Introduction: Adolescent idiopathic scoliosis (AIS) is a three-dimensional deformity of the spine occurring in the general population with prevalence between 2 and 3%, the aetiology and pathogenesis of which are poorly understood¹. Investigating the relationship between scoliosis and biomechanical measures such as trunk muscle activation would provide valuable information for the understanding of the pathomechanics of the AIS spine. Unfortunately, this measure is difficult to be acquired in vivo due to the invasiveness of the measurement procedure, and few data (from surface EMG) are available from the literature^2,3. An alternative strategy is represented by computational simulation based on musculoskeletal modelling approach, which allows calculating muscle activation in assigned postures and movements by means of inverse dynamic analysis. In this regard, a thoracolumbar model with articulated ribcage, developed in AnyBody software⁴ (AnyBody Technology, Denmark), has been recently adapted to replicate the spine alignment in AIS^5,6. The present study exploits the available model to replicate the subject-specific spine alignment in a dataset of 66 AIS subjects with scoliotic curve with apex in the thoracolumbar/lumbar region (Lenke type 3, 5, and 6) and Cobb angle of the curve ranging from 10° to 45°. Movement of trunk forward flexion from 0° (relaxed standing) to 45° is simulated (distributed from T12 to the fixed sacrum, Fig.1e). The asymmetry in erector spinae (ES) and multifidus (MF) muscle activation in the lumbar region (T12-L5) is calculated between the convex and concave side of the scoliotic curve, and distinguished between mild scoliosis (Cobb 10°-25°, 32 subjects) and moderate (Cobb 25°-45°, 34).

Methods: The exploited dataset was acquired by our group in a previous clinical study⁷. The subjects underwent radiological examination in orthostatic position by EOS system (EOS Imaging, France), providing the simultaneous acquisition of the frontal and lateral plane images (Fig.1a,b). The 3D spine alignment was replicated in the musculoskeletal model (Fig.1c,d) and scaling by subject’s weight and height was performed³. The asymmetry in ES and MF muscle activation was evaluated as (convex - concave)/(convex + concave) side of the scoliotic curve, providing zero value in case of balanced activation, and positive and negative values (from 0 to ±1) for larger activation in the convex and concave side, respectively.

Results: During flexion, the asymmetry of ES and MF activation showed decreasing and increasing trend, respectively, both for mild and moderate scoliosis (Fig.2a,b and d,e). Compared to relaxed standing, ES activation was found more negative at maximum flexion (Fig.2c,f), with mean±sd equal to 0.02±0.11 and -0.09±0.21 (mild scoliosis), and -0.01±0.15 and -0.11±0.25 (moderate), respectively. Conversely, MF exhibited larger positive values: 0.03±0.19 and 0.29±0.14 (mild), and -0.02±0.18 and 0.29±0.13 (moderate).

Discussion: The results pointed out concurrent opposite activation of ES and MF muscle in the lumbar region during forward flexion. This novel finding, based on computational simulation, suggests the presence of muscle synergy between ES (more involved to straighten the trunk) and MF (stabilizing the motion segments) in presence of scoliosis. Larger datasets and the simulation of additional movements in the other planes are needed to provide deeper understanding.