Is It Possible to Correct Congenital Spinal Deformity Associated With a Tethered Cord Without Prophylactic Intradural Detethering?

Abstract

Background

Patients who have a congenital spinal deformity with a tethered cord generally are treated with prophylactic intradural detethering before deformity correction. However, the detethering procedure carries substantial risk, and it is not clear whether deformity correction can be performed without detethering.

Questions/purposes

To determine the (1) correction rate, (2) proportion of patients who experienced complications after surgery, and (3) neurological status after recovery from surgery in a group of patients with congenital spinal deformity and a tethered cord who were treated either with posterior spinal fusion only (PSF), pedicle-subtraction osteotomy (PSO), or a vertebral column resection (VCR), based on an algorithmic approach.

Methods

Between 2006 and 2016, we treated 50 patients surgically for a congenital spinal deformity and a tethered cord. We defined a congenital spinal deformity as one that was caused by failure of vertebral segmentation, failure of vertebral formation, or both, and we made the diagnosis of a tethered cord based on a conus medullaris lower than L2 level, or a diameter of the filum terminale greater than 2 mm, as shown on magnetic resonance image. Of those, nine patients were lost to followup before the 2-year minimum, leaving 41 for analysis at a mean followup of 47 months (range, 24 to 92 months) in this single-institution retrospective study. The treatment algorithm involved one of three approaches: PSF, PSO, or VCR. A total of 15 patients underwent PSF; we used this approach for patients with moderate curves (Cobb angle < 80°) and intact neurological status both previously and during a bending and traction test. Eleven patients underwent PSO; we performed PSO when patients had neurological symptoms (in daily life or during the traction/bending test) and a magnitude of the curve less than 80°. Finally, 15 patients underwent VCR, which we used in patients with a magnitude of the curve more than 80° and/or flexibility less than 20%, with/without neurological symptoms. No patient in any group underwent intradural detethering. We report on the correction rate, defined as the ratio between the corrected magnitude and preoperative magnitude of a curve at a given postoperative time point (correction rate = 1- (Cobb angle at a given time point/preoperative Cobb angle) x 100%); complications, that is, postoperative/recurrent neurological symptoms, cerebrospinal fluid leakage, infection, blood loss > 5000 mL, as determined by chart review performed by an individual not directly involved in patient care; and a detailed neurological exam, including evaluations of sensory function, extremity muscle strength, pain, gait, physiological reflexes, and pathological signs, both before surgery and at most recent followup, as performed by the surgeon. All neurologically symptomatic patients were evaluated with a neurologic scoring system.

Results

The overall mean ± SD correction rate in this series was 63% ± 14%. It was 70% ± 12% in the PSF group, 64% ± 17% in the PSO group, and 56% ± 12% in the VCR group. Seven patients in those three groups experienced major complications, including blood loss more than 5000 mL, temporary neurological symptoms, cerebrospinal fluid leakage, and infection. The most severe complications included one patient in the VCR group who had temporarily decreased strength in the lower limb, and one patient in the PSO group with temporary numbness in the lower limb. Finally, no patients in PSF group had postoperative neurological complications, and all patients with neurological symptoms in the PSO/VCR group improved to varying degrees. For neurologically symptomatic patients in PSO group (n = 6), the neurological score improved slightly, from 22.5 ± 1.9 preoperatively to 24.2 ± 0.8 at the most recent followup (p = 0.024) with a mean difference of 1.7. For neurologically symptomatic patients in VCR group (n = 10), the neurological score improved slightly from 23.1 ± 1.1 preoperatively to 24.2 ± 0.6 at most recent followup (p = 0.009) with a mean difference of 1.1.

Conclusions

Congenital spinal deformity with a tethered cord may be treated without prophylactic intradural detethering. In the current series treated according to this treatment algorithm, good correction and neurological improvement were achieved, and few complications occurred. However, such a small series cannot prove the safety of this treatment; for that, larger, multicenter studies are necessary.

Level of Evidence

Level IV, therapeutic study.

Introduction

Congenital scoliosis is a complex spinal deformity associated with other intraspinal anomalies in 25% to 37% patients [1, 3, 7, 14, 16], and a tethered cord is especially common, occurring in up to 12% to 17% of patients with congenital scoliosis [1, 14, 16]. A tethered cord presents with excessive tension of the spinal cord [19], and it may be a cause of scoliosis [11]. Under these circumstances, treatment may be challenging because the scoliosis correction might extend the length of the spinal canal, further increasing the tension of the cord, leading to a poor prognosis.

In the past, congenital spinal deformity with a tethered cord has been treated with prophylactic intradural detethering either before deformity correction in a concurrent procedure or as the first stage of a two-stage procedure [11, 13, 15]. A two-stage approach has the disadvantages of a second anesthesia, an increased time to full recovery, and a higher risk of complications [13, 15]. In addition, the detethering procedure itself carries substantial surgical risk. In 2009, Matsumoto et al. [9] reported serious neurological complications in a prophylactic intradural detethering procedure; multiple neurologic deficits occurred in a patient who had been neurologically asymptomatic, even though the detethering procedure was conducted under a surgical microscope and neurologic electrophysiological monitoring. According to previous reports, 26% to 29% of patients experienced a surgical complication from this intradural detethering procedure [12, 13], and 10% to 37% of them experienced retethering after the detethering procedure [10, 12]. Furthermore, conventional approaches also achieve only limited scoliosis correction, resulting in 23% to 27% reduction of preoperative curve magnitude [2, 12].

Therefore, there is a potential benefit to exploring alternative strategies to try to minimize these complications. In 1995, Kokubun [7] first presented the feasibility of treating tethered cord syndrome by using spine-shortening osteotomy without cord detethering. In addition, Huang et al. [4] reported satisfying outcomes for a severe and rigid congenital scoliosis and tethered cord population treated with spine-shortening osteotomy without cord detethering. Nevertheless, spine-shortening osteotomy was suitable for severe and rigid curves, but for a mild or moderate curve with good flexibility, its complication risk may overweigh its benefit. As these studies suggested, in the treatment of congenital scoliosis associated with a tethered cord, an alternative strategy without prophylactic intradural detethering may be possible, while a systematic algorithm was needed in addition to the spine-shortening osteotomy.

We therefore sought to determine the (1) correction rate, (2) proportion of patients who experienced complications after surgery, and (3) neurological status after recovery from surgery in a group of patients with congenital spinal deformity and a tethered cord who were treated either with posterior spinal fusion only (PSF), pedicle-subtraction osteotomy (PSO), or a vertebral column resection (VCR) without intradural detethering, based on an algorithmic approach.

Patients and Methods

This is a retrospective case series of patients with congenital scoliosis and a tethered cord treated at a single institution, with approval from the institutional review board. Between 2006 and 2016, we treated 50 patients surgically for congenital spinal deformity and tethered cord.

Three-dimensional CTs and MRIs of the whole spine, which were routine for patients with scoliosis in our center, were evaluated to identify both vertebral and intraspinal anomalies. All these medical images were evaluated by the medical imaging department in our hospital and then double-checked by two experienced spine surgeons (HT, CD) before surgery. We defined a congenital spinal deformity as one caused by failure of vertebral segmentation, failure of vertebral formation, or both (as seen on computed tomography). We made the diagnosis of tethered cord based on a conus medullaris lower than L2 level, or a diameter of the filum terminale greater than 2 mm as shown on MRI.

Of the 50 potentially eligible patients, nine were lost to followup before the 2-year minimum, leaving 41 for analysis at a mean followup of 47 months (range, 24 to 92 months) in this single-institution retrospective study. Informed consents were obtained from all individual participants.

Treatment Algorithm and Surgical Procedures

All patients received detailed neurological exams and all asymptomatic patients received a bending test and a traction test simulating the post-correction condition to evaluate their tolerance for correction. The bending test was conducted during a detailed neurological exam with the patient in a bending position. The traction test was performed during a detailed neurological exam throughout a halo-gravity traction course. The treatment algorithm involved one of three approaches: PSF, PSO, or VCR. A total of 15 patients underwent PSF; we used this approach for patients with moderate curves (Cobb angle < 80°) and intact neurological status both before and during a bending and traction test. Eleven patients underwent PSO; we performed PSO when patients had neurological symptoms (in daily life or during the traction/bending test) and magnitude of the curve less than 80°. Finally, 15 patients underwent VCR, which we used in patients with magnitude of the curve more than 80° and/or flexibility less than 20%, with/without neurological symptoms.

No patient in any group underwent intradural detethering. All operations were performed under general anaesthesia on an open frame Jackson table. After a curved incision to expose the spine, the surgeon placed the pedicle screws using a standard free-hand technique. VCRs and PSOs were performed at the apical vertebrae to shorten the spine. In brief, the surgeon first performed laminectomy and then completely excised the facets. One senior surgeon (HT) performed all surgical procedures. The surgeon removed the pedicles at the apical level in their entirety. Then an osteotomy was performed in an eggshell fashion followed by removal of the lateral vertebral body wall. The surgeon placed temporary rods onto the pedicle screws before initiating the lateral wall resection. After bone chips were grafted into the osteotomy defect, the surgeon achieved bone-to-bone contact at the posterior vertebral wall by compression in all PSOs and VCRs. The spine-shortening length was defined as the height of the osteotomy gap at the posterior vertebral wall on the convex side, which was measured by a compass and ruler during surgery. The temporary rods were finally changed to final rods that were precontoured to the normal sagittal curve. None of these osteotomy procedures were performed for patients in the PSF group. Then, an in situ bender was used to correct scoliosis. After the spine shortening and correction procedures, the surgeon completed spinal fusion with a bony graft. Finally, the surgeon closed the wound in a standard fashion. All the operations were completed under both somatosensory-evoked potential monitoring and motor-evoked potential monitoring.

Evaluation of Scoliosis Correction

All patients had standing posteroanterior and lateral radiographs of the whole spine taken preoperatively, postoperatively, and at most recent followup and preoperative prone lateral bending radiographs. One observer (KY) measured the radiographs, independent from the treatment. Coronal curves and sagittal curves were measured by the Cobb method in standing posteroanterior, lateral, and bending radiographs to evaluate the severity of scoliosis/kyphosis, spine flexibility, and correction rate. We defined spine flexibility as the reduced percent of the curve in a prone bending position. We reported on the correction rate to evaluate the outcome of scoliosis correction, defined as the ratio between the corrected magnitude and preoperative magnitude of a curve at a given postoperative time point. Flexibility and correction rates at each time point were defined as follows:

Complications

Complications were defined as postoperative/recurrent neurological symptoms, cerebrospinal fluid leakage, infection, and blood loss greater than 5000 mL. Complication data was determined by chart review performed by an individual (CF) not directly involved in patient care, and detailed neurological exams (including evaluations of sensory function, extremity muscle strength, pain, gait, physiological reflexes, and pathological signs), both before and at most recent followup, as performed by the surgeon (HT).

Recovery of Neurological Status

The surgeon performed detailed neurological exams, both before treatment and at most recent followup. All neurologically symptomatic patients were evaluated with a neurologic scoring system. The neurologic scoring system graded the neurological status according to pain intensity, sensory disturbance or dysesthesias, motor weakness, gait ataxia and sphincter function, in which each item could be given 0 (worst) to 5 (normal) scores and a result of 25 scores represents a normal neurologic status [6]. We compared the neurological status at most recent followup with the preoperative neurological status for each patient.

The current series consisted of 30 female patients and 11 male patients with a mean age of 14 years (range, 7-24 years) and an average followup of 47 months (range, 27-92 months) (Table 1). There were 20 patients who had kyphosis: four in the PSF group, five in the PSO group, and 11 in the VCR group. Eleven patients had congenital scoliosis caused by failure of formation solely, 15 patients had congenital scoliosis caused by failure of segmentation solely and the remaining 15 patients had mixed vertebral deformity. Most of their conus medullaris were located at the L3-L5 levels, and four patients had conus medullaris in the sacral region (Table 2).

Table 1.

Preoperative and intraoperative characteristics

Table 2.

Vertebral deformities, locations of conus medullaris, and associated intraspinal anomalies

Statistical Analysis

Statistical analysis was conducted using SPSS software for Windows (Version 23.0; IBM Corp, Armonk, NY, USA). Neurologic parameters before the operation and at most recent followup were compared by Wilcoxon signed rank-test. Significance level was set at p < 0.05. All parameters are given as mean ± SD except patient numbers and age range. Ninety-five percent confidence intervals (CIs) were noted to assess precision.

Results

The overall mean ± SD correction rate at most recent followup in this series was 63% ± 14%. It was 70% ± 12% in the PSF group (Fig. 1), 64% ± 17% in the PSO group, and 56% ± 12% in the VCR group (Table 3).

Fig. 1 A-H.

A 19-year-old woman with congenital scoliosis and a tethered cord received posterior spinal fusion without detethering. No neurologic deficits were found before or after surgery. (A, B) Preoperative 3-D CT showing failure of segmentation and posterior deformity. (C, D) Preoperative radiographs showing moderate scoliosis (47°) with no kyphosis. (E) Preoperative MRI of the whole spine showing the conus medullaris at the L4 level. (F) Good flexibility (32° on bending radiograph, flexibility: 32%) on a bending radiograph. (G, H) Posteroanterior and lateral radiographs at a 3-year followup visit.

Table 3.

Correction in coronal and sagittal planes

Seven patients experienced major complications. In the VCR group, two patients had blood loss greater than 5000 mL; one patient had decreased strength in the lower limb, which had improved by the most recent followup. In addition, one patient had a cerebrospinal fluid leakage that resolved in 5 days, and one patient had a urinary tract infection, which resolved in 1 week. In the PSO group, one patient developed numbness in the lower limb (which resolved by the 6-month followup), and another patient had a urinary tract infection, which resolved quickly. One patient in the PSF group developed a superficial wound infection that resolved in the first week.

Finally, no patients in the PSF group had postoperative neurological complications, and all patients with neurological symptoms in the PSO/VCR group had improved to varying degrees (Table 4). Six patients had neurological symptoms before surgery in the PSO group, and their neurological status (as measured by neurologic scoring system) improved slightly from before surgery (22.5 ± 1.9) to the most recent followup (24.2 ± 0.8) (p = 0.024, mean difference: 1.7 ± 1.2; 95% CI, 0.4-2.9). Ten patients had neurological symptoms preoperatively in the VCR group and their neurological status improved slightly from before surgery (23.1 ± 1.1) to the most recent followup (24.2 ± 0.6; p = 0.009; mean difference: 1.1 ± 0.7; 95% CI, 0.6-1.6) (Fig. 2). The PSF group had no neurological symptoms preoperatively, and we found no intraoperative monitoring changes or postoperative neurological changes.

Table 4.

Neurologic improvements after surgery

Fig. 2 A-H.

A 17-year-old girl with congenital scoliosis and a tethered cord received vertebral column resection without detethering. Claudication and weakness in the left lower limb were identified preoperatively. The gait turned normal at 24-month followup. (A, B) Preoperative radiographs showing severe scoliosis (114°) and kyphosis. (C, D) Posteroanterior and lateral radiographs at 35-month followup. (E) Preoperative MRI of the whole spine showing the conus medullaris at the L4 level. (F) Preoperative 3-D CT showing failure of segmentation in T7-8. (G) Intraoperative images before spine shortening. (H) Intraoperative images after spine shortening.

Discussion

The standard treatment for tethered cord syndrome has long been intradural detethering, and prophylactic detethering has generally been performed before scoliosis correction for patients with both scoliosis and a tethered cord [18]. Nevertheless, the intradural detethering procedure carries substantial risk of neurologic complications and recurrence, and the correction rate of concomitant scoliosis has been limited when such a strategy has been employed [9, 10, 12, 13]. Therefore, there is a potential benefit to establishing the efficacy of an alternative strategy to minimize these complications. We sought to evaluate the correction rate, complications, and neurological findings using an algorithmically driven approach to the care of the patient with congenital scoliosis and tethered cord, in which patients with milder scoliosis and intact neurological status both before and during a bending and traction test were treated with PSF only, those with neurological symptoms (in daily life or during the traction/bending test) and magnitude of the curve less than 80° were treated with PSO, and those with the severe scoliosis (magnitude of the curve over 80° and/or flexibility < 20%, with/without neurological symptoms) were treated with VCR. We found that patients in our current series treated according to this treatment algorithm achieved good correction and neurological improvement, and few complications occurred.

This study had several limitations. First, nine of the 50 patients (18%) were lost to followup before 2 years. This included four patients in the PSF group, two patients in the PSO group, and three patients in the VCR group. In general, patients who are lost to followup may not be doing as well as those who are accounted for, and we caution readers to consider this when interpreting our findings. In addition, this study—though large compared with others [11, 13, 15]—includes only 41 patients, and so uncommon complications may not have been observed in this group; for that reason, we cannot make broad claims about safety. Future studies are needed to validate our results, and to determine whether other thresholds should be used to decide when to apply each of the three procedures we employed. In addition, we did not assess the natural history of the tethered cord or its impact on patients’ neurologic function at skeletal maturity, especially for the asymptomatic patients treated with PSF only. For patients who have reached skeletal maturity, the surgical outcome after 2-year followup would be unlikely to change since the spinal cord tension would hardly increase with a stable length of spinal canal. However, spinal cord tension may change in younger patients as the spine grows, and so longer followup will be important for these patients. Because there were few children who had both early-onset scoliosis and a tethered cord, the treatment strategy for these growing spines requires further study. A larger series, which may be achieved by multicenter cooperation, and longer followup until skeletal maturity is needed to address this limitation.

Our treatment algorithm resulted in good scoliosis correction in all three groups (Table 3). The correction rate was apparently higher than previous studies using conventional treatment (63% in this series versus 23%-27% in conventional treatment) [2, 12]. In our opinion, there were two reasons for this. First, for patients in the VCR and PSO groups, the use of spine-shortening osteotomies to relieve spinal cord tension allows us to correct the deformity without worrying about the possible neurologic deficits caused by stretching the tensioned spinal cord. Second, for patients with relatively moderate and flexible curves, posterior spinal fusion only achieved substantial curve correction after the tolerance of cord correction was tested by bending and traction tests [17]. Based on these mechanisms, scoliosis in such a population can be substantially corrected and the tethered cord can be addressed by spine-shortening for tension relief and preoperative tests for tolerance of correction. The actual tension change of spinal cord remained unclear in current series. However, it is hard to analyze since monitoring of spinal cord tension potentially harms the cord and causes surgical complications. Future studies using an applicable animal model of scoliosis may be possible to address this question.

As with any major spinal procedure, complications occurred; however, we believe our approach compares favorably to the published alternatives. The complications in this series did not exceed those typically observed with more conventional treatments (17% in the current series versus 26%-29% in other studies of more conventional treatments) [12, 13], while the proportion of neurological complications in this series was low (5%). In our treatment algorithm, intradural procedures were avoided, perhaps contributing to the relatively low frequency of neurological complications. Due to the neurological complications during the detethering procedure and the retethering phenomenon after intradural detethering, this treatment algorithm may be beneficial in reducing surgical complications in such a patient population.

We observed neurological improvement after surgery in the PSO and the VCR groups. Possible explanations for the improvement may be the benefit of the spine-shortening osteotomy. After the spine-shortening osteotomy, namely PSO and VCR, the spinal canal was shortened, resulting in theoretical tension relief in the spinal cord. Others have found that the pathophysiology of a tethered cord is associated with impaired oxidative metabolism in the affected spinal cord [20]. Correspondingly, spinal cord blood flow was found to increase after spine-shortening osteotomy in animal experiments [5]. These studies suggested that spine-shortening osteotomy may contribute to the improvement of spinal cord blood flow in patients, and consequently, may benefit the neurological function.

In summary, our treatment approach, which selectively applied PSF alone (for moderate curves and intact neurological status), PSO (for neurological symptoms and moderate curves), and VCR (for severe curves), achieved good scoliosis correction, and resulted in few complications in a small retrospective series. We hope that future, larger series with longer followup on patients treated according to this algorithm will either support our findings or identify important shortcomings that can improve care. Based on our findings, we believe this treatment algorithm can be considered as a viable alternative strategy for patients with congenital scoliosis and a tethered cord.