Pediatr Radiol (1999) 29: 360363 Springer-Verlag 1999

Yu-Leung Chan Jack C. Y. Cheng Xia Guo Ann D. King James F. Griffith Constantine Metreweli

MRI evaluation of multifidus muscles in adolescent idiopathic scoliosis

Received: 9 June 1998 Accepted: 2 November 1998 Presented at 35 th Annual Congress of ESPR, Rhodes, Greece, May 1998

Y. L. Chan ( ) A. D. King J. F. Griffith C. Metreweli Department of Diagnostic Radiology & Organ Imaging, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong J. C. Y. Cheng X. Guo Orthopaedics & Traumatology, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong

Abstract Background. The role of the multifidus muscles in the initiation and progression of curve in adolescent idiopathic scoliosis is not fully understood and controversy exists as to the side of the abnormality. Objective. To evaluate on MRI the multifidus muscles at the apex of the major curve in adolescent idiopathic scoliosis to ascertain if the multifidus muscles on the convex or concave side are abnormal and the relationship to curve severity. Materials and methods. Forty-six patients with adolescent idiopathic scoliosis, separated into two groups, were studied using a 1.5-T MR scanner with the synergy spine coil, employing a modified STIR (short tau inversion recovery) axial sequence obtained at the apex of the major scoliotic curve.

Results. No hyperintense signal change was demonstrated in the convex side multifidus muscles in any patient. In group I, 16 of 18 patients with severe or rapidly progressive curve showed increase in signal intensity in the multifidus muscle on the concave side of the apex of the curve. In group II, of the 15 patients with mild curve (Cobb angle 1030 ), 4 had increased signal intensity in the multifidus muscle on the concave side; of the 13 with more severe curve (Cobb angle greater than 30 ), 10 had increase in multifidus signal intensity on the concave side. Conclusions. The concave-side multifidus muscle at the apex of a scoliotic curve was morphologically abnormal. A significant association between abnormal signal change and curve severity was also established.

Introduction
The aetiology of adolescent idiopathic scoliosis (AIS) is unresolved. Hypotheses on the pathogenesis of AIS include damage to the thalamocortical tract [1], equilibrium dysfunction [2, 3], syrinx and Chiari I malformation [4, 5], asymmetrical loading of the spine [6] and thoracic hypokyphosis [7]. The paraspinal muscles are considered to have a significant effect on the pathomechanics of the spinal curvature and may be involved either in the initiation or progression of the scoliotic curve in AIS. Asymmetry of the paraspinal muscles in idiopathic scoliosis has been demonstrated using biochemistry [813], histopathology [9, 14, 15]

and electromyography [1619]. However, the results with regard to the side of abnormality in these studies remain controversial. The multifidus muscles originate from the transverse processes from C 4 to L 5 and run obliquely upwards and medially to insert on the spinous processes two to four segments higher. They are involved in the dynamic maintenance of the spinal curvature in the coronal and axial planes. MRI is a sensitive technique for the detection of abnormalities in muscles [20], and it has been employed in the study of the multifidus muscle. Hyperintense signal change in the multifidus muscle has been demonstrated on MRI after trunkextension exercise in patients with chronic low-back pain and in normal controls [21]. The objectives of the

361

1
Fig. 1 A 20-year-old male with left thoracolumbar scoliosis with a severe curve having a Cobb angle of 60 . At the apex, at T 12 level, high signal intensity was demonstrated in the multifidus muscle on the concave right side (white arrows). (L laminae of the T 12 vertebra, E erector spinae muscle) Fig. 2 A 13-year-old female with right thoracic curve with a Cobb angle of 32 . At the apex, at T 8 level, high signal intensity was demonstrated in the multifidus muscle on the concave left side (black arrows). (L laminae of the T 8 vertebra, E erector spinae muscle)

2
This was useful because the higher signal in normal muscles allows more accurate localisation of the small multifidus muscles. The axial images were evaluated for the relative signal intensities of the multifidus and erector spinae muscles on both sides at the apex of the curve without knowledge of the Cobb angle of the curve. The duration of the STIR sequence was 2.5 min, whilst that of the entire MRI examination was about 25 min. Sedation was not required for any patients. The study was approved by the ethics committee of the university.

present study were to evaluate on MRI the multifidus muscles at the apex of the major curve in AIS to ascertain if the multifidus muscles on the convex or concave side are abnormal and the relationship to curve severity.

Results
Sixteen of 18 patients in group I with severe curve planned for operative treatment showed increase in signal intensity in the multifidus muscle on the concave side (Fig. 1) of the apex of the curve (Table 1). In group II, of the 15 patients with mild curve (Cobb angle 1030 ), 4 had increased signal intensity in the multifidus muscle on the concave side (Table 2); of the 13 with more severe curve (Cobb angle greater than 30 ), 10 had increase in multifidus signal intensity (Fig. 2) on the concave side (Table 3). The multifidus muscles on the convex side of the major curve in all patients did not show any increase in signal intensity, irrespective of group or degree of curvature. No significant difference in the signal intensity in the erector spinae muscles between the concave and convex sides was found in either group I or group II and irrespective of the curve severity.

Materials and methods

This was a prospective cross-sectional study on 46 patients with documented AIS recruited from two groups of patients. Group I comprised 18 patients (14 girls, 4 boys; aged 1020 years, mean 15.2 years) with severe or rapidly progressive curve planned for operative treatment. The Cobb angles were over 40  or there was an increase of more than 5  per year. Group II consisted of 28 patients (27 girls, 1 boy, aged 1015 years, mean 12.5 years) recruited in a period of 2 months from the scoliosis clinic and included both follow-up patients and new consultations. Patients in this second group were selected for different degrees of curve varying from mild to severe with a Cobb angle of 1055 . MRI examinations were performed in a 1.5-T MR scanner, using a synergy spine coil. Sagittal T 2-weighted (T 2-W) sequence of the whole spine and coronal T 2-W sequence centred at the curvature were first performed for location of the level of the apex of the primary curve, or the more severe curve when there was a double curve. An axial STIR (short tau inversion recovery) sequence was subsequently obtained at the apex of the curve with the following parameters: repetition time 1400 ms, echo time 15 ms, inversion time 140 ms, 4-mm slice thickness with 0.8 mm gap, 20 slices, number of excitations 2, field of view 230 mm and 256 256 matrix. A short echo time was chosen specifically for this study as compared to the more T 2-W STIR sequence performed generally.

Discussion
Hyperintense signal could be demonstrated on STIR sequence on MRI in the concave-side multifidus muscles at the apex of the major curve in a significant proportion of patients suffering from AIS. The prevalence of hyperintense signal intensity increased with greater degrees of curve severity. The findings suggest a pathological pro-

362

Table 1 Multifidus signal at apex of major curve in patients with severe or rapidly progressive curve (group I) Signal intensity in multifidus muscle Increase Concave side Convex side n = 16 n=0 No increase n=2 n = 18 P < 0.0001 Fisher Exact Test

Table 2 Multifidus signal at apex of major curve in first presenting patients (group II) with Cobb angle 10 to 30  Signal intensity in multifidus muscle Increase Concave side Convex side 2 0 No increase 9 11 Not significant Fisher Exact Test

Table 3 Multifidus signal at apex of major curve in first presenting patients (group II) with Cobb angle greater than 30  Signal intensity in multifidus muscle Increase Concave side Convex side 9 0 No increase 3 12 P = 0.002 Fisher Exact Test

cess involving the multifidus muscle on the concave side at the apex of a major scoliotic curve, and there was also an association of the process with the severity of the curve. Hyperintense signal change in muscles on STIR sequence is a non-specific finding and may be seen in muscle injury, following exercise, oedema, inflammation, myopathy, chronic muscle overuse syndrome [20], and denervation [22]. Histological studies in paraspinal muscles in AIS have so far not revealed any evidence of inflammatory cellular infiltration [10], muscle denervation [10, 23] or primary muscle disease [23]. Kuno et al. [24] studied the relationship between MR relaxation time and muscle fibre composition in humans in vivo. Prolongation of T 2 relaxation time increased with the percentage of fast twitch type II muscle fibres.

Several previous studies have shown an increase in type I fibres in the paraspinal muscles on the convex side of the curve in AIS [810, 12, 14] and also a smaller relative proportion of type I fibres on the concave side of the curve [13]. More type II fibres were found in muscle biopsies of the multifidus muscle on the concave side at the apex of the scoliotic curve in AIS [25]. The increase in multifidus muscle signal on the concave side at the apex of the curve in our patients with AIS may therefore be explained by a relative increase of type II muscle fibres, with consequent longer T 2 relaxation time and higher signal on STIR sequence. Although controversy exists and there are also histological and biochemical studies suggesting the convex side to be the abnormal side [1012], MRI is comprehensive and not subject to the limitation of sampling error as may occur in muscle biopsy. Different groups of multifidus, deep erector spinae or superficial erector spinae, or their combinations have been targeted in these histological and biochemical studies, which was also a possible source of error. Ultrastructural changes have been found in the myotendinous junction of the multifidus muscles on the concave side of a scoliotic curve, and it has been suggested that this is of primary aetiological importance [15]. Increase in muscle signal intensity has been demonstrated on STIR sequence in recurrent exertional or chronic muscle overuse syndrome, with mild oedema-like changes being evident in muscle-tendon units [20]. Another postulate of the signal change in the concave-side multifidus muscle therefore may be the chronic overuse of the multifidus muscle in the dynamic maintenance of balance of a scoliotic curve. In conclusion, high-signal intensity change was detected in the concave-side multifidus muscle at the apex on STIR sequence, suggesting that the concave side muscles were the morphologically abnormal ones. A significant association between abnormal signal change and curve severity is also established. Whilst the primary or secondary nature of the increase of signal is not addressed by the present study, the signal increase may be explained by a change in muscle fibre composition, or chronic muscle overuse. MRI, being a comprehensive and non-invasive study, is well suited to a longitudinal study on the changes in the multifidus muscles in patients with AIS to uncover the possible pathogenetic role of the multifidus muscles in AIS.

References
1. Machida M, Dubousset J, Imamura Y, et al (1993) An experimental study in chickens for the pathogenesis of idiopathic scoliosis. Spine 18: 16091615 2. Yekutiel M, Robin GC, Yarom R (1989) Proprioceptive function in children with adolescent idiopathic scoliosis. Spine 6: 560566 3. Byl NN, Gray JM (1993) Complex balance reactions in different sensory conditions: adolescents with and without idiopathic scoliosis. J Orthop Res 11: 215227

363

4. Schwend RM, Henniikus W, Hall JE, et al (1995) Childhood scoliosis: clinical indications for magnetic resonance imaging. J Bone Joint Surg Am 77: 4653 5. Evans SC, Edgar MA, Hall-Craggs MA, et al (1996) MRI of idiopathic juvenile scoliosis. J Bone Joint Surg Br 78: 314317 6. Trontelj JV, Recak F, Dimitrijevic MR (1979) Segmental neurophysiological mechanisms in scoliosis. J Bone Joint Surg Br 61: 310313 7. Dickson RA (1988) The aetiology of spinal deformities. Lancet 1: 11511155 8. Fidler MW, Jowett RL (1976) Muscle imbalance in the aetiology of scoliosis. J Bone Joint Surg Br 58: 200201 9. Yarom R, Robin GC (1979) Studies on spinal and peripheral muscles from patients with scoliosis. Spine 4: 1221 10. Zetterberg C, Aniansson A, Grimby G (1983) Morphology of the paravertebral muscles in adolescent idiopathic scoliosis. Spine 8: 457462 11. Sahgal V, Shah A, Flanagan N, et al (1983) Morphologic and morphometric studies of muscle in idiopathic scoliosis. Acta Orthop 54: 242251

12. Ford DM, Bagnall KM, McFadden KD, et al (1984) Paraspinal muscle imbalance in adolescent idiopathic scoliosis. Spine 9: 373376 13. Bylund P, Jansson E, Dahlberg E, et al (1987) Muscle fiber types in thoracic erector spinae muscles. Clin Orthop 214: 222228 14. Spencer GS, Zorab PA (1976) Spinal muscle in scoliosis. Part 1: histology and histochemistry. J Neurol Sci 30: 137142 15. Khosla S, Tredwell SJ, Day B, et al (1980) An ultrastructural study of multifidus muscle in progressive idiopathic scoliosis changes resulting from a sarcolemmal defect of the myotendinous junction. J Neurol Sci 46: 1331 16. Zuk T (1962) The role of spinal and abdominal muscles in the pathogenesis of scoliosis. J Bone Joint Surg Br 44: 102105 17. Butterworth TR, James C (1969) Electromyographic studies in idiopathic scoliosis. South Med J 62: 10081010 18. Reuber M, Schultz A, McNeill T, et al (1983) Trunk muscle myoelectric activities in idiopathic scoliosis. Spine 8: 447456 19. Alexander MA, Season EH (1978) Idiopathic scoliosis: an electromyographic study. Arch Phys Med Rehabil 59: 314315

20. Shellock FG, Tyson LL, Fleckenstein JL (1996) Diagnostic imaging of skeletal muscle exercise physiology and pathophysiology. In: Fleckenstein JL, Crues JV III, Reimers CD (eds) Muscle imaging in health and disease. Springer, Berlin Heidelberg New York, pp 13132 21. Flicker PL, Fleckenstein JL, Ferry K, et al (1993) Lumbar muscle usage in chronic low back pain magnetic resonance image evaluation. Spine 18: 582586 22. Uetani M, Hayashi K, Matsunaga N, et al (1993) Denervated skeletal muscle: MR imaging. Radiology 189: 511515 23. Fidler MW, Jowett RL, Troup JD (1974) Histochemical study of the function of multifidus in scoliosis. In: Zorab PA (ed) Scoliosis and muscle. Heinemann, London, pp 184192 24. Kuno S, Katsuta S, Inouye T, et al (1988) Relationship between MR relaxation time and muscle fiber composition. Radiology 169: 567568 25. Meier MP, Klein MP, Krebs D, et al (1997) Fiber transformations in multifidus muscle of young patients with idiopathic scoliosis. Spine 20: 23572364