Introduction

Butterfly vertebra is a rare congenital anomaly, presenting as a sagittal cleft in the vertebral body due to failure of fusion of the two chondrification centers. It has been mainly described as an isolated finding in the literature, but it can be associated with various syndromes, such as Alagille, Jarcho-Levin, Crouzon and Pfeiffer syndrome [1]. Usually the abnormality occurs in the thoracic spine, rarely in the lumbar or cervical spine [2]. It is usually asymptomatic, but it could lead to chronic back pain due to alteration of spine biomechanics, disc herniation [3] and rarely deformity such as kyphosis or kyphoscoliosis with or without neurological symptoms [4]. Most reported cases on butterfly vertebra focused mainly on the imaging findings [5,6,7,8] and only few papers discussed a treatment strategy in case surgery was indicated [9, 10]. We present the case of a male patient, almost reaching his 4th decade, who has been complaining of chronic back pain and legs pain caused by thoracolumbar kyphoscoliosis in relation to a butterfly vertebra at the level of T11.

Case report

A 39-years old male has been complaining of middle and low back pain for over 10 years with progressive onset of bilateral radicular pain with legs heaviness, forcing him to use a case during walking. His past history revealed the use of a brace during childhood for a congenital spinal condition that he could not specify.

Physical examination revealed back pain at the thoracolumbar level with a significant gibbosity noticed clinically, associated to a low back pain probably due to a facet joint syndrome as hyperlordosis could be noted. Neurological assessment showed an unsteady gait with bilateral quadriceps weakness (4/5), altered sensation below the belly button (T11 level) mainly on both thighs, and enhanced lower extremities reflexes with positive Babinski sign and ankle clonus, but normal bowel and bladder functions.

Full spine anteroposterior and lateral X-rays at the time of presentation revealed a pelvic incidence of 41°, a sacral slope of 31°, a thoracolumbar kyphosis with an angulation of 60° between T10 and T12, with compensatory lumbar hyperlordosis of 71° (including an L4S1 angle of 34°), a thoracic hypokyphosis of 20°, and a global tilt angle [11] of 10°, also a short thoracolumbar scoliosis of 32 degrees was seen (Fig. 1a, b). Preoperative Global Alignment and Proportion (GAP) score [12] was calculated based on the previously mentioned spinopelvic parameters, it resulted in a total score of 4 which corresponds to a moderately disproportioned sagittal spinopelvic state (Table 1). CT scan confirmed the deformity (Fig. 2a, b) and revealed its congenital nature with the presence of a butterfly vertebra at the level of T11, where failure of fusion of the two chondrification centers of the vertebral body with anterior and median aplasia could be seen (Figs. 2c, 3a, b), with posterior arch fusion between T10 and T12 (Fig. 3c). MRI showed cord compression at the apex of the kyphosis associated to syringomyelia (Fig. 4a, b) which was an incidental finding rather than the primary pathology, and no Chiari malformation could be detected.

Fig. 1
figure 1

Full spine anteroposterior and lateral X-rays at the time of presentation revealing a thoracolumbar kyphosis with an angulation of 60° between T10 and T12, with compensatory hyperlordosis of 71°, thoracic hypokyphosis of 20 degrees and coronal thoracolumbar deformity of 32 degrees (a and b)

Full size image
Table 1 Global alignment and proportion (GAP) score based on the spinopelvic sagittal parameters
Full size table
Fig. 2
figure 2

CT scan confirming the sagittal and coronal deformity (a and b), and revealing a typical aspect of a butterfly vertebra of T11 on the axial view (c)

Full size image
Fig. 3
figure 3

3D reconstruction detailing the left and right parts of the T11 butterfly vertebra (a and b, black arrows), and showing a posterior fusion mass between T10 and T12 (c, black arrow)

Full size image
Fig. 4
figure 4

MRI T2-weighted images revealing cord compression at the apex of the kyphosis associated to syringomyelia (a, black arrow and b)

Full size image

Given the presented deformity with neurological impairment, the patient underwent a posterior resection of the butterfly vertebra (T11) with instrumentation from T8 to L2.

Surgical technique

The patient was installed in a prone position, on four cushions (Fig. 5a). During the surgery, we used transcranial motor evoked potentials, somatosensory evoked potentials, and free running electromyography (EMG) of the lower extremities as well as evoked EMGs with pedicle screw stimulation.

Fig. 5
figure 5

Intra-operative images showing the thoracolumbar hump (a), spinal exposure and screws insertion from T8 to L2 but sparing T11 and left T12 (b), resection of the T11 butterfly vertebra (c, black arrows), insertion of the cage on the left side (d, black arrow), reduction of the kyphosis with placement of a domino on the right side (e), placement of autologous bone graft (f), clinical aspect at the end of the procedure (g)

Full size image

The operative field was exposed from T8 to L2; a posterior cutaneous midline incision was made. The spine was exposed subperiosteally, going laterally to the costotransverse junction. Resection of the inferior articular processes at all levels was performed bilaterally to provide maximum flexibility to the spine. We next used the free-hand technique to place the pedicle screws from T8 to T10 and from T12 to L2, but left T12 pedicle was congenitally absent; therefore, no screw was inserted (Fig. 5b), two sublaminar offset hooks were put at the distal level of the construct to increase the resistance to pull out forces. To expose the lateral wall of T11, the transverse processes were removed with a rongeur, and the proximal 3 cm of the right and left ribs (including the rib head) were removed. A cobb elevator was than placed on the lateral wall of the vertebra, bilaterally, and moved anteriorly to the anterolateral quadrant, to retract all the lateral soft tissues, a complete surgicel was left in place to maintain a safe distance between the bone and the soft tissues. Two complete foraminotomies both cephalad and caudad to the T11 pedicles on both sides were made, completed by complete laminectomy of the concerned level, with partial laminectomies of the levels just above and just below, this enabled surrounding of the pedicles. Both pedicles were than removed exposing the posterior walls of each part of the butterfly vertebra, and the nerve roots above and below were identified. Both bodies of the butterfly vertebra were then removed with the use of osteotoms and pituitary rongeurs, including the discs above and below, and no retraction of the dural sac as it is prohibited in the thoracic region (Fig. 5c). A small mesh cage filled with bone graft was inserted anteriorly on the left side, between T10 and T12 vertebral edges, to act as a hinge to enable asymmetrical correction in the coronal plane, and maintain vertebral height to avoid spine over-shortening during correction (Fig. 5d). For the correction technique, at first, cantilevering of the spine with two prebended titanium alloy rod connected by a domino on the right side was performed, and then, further closure of the resection site by compression with the use of the domino was applied, the rod was than completely locked (Fig. 5e). The bone on bone contact on the right side was checked, the spinal cord was carefully controlled as kinking could occur in case of important reduction, and the cage showed a strong primary stability. Contralateral rod was placed and secured. The fixation was completed by the placement of one crosslink connector between the two rods. The prepared autologous bone grafts were placed to cover the maximum surface (Fig. 5f). Postoperative improvement of the gibbosity could be seen (Fig. 5g). Operative time was 320 min, and total blood loss was 800 mL.

The patient could walk on day 2, with a Thoracic Lumbar Orthosis, to be kept for 3 months, and with assistance from a physical therapist. He was discharged on Day 10. Postoperative X-ray and regular X-rays later on showed a stable correction of the thoracolumbar kyphosis with angulation of 10° between T10 and T12 with correction of the compensatory hyperlordosis to 48° (including an L4S1 angle of 32°) and hypokyphosis to 44 degrees. Sacral slope was 32° and global tilt 10°. In addition, scoliosis was corrected to 6°. Postoperative GAP score was 0, which corresponds to a sagittally proportioned spinopelvic state (Table 1). He could walk without a cane at 3 months and showed satisfactory clinical and radiological results at 2 years (Figs. 6a, b, 7a, b).

Fig. 6
figure 6

Two-year full spine X-rays showing stable correction of the thoracolumbar kyphosis with angulation of 10° between T10 and T12, correction of the compensatory hyperlordosis to 48° and hypokyphosis to 44°. In addition, coronal deformity was corrected to 6°

Full size image
Fig. 7
figure 7

Two-year postoperative CT scan confirming the complete fusion and a stable correction in both planes at the level of the resection (a and b). In addition, the axial image (c) shows how the lateral part of the cage lies in the fatty area between the vertebral body and the left diaphragmatic crus (postero-medially to the latter, black arrows), with no proximity to any vascular or neurological structures

Full size image

Discussion

Butterfly vertebra is considered a benign spinal anomaly and its discovery is usually coincidental in patients evaluated for routine back ailments. It is frequently asymptomatic, but it may be the cause of back pain due to alteration of spine biomechanics. Sciatica may also be part of its presentation when disc herniation from the sagittal cleft occurs [3].

This congenital condition can be confused with a pathological fracture in the lateral view due to wedging [13], it may also be confused with infection or metastases [1], but MRI would rule out these differential diagnoses, as it does not show any altered signal intensity or soft tissue enhancement, which usually occur in these pathologies. Computerized tomography scan, with the help of 3D reconstruction, typically shows the split vertebra with two lateral halves.

Butterfly vertebra may be associated with an anterior column defect of its two lateral halves, this anterior aplasia leads to kyphotic deformity, and depending on the asymmetry between the 2 halves, in terms of size, scoliotic deformity component may also come into play [4]. As described in the literature, congenital kyphosis (or kyphoscoliosis) is a much less common deformity than pure congenital scoliosis [14], and previous efforts to control the former by physiotherapy and bracing have been ineffective [15]. Therefore, surgical management is recommended, though challenging, consisting traditionally of a combined anterior and posterior approaches in one or two stage procedures [16], but during the past decade, the use of a single-stage posterior approach in combination with vertebral osteotomies became very popular, avoiding the morbidity of an additional anterior approach, with satisfying correction results [17].

Few reports in the literature investigated the surgical outcome of congenital thoracolumbar kyphosis, focusing mainly on the pediatric and adolescent population. In a series of 10 patients treated by a single posterior approach and using either pedicle subtraction osteotomy or vertebral column resection as a correction technique, Spiro [17] showed an improvement of the thoracolumbar kyphosis from 60° preoperatively to 17.5° postoperatively with no neurological complications, which is similar to the results of our case report. In another case series by Shi [18], 38 congenital thoracolumbar kyphosis patients were all surgically managed by SRS-Schwab grade 4 osteotomy, kyphosis improved from 49.5° to 6.8°, mean operative time was 242 min and mean blood loss was 634 mL.

In addition, only few case reports describing congenital bony spinal anomalies treated in adulthood could be retrieved. Polly [19] reported two adult patients (24 and 42-years old) with thoracic kyphoscoliosis due to fully segmented thoracic hemivertebra, they were surgically treated by posterior transpedicular lateral extracavitary excision and spinal instrumentation with satisfactory results and no neurological complications. Another case report by Ansari [20] described a 31-years old patient presenting with a waddling gait, legs numbness and urinary incontinence in relation to a dorsal midline hemivertebra causing kyphosis at the lumbosacral junction, he underwent resection of the hemivertebra with posterolateral instrumented fusion from L2 to the pelvis with no deformity correction but significant clinical improvement. In addition, Ruf [21] described a 42-years old patient presenting with head coronal malalignment who underwent C2 hemivertebra resection by combined anterior (transoral) and posterior approaches with good clinical and radiological outcome.

A case in which an isolated butterfly vertebra caused symptomatic kyphoscoliosis at the thoracolumbar junction with neurological impairment has, to our knowledge, never been reported. In fact, few cases reported the association of butterfly vertebra with spinal deformity and discussed its surgical management. Cui et al. [9] suggested, in a case of lumbosacral butterfly vertebra causing lumbar scoliosis and spondylolisthesis, a single-stage partial vertebra resection, interpedicular graft with a cage, and instrumentation via a posterior approach, to preserve the two distal concave nerve roots and correct the deformity. In addition, in a previously operated patient for L2 butterfly vertebra with secondary kyphosis, Zhan et al. [10] described the use of a pedicle subtraction osteotomy at the level of the butterfly vertebra combined to multiple levels Ponte osteotomies, to correct the sagittal deformity. The aforementioned cases are summarized in Table 2.

Table 2 Previous reports of butterfly vertebra associated to spinal deformity with surgical management
Full size table

The current report described a case of thoracolumbar kyphoscoliosis secondary to a butterfly vertebra of T11 causing progressive neurological impairment. Complete resection of the congenital anomaly, given its sharp angulation, with an asymmetrical coronal correction by the insertion of a mesh cage on the concave side, and a domino on the convex side was performed. It was decided to go 3 levels down, as the sagittal stable vertebra concept [22] was used to prevent distal junctional kyphosis; therefore, the most proximal lumbar vertebra body touched by the vertical line from the posterior—superior corner of the sacrum, which was L2, was included. In addition, because of the sharp and stiff deformity with vertebral resection, and the long lever arm required for correction, fixation was extended 3 levels up. Two sublaminar offset hooks were put at the distal level of the construct as biomechanical studies showed that a pedicle screw combined with an infra-laminar hook offers greater fixation strength (resistance to loosening tests) and a stiffer construct, when compared with fixation by pedicle screws alone [23]. In addition, supplemental offset sublaminar hooks significantly increase the construct stiffness without sacrificing an additional level distally, and adsorb some part of the construct strain, thereby reducing pedicle screw bending moments [24]. The main disadvantage of sublaminar hooks is the canal intrusion on both sides, nevertheless in our case, the canal was wide at the L2 level which enabled smooth insertion of the hooks with no neurological compromise. Another disadvantage is the possibility of hook migration into the canal, which generally occurs in case there is not a tight fit of the hook throat to the lamina; such complication can be avoided by selecting a proper hook size and by keeping a hook holder on the hook during manipulation such as rod insertion [25]. The titanium mesh cage was intended to be placed on the edges of the vertebral endplates of T10 and T12 to be as asymmetrical as possible for the coronal correction, nevertheless the inserted cage was 15 mm in diameter and the ideal width would have been 10 mm which was unfortunately not an available size at the time of the surgery, this is why it overstepped the edges laterally by 5 mm; nonetheless, axial CT image (Fig. 7c) showed that its lateral part lies in the fatty area between the vertebral body and the left diaphragmatic crus (postero-medially to the latter), with no proximity to any vascular or neurological structures.

In summary, complete resection of the butterfly vertebra in combination to asymmetrical coronal correction, was effective for the treatment of this rare pathology in both planes, with a favorable long-term outcome.