Limited dorsal myeloschisis associated with intramedullary infantile hemangioma in the conus medullaris: illustrative case

Abstract

BACKGROUND

Limited dorsal myeloschisis (LDM) and intramedullary infantile hemangioma rarely coexist in the spinal cord.

OBSERVATIONS

The authors describe the case of a 3-month-old girl who, despite lacking neurological symptoms or signs, had a cigarette burn-like mark at the lumbosacral area and skin dimpling in the gluteal area. Magnetic resonance imaging showed a low-set conus due to a thickened filum and an abnormal subcutaneous stalk connected to the conus medullaris. In combination with the skin lesions, these findings strongly implied nonsaccular-type LDM. An intramedullary mass in the conus medullaris was also shown on magnetic resonance imaging and was homogenously enhanced with isointensity on T1- and T2-weighted images. We prophylactically untethered the spinal cord and partially removed the intramedullary mass, which had no clear borders, for a safe surgical dissection. Histologically, the intramedullary mass was an infantile hemangioma, and the subcutaneous stalk was a lesion associated with LDM. The patient remained neurologically intact after surgery, and then 2 years later, there was spontaneous regression of the residual tumor.

LESSONS

Although rare, nonsaccular type LDM may appear concurrently with intramedullary infantile hemangioma at the conus medullaris. The authors present a possible mechanism behind this concurrent presentation in the same area.

Limited dorsal myeloschisis (LDM), classified as occult spinal dysraphism, can be associated with intramedullary mass lesions, such as spinal lipoma and epidermoid tumor.^1,2 Here, we report a rare case of LDM associated with intramedullary infantile hemangioma in the conus medullaris. Infantile hemangioma is a benign tumor classified by the International Society for the Study of Vascular Anomalies as a vascular tumor.³ Although one of the most common tumors in the skin and soft tissues in infancy, it rarely occurs in the intramedullary spinal cord.⁴ We discuss the diagnosis and treatment of LDM presenting concurrently with intramedullary infantile hemangioma in the conus medullaris, and we explore the mechanism behind it.

Illustrative Case

A 3-month-old girl was referred to Osaka City General Hospital with skin dimpling at the gluteal fissure. She was born at 40 weeks of gestation. At birth, the pediatrician noted two red skin rashes (later determined by a dermatologist to be cutaneous capillary malformation) and skin dimpling in the gluteal fissure. She had good movement in both lower limbs without any foot deformity. There was no history of urinary tract infection or meningitis, and there were no noted abnormalities of the lower gastrointestinal tract or lower urinary tract. Her family history was unremarkable. On physical examination, she had a dimpling in the gluteal fissure surrounded by a red skin rash (cutaneous capillary malformation; Fig. 1, black arrow) and a lumbosacral cigarette burn-like mark, with another red skin rash approximately 1.5 cm cranial from the gluteal fissure (Fig. 1, black arrowhead). She had no obvious neurological deficit.

FIG. 1.

Photograph showing lumbosacral skin lesions. Black arrow indicates a dimple in the gluteal fissure with a red skin rash around it. Black arrowhead indicates a lumbosacral cigarette burn-like mark with another red skin rash approximately 1.5-cm cranial from the gluteal fissure.

We suspected nonsaccular-type LDM based on the characteristic skin lesions and performed lumbar magnetic resonance imaging (MRI).² The caudal end of the conus medullaris was at the level of L5–S1. Conus medullaris below the lower border of L2 is thought to be abnormally low (Fig. 2A and B).⁵ A lower subcutaneous stalk continued into the spinal canal from a dimple in the gluteal fissure (Fig. 2C, white arrow). An upper subcutaneous stalk continued into the spinal canal from a lumbosacral cigarette burn-like mark (Fig. 2C, black arrow). This was continuous with the conus medullaris and the mass lesion inside (Fig. 2C). The mass lesion was in the intramedullary region of the conus medullaris and had a maximum diameter of 9 mm on isointensity signal on both T1- and T2-weighted images (Fig. 2A, B) and uniform contrast enhancement (Fig. 2D and E). The spinal cord around this mass lesion was not edematous, and there was no flow void. Diffusion-weighted images showed no high-signal lesions. In addition, a thickened filum terminale (2 mm in maximum diameter at the level of S1–2) was continuous with the conus medullaris. LDM has been reported to be associated with a thickened filum terminale,^1,2 so we considered all of these findings, with the exception of an enhanced intramedullary mass in the conus medullaris, to be typical lesions associated with nonsaccular-type LDM.

FIG. 2.

A: Preoperative sagittal T1-weighted MRI. B: Preoperative sagittal T2-weighted MRI. C: Preoperative sagittal constructive interference in steady-state (CISS) image. The upper stalk (black arrow) connects from the base of the cigarette burn-like mark to the conus medullaris. The lower stalk (white arrow) connects from the skin dimple to the end of the dural sac. The black arrowhead indicates a thickened filum terminale. D: Preoperative sagittal contrast-enhanced T1-weighted MRI. E: Preoperative axial contrast-enhanced T1-weighted MRI at the L5–S1 level. A white arrowhead indicates a contrasted intramedullary mass lesion in the conus medullaris. F: Sagittal contrast-enhanced T1-weighted MRI 1 year after surgery. A white arrowhead indicates a contrasted residual intramedullary mass lesion. G: Axial contrast-enhanced T1-weighted MRI at the L4–5 level 1 year after surgery. A white arrowhead indicates a contrasted residual mass lesion. H: Sagittal contrast-enhanced T1-weighted MRI at 2 years after surgery. A white arrowhead indicates a contrasted residual intramedullary mass lesion.

LDM and the enhanced intramedullary mass lesion seemed asymptomatic, but MRI showed that the conus medullaris was at the L5–S1 level, so we considered that to be tethered cord. We thought it was necessary to confirm the histological diagnosis by removal of the intramedullary neoplastic lesions and to perform untethering of the spinal cord due to the LDM.

Split laminotomy was performed at L5 and S1 and a midline fascial incision at S2. Below the S2 level, the lower subcutaneous stalk penetrated the midline fascia defect. The upper subcutaneous stalk (Fig. 2C, black arrow) penetrated the midline fascia defect and dura mater and migrated to a mass lesion in the dorsal conus medullaris. The mass lesion in the conus medullaris, continuous with the upper stalk, was a subpial mass and a firm, reddish-white mass. These findings did not indicate spinal lipoma or epidermoid tumor, which may be occasionally associated with LDM (Fig. 3A). A thickened filum terminale was observed caudodorsal to the mass (Fig. 3B). The thickened filum terminale connecting with the lower subcutaneous stalk was continuous with the conus medullaris (Fig. 3C). We dissected the upper stalk and thickened filum terminale and untethered the spinal cord. Intraoperative histopathological diagnosis of the upper stalk showed no epithelial component suggestive of congenital dermal sinus, so we diagnosed nonsaccular-type LDM. Next, the subpial mass lesion of the conus medullaris was removed incrementally. Detachment at the boundary between the mass lesion and the intact spinal cord (Fig. 3D) was difficult, so when we performed a partial resection of the intramedullary mass lesion, a thin layer of the mass lesion remained on the dorsal side of the spinal cord (Fig. 3E). After removing the mass lesion, the ends of the pia mater on both sides of the conus medullaris were sutured to reconstruct a form of the conus medullaris. Intraoperative neurophysiological monitoring consisted of evoked electromyography of the lower extremities and bulbocavernosus reflex, and these were uneventful.

FIG. 3.

A: Intraoperative photograph after opening the dura. The upper stalk (asterisk) from the cigarette burn-like skin lesion had penetrated the dura mater and migrated to a mass lesion (T) on the dorsal side of the spinal cord. The mass was firm and in a subpial location. B: When the mass (T) was lifted, a thickened filum (F) caudal and dorsal to the mass was observed. C: The lower stalk (double asterisks) connected from the skin dimple to the filum terminale (F) covered by the epineurium running caudally from the end of the dural sac. D: After cutting the thickened filum, a spinal cord (S) continuous with the filum terminale (F) could be seen on the ventral side of the mass (T). E: After removal, the mass (T) remained in a thin layer on the dorsal side of the spinal cord (S).

Histopathologically, the intramedullary mass comprised lobulated, round nucleated cells, forming a luminal structure with red blood cells partially inside (Fig. 4A). The proliferating cells were immunohistochemically positive for GLUT-1 (Fig. 4B), CD31, and CD34. The subcutaneous stalk comprised dense collagen fibers containing microvessels and peripheral nerves, although there was no evidence of glial tissue. The cells forming an intramedullary mass lesion in the conus medullaris were positive for GLUT-1 specific to infantile hemangioma.

FIG. 4.

Pathological findings. A: Hematoxylin and eosin staining, original magnification ×400. Lobulated arrangement of proliferating cells with a central lumen. B: Positive staining for GLUT-1, original magnification ×400.

Postoperatively, the patient remained apparently neurologically intact. In the 2 years since surgery, she has not received a formal evaluation of her bowel/bladder function because of her age, but she is growing without urinary tract infections. We followed up on the remaining infantile hemangioma without additional treatment. Lumbar MRI at 3 months, 1 year, and 2 years after surgery showed gradual regression of the contrasted residual mass lesion without hemorrhage (Fig. 2F–H, white arrowhead).

Discussion

Observations

Spinal Lesions Associated With LDM

LDM caused by spinal dysraphism is occasionally associated with other congenital lesions. Pang et al.² reported 63 cases of LDM, including six cases of dorsal lipoma, two cases of terminal lipoma, 27 cases of thickened filum terminale, six cases of split cord malformation, one case of neurenteric cyst, three cases of dermal sinus, and two cases of syringomyelia. Also, dermoid sinuses, which have a similar subcutaneous stalk on MRI, may be associated with epidermoid and teratomas as intramedullary lesions.^6,7 To the best of our knowledge, however, there are no reports of LDM and infantile hemangioma in the conus medullaris that are similar to this case.

Histopathological Diagnosis of Infantile Hemangioma

In this case, the intramedullary lesion associated with LDM was histopathologically infantile hemangioma, also known as capillary hemangioma or juvenile hemangioma. It is challenging to distinguish between infantile hemangioma and other vascular tumors or malformations by use of hematoxylin and eosin staining alone. However, infantile hemangioma can often be diagnosed by observation of the structure using a combination of special stains and immunostaining. Endothelial cells in infantile hemangioma have positive immunostaining for GLUT-1. Infantile hemangioma can be differentiated from other benign vascular tumors and malformations because endothelial cells of other benign vascular tumors and malformations are GLUT-1 negative.⁸

Rarity of Infantile Hemangioma in and Around the Spinal Cord

Infantile hemangioma is one of the most common tumors in skin and soft tissue. Reports of infantile hemangioma arising in the intramedullary region of the spinal cord of pediatric patients, as in our case, are rare. Regarding the frequency of infantile hemangioma occurring in the spinal canal in children, Viswanathan et al.⁴ reported that of 1,454 cases of infantile hemangioma, just 7 infant cases had extradural lesions in the spinal canal and none had intramedullary lesions. Drolet et al.⁹ reported 9 cases of infantile hemangioma in the spinal canal among 41 patients younger than 18 years who had undergone MRI examinations for infantile hemangioma in the midline lumbosacral skin region. Schumacher et al.¹⁰ reported on 20 patients younger than 18 years with infantile hemangiomas in the midline lumbosacral skin who had undergone MRI examinations. Nine of these patients had infantile hemangiomas in the spinal canal, 6 of whom had extradural lesions in the spinal canal, 1 had intradural lesions only, and 2 had intradural and extradural lesions. In addition, various infant cases of infantile hemangioma in the spinal canal accompanied by cutaneous infantile hemangioma have been reported.^11–15 Meanwhile, there has been only 1 reported case of infantile hemangioma of the spinal cord without cutaneous infantile hemangioma, that of a 3-year-old girl reported by Steinberger et al.¹⁶

MRI Findings of Infantile Hemangioma

General MRI findings of infantile hemangioma are as follows. In the proliferative phase, the tumor has a well-defined lobulated shape, with iso- to low intensity on T1-weighted images and high intensity on T2-weighted images. There is no arteriovenous shunt, although there is a no-signal area due to a flow void within or at the margin of the tumor. Dynamic MRI shows strong and uniform contrast in the early stage. Infantile hemangioma lacks infiltration of the surrounding area and no surrounding edema.¹⁷

In the currently reported case, there was an intramedullary neoplastic lesion with uniform enhancement by contrast media. It was located at the end of the conus medullaris and was not accompanied by cysts or surrounding edematous changes. Intramedullary infantile hemangioma in the spinal cord is rare. Moreover, the lack of a flow void on MRI and the isointense T2-weighted image in this case made the diagnosis by preoperative imaging more difficult because the findings differed from those typical of infantile hemangioma. Infantile hemangioma should be a consideration if intramedullary lesions are found in LDM that do not match the findings of lipoma and epidermoid because intramedullary lesions have association with LDM.

Clinical Characteristics and Treatment of Infantile Hemangioma

The natural history of infantile hemangioma of the central nervous system is unclear. Capillary hemangioma, which has similar histology to infantile hemangioma, occurs mainly in adults and has a different clinical presentation in pediatric cases.^18,19

However, infantile hemangioma of the skin and soft tissues, which occurs most frequently and is relatively easy to follow, becomes apparent in the first few weeks of life. There is then a proliferative phase that increases until approximately 18 months of age, followed by regression by about 5 years of age, and disappearance after that, depending on the individual.^20–22 Hervey-Jumper et al.¹² reported an infantile hemangioma in the L5 dorsal root ganglion that was biopsied, although the lesion regressed after 16 months. Regarding the regression of infantile hemangioma, GLUT-1–positive endothelial cells in infantile hemangioma are facultative stem cells that can take on a stem cell or progenitor cell phenotype under certain conditions. The decrease in GLUT-1–positive endothelial cells with age is reportedly responsible for the characteristic course of the spontaneous regression of infantile hemangioma.²³

Treatment of infantile hemangioma of the skin and soft tissues with propranolol, steroids, surgical removal, and embolization/sclerotherapy has been reported to be effective.^22,24,25 Whether drug treatment is also effective for infantile hemangioma of the central nervous system is unclear. Eight cases of infantile hemangioma in the central nervous system reported by Viswanathan et al.⁴ were treated with steroids, with early regression of the lesions. Surgery was performed in one case of infantile hemangioma in the spinal canal, and there was improvement of the sensory and motor functions of the lower extremities. In some cases, the lesions reportedly grow with a steep curve.^22,26 Steinberger et al.¹⁶ reported a case of multiple infantile hemangiomas in the thoracic spinal cord, the lungs, and the liver. They removed infantile hemangiomas after embolization because the tumors had enlarged after medical treatment. In a case of intramedullary infantile hemangioma reported by Ksendzovsky et al.,¹³ reoperation was performed because of reincreased lesion size and worsening edema 3 weeks after partial removal of the lesion.

Postoperative Follow-Up of our Patient

In our patient, after partial resection, the residual infantile hemangioma tended to regress for 2 years after surgery without additional treatment. In the event that the infantile hemangioma remained in the spinal cord after removal, it would be reasonable to follow-up with caution for new neurological deficits due to enlargement of the lesion or hemorrhage, although these are less common. If symptoms due to infantile hemangioma progress, the intramedullary mass lesion enlarges, or any bleeding from it is observed, reoperation should be considered. If removal is difficult, drug treatment such as with propranolol may be an option, as in the case of infantile skin hemangiomas.

Infantile Hemangioma and Spinal Dysraphism

Drolet et al.⁹ reported spinal anomalies in children with cutaneous lumbosacral infantile hemangiomas. Their prospective study found that the relative risk for patients with infantile hemangioma overlying the lumbosacral region measuring ≥2.5 cm in diameter for spinal anomalies was 640 (95% confidence interval, 404–954). The positive predictive value of cutaneous lumbosacral infantile hemangioma for spinal dysraphism was 51.2%. They recommended MRI screening for children with cutaneous lumbosacral infantile hemangioma.

The relationship between cutaneous lumbosacral infantile hemangioma and spinal dysraphism has been examined in several studies. The spinal dysraphism varied from spinal lipoma and congenital dermal sinus, which are caused by a disorder of primary neurulation, to filum lipoma and thickened filum terminale, which are caused by a disorder of secondary neurulation.^9,27–29

Perineal hemangioma, external genitalia malformations, lipomyelomeningocele, vesicorenal abnormalities, imperforate anus, and skin tag syndrome; spinal dysraphism and anogenital, cutaneous, renal, and urological anomalies associated with an angioma of lumbosacral localization syndrome; and lower body hemangioma and other cutaneous defects, urogenital anomalies/ulceration, myelopathy, bony deformities, anorectal malformations/arterial anomalies, and rectal anomalies syndrome are cutaneous infantile hemangiomas of the perineum or lumbosacral region associated with spinal dysraphism, urogenital anomalies, anorectal anomalies, and others. The most common spinal dysraphism is lipomyelomeningocele, but there are also filum lipoma and thickened filum terminale.^30–32

Schumacher et al.¹⁰ reported 20 cases of cutaneous lumbosacral infantile hemangioma associated with spinal dysraphism; 50% of the cases had spinal lipoma, 65% had filum abnormality, and 45% had extradural and/or intradural infantile hemangioma in the spinal canal. The group was unable to speculate a time during primary or secondary neurulation when the infantile hemangioma forms.

Considering the Pathogenesis of Infantile Hemangioma

Although infantile hemangioma is the most common tumor of infancy, its cause remains unknown. The most likely scenario is that hypoxic stress triggers the induction of overexpression of angiogenic factors such as VEGF. In response to that, infantile hemangioma stem cells proliferate and differentiate into immature endothelial cells, as well as pericytes, dendritic cells, and mesenchymal cells with adipogenic potential.²² Based on the expression of pericytes, neural crest, and stem cell markers within them, infantile hemangiomas could be derived from pluripotent stem cells of the neural crest.^33,34

Origins of Neural Crest Cells

The neural crest is a transient and multipotent cell population.^35,36 The neural crest cells in caudal regions derive from the tail bud.^37–39 In human embryos, a secondary neural tube arises from the mesenchyme of caudal eminence (tail bud). Derivatives of the caudal eminence also include the caudal portions of the digestive tube, blood vessels, notochord, and somites.^40,41 Copp et al.^42,43 reported that the stem cell population within the tail bud is multipotent, giving rise to all nonepidermal tissues of the postlumbar body, including the neural tube and vertebrae, and that probably for this reason, malformations and tumors of sacral and coccygeal regions comprise several tissue types.

Considering the Pathogenesis of LDM

Pang et al.¹ hypothesized that disorders of primary neurulation might cause LDM. In contrast, Kim et al. proposed that the LDM penetrating the interspinous ligament below the S1–2 level might be caused by secondary neural tube defects based on the developmental process.^27,44 From data on the morphology of secondary neurulation in chick embryos, at Hamburger and Hamilton stage (HH stage) 16, the caudal cell mass is completely fused with the future cutaneous ectoderm. Then, as the cavitation process progresses, the mass is detached from the cutaneous ectoderm at HH stage 20.^44,45 Kim et al.⁴⁴ hypothesized that if the focal point where the caudal cell mass fuses with the skin is left undetached and if the rest of the secondary neurulation occurs, a stalk may reside between the skin and the cord at the level of the secondary neurulation. This event may result in LDM in the form of the nonsaccular type.

Thickened filum terminale is believed to be caused by disorders of secondary neurulation.²⁷

Our Hypothesis About LDM and Intramedullary Infantile Hemangioma

In this case, the subcutaneous stalk of LDM penetrated the interspinous ligament below the S2 level and was directly continuous with the intramedullary infantile hemangioma in the conus medullaris. This suggests that both LDM and intramedullary infantile hemangioma can occur by a common pathogenic mechanism.

Disorders of secondary neurulation can cause thickened filum terminale and LDM with the stalk penetrating the interspinous ligament below the S1–2 level.^27,44 The neural crest cells in the sacral region derive from the caudal cell mass,^37–39 and infantile hemangioma can derive from pluripotent stem cells of the neural crest.^33,34

For these reasons, we hypothesized that errors of secondary neurulation might cause LDM with a subcutaneous stalk at the S1–2 level. After that, intramedullary infantile hemangioma in the conus medullaris might have developed with that neural crest in sacral region derived from the caudal cell mass during secondary neurulation. Because the neural crest in the sacral region developed from the caudal cell mass, which is usually not in contact with cutaneous ectoderm, infantile hemangioma in the conus medullaris might arise without the lumbosacral cutaneous infantile hemangioma.

However, as there are no other reported cases of LDM and intramedullary infantile hemangioma of the conus medullaris, we hope to analyze the pathogenesis of LDM and intramedullary infantile hemangioma by collecting more cases in the future.

Lessons

We encountered a rare case of nonsaccular type LDM with intramedullary infantile hemangioma in the conus medullaris. The possibility of infantile hemangioma should be considered when the LDM has intramedullary lesions that do not match the imaging findings of spinal lipoma and epidermoid. We hypothesized that the secondary neurulation might be associated with the occurrence of the intramedullary infantile hemangioma in the conus medullaris and the LDM with the subcutaneous stalk penetrating the interspinous ligament below the S1–2 level.

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author Contributions

Conception and design: Nakanishi, Kunihiro, Sakamoto. Acquisition of data: Nakanishi. Analysis and interpretation of data: Nakanishi, Inoue, Sakamoto. Drafting the article: Nakanishi, Sakamoto. Critically revising the article: Nakanishi, Umaba, Sakamoto. Reviewed submitted version of manuscript: Nakanishi, Umaba, Matsusaka. Approved the final version of the manuscript on behalf of all authors: Nakanishi. Statistical analysis: Nakanishi. Administrative/technical/material support: Nakanishi, Umaba. Study supervision: Nakanishi.