, and (3) achieving continence, in that order. Since preservation of renal function means avoidance of chronic pyelonephritis and hydronephrosis, the finding of dilated upper tracts, cortical thinning, persistent bacteriuria, large post-void residual volumes, or vesiculoureteral reflux requires early aggressive therapy. Even if the initial evaluation shows only mild reflux and small post-void residual volume. the likelihood of further deterioration in the upper tract as well as late increase in reflux is significant. Prevention of hydronephrosis and chronic infection can be achieved with manoeuvres that facilitate complete bladder emptying and maintain low intravesicular pressures. Methods such as the Crede manoeuvre, bladder training (if some detrusor function remains). intermittent catheterization, bladder trigonoplasty and urinary diversion with a vesicostomy or an ileal conduit have all enjoyed success when tailored to the individual needs of the patient.

Neurosurgical procedures are performed on three categories of patients; the timing of each depends on the associated caudal spinal cord lesion. In the first category, the immediate treatment group, newborns with a large protruding lumbosacral sac are treated definitively shortly after the child recovers from the initial series of abdominopelvic procedures. Thus, for infants with terminal myelocystocele and the OEIS complex, the intact skin over the sac is opened to expose the proximal meningocele. the spinal cord just rostral to the myelocystocele, the large ependyma-lined, trumpet-shaped terminal spinal cord cyst and the subcutaneous lipoma to which the cyst is attached after protruding through the neural arch defect. The myelocystocele is resected together with the lipoma; the rostral, more normal-looking cord is thereby untethered and the distal dural sac is reconstructed. Transitional lipomas are resected flush with the surface of the cord and detached from the dorsal dural attachment; the dural defect is then closed with a graft. LDM and partially skin-covered meningoceles are treated by resection of the meningeal diverticulum and sectioning of the tethering band of fibroneural tissues. Included in this immediate treatment group also are the infants with an extremely low­ lying cord and terminal hydromyelia even though they do not have an external protruding sac. The terminal conus is sharply resected from the caudal dural cul-de-sac, the terminal hydromyelia is drained and the spinal cord is untethered. In the second category of patients, the neurosurgical procedure is performed at a later age either because of late-onset neurological deficits or when incidental imaging reveals a tethered cord. The primary aim for this group is to untether the conus. Most of these patients have a thick, taut filum essentially indistinguishable from the tight filum found in patients without sacral agenesis and are treated the same way with simple division of the filum. Sometimes, a large terminal lipoma caudal to the conus-filum transition is also resected. Any associated intraspinal dermoid cyst should be resected totally. The author has also encountered a large anterior sacral meningocele that was drained through an intraspinal approach and its ventral dural opening closed with a dural patch. The third category are patients with symptomatic dural stenosis. which usually presents in late childhood or early adulthood with neurogenic claudication. The primary aim is to decompress the severely constricted cauda equina. Typically. the nerve roots herniate out abruptly following laminectomies and upon opening the slender dural tube. attesting to the tightness of the constriction. A generous decompressive duraplasty can be fashioned.

Orthopaedic management, though third in priority of timing, is no less important in caudal agenesis, especially if the patient's neurological function allows for ambulation, provided there is spinopelvic stability and functional hip joints. Spinopelvic instability is the most serious skeletal deficiency, impeding ambulation or even good sitting balance. It is particularly severe in type IN and type IIN cases, where the ilia articulate with each other below the endplate of the last vertebra. Spinal fusion should be carried out early to prevent hip flexion contractures and permanent hip dysplasia. Scoliosis may be due to hemivertebra or neuromuscular paralysis; the former kind is unlikely to respond to bracing. Hip dislocation resulting from muscle imbalance and/or acetabular malposition is often treated with a combination of procedures such as closed reduction, adductor release, innominate osteotomy, posterior iliopsoas transfer and varus derotational osteotomy. Ambulation, proper standing and sitting postures and efficient usage of prostheses and crutches are also enormously benefited by inhibitory casting and soft tissue releases when hip or knee joint contractures are present. Deformities of the feet are treated with serial casting, tendon transfer, soft tissue release. taliectomy and various forms of ankle arthrodesis.