CURVE PROGRESSION IN SCOLIOSIS ASSOCIATED WITH ARNOLD CHIARI MALFORMATION AND SYRINGOMYELIA FOLLOWING SUBOCCIPITAL DECOMPRESSION

Participants: F.A. Farley, A. Puryear, J.M. Hall, K. Murasko

Keywords: scoliosis, Chiari malformation, syringomyelia, suboccipital decompression

Introduction

There is an established association between Arnold Chiari I malformations, syringomyelia, and scoliosis. In children, scoliosis may be the only presenting sign of an underlying Arnold Chiari I malformation and syrinx, which underscores the importance of investigating subtle neurological findings and obtaining MRI studies on these patients. In the past, syringosubarachnoid shunts were used to reduce the size of the syrinx. More recently, suboccipital decompression has been used to address the Arnold Chiari I malformation. Recent literature suggests suboccipital decompression may improve or reduce the scoliosis.

We undertook a retrospective study to test the hypothesis that suboccipital decompression would stabilize the scoliosis in children with Arnold Chiari I malformation and syringomyelia, examined at short and mid term follow-up.

Materials and Methods

Children with Arnold Chiari I malformations, syringomyelia, and scoliosis who underwent suboccipital decompression were identified. Patients with less than a year follow-up or those who underwent spinal fusion within a year of suboccipital decompression were excluded. The suboccipital decompression procedure included removal of the inferior, posterior occiput, laminectomy of C1 and dural patch graft. All suboccipital decompressions except one were performed by a single neurosurgeon (KM). In three cases a syringosubarachnoid shunt was done either at the time of suboccipital decompression or subsequently.

Demographic data, MRI results, neurosurgical procedure and orthopaedic brace treatment were abstracted from the medical records. Baseline radiographs (taken at the time of suboccipital decompression) and annual follow-up radiographs (through final follow-up or spinal fusion) were examined to determine curve size using the Cobb method. The Risser sign was determined based on the radiograph taken closest to the time of suboccipital decompression and the PA film each subsequent year.

One reader of radiographs determined the end vertebrae and then two readers measured the curves on all radiographs. To provide a measure of intra-observer variability, each reader read a subset of radiographs a second time. Curves measuring from 10 to 25 degrees were designated mild; from 26 to 45 degrees, moderate; and from 46 to 90 degrees, severe. Variability of the curve measurements within and between the two readers was calculated using procedures described in Bland and Loder.¹ Using the first set of curve measurements from each patient, 95 percent confidence limits for intra and inter-observer measurement agreement were estimated for the sample. The largest of the calculated confidence intervals (± 6 degrees) was then designated as the cut-off for meaningful curve change. Curves showing less than 6 degrees of change between follow-up intervals were considered stable; curves that increased by 6 or more degrees were progressive, and curves that decreased by 6 or more degrees were improved.

Results

Nine children (7 female, 2 male; mean age 8 years) with scoliosis and Arnold Chiari I malformations were followed after suboccipital decompression. Eight of the patients also had syringomyelia. The mean thoracic curve at presentation was 46.3 +/- 18.0 degrees and the mean lumbar curve was 35.3+/-16.8 degrees. Seven children were Risser 0 at the time of suboccipital decompression. One was Risser 1 and one was Risser 3. Seven had right thoracic curves and two had left thoracic curves. At presentation, five of the nine children were candidates for spinal fusion because of the magnitude of their curves.

All curves were measured from the same levels by two observers. For thoracic curves, (N=36), the mean difference between the two observers was 0.4 ± 2.5 degrees. The 95% confidence interval for the inter-observer variability of the thoracic curve measure was —5.3 to + 4.5 degrees. For lumbar curves (N=31), the mean difference between the two observers was 1.0 degree. The 95% confidence interval for the inter-observer variability of the lumbar curve was —6.3 to + 4.5 degrees.

At the time of suboccipital decompression, 2 curves were mild, 2 moderate, and 5 severe. Seven curves were measured at one year post decompression; 3 had improved, 2 were stable and 2 had progressed. Five curves were measured at two years; 4 were stable and 1 had progressed. Of 5 curves measured at three years, 3 were stable and 2 had progressed. Of three curves measured at four years, 2 were stable and 1 had progressed. Of two curves measured at five years, both had progressed.

Three children with longer follow-ups (4-7 years) had stable or improved curves for several years following suboccipital decompression, but curve progression began to recur between ages ten and thirteen. All three had been braced by age 9, two with the Boston type TLSO and one with a Charleston Bending Brace. Despite bracing, all three curves progressed at least 21 degrees and the children were spinal fusion candidates at final follow-up. MRI studies at the time of curve progression showed reduced syrinx size in one, and stable syrinx size in the other two.

At final follow-up the average age in the full sample (N=9) was 12.3 years (range 8 to 15 years). Five had undergone a posterior spinal fusion, three were candidates for spinal fusion (curve range 60 to 65 degrees) and one had a 20 degree curve at age 11 years. The mean thoracic curve at final follow-up was 61.0+/-23.2 degrees and the mean lumbar curve was 45.1+/-21.1 degrees. The Risser sign at final follow-up was four in three patients; Risser two in one child, Risser one in one, and Risser zero in four. Of the four Risser zero children, one had a posterior spinal fusion and two were spinal fusion candidates.

Discussion

The early results of the effect of suboccipital decompression on scoliosis curve progression show promise. At one-year post suboccipital decompression, curve stabilization is the norm. Suboccipital decompression may initially cause this curve stabilization, which may last as long as seven years. By three to seven years following suboccipital decompression, curve progression is the norm. The relentless progression may be due to the physiology of the syrinx or the adolescent growth spurt. Bracing of three patients in our series had no effect on rapid curve progression during early adolescence. Further study is needed to determine the role of bracing in these patients.

References

[1] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1(8476):307-10.