Case 1. An eleven-year-old girl presented with progressive neuromuscular scoliosis. The patient’s congenital hydrocephalus had been managed with a VP shunt when she was two years old. A tethered cord release had been performed when she was five years old. At the age of eleven years, she presented with limited walking ability, inability to hold a standing position, and diminishing pulmonary function. A right thoracolumbar prominence was grossly evident on examination (Figs. 1-A and 1-B). Progressive lower-extremity weakness correlated with the increasing sagittal plane spinal deformity and may have been related to increasing compression of the conus medullaris or spinal cord over the apex of the thoracolumbar kyphosis.
Radiographs demonstrated a long right thoracolumbar scoliosis extending from T4 to L4 with a Cobb angle of 40° (Fig. 2-A). There was reversal of both normal thoracic kyphosis and lumbar lordosis with Cobb angle measurements of 20° of lordosis from T2 to T12, 28° of kyphosis from L1 to S1, and 57° of kyphosis at the thoracolumbar junction (Fig. 2-B). Computed tomography (CT) revealed that the tubing for the VP shunt was in a normal position. Magnetic resonance imaging (MRI) did not provide evidence of recurrent tethering of the spinal cord.
Because this patient had progressing neuromuscular scoliosis with worsening restrictive lung disease and progressive loss of ambulatory function, surgical correction of the deformity was recommended. After clearance by cardiology, pulmonology, neurology, neurosurgery, and pediatric consultants, she underwent a posterior spinal fusion with instrumentation from T2 to the pelvis with wide facetectomies performed at the apex of the deformity. Supine intraoperative radiographs after wound closure demonstrated scoliosis correction, with the T4 to L4 Cobb angle measuring 25° (Fig. 3-A). Thoracic kyphosis was restored to 19°, and lumbar lordosis was restored to 43° (Fig. 3-B). Intraoperative electromyography recordings showed no prolonged irritation or abnormal discharges. Transcranial motor evoked potential signals were absent from the lower extremities at the beginning of the surgery. Upper and lower-extremity somatosensory evoked potential signals could not be obtained preoperatively or intraoperatively despite a trial of a total intravenous anesthesia technique, possibly as a result of prior surgery to treat a tethered spinal cord at the age of five years.
The patient remained hemodynamically stable throughout the case, with the exception of a brief decrease in blood pressure and increase in heart rate during lumbar spinal correction. This episode was transitory and was thought to be insufficiently profound to be related to subsequent neurologic injury.
An intraoperative prone wake-up test under anesthesia was not attempted because of preoperative sluggish and limited lower-extremity movement. Immediately postoperatively, the patient was unresponsive during a wake-up test, and the pupils were fixed and dilated. She was immediately taken to the pediatric intensive care unit where the VP shunt was tapped and 15 cc of cerebrospinal fluid was aspirated. A CT of the brain demonstrated diffuse swelling with a decrease in the size of the ventricles. A ventriculostomy was performed with an initial intracranial pressure of 13 mm Hg. A radiographic shunt series was normal and demonstrated redundant loops of shunt tubing in the abdomen. The electrocardiogram, echocardiogram, and troponin levels were normal, ruling out a coronary shunt problem or myocardial infarction. On postoperative day one, an MRI and magnetic resonance angiogram demonstrated brainstem herniation and no evidence of cerebral perfusion (Fig. 4). The patient failed two brain death examinations on postoperative day two and died the next day.
Case 2. An otherwise healthy twelve-year-old boy presented with progressive neuromuscular scoliosis associated with spina bifida at the L3 level. By the age of two years, he had had perinatal myelomeningocele closure and VP shunt placement for hydrocephalus, followed by two shunt revisions. On presentation to us, he walked with a walker and orthotics, and practiced bladder self-catheterization. A large right thoracolumbar scoliosis and thoracolumbar kyphosis were evident on physical examination.
Radiographs revealed a long thoracolumbar scoliosis measuring 54°, and a thoracolumbar kyphosis measuring 66°. A preoperative MRI revealed a shunted Chiari II malformation with small ventricles, tonsillar herniation through a widened foramen magnum, and a thoracic-to-lumbar spinal cord syrinx. There was no evidence of cord tethering based on neurologic or urologic symptoms, and the VP shunt was reported to be intact and functioning.
After clearance by cardiology, pulmonology, neurosurgery, urology, and neurology consultants, the patient underwent anterior discectomies and spinal fusion from T11 to L5 as well as posterior spinal instrumentation and fusion from T4 to the pelvis. Somatosensory evoked potential signals were not detectable prior to instrumentation, presumably because of preexisting spinal dysraphism and progressive neurologic deficit secondary to cerebellar herniation.
Intraoperatively, there was a brief episode of tachycardia without hypotension, which resolved after transfusion. Immediately postoperatively, the patient was responsive and moved all extremities on a wake-up test. He was transferred to the pediatric intensive care unit, where he was intubated and sedated with fentanyl. The pupils were symmetric at 3 mm and reactive. On postoperative day one, after cessation of the fentanyl drip, the patient was found to be unresponsive to verbal, tactile, and noxious stimuli. A CT scan on postoperative day two showed diffuse cerebral edema with no evidence of acute cerebral hemorrhage or infarct. A shunt series demonstrated an intact shunt, with the tubing terminating in the abdominal soft tissues but not entering into the peritoneum. A bedside shunt tap by a neurosurgeon revealed an intracranial pressure of 30 mm Hg. Distal shunt obstruction was discovered intraoperatively during an emergency shunt revision. On examination on postoperative day two, brainstem reflexes were absent.
An MRI on postoperative day five showed brainstem and cerebellar edema as well as bilateral occipital lobe infarcts, consistent with transforaminal herniation. Hemodynamic instability progressed to a fatal cardiac arrest on postoperative day ten.
Evidence of brain swelling on postoperative CT scans and signs of brainstem herniation on MRI suggested elevated intracranial pressure as a likely etiology for the catastrophic events in our two patients. A malfunctioning shunt is a probable cause for increased cerebral pressure. In both of our cases, the VP shunts were nearly a decade old and could have been scarred along their tracts, restricting tubing excursion with large spinal deformity corrections. An emergency postoperative shunt revision in Case 2 revealed a distal shunt malfunction. In Case 1, the shunt series was normal. However, a shunt series is not a test of function and does not rule out shunt failure.
To our knowledge, there is one report in the English-language literature documenting shunt failure following deformity correction in patients with myelomeningocele. Geiger et al.7 reported on seventy-seven patients with myelomeningocele who underwent scoliosis surgery. There were five instances of postoperative shunt insufficiency in the acute postoperative period, all in patients younger than fourteen years old. There was one death secondary to acute hydrocephalus that occurred eight hours after surgery, but no additional details were reported.
A second possible contributor to brainstem herniation in our two cases was undetected spinal cord tethering, which could have pulled the brainstem into the foramen magnum with deformity correction. Some of the symptoms noted in Case 1 prior to the spinal surgery could have been caused by spinal cord retethering8. Symptomatic tethering generally requires release of the tether and, in cases where the scoliotic deformity is less than 40°, the curve may improve5,9. In larger curves, detethering in symptomatic patients is required prior to spinal deformity surgery because of concerns of traction injury to the spinal cord with deformity correction6,8,10,11.
In the absence of a symptomatic tethered cord, as in Case 2, it is thought that scoliosis correction can be performed without detethering. Samdani et al.12 reported on seventeen patients with myelomeningocele and no clinical evidence of a symptomatic tethered cord. They all underwent deformity correction with no postoperative neurologic dysfunction or shunt malfunction12. Although this series is reassuring, in our cases, it is possible that undetected cord tethering contributed to the catastrophic outcomes.
Finally, a large rigid sagittal deformity, such as those seen in our cases, can cause stretching of the spinal cord when correction is performed. Moreover, if the distal cord is tethered, the correction can result in traction on the brainstem. Powerful corrective forces on the spine can occur with spinal instrumentation with use of pedicle screw constructs and can result in progressive brainstem herniation.
Cases such as those presented here are rare but devastating. We suggest that patients with neuromuscular scoliosis and VP shunts are at risk for intraoperative cerebral edema. We recommend preoperative functional tests to evaluate the performance of VP shunts. An intraoperative electroencephalogram and intracranial pressure monitoring can facilitate the detection of early changes, which would allow surgeons to rapidly abort surgery and reposition the patient. Furthermore, it can be difficult to determine cord tethering or retethering on imaging studies alone. Patients may benefit from exploration and release of a potentially tethered cord prior to proceeding with deformity correction, but this is currently unclear. Finally, these patients must be carefully monitored during emergence from anesthesia and postoperatively for alterations in intracranial pressure and shunt function.