Analysing the correlation between preoperative clinical and radiological shoulder balance parameters among adolescent idiopathic scoliosis (AIS) patients with Lenke type 1 and 2 curves

Abstract

Background

Previous studies reported that medial shoulder balance (MSB) with neck tilt correlated radiologically with T1 tilt, first rib angle (FRA), and cervical axis (CA), whereas lateral shoulder balance (LSB) correlated radiologically with clavicle angle (Cla-A), coracoid height difference (CHD), clavicle-rib intersection distance (CRID), and radiographic shoulder height (RSH). Nevertheless, there is a scarcity of literature correlating both the clinical and radiological shoulder balance parameters in adolescent idiopathic scoliosis (AIS) patients. This study investigates the correlations between preoperative radiological and clinical shoulder balance (MSB and LSB) in Lenke type 1 and 2 AIS patients.

Methods

We reviewed 50 Lenke type 1 and 2 AIS patients between 2021 and 2022. Preoperative clinical shoulder parameters included front/neck base angle, front/back clavicle angle, front/back trapezium angle ratio, front/back axilla angle, trapezial area, shoulder area index 1 (SAI 1), SAI 2, and inner/outer shoulder height (SHi/SHo). Radiological shoulder parameters included T1 tilt, FRA, CA, Cla-A, CHD, CRID, and RSH.

Results

Strong correlations (r ≥ 0.6) were observed between T1 tilt, FRA, and CA (MSB with neck tilt), and between Cla-A, CHD, CRID, and RSH (LSB) (r > 0.9). T1 tilt and FRA correlated strongly (r ≥ 0.6) with front and back neck base angles, Ln [L/R trapezial area], SAI 1, SAI 2, and SHi, with the highest coefficients for front/back neck base angles (r ≥ 0.8). CA correlated strongly with front and back neck base angles, SAI 2, and SHi. Cla-A, CHD, CRID, and RSH correlated strongly with back clavicle angle.

Conclusion

MSB (T1 tilt and FRA) demonstrated very strong correlation with front and back neck base angles, and strong correlation with Ln [L/R trapezial area], SAI 1, SAI 2, and SHi. Neck tilt (CA) correlated strongly with front and back neck base angles, SAI 2, and SHi. Meanwhile, LSB (Cla-A, CHD, CRID, and RSH) correlated strongly with back clavicle angle.

Keywords

Lenke type 1 Lenke type 2 clinical shoulder balance radiological shoulder balance

Introduction

Neck and shoulder imbalance among adolescent idiopathic scoliosis (AIS) patients is a prevalent phenomenon that should be adequately addressed in corrective surgeries.^1–3 The incidence of preoperative shoulder imbalance among AIS patients was reported to be 23% and 32% for Lenke 1 and Lenke 2, respectively.⁴ Kuklo et al.⁵ proposed a clinical shoulder grading system based on radiographic shoulder height (RSH) difference where balanced was defined as <1 cm, minimal imbalance as 1–2 cm, moderate imbalance as 2–3 cm, and significant imbalance as >3 cm. Subsequently, Qiu et al.⁶ identified a discrepancy between radiological and clinical shoulder balance and suggested that shoulder balance should be further subdivided into inner and outer shoulder height. Building on this concept, Ono et al.⁷ described clinical shoulder as two distinct entities: medial shoulder balance (MSB) and lateral shoulder balance (LSB). MSB was reflected by the trapezial angle as well as Ln [L/R trapezial angle], and correlated well with radiological parameters, i.e., T1 tilt, first rib angle (FRA), and upper thoracic curve size. In contrast, these radiological parameters correlated weakly with LSB, as represented by the clinical clavicle angle.⁷ Kwan et al.⁸ further reported that clinical shoulder grading correlated with coracoid height difference (CHD), RSH, clavicle-rib intersection distance (CRID), and clavicle angle (Cla-A), parameters that may predominantly reflect LSB. Additionally, the authors introduced a clinical neck tilt classification based on trapezius muscle height difference, ranging from Grade 0 (no neck tilt) to Grade 3 (uncorrectable neck tilt with trapezius muscle height difference of >1 cm), and concluded that neck tilt with MSB correlated with cervical axis (CA) and T1 tilt.⁸ Deng et al.¹ later emphasized that MSB and LSB represent two disparate features of shoulder balance. Among Lenke 2 AIS patients who achieved LSB (RSH <10 mm) postoperatively, 69.2% of them experienced medial shoulder imbalance (MSI) (T1 tilt <−3° or >3°).¹

With regards to the relationship between clinical and radiological shoulder balance, Qiu et al.⁶ reported that none of the radiological parameters could precisely reflect the clinical shoulder appearance (none had correlation efficient, r > 0.8). In contrast, Matamalas et al.⁹ identified statistically significant correlations between clinical and radiological shoulder parameters, although they ranged from poor to moderate (r < 0.6). Taken together, these findings illustrate the ongoing inconsistency in the literature concerning the strength of clinical and radiological correlation in the shoulder balance assessment. In this study, we hypothesized MSB with neck tilt as represented radiologically by T1 tilt, FRA, and CA, may correspond to specific clinical parameters. Meanwhile, LSB as represented radiologically by Cla-A, CHD, CRID, and RSH, could correspond to another specific clinical shoulder parameters. Hence, this study was designed to investigate the correlations between preoperative radiological and clinical shoulder parameters for MSB with neck tilt and LSB, among Lenke type 1 and 2 AIS patients.

Materials and methods

This retrospective study was conducted in a single institution with institutional review board approval (MREC ID No.: 202214-10890). Sample size calculation was performed using G* Power (version 3.1.9.6; Heinrich Heine University), based on the correlation coefficient reported by Ono et al.⁷ (for trapezial angle and T1 tilt). Using an alpha of 0.05 and a power of 0.95, a minimum of 15 patients was required. A total of 73 AIS patients with Lenke type 1 or 2 curve, who underwent posterior spinal fusion (PSF) surgery between November 2021 and April 2022 were consecutively recruited. 23 patients who had poor quality radiographs or clinical photographs were excluded (Figure 1). All 50 patients recruited in this study had complete preoperative clinical photographs and radiographs. Preoperative clinical and radiological shoulder parameters were measured.

Figure 1.

CONSORT flow diagram illustrating the patient selection.

Clinical parameters

All patients were given detailed explanations and instructions on positioning, with written consent obtained prior. During clinical photography session, male patients were required to remove all upper-body clothing, whereas female patients were provided a white singlet to conceal the front of their bodies to preserve privacy. Patients were instructed to stand in an upright neutral position, defined as a comfortable natural standing posture, with both upper limbs relaxed by the sides and the head facing the camera. Clinical photographs were obtained against a grid background, which served as a reference to facilitate visual assessment of symmetry and alignment. The hips and knees were fully extended, with the feet positioned plantigrade and equidistant from the midline at approximately shoulder width.

A tripod-mounted camera was placed directly in front of the patient at shoulder level, aligned parallel to the floor, and placed at a standardised distance of one metre from the patient’s footwear. Standardised photographs were obtained sequentially in both front and back views under identical condition during quiet standing, without breath-holding instructions. To minimise posture-related variability, patients were instructed to remain relaxed and avoid voluntary shoulder elevation or neck tilting, and maintain a forward-facing head position during image acquisition.

Definitions of clinical parameters applied were mentioned in previous studies^6,7,9,10 (Figure 2). Positive values indicated left-side elevation. The metric for height difference was in millimetres whereas angles were measured in degrees.

(1) Front neck base angle: the angle between a horizontal reference line and the line connecting the intersections of the sternocleidomastoid muscle and trapezius muscle profiles. This was known as trapezial angle by Ono et al.⁷

(2) Back neck base angle¹⁰: the angle between a horizontal reference line and the line connecting the base of the neck.

(3) Front clavicle angle⁷: the angle between a horizontal reference line and the line connecting the upper margin of both acromial processes.

(4) Back clavicle angle: the angle between a horizontal reference line and the line through two inflection points of the shoulders and upper arms. This was known as the shoulder angle by Qiu et al.⁶

(5) Front/back trapezium angle⁹: the angle between a horizontal reference line and the line following the external border of the trapezium muscle. The left/right trapezium angle ratio was calculated.

(6) Front/back axilla angle^6,9: the angle between a horizontal reference line and the line through both axillae.

(7) Trapezial area⁷: area enclosed by the following borders:

(a) the line connecting the top margin of the acromial processes

(b) the perpendicular line to it through the intersection of the sternocleidomastoid muscle and the trapezius muscle

(Measurement from the front view.)

(For normalization purposes, the natural log of the ratio of left to right trapezial area (Ln[L/R trapezial area]) was used.)

(8) Shoulder area index 1 (SAI 1)⁶: a horizontal line (l₁) was drawn through the higher axilla, followed by the plumb line through the midpoint of neck to divide between two sides. Line m was the line connecting the inflection points between shoulder and neck. Two areas were marked, both surrounded by the superior margin of the shoulders, the outer margin of the upper arms, l₁, plumb line, and line m. Left area was identified as A1 and right was A2. The ratio A1/A2 was defined as SAI 1.

(9) Shoulder area index 2 (SAI 2)⁶: A horizontal line was drawn through the lower inflection points between the shoulder and the upper arm (l₂) followed by the plumb line through the midpoint of neck to divide between two sides. Two areas were marked, both surrounded by the superior margin of the shoulders, l₂, plumb line, and line m. Left area was identified as B1 while right was B2. The ratio B1/B2 was defined as SAI 2.

(10) Shoulder height (SH)⁶: A horizontal line through the higher axilla intersects the arms at P (left) and Q (right) was drawn, followed by the plumb line through the midpoint of neck intersecting the horizontal line (O). Two trisection lines were drawn from the horizontal line at equal space from the plumb line at both right and left sides, intersecting the shoulders at A, B (left) and A’, B’ (right).

(a) Inner shoulder height (SHi): the height difference between the inner horizontal lines (A and A’).

(b) Outer shoulder height (SHo): the height difference between the outer horizontal lines (B and B’).

Figure 2.

Photographs demonstrated all clinical parameters measured preoperatively.

Radiological parameters

All AIS patients had routine whole spine erect anteroposterior and lateral radiographs performed as part of the preoperative assessment. Similarly, radiographs were obtained with patients in an upright neutral standing posture, with both upper limbs relaxed by the sides, hips and knees fully extended, and the feet positioned plantigrade and equidistant from the midline at approximately shoulder width. Patients were instructed to face the imaging apparatus with a forward-facing neutral head position, without breath-holding, and to remain relaxed while avoiding voluntary shoulder elevation or neck tilting during radiographic acquisition. Radiological measurements were performed using Centricity™ Universal Viewer Zero Footprint, version 6.0 SP11.2.2, GE Healthcare. Preoperative proximal thoracic (PT) Cobb angle, main thoracic (MT) Cobb angle, lumbar (L) Cobb angle, and radiological shoulder parameters were measured from the available radiographs.

Preoperative radiological shoulder parameters measured are defined below.^5,7,11 Positive values signified inclination to the right side of the reference line (left side elevated) and contrariwise for negative values.

(1) T1 tilt: The angle between a horizontal reference line and the line parallel to the upper endplate of T1 vertebra (degree).

(2) First rib angle (FRA): The angle between a horizontal reference line and the line connecting both superior borders of the first ribs (degree).

(3) Clavicle angle (Cla-A): The angle between a horizontal reference line and the line connecting the most cephalad portion of both right and left clavicles (degree).

(4) Cervical axis (CA): The angle between a vertical reference line and the longitudinal axis of cervical spine (bisecting the centre of C2 odontoid process and C7) (degree).

(5) Coracoid height difference (CHD): A horizontal line was traced at the superior border of both right and left coracoid processes. The height difference was identified between the right and left lines measured (millimetres).

(6) Clavicle-rib intersection distance (CRID): The height difference between the horizontal lines at the level of intersection between the clavicle and the outer border of second rib (millimetres).

(7) Radiographic shoulder height (RSH): the difference between the horizontal lines at the level of soft tissue shadow superior to the acromioclavicular joint (millimetres).

Statistical analysis

Data analysis was conducted using SPSS Version 26.0 (SPSS Statistics for Windows, IBM Corp., Armonk, New York, USA). Continuous variables were assessed for normality using the Shapiro-Wilk test and by visual inspection of histograms. Continuous data are presented as mean and standard deviation for parametric data and as median and interquartile range (IQR) for non-parametric data. Categorical data are presented as frequency and percentage. Comparison between Lenke subgroups were performed using the independent t-test for parametric continuous data, Mann-Whitney U test for non-parametric continuous data, and Fisher’s exact test for categorical variables. Pearson correlation analysis was used for normally distributed continuous variables while Spearman correlation analysis was applied for non-normally distributed continuous variables, to assess the relationships between preoperative clinical and radiological shoulder balance parameters. In this study, we regarded r, <0.20 as very weak, 0.20 – 0.39 as weak, 0.40 – 0.59 as moderate, 0.60 – 0.79 as strong, and 0.80 – 1 as very strong correlation.^9,12 P value <0.05 was the cut-off value for statistical significance.

Results

Fifty Lenke 1 and 2 AIS patients were analysed. Majority had Lenke type 1 curve (66.0%) (Table 1). In comparison between Lenke 1 and Lenke 2 patients, significant differences were observed for front and back neck base angles, Ln [L/R trapezial area], SAI 1, SAI 2, and SHi (Table 2). In general, Lenke 2 patients exhibited higher magnitude/more positive inclination in both clinical and radiological shoulder parameters.

Table 1.

Demographic data for 50 Lenke type 1 and 2 AIS patients.

Demographic data	N = 50 [N (%)/mean ± standard deviation/median [interquartile range]]
Age (years)	15.0 [13.0 – 19.5]
Gender
Male	7 (14.0%)
Female	43 (86.0%)
Height (cm)	158 ± 7
Weight (kg)	45 [41 – 50]
Body mass index (BMI) (kg/m²)	18.2 [16.4 – 20.3]
Lenke curve type
1	33 (66.0%)
2	17 (34.0%)

Parametric variables are presented as median ± standard deviation, while non-parametric variables are reported as median [interquartile range].

Height and weight are recorded to the nearest whole unit to minimise false precision.

Table 2.

Comparison of clinical and radiological shoulder balance parameters between Lenke type 1 and 2 AIS patients.

	All (N = 50, 100.0%)	Lenke 1 (N = 33, 66.0%)	Lenke 2 (N = 17, 34.0%)	P value
Clinical parameters
Neck base angle (°)
Front	−2.9 ± 7.0	−5.5 ± 5.6	2.3 ± 6.7	<0.001^a
Back	−0.9 ± 5.9	−3.4 ± 4.1	3.8 ± 6.0	<0.001^a
Clavicle angle (°)
Front	−2.3 ± 2.5	−2.8 ± 2.4	−1.4 ± 2.5	0.071
Back	−0.7 ± 2.3	−1.0 ± 2.3	−0.3 ± 2.4	0.327
Trapezium angle ratio
Front	1.2 [1.1 – 1.4]	1.2 [1.1 – 1.5]	1.2 [1.1 – 1.3]	0.346
Back	1.2 [1.1 – 1.3]	1.2 [1.1 – 1.4]	1.2 [1.1 – 1.3]	0.935
Axilla angle (°)
Front	−1.7 ± 3.4	−2.0 ± 3.4	−0.9 ± 3.6	0.275
Back	−4.9 ± 2.6	−5.1 ± 2.6	−4.6 ± 2.6	0.457
Ln[L/R trapezial area]	−0.1 ± 0.4	−0.2 ± 0.3	0.2 ± 0.3	<0.001^a
SAI 1	0.9 ± 0.1	0.9 ± 0.1	1.0 ± 0.1	0.007^a
SAI 2	0.9 ± 0.2	0.9 ± 0.1	1.0 ± 0.2	0.002^a
Shoulder height (SH) (mm)
Inner (SHi)	0.0 [−7.4 – 7.6]	0.0 [−10.4 – 4.5]	7.5 [4.8 – 10.5]	<0.001^a
Outer (SHo)	−4.5 [−14.6 – 0.0]	−7.0 [−16.3 – 0.0]	0.0 [−7.7 – 3.1]	0.100
Radiological parameters
Proximal thoracic cobb angle (°)	35 ± 13	27 ± 7	49 ± 9	<0.001^a
Main thoracic cobb angle (°)	69 ± 13	66 ± 12	74 ± 13	0.040^a
Lumbar cobb angle (°)	40 ± 13	42 ± 13	38 ± 11	0.299
T1 tilt (°)	−0.3 ± 8.2	−4.0 ± 5.7	6.8 ± 7.8	<0.001^a
FRA (°)	−2.0 ± 6.4	−4.7 ± 4.6	3.3 ± 6.1	<0.001^a
CA (°)	2.3 ± 4.1	0.8 ± 3.3	5.2 ± 4.0	<0.001^a
Cla-A (°)	−1.8 ± 2.7	−2.7 ± 2.4	0.0 ± 2.5	0.001^a
CHD (mm)	−4.6 ± 12.3	−8.4 ± 10.9	2.9 ± 11.5	0.001^a
CRID (mm)	−0.4 ± 0.8	−0.7 ± 0.8	0.1 ± 0.7	<0.001^a
RSH (mm)	−11.5 ± 15.9	−15.9 ± 14.8	−2.9 ± 14.7	0.005^a

FRA: First rib angle; Cla-A: Clavicle angle; CA: Cervical axis; CHD: Coracoid height difference; CRID: Clavicle-rib intersection distance; RSH: Radiographic shoulder height; Ln[L/R trapezial area]: natural log of the ratio of left to right trapezial area; SAI 1: Shoulder area index 1; SAI 2: Shoulder area index 2; SHi: Inner shoulder height; SHo: Outer shoulder height.

Parametric variables are presented as median ± standard deviation, while non-parametric variables are reported as median [interquartile range].

Cobb angles are recorded to the nearest whole unit to minimise false precision.

^aStatistically significant.

In the correlation analysis for radiological shoulder parameters, T1 tilt and FRA exhibited very strong correlations with each other (r = 0.938) (p < 0.001) (Table 3). There were strong correlations identified between CA, T1 tilt (r = 0.669) and FRA (r = 0.634) (p < 0.001). These parameters were indicative of neck tilt and MSB. Meanwhile, there were very strong correlations between Cla-A, CHD, CRID, and RSH (r ≥ 0.8), in which these parameters were suggestive of LSB (p < 0.001) (Table 3).

Table 3.

Pearson correlation analysis for radiological shoulder balance parameters.

	T1 tilt	FRA	CA	Cla-A	CHD	CRID	RSH
T1 tilt
r	1	0.938	0.669	0.614	0.529	0.618	0.545
P value	-	<0.001 ^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	-	0.890-0.970	0.441-0.800	0.400-0.782	0.287-0.724	0.438-0.764	0.337-0.729
FRA
r	0.938	1	0.634	0.670	0.613	0.675	0.622
P value	<0.001 ^a	-	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.890-0.970	-	0.430-0.777	0.491-0.816	0.405-0.774	0.493-0.810	0.440-0.773
CA
r	0.669	0.634	1	0.579	0.545	0.523	0.531
P value	<0.001^a	<0.001^a	-	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.441-0.800	0.430-0.777	-	0.387-0.714	0.315-0.710	0.308-0.681	0.334-0.673
Cla-A
r	0.614	0.670	0.579	1	0.941	0.887	0.948
P value	<0.001^a	<0.001^a	<0.001^a	-	<0.001 ^a	<0.001 ^a	<0.001 ^a
95% CI	0.400-0.782	0.491-0.816	0.387-0.714	-	0.903-0.965	0.812-0.939	0.901-0.974
CHD
r	0.529	0.613	0.545	0.941	1	0.905	0.959
P value	<0.001^a	<0.001^a	<0.001^a	<0.001 ^a	-	<0.001 ^a	<0.001 ^a
95% CI	0.287-0.724	0.405-0.774	0.315-0.710	0.903-0.965	-	0.825-0.951	0.924-0.980
CRID
r	0.618	0.675	0.523	0.887	0.905	1	0.910
P value	<0.001^a	<0.001^a	<0.001^a	<0.001 ^a	<0.001 ^a	-	<0.001 ^a
95% CI	0.438-0.764	0.493-0.810	0.308-0.681	0.812-0.939	0.825-0.951	-	0.836-0.951
RSH
r	0.545	0.622	0.531	0.948	0.959	0.910	1
P value	<0.001^a	<0.001^a	<0.001^a	<0.001 ^a	<0.001 ^a	<0.001 ^a	-
95% CI	0.337-0.729	0.440-0.773	0.334-0.673	0.901-0.974	0.924-0.980	0.836-0.951	-

FRA: First rib angle; Cla-A: Clavicle angle; CA: Cervical axis; CHD: Coracoid height difference; CRID: Clavicle-rib intersection distance; RSH: Radiographic shoulder height.

r: correlation coefficient.

95% CI: 95% confidence interval.

Values with bold fonts indicated variables with very strong correlations, r ≥ 0.8.

^aStatistically significant.

The correlation analysis between preoperative clinical and radiological shoulder parameters was summarised in Table 4. Front and back neck base angles, SAI 2, and SHi had strong correlations (r ≥ 0.6) with radiological MSB with neck tilt parameters (T1 tilt, FRA, and CA) whereas Ln [L/R trapezial area] and SAI 1 had strong correlations with only T1 tilt and FRA (p < 0.001). Interestingly, front and back neck base angles had the highest correlation with T1 tilt (r = 0.831; r = 0.818) and FRA (r = 0.839; r = 0.840) (p < 0.001). The back neck base angle had strong correlations with all radiological shoulder balance parameters except for CHD (r = 0.573). SAI 2 had strong correlations with all radiological shoulder parameters. CA, which represents neck tilt, has a strong correlation (r ≥ 0.6) with front and back neck base angles, SAI 2, and SHi.

Table 4.

Correlation analysis between clinical and radiological shoulder balance parameters.

Clinical Shoulder Balance Parameters	Radiological Shoulder Balance Parameters
Clinical Shoulder Balance Parameters	T1 tilt	FRA	CA	Cla-A	CHD	CRID	RSH
Front measurements
Front neck base angle
r	0.831	0.839	0.671*	0.577	0.528	0.596	0.577
P value	<0.001 ^a	<0.001 ^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.729-0.897	0.754-0.902	0.440-0.795	0.389-0.730	0.304-0.717	0.389-0.753	0.385-0.731
Front clavicle angle
r	0.458	0.497	0.459	0.527	0.540	0.536	0.584
P value	0.001^a	<0.001^a	0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.198-0.650	0.282-0.668	0.184-0.647	0.287-0.722	0.284-0.746	0.296-0.729	0.350-0.751
Front trapezium angle ratio#
r	−0.286	−0.267	−0.335	−0.435	−0.443	−0.464	−0.483
P value	0.044^a	0.061	0.017^a	0.002^a	0.001	<0.001^a	<0.001^a
95% CI	−0.525 to −0.032	−0.490 to −0.022	−0.563 to −0.067	−0.652 to −0.197	−0.660 to −0.163	−0.657 to −0.231	−0.692 to −0.236
Front axilla angle
r	0.429	0.503	0.258	0.464	0.409	0.440	0.485
P value	0.002^a	<0.001^a	0.070	0.001^a	0.003^a	0.001^a	<0.001^a
95% CI	0.183-0.606	0.292-0.665	−0.046 to −0.536	0.166-0.719	0.082-0.694	0.152-0.686	0.190-0.716
Ln [L/R trapezial area]
r	0.708*	0.701*	0.575	0.453	0.409	0.488	0.403
P value	<0.001^a	<0.001^a	<0.001^a	0.001^a	0.003^a	<0.001^a	0.004^a
95% CI	0.493-0.848	0.485-0.841	0.332-0.741	0.213-0.640	0.167-0.603	0.299-0.644	0.161-0.607
Back measurements
Back neck base angle
r	0.818	0.840	0.715*	0.611*	0.573	0.612*	0.612*
P value	<0.001 ^a	<0.001 ^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.699-0.900	0.750-0.917	0.474-0.839	0.417-0.766	0.323-0.763	0.413-0.773	0.419-0.781
Back clavicle angle
r	0.381	0.447	0.395	0.635*	0.629*	0.617*	0.709*
P value	0.006^a	0.001^a	0.005^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.104-0.614	0.186-0.657	0.142-0.593	0.394-0.809	0.385-0.815	0.386-0.793	0.504-0.848
Back trapezium angle ratio#
r	−0.218	−0.250	−0.156	−0.548	−0.528	−0.474	−0.591
P value	0.129	0.080	0.280	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	−0.497-0.099	−0.551-0.083	−0.458-0.155	−0.775 to −0.237	−0.754 to −0.206	−0.694 to −0.172	−0.764 to −0.309
Back axilla angle
r	0.429	0.396	0.203	0.512	0.498	0.551	0.569
P value	0.002^a	0.004^a	0.157	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.218-0.620	0.153-0.610	−0.091-0.438	0.255-0.724	0.238-0.700	0.301-0.746	0.331-0.754
SAI 1
r	0.672*	0.675*	0.565	0.553	0.499	0.537	0.559
P value	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.413-0.823	0.380-0.849	0.233-0.784	0.186-0.785	0.117-0.779	0.176-0.783	0.237-0.787
SAI 2
r	0.704*	0.758*	0.652*	0.676*	0.645*	0.680*	0.694*
P value	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.511-0.840	0.644-0.853	0.416-0.804	0.475-0.814	0.405-0.824	0.456-0.825	0.513-0.816
SHi#
r	0.694*	0.736*	0.643*	0.564	0.538	0.582	0.558
P value	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.456-0.866	0.498-0.885	0.419-0.793	0.328-0.740	0.309-0.722	0.329-0.764	0.301-0.744
SHo#
r	0.409	0.479	0.347	0.588	0.588	0.617*	0.630*
P value	0.003^a	<0.001^a	0.014^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
95% CI	0.121-0.660	0.229-0.677	0.070-0.598	0.369-0.740	0.377-0.752	0.401-0.767	0.426-0.769

r: correlation coefficient.

95% CI: 95% confidence interval.

Correlation coefficients labeled (*) indicate strong correlations (r ≥ 0.60 or ≤ −0.60), whereas bolded values indicate very strong correlations (r ≥ 0.80 or ≤ -0.80).

^aStatistically significant; #: non-normally distributed variables analysed using Spearman correlation analysis.

On the contrary, only the back clavicle angle demonstrated a strong correlation with all LSB parameters, i.e., RSH (r = 0.709), Cla-A (r = 0.635), CHD (r = 0.629), and CRID (r = 0.617) (p < 0.001). SHo correlated strongly with CRID (r = 0.617) and RSH (r = 0.630) (p < 0.001). Meanwhile, front clavicle angle, front/back trapezium angle ratio, front/back axilla angle, did not have any strong correlation with MSB with neck tilt or LSB parameters (r < 0.6).

Discussion

In the recent years of research, shoulder balance has been evaluated as two distinct entities (MSB and LSB) whereas neck tilt as represented by CA was another constituent.^{1,3,7,8,10,13,14} Ono et al.⁷ reported that clinical MSI (trapezial angle and Ln[L/R trapezial area]) correlated well with T1 tilt, FRA, and upper thoracic curve size while lateral shoulder imbalance (LSI) exhibited discrepancy to T1 tilt. Kuklo et al.⁵ proposed Cla-A as a radiographic predictor for clinical shoulder balance as represented by RSH. Similarly, Kwan et al.⁸ also reported significant correlations between clinical shoulder height, which represents LSB, and the radiological parameters (Cla-A, CHD, CRID, and RSH). The authors elaborated on clinical neck tilt (with/without MSI) and deduced that it correlated well with CA and T1 tilt.⁸ Our results further affirmed that MSB with neck tilt and LSB were represented by different radiological parameters.

The results in this study supported that MSB, as represented radiologically by T1 tilt and FRA, had very strong correlations (r ≥ 0.8). This is consistent with the findings reported by Ono et al.⁷ Furthermore, T1 tilt and FRA (MSB) correlated strongly with clinical parameters, i.e., front and back neck base angles, Ln [L/R trapezial area], SAI 1, SAI 2, and SHi. Interestingly, front and back neck base angles demonstrated very strong correlations (r ≥ 0.8) with T1 tilt and FRA, and they may be the best clinical indicators for MSB. However, back neck base angle and SAI 2 also demonstrated significant correlations with other radiological LSB parameters, suggesting that these parameters may lack specificity. In conjunction with other studies, our findings demonstrated a very strong correlation (r ≥ 0.8) between CHD, Cla-A, CRID, and RSH, which were indicative of LSB.^8,11,15 We identified that the back clavicle angle demonstrated a strong correlation to radiological LSB, especially RSH (r = 0.709).

The feasibility of the shoulder balance measurements evaluated in this study supports their use in routine clinical assessment. Front-view evaluation appears to be more appropriate for assessing MSB. Based on our findings, the front neck base angle demonstrated stronger correlations with MSB-related parameters (r = 0.831 with T1 tilt; r = 0.839 with FRA), and may therefore better represent MSB. This parameter was previously described as trapezial angle by Ono et al., and was reported to have excellent intra-observer reliability (intraclass correlation coefficient (ICC) = 0.997).⁷ In contrast, LSB appeared to be better represented by posterior clinical assessment, with the back clavicle angle showing stronger associations with LSB (CHD, Cla-A, CRID, and RSH). This parameter has demonstrated good inter-observer and intra-observer reliability (ICC = 0.80 and 0.88, respectively), as reported by Matamalas et al.⁹ Collectively, these findings highlight the complementary roles of anterior and posterior assessments in characterizing shoulder balance. From a surgical planning perspective, UIV selection is commonly influenced by preoperative shoulder tilting, and previous studies have suggested that an elevated left shoulder may warrant a higher fusion level.^16,17 Consequently, future studies investigating the targeted assessment of the MSB and LSB, establishing clinically pertinent thresholds for imbalance, may further facilitate the selection of UIV, potentially mitigating postoperative shoulder imbalance.

Nevertheless, our results were contrary to the findings of previous reports. We identified that most of the clinical shoulder parameters had moderate to strong correlations with radiological shoulder parameters for both MSB and LSB. A study of 80 idiopathic scoliosis patients by Matamalas et al.,⁹ reported weak to moderate correlations (r < 0.6) between clinical and radiological parameters despite being statistically significant. Qiu et al.⁶ highlighted that none of the radiological shoulder parameters could precisely reflect the clinical parameters. However, the cut-off value applied in their study for a strong correlation coefficient was 0.8, which was higher than the common references. Yang et al.¹⁸ evaluated the clinical and radiological shoulder balance of 74 Lenke 1 and 2 AIS patients. They observed that none of the radiological measures demonstrated a strong correlation (r > 0.8) with clinical shoulder balance.¹⁸

Previous studies had investigated the effects of neck and shoulder imbalance on patients’ perception of clinical outcomes and deduced that it was associated with patients’ dissatisfaction.^19,20 Postoperative shoulder imbalance not only led to patients’ dissatisfaction but might lead to possible complications, e.g., distal adding-on, lumbar curve progression, or postoperative truncal shift.⁴ Therefore, recognising the potential clinical and radiological predictive factors is essential to avoid postoperative shoulder imbalance. For instance, preoperative shoulder balance, skeletal maturity, MT curve flexibility, UIV level selection, or UIV tilt angle, should be considered in the preoperative planning.^21–24 Therefore, our findings could be applied as a guide in the preoperative measurement of clinical and radiological MSB with neck tilt and LSB.

This study solely concentrated on the preoperative data. We acknowledged that it was one main limitations in our study. Future studies could incorporate postoperative and long-term follow-up to further investigate the progression of neck and shoulder balance at different stages. Another limitation would be the discrepancy in the patients’ positioning during the clinical photography and radiography sessions. Patients might position themselves differently in both settings. A further limitation is the absence of a study-specific intra-observer and inter-observer reliability assessment using intraclass correlation coefficients. However, all clinical shoulder balance parameters reported in this study were derived from the previously published methodologies. Nevertheless, a previous study by Chiu et al.²⁵ concluded that T1 tilt had good reliability, CA, CRID and CHD had moderate reliability, whereas RSH and Cla-A had poor reliability.

Conclusion

MSB, as represented by T1 tilt and FRA, demonstrated very strong correlations with the front and back neck base angles (r ≥ 0.8), and strong correlations with Ln [L/R trapezial area], SAI 1, SAI 2, and SHi. Neck tilt, as represented by CA, correlated strongly with the front and back neck base angles, SAI 2, and SHi. LSB, as represented by Cla-A, CHD, CRID, and RSH, demonstrated a strong correlation with the back clavicle angle.

Footnotes

Acknowledgements

The authors would like to acknowledge Universiti Malaya Medical Centre for their invaluable support and assistance, which has contributed to the successful completion of this work.

ORCID iDs

Mun Keong Kwan

Sin Ying Lee

Woon Theng Heng

Chee Kidd Chiu

Chris Yin Wei Chan

Ethical considerations

This retrospective review was approved by our institutional review boards (MREC ID No.: 202214-10890).

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Mun Keong Kwan, Sin Ying Lee, and Woon Theng Heng. The first draft of the manuscript was written by Mun Keong Kwan, Sin Ying Lee, and Woon Theng Heng, Chee Kid Chiu, and Chris Yin Wei Chan. All authors commented on and reviewed previous versions of the manuscript. Mun Keong Kwan was responsible for conceptualization, methodology, supervision, and providing resources for the study. All authors read and approved the final manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Deng

Wang

, et al. Incidence and risk factors of postoperative medial shoulder imbalance in Lenke Type 2 adolescent idiopathic scoliosis with lateral shoulder balance. BMC Muscoskelet Disord 2022; 23(1): 947, Epub 2022/11/04.

Zhang

Feng

, et al. Incidence and risk factors for postoperative shoulder imbalance in scoliosis: a systematic review and meta-analysis. Eur Spine J 2018; 27(2): 358–369, Epub 2017/09/11.

Zhao

Yang

, et al. (eds) Cervical axis tilt phenomenon in Lenke type 1 and 2 adolescent idiopathic scoliosis. Int J Clin Exp Med, 2018, 11(5), pp. 5005–5011.

Nikouei

Ghandhari

Ameri

, et al. Shoulder imbalance in adolescent idiopathic scoliosis: a systematic review of the current state of the art. Arch Bone Jt Surg 2022; 10(12): 992–1003, Epub 2023/02/02.

Kuklo

Lenke

Graham

, et al. Correlation of radiographic, clinical, and patient assessment of shoulder balance following fusion versus nonfusion of the proximal thoracic curve in adolescent idiopathic scoliosis. Spine 2002; 27(18): 2013–2020, Epub 2003/03/14.

Qiu

, et al. Discrepancy between radiographic shoulder balance and cosmetic shoulder balance in adolescent idiopathic scoliosis patients with double thoracic curve. Eur Spine J 2009; 18(1): 45–51, Epub 2008/12/02.

Ono

Bastrom

Newton

. Defining 2 components of shoulder imbalance: clavicle tilt and trapezial prominence. Spine 2012; 37(24): E1511–E1516, Epub 2012/08/17.

Kwan

Wong

Lee

, et al. Is neck tilt and shoulder imbalance the same phenomenon? A prospective analysis of 89 adolescent idiopathic scoliosis patients (Lenke type 1 and 2). Eur Spine J 2016; 25(2): 401–408, Epub 2015/05/13.

Matamalas

Bago

D’Agata

, et al. Reliability and validity study of measurements on digital photography to evaluate shoulder balance in idiopathic scoliosis. Scoliosis 2014; 9(1): 23, Epub 2014/12/19.

10.

Menon

Pillay

, et al. Post-operative shoulder imbalance in adolescent idiopathic scoliosis: a study of clinical photographs. Scoliosis 2015; 10: 31, Epub 2015/11/20.

11.

Akel

Pekmezci

Hayran

, et al. Evaluation of shoulder balance in the normal adolescent population and its correlation with radiological parameters. Eur Spine J 2008; 17(3): 348–354, Epub 2007/11/21.

12.

Wechsler

. Statistics at Square One. Ninth Edition, revised by M. J. Campbell, T. D. V. Swinscow, BMJ Publ. Group, London, 1996. No. of pages: 140. Price: £11. ISBN 0‐7279‐0916‐9. BMJ Publ. Group, 1996. No. of pages: 140. ISBN 0‐7279‐0916‐9. Statistics in Medicine. 1997;16:2629–30.

13.

Yang

, et al. Postoperative neck tilt in Lenke 1 and 2 AIS patients after correction surgery: a novel predictive index. BMC Muscoskelet Disord 2019; 20(1): 405, Epub 2019/09/06.

14.

Jiang

Qian

Qiu

, et al. Full fusion of proximal thoracic curve helps to prevent postoperative cervical tilt in Lenke type 2 adolescent idiopathic scoliosis patients with right-elevated shoulder. BMC Muscoskelet Disord 2017; 18(1): 362, Epub 2017/08/25.

15.

Yang

, et al.

Shoulder balance in Lenke type 2 adolescent idiopathic scoliosis: should we fuse to the second thoracic vertebra?

Clin Neurol Neurosurg 2017; 163: 156–162, Epub 2017/11/06.

16.

Lee

Denis

Winter

, et al. Analysis of the upper thoracic curve in surgically treated idiopathic scoliosis. A new concept of the double thoracic curve pattern. Spine 1993; 18(12): 1599–1608, Epub 1993/09/15.

17.

Suk

Kim

Lee

, et al. Indications of proximal thoracic curve fusion in thoracic adolescent idiopathic scoliosis: recognition and treatment of double thoracic curve pattern in adolescent idiopathic scoliosis treated with segmental instrumentation. Spine 2000; 25(18): 2342–2349. https://doi.org/10.1097/00007632-200009150-00012

18.

Yang

Feuchtbaum

Werner

, et al.

Does anterior shoulder balance in adolescent idiopathic scoliosis correlate with posterior shoulder balance clinically and radiographically?

Eur Spine J 2012; 21(10): 1978–1983, Epub 2012/07/31.

19.

Lee

Ch’ng

Wong

, et al. Patients’ perception and satisfaction on neck and shoulder imbalance in adolescent idiopathic scoliosis. Glob Spine J 2023; 13(3): 752–763, Epub 2021/04/08.

20.

Smyrnis

Sekouris

Papadopoulos

. Surgical assessment of the proximal thoracic curve in adolescent idiopathic scoliosis. Eur Spine J 2009; 18(4): 522–530, Epub 2009/02/17.

21.

Hong

Suh

Modi

, et al. Analysis of factors that affect shoulder balance after correction surgery in scoliosis: a global analysis of all the curvature types. Eur Spine J 2013; 22(6): 1273–1285, Epub 2013/03/05.

22.

Lee

Hwang

Lee

, et al.

Does fusion to T2 compared with T3/T4 lead to improved shoulder balance in adolescent idiopathic scoliosis with a double thoracic curve?

J Pediatr Orthop B 2019; 28(1): 32–39, Epub 2018/06/01.

23.

Kwan

Lee

Ch’ng

, et al. A positive (+ve) postoperative upper instrumented vertebra tilt angle (≥0°) significantly increases the risk of medial shoulder and neck imbalance in Lenke 1 and 2 adolescent idiopathic scoliosis patients. Spine 2020; 45(12): E694–e703, Epub 2020/02/08.

24.

Isogai

Yagi

Otomo

, et al. Upper end vertebra of proximal thoracic curve at T1 is a novel risk factor of postoperative shoulder imbalance in Lenke type 2 adolescent idiopathic scoliosis. Glob Spine J 2021; 13(5): 1223–1229. https://doi.org/10.1177/21925682211023049

25.

Chiu

Chan

CYW

Tan

, et al. Conformity and changes in the radiological neck and shoulder balance parameters throughout 3-Year Follow-up period: do they remain the same? Spine. 2020;45(6):E319. Epub 2019/10/09.