Correct initial diagnosis of patients with scoliosis is important for decisions on treatment options. Misdiagnosis of scoliosis with underlying syringomyelia may result in complications following operative therapy or even brace therapy 123. However, recent studies have reported that even patients with both scoliosis and syringomyelia have a small risk of neurological complications if they are neurologically normal or have a small syrinx 4567. Therefore, whether all patients with scoliosis should be examined by magnetic resonance imaging (MRI) remains controversial from a cost-effectiveness perspective 46789. The prevalence of underlying syringomyelia in cases of "idiopathic" scoliosis also differs between previous reports (range, 2-28%) 4101112. Several indicators of syringomyelia have been reported in scoliosis patients: left thoracic curve 313141516, juvenile-onset scoliosis 111517, rapid curve progression 318, pain 14, thoracic hyperkyphosis 192021, male sex 5, and abnormal neurologic signs 4591322. Of these indicators, neurologic signs, particularly an abnormal superficial abdominal reflex (SAR), is thought to be one of the most important clinical signs of syringomyelia49. In a study of 72 patients with scoliosis and atypical features, Morcuende reported that all patients with abnormal SAR had Arnold-Chiari malformations 9. In a study of 327 patients with adolescent idiopathic scoliosis, Do emphasized the importance of abnormal SAR, even though the sensitivity and specificity of this sign was questionable 4. Thus, although abnormal SAR has been regarded as an important neurological sign in the initial diagnosis of scoliosis, the prevalence of abnormal SAR and the diagnostic utility of abnormal SAR in predicting scoliosis with syringomyelia or other types of scoliosis has not been well known. We therefore examined the prevalence and diagnostic utility of abnormal SAR in all patients with scoliosis treated by a single spinal surgeon. In corrective surgery for scoliosis with syringomyelia, there is a potential risk of paraplegia: Overcorrection, tenuous blood supply, and changes in cerebrospinal fluid pressures are thought to be the causes of neurological complications 1323. In our experience, the curves of scoliosis with syringomyelia are likely to be more flexible than those of idiopathic scoliosis. However, few studies have reported differences in scoliosis curve flexibility relating to the presence or absence of syringomyelia. The purpose of this study was to determine the utility of abnormal SAR and curve flexibility in the initial diagnosis of scoliosis, particularly for scoliosis with syringomyelia.

A potential risk of neurologic complications has been reported when corrective surgery is performed for patients with scoliosis and syringomyelia 123. Several studies have found that neurosurgery before corrective surgery reduces the risk of neurological complications and improves or stabilizes the curve, particularly in patients younger than 10 years old 242526272829. Previous reports have agreed with the consensus that MRI should be performed for infantile and juvenile scoliosis 101112. However, the necessity of MRI for "adolescent idiopathic" scoliosis remains controversial 11303132. Some studies have suggested that MRI is not indicated in "adolescent idiopathic" scoliosis if the patients are neurologically normal 467. Inoue reported that only one of 12 neurologically asymptomatic patients with syringomyelia and scoliosis encountered neurological complications 5. However, once paraplegia occurred, the prognosis was very poor. We believe corrective surgery for scoliosis should be performed with utmost care so as to minimize the risk of neurological complications. Of the neurological abnormalities thought to indicate syringomyelia, SAR is considered one of the most important 49. However, in early studies, it was unclear how often patients with syringomyelia had normal SAR and how often patients with normal SAR had syringomyelia 1322. In Saifuddin's prospective study of 73 patients with "presumed idiopathic scoliosis", the prevalence of abnormal SAR was 11%. Saifuddin concluded that unilateral absence of SAR was a significant sign suggestive of syringomyelia and that bilateral absence of SAR might also be associated with neuraxis anomalies 32. Morduende reported that 11 patients (15%) of 72 patients with atypical features had abnormality on MRI. In their report, 5 of the 11 patients with abnormal SAR had MRI abnormality and therefore, abnormal SAR had a sensitivity of 45% and PPV of 45%. They concluded that the importance of SAR could not be overemphasized 8931. However, as a limitation of these studies, they might have differed in their definition of "idiopathic" or "abnormal SAR". If patients without any indicators of syringomyelia were regarded as genuine "idiopathic" cases, this might increase the PPV of SAR. Some studies have reported that absence or hyporeflexia can occur in normal individuals 33. In the present study, "idiopathic" scoliosis included 46 patients with idiopathic scoliosis and 9 patients with scoliosis and syringomyelia. Data from previous studies regarding asymmetric SAR as abnormal in "idiopathic" scoliosis and those from the present study are listed in Table 9. The sensitivity of abnormal SAR for syringomyelia in the present study was higher than that of previous studies, partly because we included bilateral absence and hyporeflexia of SAR as abnormal signs. When considering SAR as a screening test, sensitivity is more important than PPV, because final diagnosis can be made by MRI. Therefore, it could be recommended that subtle findings such as bilateral absence or hyporeflexia of SAR should be regarded as abnormal to give a useful screening test. In the present study, we found that abnormal SAR (defined as unilateral or bilateral absence of SAR or hyporeflexia as found by a single experienced examiner) yielded 89% sensitivity for syringomyelia. Furthermore, if SAR of a patient with scoliosis is abnormal, the patient has a 90% of probability of being non-idiopathic scoliosis (Table3). If patients with scoliosis had abnormal SAR, we could recommend physicians to check MRI of the whole spine in order to differentiate syringomyelia as well as to check myogenic enzyme in order to differentiate myopathy.

Some studies have reported the relationship between syrinx type and neurological findings or radiographic characteristics 53435. Inoue reported that all patients with swelling-type syrinx had some abnormal findings and our results supported these findings. In our study, male sex was a risk factor for syringomyelia with a relative risk of 6.5 (CI 2.4-18). However, there was no significant difference in left-thoracic curve pattern between patients with syringomyelia and those with idiopathic scoliosis (P = 0.3). When reporting a good prognostic model for abnormal MRI, Morcuende et al. reported that left-thoracic curve pattern by itself may not be suggestive of MRI abnormality 9. They concluded that left-thoracic curve pattern was associated with abnormal MRI when combined with neurologic abnormality or a severe curve despite immaturity. Our findings corresponded with theirs.

To our knowledge, there have been no previous reports describing the relationship between curve flexibility and syringomyelia. Some authors have alerted clinicians to the risk of paraplegia due to overcorrection or vulnerability of the spinal cord in corrective surgery for scoliosis with syringomyelia 1323. Generally, increasing age and larger Cobb angle make the curve less flexible 36. However, in the current study, curve flexibility of patients with syringomyelia was significantly greater than that of those with idiopathic scoliosis, although mean age and Cobb angle of patients with syringomyelia were larger than those of patients with idiopathic scoliosis (P = 0.02). This suggests that patients with scoliosis and syringomyelia have a more flexible curve than those with idiopathic scoliosis. The logistic regression model was close to being significant (P = 0.08). Based on this logistic regression analysis, curve flexibility ≥50% could be another risk of syringomyelia. Although we cannot explain the pathophysiology of hyperflexibility in syringomyelia, any abnormality in the pyramidal tract would make the curve more flexible in scoliosis with syringomyelia than in idiopathic scoliosis.

The sensitivity of abnormal SAR for non-idiopathic scoliosis was 38% with 96% specificity, 90% PPV, and 60% NPV. All cases of myopathic scoliosis had bilateral absence of SAR. There were significant differences between patients with syringomyelia and those with idiopathic scoliosis in gender, abnormal neurological findings (SAR), and curve flexibility. The sensitivity of abnormal SAR for syringomyelia in patients with "idiopathic " scoliosis was 89%, with 95% specificity, 80% PPV, and 98% NPV. Curve flexibility ≥50% could be another risk of syringomyelia.

The authors declare that they have no competing interests.

All authors read and approved the manuscript, which has not been submitted or published anywhere else. TF and MI designed the study. HS, KO, and YN revised it critically. All authors read and approved the final manuscript.