Rudimentary Meningocele: Remnant of a Neural Tube Defect?

Background  Rudimentary meningocele, a malformation in which meningothelial elementsare present in the skin and subcutaneous tissue, has been described in thepast under a variety of different terms and has also been referred to as cutaneousmeningioma. There has been debate as to whether rudimentary meningocele isan atretic form of meningocele or results from growth of meningeal cells displacedalong cutaneous nerves

Objective  We reviewed the clinical, histological, and immunohistochemical characteristicsof rudimentary meningocele in an attempt to assess the most likely pathologicmechanism for it.

Design  Retrospective study.

Setting  University hospitals.

Patients  Thirteen children with rudimentary meningocele.

Main Outcome Measures  Medical records were reviewed and histopathologic examination as wellas immunohistochemistry studies were performed for each case. A panel of immunoperoxidasereagents (EMA, CD31, CD34, CD57, S-100, and CAM 5.2) was used to assess lineageand to confirm the meningothelial nature of these lesions.

Results  Recent evidence indicating a multisite closure of the neural tube inhumans suggests that classic meningocele and rudimentary meningocele are ona continuous spectrum.

Conclusion  Rudimentary meningocele seems to be a remnant of a neural tube defectin which abnormal attachment of the developing neural tube to skin (comparableto that in classic meningocele) could explain the presence of ectopic meningealtissue. In the majority of cases, no underlying bony defect or communicationto the meninges could be detected. However, in light of the probable pathogenesis,imaging studies to exclude any communication to the central nervous systemshould precede any invasive evaluation or intervention.

RUDIMENTARY meningocele is an uncommon developmental anomaly in whichmeningothelial elements are displaced into the skin and subcutaneous tissue.In the past it has been described under a variety of different terms, suchas cutaneous meningioma, hamartoma (of the scalp) with ectopic meningothelialelements, sequestrated meningocele, acoelic meningeal hamartoma, and cutaneousheterotopic meningeal nodules. ^1-8Controversy exists as to whether this entity represents a form of meningocelein which the underlying connection to the meningeal space is obliterated,or whether lesions develop because of proliferation of meningeal cells alongthe routes of cutaneous nerves.⁹

There is recent evidence for a multi-site closure of the neural tubein humans similar to that observed in experimental animals.^10,11Multiple distinct closure sites of the neural tube explain the most frequentlocations of neural tube defects. Failure to achieve complete fusion resultsin various forms of dysraphia.

We present 13 cases of rudimentary meningocele that show a strikinglysimilar anatomical distribution to the neural tube fusion sites observed experimentally.The multisite closure model explains the congenital nature and distributionof these lesions and also coincides with the distribution of classic meningoceles.

The clinical data are summarized in Table 1. All 13 patients were children from 0 to 6 years of age(median, 19 months). Eight were boys and 5 were girls. All lesions, variablydescribed as patches, papules, nodules, or exophytic masses, were congenitaland ranged in size from 0.5 to 8 cm (Figure1A). Some lesions had increased insize with the growth of the child. All were midline lesions and all but 2occurred on the scalp, preferentially involving the occiput and vertex. Theremaining 2 lesions occurred overlying the cervical spine (C2-C3) and lumbosacralspine (Figure 1B). The "hair collar"sign, a ring of coarse hair surrounding the malformation and thought to becharacteristic of cranial dysraphism, was found in 3 infants. Eleven patientshad solitary defects. In each of the remaining 2 cases, 2 lesions were apparent.Other clinical abnormalities included port wine stains at a separate locationin 2 patients and a congenital melanocytic nevus in patient 1. The clinicaldiagnoses included aplasia cutis, epidermal inclusion cyst, congenital melanocyticnevus, hemangioma, skin tag, dermal sinus tract, meningocele, and encephalocele.Small bony defects were identifiable in 2 patients and a connecting fibroustract to the dura was identifiable in one based on radiologic imaging resultsor direct surgical visualization.

The 2 patients with lesions along the spine showed more complex underlyingabnormalities. Both had bifid laminae and tethered spinal cords. In one ofthese patients, there was associated diastematomyelia and syringomyelia. Thepatient also had attachment of the penis to the scrotum (penis palma) withvesicoureteral reflux. In the second patient, a fibrovascular bundle piercedthe dura and extended into the subdural space (Figure 1C).

All defects were locally excised. No recurrence was observed, and therewas no evidence of postsurgical meningitis or cerebrospinal fluid leakageafter a median follow-up of 28 months.

Formalin-fixed and paraffin-embedded tissue from 13 skin lesions wasexamined in conventional and immunoperoxidase sections via a light microscope.Immunoperoxidase stains were prepared with an avidin-biotin method using monoclonalantibodies directed against epithelial membrane antigen (EMA) (Dako, Capenteria,Calif), CD31 (Biogenex, San Ramon, Calif), CD34 (Novocastra, Newcastle-upon-Tyne,England), CD57 (Becton-Dickinson, San Jose, Calif), and CAM 5.2 (Becton-Dickinson),as well as polyclonal antibodies directed against S-100 protein (Dako). Twocases were evaluated further with anti–glial fibrillary acidic protein(GFAP) (Boehringer-Mannheim; Indianapolis, Ind) and vimentin (Monosan; Uden,the Netherlands).

The microscopic findings are summarized in Table 2. In general, skin appendages, vessels, nerves, and melanocyteswere quantified, and remnants of brain tissue were sought. Psammoma and collagenbodies and calcification were also sought.

The overall architecture varied in that 8 lesions showed replacementof both the dermis and subcutis by the malformation, while in 4 lesions only1 of the compartments was involved. In case 5 the biopsy was too superficialto exclude deep involvement. In some specimens a loose network of pseudovascularspaces in the superficial dermis (Figure 2A) contrasted with dense collagenoustissue containing clusters of meningocytes in the deeper parts. Cystic structureslined by meningothelial cells were present in 5 cases (Figure 2B). The meningocytes, recognizable by their ovoid to spindleshape with scant eosinophilic cytoplasm and small nuclei with finely stippledchromatin and indistinct cell borders, formed cords, strands, and small nestsand had the tendency to encompass collagen fibers and adnexal structures.We observed decreases as well as increases in eccrine glands and hair follicles.There were also distorted follicles and an increase in apocrine glands andsmooth muscle bundles. In one case, a follicular cyst was evident. Prominentvessels were noted in 8 cases. Necrotic glial tissue was absent. Calcificationwas noted either in the form of psammoma bodies (4 patients) or in an unusualreticulate pattern (5 patients) (Figure 2C).

EMA immunostaining highlighted the meningothelial cells (Figure 3) but revealed a heterogeneous pattern. Two cases failedto express EMA in the superficial aspect of the lesion but showed strong expressionin the deep portion. Similarly, in specimen 5, which was transected at thelevel of the midreticular dermis, the loose network of pseudovascular spacesdid not express EMA, but vimentin nicely outlined these spaces. CD31 and CD34,known to be vascular markers, did not react with these anastomosing channels,confirming their nonvascular nature.

Antibodies directed against S-100 protein, CD57, and CAM 5.2 failedto label meningothelial cells. S-100 was used to quantify melanocytes, whichwere increased at the dermo-epidermal junction in 1 specimen and within therudimentary meningocele in 2 patients. Adjacent nerves were also identifiedvia S-100 immunostaining in 4 patients. Anti-GFAP detected small nests ofapparent glial tissue in patient 7 and also stained a few cells in patient12. However, results of analysis for neurons with neuron-specific enolaseand for axons with neurofilament in these 2 cases were negative, unlike whathas been reported in most cases of heterotopic glial tissue. Meningothelialelements like those observed in both cases are not a feature of heterotopicglial tissue, but rather favor a diagnosis of rudimentary meningocele.

The pathogenesis of human neural tube defects is complex and poorlyunderstood. Minor deviations not serious enough to significantly arrest ordistort the formation of the central nervous system undoubtedly occur. Rudimentarymeningocele may be this type of minor deviation and poses a problem to dermatologistsin terms of diagnosis and management.

The 13 cases of rudimentary meningocele presented herein were congenital,situated either over the scalp or midline spine. The clinical appearance washighly variable, leading to difficulty in clinical diagnosis; clinical diagnosessuch as melanocytic nevus, epidermal cyst, dermal sinus tract, hemangioma,meningocele, encephalocele, and aplasia cutis were offered. Since the correctclinical diagnosis of cranial/spinal dysraphism was suspected in only 3 ofthe 13 patients, there appears to be relative unfamiliarity with these lesions.One of the most useful clinical features of cranial dysraphism, the so called"hair collar" sign, was observed in 3 cases.^10,12-14Described by Commens et al,¹³ this finding consists of a collaretteof hair, which encircles the malformation. This finding suggests an underlyingmeningocele, but the pattern is not specific, since the same clinical patternoccurs in association with an underlying encephalocele, heterotopia of braintissue, or membranous aplasia cutis.¹⁵ Membranousaplasia cutis refers to congenital, sharply marginated ovoid scalp defects.In contrast to rudimentary meningocele, this condition often presents as multiplelesions with a tendency to regress from an initial cystic or bullous appearanceto a scar. Despite these differences, there seems be an overlap with rudimentarymeningocele in terms of the clinical picture. One of our patients presentedwith 2 bullous scalp defects that could not be differentiated from membranousaplasia cutis on clinical grounds. The histopathologic features in our patient,however, were clearly those of rudimentary meningocele based on the presenceof meningeal tissue, which by definition is not found in membranous aplasiacutis.

In view of the relative rarity of rudimentary meningocele, the histopathologicdiagnosis is difficult. The microscopic features are often subtle, with meningealtissue simulating the appearance of vascular or connective tissue. Meningocytestend to encircle collagen bundles and are sometimes accompanied by psammomabodies; these signs should be sought as clues to the microscopic diagnosis.An additional helpful clue was a decrease or increase in adnexal structures,similar to the pattern that can be seen in some forms of epidermal nevi andaplasia cutis. Cystic structures lined by meningothelial cells, as would beseen in a classic meningocele, were occasionally evident, highlighting thesimilarity between the classic and rudimentary forms of this disorder. Classicmeningocele can be separated from rudimentary meningocele only on the basisof clinical data.¹⁶ In general, meningocyteswere few in number, were positioned between collagen bundles, and createda microscopic pattern resembling vascular spaces. These pseudovascular spacescan easily be misinterpreted as a vascular neoplasm such as lymphangioma oreven angiosarcoma.⁵ Apart from the clinicalpicture, absence of nuclear pleomorphism and mitotic figures, and the lackof reactivity to vascular markers, such as CD31 and CD34, should enable exclusionof a vascular neoplasm.

Another histopathologic finding that could contribute to error in microscopicdiagnosis is the presence of syncytial giant cells. These cells result fromfusion of meningocytes and may simulate the pattern of giant cell fibroblastoma,a form of dermatofibrosarcoma protuberans usually observed in children.^16,17 Myxoid stroma noted in either casemay further obscure the diagnosis.^18-20Immunoperoxidase staining is vital in distinguishing these disorders, as dermatofibrosarcomaprotuberans typically expresses CD34 and fails to express EMA, while rudimentarymeningoceles display opposite reactivity.¹⁶Although EMA expression is vital to the identification of cells of meningotheliallineage, it is important to note that EMA failed to label the network of pseudovascularspaces located in the superficial dermis in 3 of our patients. Only thoroughexamination revealed EMA-positive meningocytes in the deeper parts of 2 specimens.Therefore, in specimens from superficial biopsies that do not extend belowthe level of the middermis, as was the case in patient 5, meningothelial elementsmay not be detectable. With respect to superficial biopsies, some cases describedas membranous aplasia cutis with no meningeal or brain tissue detected byimmunohistochemistry may actually be examples of rudimentary meningocele.¹⁵ We observed only focal EMA positivity in normal leptomeningealtissue, as reported by others.^8,20,21The diffuse meshwork of subarachnoid cells shares many features with the loosenetwork seen in rudimentary meningocele, and this meshwork does not labelat all with EMA. An alternative explanation is that the dilution of the EMAantibody that usually works well with epithelial cells might not work as wellin cells of meningothelial lineage.

Speculation about the etiology of rudimentary meningocele started withthe first observation of this malformation by Winkler in 1904, in which hedescribed a "peculiar disease of the skin and subcutaneous fat."¹An increasing number of reports of this entity have appeared in the more recentliterature, and with them a variety of divergent opinions on the pathogenesishave been suggested. Some authors consider these lesions to be a form of meningocelewith an obliterated intracranial communication, while others refer to themas remnants of the neural crest. A more historical theory suggests that theymight represent intradermal nevi containing psammoma bodies.^3,7,16,22-26In addition, these lesions were also postulated to arise from the sheath cellsof cutaneous nerves or to reflect continued growth of meningeal cells withina perineural environment.^8,27

A fairly new theory about closure of the neural tube may yield greaterinsight into the pathologic mechanism of rudimentary meningocele, as it seemsto explain some of the clinical findings.^10,11Traditionally, neural tube closure was believed to begin in the cervical regionand to proceed from there in a continuous bidirectional way¹¹(Figure 4A). More recent observationsin humans and animals, however, indicate multisite initiation of neural tubeclosure¹⁰ (Figure 4B). The failure of closure to take place partially or completelywould thus result in neural tube defects.¹¹Mouse models have shown 4 separate sites of neural tube fusion. Homologiesof early embryogenesis in mice and humans as well as previous illustrationsand photographs of human embryos suggest that neural tube fusion occurs atsites similar to those in mice. Of particular interest are the locations ofsuch neural tube defects, as they coincide with the distribution observedin our cases of rudimentary meningocele (Figure 4C). Closure in the occipital area appears to be achievedby a membrane rather than a midline fusion of folds, which explains why suchdefects may or may not appear in the midline.¹⁰Another feature of neural tube defects is genetic susceptibility, with a riskof repetitive events in affected families.¹⁰This suggests that specific genes control individual closure sites.¹⁰ Interestingly, familial occurrence of rudimentarymeningocele has been noted in an autosomal dominant inherited pattern andamong two siblings.^20,21 Simpledisplacement of meningocytes along peripheral nerves is an unlikely alternateexplanation for such observations.

Another finding that supports classifying rudimentary meningocele asa form of meningocele is the identification of connections, albeit rudimentaryones, that extend from the lesion to dura, and in some instances are associatedwith minute bony defects.^3,6,23,25We were able to detect such fibrous tracts in 2 cases and a small osseousdefect in 1 infant.

In light of the likely pathogenesis of these lesions, prior to completesurgical excision, we believe that imaging studies should be obtained to excludethe possibility of a communication to the central nervous system. If thereis evidence of an underlying connection, the patient should be referred forneurosurgical evaluation. If a communication to the central nervous systemcan be excluded, simple excision would be appropriate, and neither recurrencenor neurologic consequences would be expected to occur.

The pathologic mechansim of rudimentary meningocele remains enigmaticand has not been completely elucidated. However, our clinical and histopathologicfindings and those of other investigators support the conclusion that thisentity is a form of dysraphism. We believe this process is a non-neoplasticcondition, and thus the designation of cutaneous meningioma should not beused as a synonym for rudimentary meningocele.

Accepted for publication October 11, 2000.

Corresponding author and reprints: Laila El Shabrawi-Caelen, MD,Department of Dermatology, University Graz, Auenbruggerplatz 8, A-8036 Graz,Austria (e-mail: lailaelshabrawi@hotmail.com).