Neuromuscular disturbances in adolescent idiopathic scoliosis observed from the anticipatory muscle activations in scapula stabilizers: a cross-sectional study

Abstract

Background

While it is well-stablished that scoliosis can lead to neuromuscular control disorders, the specific characteristics of these impairments remain unclear. This study aimed to explore the neuromuscular features of scapula stabilizers in adolescents with idiopathic scoliosis (AIS) through an analysis of anticipatory muscle activations (AMAs).

Methods

A cross-sectional observational study was conducted with 17 AIS and 19 age-matched healthy subjects. Both the AIS and healthy groups performed right and left upper limb reaching tasks at three different heights while surface electromyography monitored the activity of scapular stabilizers. The muscles examined included the bilateral infraspinatus, upper trapezius, lower trapezius, serratus anterior, latissimus dorsi and the anterior deltoid. The incidence, onset latencies, and amplitude of AMAs were compared between the AIS and healthy groups.

Results

Among scapular stabilizers, ipsilateral upper trapezius (iUT) and ipsilateral infraspinatus (iIS) exhibited the highest AMAs incidence at 72.59% and 70.06%, respectively. However, AMAs incidence significantly declined on the concave side of the thoracic curve in AIS group, particularly in iUT (55.63 ± 14.74 vs. 45.21 ± 19.92, t = 2.330, P = 0.034) and iIS (59.38 ± 20.16 vs. 48.13 ± 22.11, t = 2.316, P = 0.035). Regarding onset latencies, the AIS group exhibited delayed activation of ipsilateral lower trapezius (F = 3.586, P < 0.05, η²_p = 0.006) and advanced activation of contralateral upper trapezius (F = 7.753, P < 0.001, η²_p = 0.027) and contralateral infraspinatus (F = 6.554, P < 0.01, η²_p = 0.024) on concave side compared to the healthy group. Additionally, almost all scapula stabilizers in AIS group exhibited reduction of AMAs amplitude compared to the healthy group.

Conclusions

This preliminary study suggests potential neuromotor control impairments in the AIS population, as indicated by observed trends in the incidence, amplitude, and timing of AMAs in scapular stabilizers. These preliminary insights may inform the design of future rehabilitation interventions, with attention to neurodevelopmental needs during adolescence.

Trial registration number

ChiCTR2300075167, Date: 2023-08-28.

Introduction

Adolescent idiopathic scoliosis (AIS) is a three-dimensional spinal deformity unknown etiology in adolescents aged of 10 to 18 years [1, 2]. It is the most common skeletal-muscular deformity in adolescents, with a prevalence of 2–3%, and a female-to-male ratio ranging from 1.5:1 to 11:1 [2–5]. Beyond its impact on posture, AIS is associated with significant mental health challenges and a decreased quality of life [2, 6]. Additionally, AIS is also accompanied by alterations in neuromuscular control [7, 8].

Neuromuscular control of the scapula is particularly important for the AIS population. Adolescents engage in numerous upper extremity functional activities in their daily lives, where shoulder stability plays a fundamental role. This stability heavily relies on the subconscious pre-activation of scapula stabilizers [9], a process known as anticipatory muscle activations (AMAs) [10]. The AMAs embodies the process of neuromuscular control in the preparation for movement [11, 12]. Delays or advancements in AMAs are usually recognized as indicators of motor control dysfunction [13]. Adolescence is a critical period of growth and development, yet AIS introduces disruptions in neuromuscular control [14–19]. Abnormal activation patterns of scapular stabilizers are often reinforced through repetitive movements, potentially exacerbating the spinal deformity in a vicious cycle [20, 21]. Studies have shown that the magnitude of spinal curvature is a strong predictor of AIS progression [22], with an initial Cobb angle of 15° or greater posing a higher risk of deterioration [22–24]. We hypothesize that significant changes in the neuromuscular control patterns of the scapula stabilizers may contribute to the rapid progression of AIS once spinal deformities reach 15 ° or more. A thorough understanding of these neuromuscular control mechanisms of the scapula stabilizers in AIS populations at high risk for progression is crucial for mitigating disease progression and guiding the development of effective rehabilitation interventions. However, research in this area remains limited.

To date, only a few studies have focused on neuromuscular control of the scapula stabilizers in AIS populations [25]. For example, Lin et al. reported delayed muscle activation onset in the lower trapezius (LT), serratus anterior (SA) the convex side, as well as delayed activation of the upper trapezius (UT), LT and SA on the concave side. They also emphasized that altered motor control in the kinematic connection between the thoracic spine, scapulae, and arms is responsible for the changes in muscle activity [25]. However, they predominantly focused on muscle activation timing, neglecting other critical metrics of neuromuscular control, such as AMAs incidence and amplitude. Furthermore, these studies rarely consider the dynamic nature of daily shoulder activities, which are essential for understanding muscle performance under real-world conditions.

Investigating the AMAs patterns of scapula stabilizers in both AIS and healthy populations may be valuable for enhancing physiotherapeutic scoliosis-specific exercises (PSSE) and improving treatment outcomes. Therefore, the objectives of this study were: [1] To investigate the presence of AMAs abnormalities in the scapula stabilizers of AIS subjects [2] to identify specific muscles exhibiting AMAs changes compared to a control group [3] to examine alterations in AMAs during different reaching tasks.

Materials and methods

Participants

This cross-sectional observational study aimed to investigate scapular muscle anticipatory activation patterns in AIS. A total of seventeen outpatients and nineteen age-matched healthy controls were recruited from the Department of Rehabilitation Medicine and advertisements between September 2023 and December 2023. Written informed consent was obtained from all participants and their guardians, following the latest Declaration of Helsinki guidelines. Ethical approval was granted by the Clinical Trials Ethics Committee (certificate no. TJ-IRB20230512) on May 18, 2023. Demographic characteristics of participants were summarized in Table 1. Inclusion criteria for AIS subjects: (1) age between 10 and 18 years (2) clinical diagnosis of idiopathic scoliosis (right thoracic and left lumbar type), classified as Type 1 (double curve with a larger and stiffer lumbar curve) or Type 2 (double curve with a larger and stiffer thoracic curve) according to King’s classification [26] (3) thoracic curve Cobb angle between 15° and 45° (4) no previous scoliosis treatment (5) right-handedness (6) absence of shoulder pain or limitation in shoulder motion. Inclusion criteria for healthy subjects: (1) age between 10 and 18 years (2) angle of trunk rotation (ATR) < 3° measured by scoliometer (3) no other neurological, musculoskeletal disorders (4) right-handedness (5) absence of shoulder pain or limitations in shoulder motion. Exclusion criteria: (1) spinal tumors, spinal tuberculosis, other spine-related diseases or spinal surgery (2) shoulder pain, limited shoulder motion or history of shoulder injury (3) presence of other serious diseases (4) any conditions deemed unsuitable for trial participation.

Table 1.

Demographic characteristics of the participants

Subjects	Age (yr.)	BMI (kg/m²)	Apex-T	Cobb-T (deg)	Apex-L	Cobb-L (deg)	Risser	Female/Male
#1	12	17.83	T9	39	L2	24	4	F
#2	13	17.09	T6	20	L2	30	3	M
#3	14	16.61	T8	28	L2	35	3	F
#4	11	21.62	T8	31	L3	32	3	F
#5	13	16.9	T9	26	L3	8	3	M
#6	18	16.94	T8	42	L2	30	5	F
#7	13	20.7	T7	20	L2	40	3	F
#8	13	18.52	T8	26	L3	22	0	F
#9	15	25.35	T9	39	L2	30	4	M
#10	11	16.61	T8	40	L2	24	1	F
#11	13	16.8	T9	30	L2	46	1	F
#12	17	16.33	T7	24	L3	28	5	M
#13	15	18.34	T9	26	L2	17	3	M
#14	16	14.69	T8	50	L2	43	4	F
#15	13	19.17	T11	24	L3	12	3	F
#16	13	16.76	T9	18	L2	7	2	M
#17	14	18.03	T11	43	L4	15	2	F
#AIS group (n = 17)	13.76 ± 1.92	18.13 ± 2.50	T6-T11	30.94 ± 9.45	L2-L4	26.06 ± 11.62	0–5	11/6
Healthy group (n = 19)	14.58 ± 2.19	19.75 ± 2.82	NA	NA	NA	NA	NA	13/6
p	0.247	0.078	/	/	/	/	/	0.813

BMI body mass index

#Represents the AIS participations. The yr. is same as years old. Apex-T and Apex-L represent the apex vertebral of thoracic and lumbar respectively. Cobb-T and Cobb-L represent the Cobb angle of curves in thoracic and lumbar respectively. NA means not applicable

Experimental Procedure

Instruments

Surface electromyography (sEMG) data were acquired using a real-time Ultimu sEMG system (Noraxon USA Inc, Scottsdale, AZ, USA) at a sampling rate of 2000 Hz.

Pre-test preparation

Participants’ skin was cleaned with alcohol to ensure proper electrode adhesion. Surface electrodes (Ag/AgCl) were placed on the specified muscles following SENIAM guidelines [27]. The muscles tested included: the anterior deltoid (AD) (at 1.5 cm distal and anterior to the acromion) on the motor side, bilateral UT (two-thirds of the distance from the spinous process of the seventh cervical vertebra to the acromion process), LT (one-fourth of the distance from the thoracic spine to the inferior angle of the scapula with the arm elevated in the sagittal plane), latissimus dorsi (LD) (approximately 4 cm below the inferior tip of the scapula and midway between the spine and lateral edge of the torso at a 25° oblique angle), infraspinatus (IS) (4 cm below the spine of the scapula on the lateral aspect over its infra-scapular fossa), and SA (over the muscle fibers anterior to the LD muscle with the arm elevated 90° in the sagittal plane).

The environmental preparation ensured a warm, quiet, and comfortable setting to minimize external distractions. Participants were instructed to seated on armless, backless chairs positioned 1.5 m from a whiteboard, maintaining 90° hip and knee flexion while resting their arms on either side of the body in a relaxed posture. A tripod with three height-adjustable discs was positioned at a 40° angle (scapular plane) [28] relative to the subject’s shoulder crest, with the discs set at 120% of the participant’s arm length. The disc heights were individually adjusted before the experiment to accommodate each participant. Participants were instructed to complete the target task promptly and accurately based on the beeps emitted by the headphones they wore (Sennheiser HD25-I; Wedemark, Germany).

Test tasks

Participants were asked to perform low, horizontal, and high reaching tasks, defined as shoulder flexion to 60%, 100%, and 140% of acromion height, respectively, with the middle finger touching the disc’s center. Figure 1(a) schematizes the task paradigm. Each of the three tasks was repeated 10 times, randomly ordered within a 30-trial test, with a 1-minute rest between every 10 consecutive repetitions. Figure 1(b) illustrates the auditory command sequence. Five seconds into each test, participants received verbal instructions for the task, such as “Reaching high.” Subsequently, participants were prompted to assume a ready state upon hearing a continuous series of beep-beep-beep signals (80 dB, 500 ms, 1000 Hz). Within the next 2.5 to 3 s, a 40-ms 80-dB sound was randomly delivered through the headset at any point in time to signal the initiation of the task, preventing premature task initiation. Participants were required to initiate the task immediately upon hearing the sound. A 15-second interval was observed between tasks to ensure adequate relaxation. Participants were instructed to perform 30 repetitions of the reaching task right side first, followed by the left. Prior to data collection, participants practiced the task 5–10 times to familiar. The design and execution of all tests were conducted using the Psychtoolbox-3 toolkit based on MATLAB (2017b, MathWorks, Natick, MA, USA).

Fig. 1.

An illustration of task paradigm in this study. a An illustration of the task “horizontal reaching” in our study. b An illustration of one task’s audio command from start to the end in our study

Data preprocessing

Tasks with significant movement pre-occurrence and incorrect execution were excluded based on video recordings. Raw sEMG data were processed using MATLAB software (2017b, MathWorks, Natick, MA, USA). Initial preprocessing involved segmentation and filtering, employing a bandpass filter set at 30–300 Hz and a notch filter at 50 Hz. Teager-Kaiser energy operations were applied to the filtered data to synchronize sEMG amplitude and instantaneous frequency as a reference basis, demonstrating higher reliability compared to conventional methods [29]. Following data correction and normalization, a threshold detection method was utilized to determine the onset of muscle activation. The threshold (T) was defined as:

$$T=\mu +3\sigma$$

Where µ and σ represent the mean and standard deviation of the baseline amplitude in the 2,500–500 ms time window before each “Go” signal. False-positive activation points were subsequently eliminated through morphological operations [30].

Data from healthy subjects on the left and right sides were standardized to the same model, while data from AIS subjects were transformed into convex and concave formats. The classification of convex and concave sides was determined by the thoracic curve, with the apex vertebra positioned between the sixth (T6) and eleventh thoracic vertebrae (T11). The onset time of muscle activation for the AD was considered as time zero (T₀) for all tasks. Detailed descriptions of parameters and outcome variables, including premotor reaction time, valid trial numbers, AMAs onset, the incidence of AMAs of valid trials, muscle onset latency, and AMAs amplitude, are provided in Table 2, referencing our previous study on AMAs [31].

Table 2.

The definition and calculation process of the parameters and outcome variables

Parameters/outcomes	Definition	Calculation methods
T0	The muscle activation onset time point of anterior deltoid in each trial	Over 10 consecutive samples of the smoothed signal exceeding the threshold (the mean with 3 SD of baseline amplitude of for anterior deltoid in the time window 2,500 to 500ms before each “Go” cue).
Premotor reaction time	The response time of motor initiation (shoulder flexion)	The interval between the rise of “Go” signal to T0 at each trial.
Valid trials numbers	Valid trials numbers in the total 60 trials for each subject	Trials were excluded if the premotor reaction time is within 30ms after the “Go” cue or exceeds 400ms for muscle response.
AMAs onset	Anticipatory muscle activation onset of the testing muscles	The onset time point calculation through the threshold method (T = µ + hσ). In valid trials, the number of muscle activation onset at the anticipatory posture adjustment (APA) window: −100 to + 50ms to T0 for each muscle except AD.
Proportion (incidence) of AMAs in valid trials	The percentage of AMAs onset trial numbers in total valid trial numbers for each muscle of each subject	“AMAs onset trial numbers for each muscle of each subject” / “total valid trial numbers for each muscle of each subject”
Muscle onset latency	The interval of muscle onset time to T0 in each trial with positive	Use the actual activation time of the target muscle minus the time point of T0.
AMAs amplitude	The muscle activation amplitude of each muscle at the APA window in each valid trial with AMAs onset	The integral of the amplitude of sEMG signal after the preprocessing step in the time window − 100 to + 50ms to T0.

Statistical analyses

All statistical analyses were performed using IBM SPSS 22.0 software (IBM Corp, Armonk, NY, USA), with a significance level set at P < 0.05. Normality of the distribution of demographic data and proportional variables was assessed through the Shapiro-Wilk test. Independent Student’s t-tests and chi-square test compared demographic data. Paired t-tests were utilized to examine differences in proportional variables on two sides of AIS subjects. Two-way ANOVA and post-hoc comparisons with Bonferroni correction were adopted to analyze the onset latencies and AMAs amplitude.

Results

The proportion of valid trials and trials with valid AMAs onset

We compared the proportion of valid trials and valid AMAs onset between the two groups and summarized the data in Table 3. A total of 1,660 trials were successfully screened from the tests of 36 participants (out of 2,160 trials). The number of valid trials was significantly higher in the healthy than in the AIS (50.26 ± 6.67 vs. 42.63 ± 9.29, F = 1.996, P = 0.008). However, there was no difference in the number of valid trials between concave and convex sides in AIS (14.00 ± 6.49 vs. 11.06 ± 5.47, P > 0.05). This finding suggests that there is a reduction in task responsiveness in the AIS, and this effect is widespread and not a centralized impairment on one side of the spine.

Table 3.

Proportion of anticipatory muscle activations（AMAs）of subjects

	Valid trial numbers	Proportion of AMAs in valid trials (%)
		iUT	cUT	iLT	cLT	iSA	cSA	iIS	cIS	iLD	cLD
AIS group, mean (SD)	42.63(9.29)	50.42(15.07)	22.08(11.52)	47.60(15.48)	26.15(10.07)	41.77(16.46)	22.19(7.47)	53.75(18.79)	21.04(9.58)	36.25(13.52)	22.60(7.30)
Healthy group, mean (SD)	50.26 (6.67)	57.98(9.82)	29.82(11.89)	55.70(10.83)	28.33(10.05)	37.63(12.67)	19.47(7.82)	58.95(11.87)	29.04(11.55)	38.86(12.36)	24.82(7.76)
F-value	1.996	3.178	0.182	0.856	0.037	2.185	0.104	4.313	0.322	0.136	0.001
P-value	0.008**	0.083	0.06	0.079	0.526	0.407	0.304	0.327	0.035*	0.555	0.393
Convex side, mean (SD)	14.00(6.49)	55.63(14.74)	22.29(14.44)	49.58(22.47)	24.79(14.35)	46.67(21.64)	16.46(8.98)	59.38(20.16)	22.08(12.22)	38.54(21.70)	20.42(11.01)
Concave side, mean (SD)	11.06 (5.47)	45.21(19.92)	21.88(12.23)	45.63(18.04)	27.50(14.17)	36.88(18.24)	27.92(12.58)	48.13(22.11)	20.00(11.86)	33.96(15.36)	24.79(10.95)
t-value	1.721	2.33	0.123	0.597	−0.537	1.721	−2.871	2.316	0.571	0.702	−1.067
P-value	0.106	0.034*	0.904	0.559	0.599	0.106	0.012*	0.035*	0.577	0.494	0.303

i/cUT, i/cLT, i/cSA, i/cIS, i/cLD represents ipsilateral/contralateral upper trapezius, ipsilateral/contralateral lower trapezius, ipsilateral/contralateral serratus anterior ipsilateral/contralateral infraspinatus and ipsilateral/contralateral latissimus dorsi

*P-value < 0.05, **P-value < 0.01

The number of valid AMAs onset trials per muscle ranged from 441 to 1205 in the 10 tested muscles. Ipsilateral IS (iIS) and ipsilateral UT (iUT) had the top AMAs incidences, 72.59% and 70.06%, respectively. For the AIS, the proportion of valid AMAs trials in iUT (55.63 ± 14.74 vs. 45.21 ± 19.92, t = 2.330, P = 0.034) and iIS (59.38 ± 20.16 vs. 48.13 ± 22.11, t = 2.316, P = 0.035) indicated a lower incidence in the concave side. These suggest that the IS and UT muscles play an important role in postural preparation for the reaching tasks. However, significant changes occurred in the advance activation of these two muscles on the concave side of the AIS.

Muscles onset latencies

To compare the timing of the scapula stabilizers, we compared the onset latencies on different moving sides and different tasks and summarized in Fig. 2A. There was a significant main effect (P < 0.05) of the side factor on the onset latencies of iLD (F = 3.637, P = 0.027, η²_p = 0.009), iLT (F = 3.581, P = 0.028, η²_p = 0.007), cUT (F = 7.015, P = 0.001, η²_p = 0.025) and cIS (F = 6.627, P = 0.001, η²_p = 0.024). Post-hoc comparisons were conducted on different sides and presented in Fig. 2B. Compared to the healthy, there was a significant delay on iLT activation but an early onset of cIS and cUT during moving the concave side. Compared to the convex side, there was a significant delay of iLD activation but an early onset of cIS and cUT during moving on the concave side. These findings suggest that the AMAs timing is indeed altered in the AIS population, and it is mainly concentrated in the UT, LT, and IS muscles.

Fig. 2.

Onset latencies of peri-shoulder stabilizing muscle groups in reaching tasks of the healthy, convex, and concave side. a This illustrates the tunes of AMAs onset in high reaching, horizontal reaching, and low reaching with different move sides, including healthy subjects, convex side, and concave side of AIS subjects, respectively. Horizontally arranged from left to right are the mean with SEM of AMAs onset latency in healthy, convex, and concave sides; vertically arranged from top to bottom are the different AMAs onset latency (mean with SEM) in tasks reaching high, horizontal, and low. b This demonstrates the results of Bonferroni post-hoc comparison on different sides in iLT, iLD, cIS, cUT muscles. Arranged horizontally from left to right are the healthy, convex, and concave sides of the AMAs latency (mean with SEM); *P-value < 0.05; **P-value < 0.01

AMAs amplitude of muscles

And we similarly compared the AMAs amplitude at different sides (healthy, convex, concave) while performing different tasks, summarized in Fig. 3. The results revealed significant main effects for both the side and task factors. For sides, significant main effects were observed in the iIS (F = 14.792, P < 0.001, η²_p = 0.025), iLD (F = 16.655, P < 0.001, η²_p = 0.041), iLT (F = 9.945, P < 0.001, η²_p = 0.019), iSA (F = 4.823, P = 0.008, η²_p = 0.011), iUT (F = 5.615, P = 0.004, η²_p = 0.010), cLD (F = 7.749, P < 0.001, η²_p = 0.031), cSA (F = 10.319, P < 0.001, η²_p = 0.046) and cIS (F = 3.092, P = 0.046, η²_p = 0.012). And for almost all muscles, the AMAs amplitude was generally greater on the healthy than on the concave and convex sides (Table 4). These findings suggest a generalized decline in AMAs amplitude in the AIS.

Fig. 3.

Anticipatory muscle activations (AMAs) amplitude of shoulder stabilizers in reaching tasks of the healthy, convex, and concave side. This illustrates the AMAs amplitude of shoulder stabilizers in high reaching, horizontal reaching, and low reaching with different move sides, including healthy subjects, convex side, and concave side of AIS subjects, respectively. Horizontally arranged from left to right are the mean with SEM of AMAs amplitude in healthy, convex, and concave sides; vertically arranged from top to bottom are the different AMAs amplitude (mean with SEM) in tasks reaching high, horizontal, and low

Table 4.

Post hoc comparison of anticipatory muscle activations amplitude for sides and tasks

		AMAs amplitude
		iUT	cUT	iLT	cLT	iSA	cSA	iIS	cIS	iLD	cLD
Moving sides	H; mean (SD)	37.03(25.99)	30.96(28.49)	27.19(24.60)	36.72(28.79)	37.61(29.54)	34.46(27.27)	44.75(28.53)	38.34(29.58)	45.68(29.06)	50.80(30.44)
	Cv; mean (SD)	35.13(27.37)	29.72(30.65)	18.95(20.38)	32.05(28.10)	32.23(28.33)	19.00(20.32)	38.49(27.74)	30.64(26.37)	32.68(28.88)	41.52(30.56)
	Cc; mean (SD)	30.00(25.10)	25.20(29.10)	24.52(24.25)	31.88(27.64)	30.94(27.01)	30.66(29.13)	34.04(27.41)	37.37(27.85)	34.48(31.47)	38.41(30.55)
	p-value (H vs. Cv)	0.902	1	<0.001**	0.38	0.051	<0.001**	0.004**	0.047*	<0.001**	0.022*
	P-value (H vs. Cc) Concave)	0.001**	0.222	0.448	0.296	0.022*	0.547	<0.001**	1	<0.001**	0.001**
	P-value (Cv vs. Cc)	0.084	0.744	0.041*	1	1	0.004**	0.202	0.276	1	1
Tasks	Task 1 mean (SD)	45.27(27.42)	32.64(31.24)	28.56(24.46)	37.24(28.60)	44.23(29.64)	32.43(28.96)	51.40(28.17)	42.64(30.32)	47.59(32.11)	41.59(28.79)
	Task 2 mean (SD)	30.31(24.02)	24.14(25.19)	22.83(23.97)	30.71(28.37)	27.94(26.90)	21.67(20.71)	35.47(28.21)	29.22(26.89)	35.53(28.22)	43.65(30.58)
	Task 3 mean (SD)	29.98(24.23)	32.60(29.73)	23.46(22.92)	36.47(27.80)	30.86(26.91)	37.63(29.02)	36.18(25.94)	38.61(27.24)	37.76(28.46)	55.90(32.42)
	p-value (1 vs. 2)	<0.001**	0.008**	0.004**	0.052	<0.001**	0.001**	<0.001**	<0.001**	<0.001**	1
	P-value (1 vs. 3)	<0.001**	1	0.013*	1	<0.001**	0.272	<0.001**	0.573	<0.001**	<0.001**
	P-value (2 vs. 3)	1	0.024*	1	0.186	0.667	<0.001**	1	0.007**	1	0.001**

i/cUT, i/cLT, i/cSA, i/cIS, i/cLD represent ipsilateral/contralateral upper trapezius ipsilateral/contralateral lower trapezius, ipsilateral/contralateral serratus anterior, ipsilateral/contralateral infraspinatus and ipsilateral/contralateral latissimus dorsi

H, Cv and Cc represent Healthy, Convex and Concave group respectively

Task 1, 2 and 3 represent high, horizontal and low reaching correspondingly

*P-value < 0.05, **P-value < 0.01

For the tasks, significant main effects were observed for all ten muscles. Post-hoc analyses found that the AMAs amplitude of horizontal reaching was lower than high and low reaching universally (Table 4). Particularly, AMAs amplitude during low reaching and horizontal reaching were significantly lower than high reaching for iIS, iLD, iLT, iSA and iUT (Table 4). This suggests that for the scapula stabilizers, the amplitude of the high reaching was significantly higher than the low and horizontal reaching.

Discussion

To investigate the neuromuscular control in AIS at high risk for progression, this exploratory study examined the activation patterns of the scapula stabilizers during reaching tasks. The findings suggested a potential reduction in AMAs incidence on the concave side of AIS subjects, particularly in the IS and UT muscles, indicating a possible alteration in neuromuscular control that could affect task preparation. Notably, we observed that cUT and cIS on the concave side activated earlier than on both healthy and convex sides, indicating a distinct neuromuscular control pattern in AIS. What’s more, AIS subjects exhibited an extensive decrease in AMA amplitudes bilaterally. However, given the study’s small sample size and exploratory nature, these findings are preliminary and require further investigation with larger cohorts.

As the key structure connecting the trunk and upper extremities, the scapula is a key component of upper extremity movement [32, 33]. During reaching tasks, anticipatory control of scapula stabilizers is essential to ensure the compatible scapular and humeral positions [34]. Consistent with previous findings [25, 35–38], our findings suggest impaired neuromuscular control in AIS, but with specific alterations in scapula stabilizer AMAs not extensively documented previously. The observed decrease in IS AMAs incidence may imply reduce shoulder stability in the AIS population. It is known that IS plays a crucial role in stabilizing the shoulder. Typically, early and sustained activation of IS muscle helps maintain adequate subacromial space by pulling the humerus downward during forward flexion [39]. However, disruptions in IS activation patterns, as observed in AIS individuals, could increase their susceptibility to subacromial impingement syndrome and rotator cuff pathology [35, 40, 41]. Consequently, the reduced incidence of IS AMAs in AIS may carry clinical implications, underscoring the need for targeted rehabilitation strategies aimed at optimizing scapular function and minimizing shoulder-related complications.

Additionally, AIS subjects displayed a higher incidence of UT, IS, and SA AMAs on the convex side. Previous studies [35, 42] revealed that in AIS population the scapula was significantly more downwardly rotated and anteriorly tilted at rest on the convex side. These altered scapular positions are further aggravated during arm elevation, which requires scapular upward rotation and posterior tilt [43]. The increased and earlier activation of UT, IS, and SA on the convex side could represent an adaptive mechanism to stabilize the scapula during the reaching tasks, compensating for the altered kinematics and increased risk of instability in the convex-side scapula. This highlights the dynamic nature of scapular control in AIS and underscores the importance of addressing muscle imbalances during rehabilitation.

The findings related to muscle onset latencies indicated a complex interaction between delayed and early muscle activation. Consistent with previous findings, our study found delayed onset of concave LT muscle [25]. This suggests that concave-side LT may be difficult to recruit in a timely manner during anticipatory posture adjustment (APA) due to biomechanical changes related to spinal curvature. Since changes in postural muscle length affect the AMAs [44], in the resting position the concave-side scapula was more upwardly rotated so that the LT was not at its optimal initial length [34, 35, 42], and this may account for its delayed initiation. In contrast, the onset of IS and UT on the stabilized side was significantly earlier during concave-side movements, implying an overactive response and possibly a compensatory mechanism for homeostasis. It suggests that subjects tried to maintain stability by activating the muscle earlier than usual. Unlike the results of Lin et al. [25] we did not find delayed activation of SA in the AIS population, and we speculate that this may be related to the experimental paradigm. Reaching produces a greater postural perturbation compared to shoulder forward flexion, and APA can be adjusted according to the magnitude of the perturbation [45], and this may force the AIS subjects to change the APA strategy that activates SA in advance to maintain scapular stabilization. These findings highlight the complexity of neuromuscular adaptations associated with AIS and emphasize the need for targeted rehabilitation strategies for specific muscle imbalances to optimize postural readiness and improve motor performance during PSSE.

Our study also revealed a general reduction in AMAs amplitude across multiple muscles, potentially indicating a broader reduction in anticipatory motor activity in AIS subjects. This widespread decline is unlikely to be solely caused by biomechanical factors, but more likely by changes in the central nervous system (CNS). Neurodevelopmental theories propose that the central motor pathways may mature at a slower rate in individuals with AIS, leading to delayed or diminished AMAs responses [46]. Given that AMAs is modulated by multiple sensorimotor cortical regions, including the primary motor cortex, lateral premotor area, and supplementary motor area [47, 48], this decline in anticipatory control may have far-reaching implications for both posture and movement coordination in AIS. These findings highlight the need for targeted training interventions that address CNS development and improve anticipatory control, which could enhance rehabilitation outcomes for AIS.

This study, as a preliminary exploratory investigation, has several limitations. First, the strict inclusion criteria specifying ‘S’-shaped scoliosis limited recruitment. As a result, the findings may not be generalizable to the broader AIS population or other scoliosis classifications. Second, while the effect sizes for muscle onset latency and amplitude were found to be small (η²_p > 0.01), they did not reach the medium effect threshold (η²_p ≥ 0.06) [49]. This suggests that a larger sample size may be needed to detect more substantial effects. Furthermore, variability in scoliosis severity and individual differences could have contributed to the heterogeneity in muscle activation patterns, as scoliosis involves complex spinal deformities and neuromuscular factors that were not fully captured in this study. The relatively small sample size further limits the generalizability of our findings. While it aligns with typical exploratory studies, larger, multi-center studies are needed to validate these preliminary results and better understand the relationship between scoliosis severity, neuromuscular control, and functional outcomes. Finally, the cross-sectional design limits our ability to establish causality or track changes over time. Longitudinal studies are needed to assess how muscle onset characteristics evolve and their potential impact on disease progression and treatment outcomes.

Conclusion

This exploratory study identifies significant alterations in scapular muscle preactivation patterns in AIS, including changes in activation timing, amplitude, and task-specific patterns. While these changes suggest potential disruptions in neuromuscular control that could impair scapular stability and increase the risk of shoulder instability, the small sample size and low effect size warrant caution in drawing definitive conclusions. Additionally, the observed compensatory early activation of muscles on the convex side of the spine points to potential scapular muscle imbalances in AIS progression. These preliminary findings offer valuable insights that may inform the development of targeted rehabilitation strategies, but further research with larger sample sizes and more robust methodologies is needed to confirm these results and explore their therapeutic potential.

Abbreviations

AD

anterior deltoid

AIS

adolescent idiopathic scoliosis

AMAs

anticipatory muscle activations

APA

anticipatory posture adjustment

ATR

angle of trunk rotation

cIS

contralateral infraspinatus

cLD

contralateral latissimus dorsi

cLT

contralateral lower trapezius

cSA

contralateral serratus anterior

cUT

contralateral upper trapezius

CNS

central nervous system

iIS

ipsilateral infraspinatus

iLD

ipsilateral latissimus dorsi

iLT

ipsilateral lower trapezius

iSA

ipsilateral serratus anterior

iUT

ipsilateral upper trapezius

IS

infraspinatus

LD

latissimus dorsi

LT

lower trapezius

PSSE

physiotherapeutic scoliosis-specific exercise

SA

serratus anterior

sEMG

surface electromyography

UT

upper trapezius

Authors' contributions

YW and NX designed the experiment and were major contributor in writing the manuscript; JX, LW, LFX and XLH have substantively revised it. NX and YL designed the experiment and analyzed the data. YW, MHG and ZJC collected and organized experimental data. All authors read and approved the final manuscript.

Funding

The study was supported by the National Natural Science Foundation of China (Grant Nos. U 1913601 and 91648203).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study has been approved by the Ethics Committee of Tongji Hospital Affiliated to Tongji Medical College of Huazhong University of Science and Technology. And informed consent has been obtained from all subjects and their guardians.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.