Can intraoperative T1 tilt predict postoperative shoulder imbalance in patients with Lenke 2 adolescent idiopathic scoliosis?: a retrospective study in Japan

Ethical statements

This retrospective, cohort study was approved by the Institutional Review Board (IRB) of Osaka City General Hospital (IRB approval number: 1912112). All study participants provided informed consent.

Study design and patient population

We analyzed prospectively collected data from patients with Lenke type 2 AIS who underwent posterior corrective fusion surgery at Osaka City General Hospital. Only patients with a minimum postoperative follow-up of 2 years were included in the study.

Surgical techniques

All patients underwent posterior spinal fusion using an all-pedicle screw construct. Curve correction was achieved through a combination of derotation, translation, direct vertebral rotation, and in-situ bending. For Lenke type 2 curves, T2 was generally selected as the UIV. However, in cases where the main thoracic (MT) curve apex was located more caudally, T3 was occasionally chosen as the UIV. Posterior column osteotomies were routinely performed as grade 1 osteotomies at all instrumented levels. In patients with reduced thoracic kyphosis (modifier –) or limited curve flexibility, a grade 2 osteotomy was additionally performed around the apex of the MT curve.

Radiographic data acquisition

Radiological assessments were performed at three time points: preoperatively, 1 week postoperatively, and at the final follow-up. The evaluated parameters included the Cobb angles of the PT, MT, and lumbar curves; the lateral displacement between the C7 plumb line and the central sacral vertical line (C7–CSVL); radiographic shoulder height (RSH); and apical vertebral translation of the MT curve (TAVT). T1 tilt was defined as the angle between the superior endplate of T1 and a horizontal reference line (Fig. 1A). A positive value indicated an upward tilt to the left, and a negative value indicated an upward tilt to the right. This convention was applied because the PT curve was left-sided in nearly all patients with Lenke type 2 AIS.

Fig. 1

(A) T1 tilt represented an angle between the horizontal line and the line through the upper endplate of T1. A positive value was adopted when the highest side was left. (B) Intraoperative T1 tilt and upper instrumented vertebra (UIV) tilt were defined as an angle between the baseline and the line through the upper endplate of T1 or UIV. A positive value was adopted when the highest side was left.

Risser grades were recorded preoperatively, and spinal flexibility was assessed using side-bending radiographs. Curve flexibility was quantified as the bending correction ratios of the PT, MT, and lumbar curves, denoted as the PT bending correction ratio, MT bending correction ratio, and lumbar bending correction ratio, respectively, on the preoperative radiographs. The ratio of curve magnitudes (PT/MT) and their corresponding correction rates were also calculated.

Additionally, intraoperative radiographs were obtained before and after correction to measure T1 tilt (iT1 tilt) and upper instrumented vertebra tilt (iUIV tilt) (Fig. 1B). Specifically, portable radiographs were obtained immediately before the start of surgery (prior to skin incision) and after rod insertion and correction.

Changes in T1 tilt (ΔT1 tilt) and UIV tilt (ΔUIV tilt) were calculated between the preoperative and 1-week postoperative standing radiographs. Similarly, changes in intraoperative tilt (ΔiT1 tilt and ΔiUIV tilt) were assessed between pre- and post-correction intraoperative prone radiographs. Finally, associations were examined between changes in radiographic shoulder height (ΔRSH) and changes in disc angulation immediately below the lower instrumented vertebra (LIV) (ΔLIV+1 disc wedge), both defined as the difference between early postoperative values and those at the final follow-up.

All radiographic measurements were independently performed by two spine surgeons, with 7 and 26 years of experience, respectively. The mean of their measurements was used for analysis. Image evaluation was conducted using Synapse OB-V software (Fuji Film Co., Tokyo, Japan).

Definition of postoperative shoulder imbalance

PSI was defined as an RSH greater than 15 mm on standing postoperative anteroposterior radiographs, in accordance with previous reports [9].

Definition of distal adding-on

DA was defined according to the criteria of Wang et al. [10], as either a >5 mm increase in deviation of the first vertebra below the instrumentation from the CSVL and/or a >5° increase in disc angulation at the same level.

Patient-reported outcomes

Functional outcomes were assessed using the Japanese version of the 22-item Scoliosis Research Society Outcomes Questionnaire (SRS-22), administered preoperatively and at the final follow-up.

Statistical analysis

Patients were categorized into two groups according to the presence or absence of PSI. Comparative analyses were performed to assess differences in demographic characteristics, preoperative and intraoperative radiographic parameters, and postoperative outcomes. Continuous variables were analyzed using the Mann-Whitney U test, and categorical variables were analyzed using the chi-square test or Fisher’s exact test, as appropriate. Correlation analyses were conducted to evaluate relationships between intraoperative tilt changes (ΔiT1 and ΔiUIV) and postoperative changes in RSH (ΔRSH) and clavicle angle (ΔCA). Statistical analyses were performed using R software ver. 3.5.1 (patched; The R Foundation, Vienna, Austria), with statistical significance defined as p<0.05.

Patient characteristics

A total of 50 patients (45 females and five males) were included. The mean age at surgery was 18.0±6.7 years, and the mean follow-up duration was 25.7±14.5 months. The UIV was T2 in 44 patients and T3 in six patients. Pedicle screws were used as anchors at the UIV in all patients. The distribution of the LIV was as follows: T11 (n=2), T12 (n=7), L1 (n=9), L2 (n=17), and L3 (n=15).

Overall outcomes

Of the 50 patients, 12 (24.0%) developed PSI. In most cases, PSI improved over time, with only five patients (10.0%) exhibiting persistent PSI at the final follow-up. None of the patients required revision surgery.

Univariate comparisons

Preoperative factors

There were no significant differences between the PSI and non-PSI groups in terms of demographic characteristics or baseline radiographic parameters (Table 1).

Table 1

Demographic, surgical data, and preoperative radiographic parameters (PSI group vs. non-PSI group)^a)

Intraoperative parameters

Post-correction iUIV tilt was significantly greater in the PSI group compared to the non-PSI group (5.17°±3.60° vs. 1.58°±3.86°, p=0.011), while none of the other intraoperative parameters reached statistical significance (Table 2).

Table 2

Intraoperative, postoperative, and final follow-up radiographic parameters (PSI group vs. non-PSI group)^a)

Patient-reported outcomes

SRS-22 domain scores at the final follow-up revealed no statistically significant between-group differences. The function score approached significance (4.05±0.22 vs. 4.60±0.32, p=0.051), whereas other domains (pain, self-image, mental health, satisfaction, and subtotal score) were comparable between groups (Fig. 2).

Fig. 2

(A–E) 22-item Scoliosis Research Society Outcomes Questionnaire scores (postoperative shoulder imbalance [PSI] group vs. non-PSI group).

Correlation analyses

A moderate to strong correlation was observed between intraoperative tilt changes and postoperative outcomes. ΔRSH was significantly correlated with both ΔiT1 tilt (R=0.613, p<0.001) and ΔiUIV tilt (R=0.435, p=0.002). Similarly, ΔCA demonstrated significant correlations with ΔiT1 tilt (R=0.453, p=0.002) and ΔiUIV tilt (R=0.383, p=0.007) (Fig. 3). No significant correlation was observed between ΔRSH and ΔLIV+1 disc wedge (R=0.051, p=0.721).

Fig. 3

(A–D) Linear regression of ΔRSH, ΔCA, versus iΔT1 tilt and iΔUIV tilt. RSH, radiographic shoulder height; CA, clavicle angle; UIV, upper instrumented vertebra.

In addition, strong associations were noted between intraoperative and postoperative tilt changes, including between ΔT1 and ΔiT1 (R=0.619, p<0.001) and between ΔUIV and ΔiUIV (R=0.604, p<0.001) (Fig. 4).

Fig. 4

(A) T1 tilt and intraoperative T1 tilt change. (B) Upper instrumented vertebra (UIV) tilt and intraoperative UIV tilt change. (C) Changes before and after surgery in radiographic shoulder heigh (RSH) and clavicle angle (CA).

Within the PSI subgroup, ΔRSH was strongly correlated with ΔiT1 tilt (R=0.678, p=0.015), whereas the correlation between ΔRSH and ΔiUIV tilt did not reach significance (R=0.481, p=0.114) (Table 3).

Table 3

Correlation between the different variables and ΔRSH in PSI group (sub-group analysis)^a)

Summary of key findings

This retrospective study examined intraoperative risk factors for PSI in patients with Lenke type 2 AIS. The incidence of PSI was 24.0% at 1 week postoperatively and decreased to 10.0% at the final follow-up, consistent with previous reports of spontaneous improvement. Notably, intraoperative changes in T1 tilt (ΔiT1) demonstrated a strong association with radiographic indicators of shoulder balance, highlighting ΔiT1 as a potentially modifiable intraoperative factor with predictive value for PSI.

Incidence and clinical impact of PSI

The primary aim of this study was to identify modifiable intraoperative risk factors for PSI in patients with Lenke type 2 AIS. In our cohort, the declining trend of the incidence of PSI from 1 week postoperatively to the final follow-up is consistent with previous reports describing spontaneous improvement in PSI over time, with overall incidence rates ranging from 11% to 40% [10]. Although most cases resolved without intervention, DA occurred in five patients, three in the PSI group and two in the non-PSI group, with a significantly higher incidence in the PSI group (25% vs. 5.3%, p=0.046). A representative case of a 16-year-old girl who developed DA following early PSI is presented in Fig. 5.

Fig. 5

A 16-year-old girl with postoperative adding-on. The shoulder balance was improved at the final follow-up. (A) Before surgery, (B) immediately after surgery, and (C) at final follow-up. In this case, the preoperative radiographic shoulder heigh (RSH) was 0 mm. Posterior spinal fusion from T2 to T12 were performed, and immediately after surgery, RSH was 25 mm, indicating the presence of postoperative shoulder imbalance (PSI). At final follow-up, RSH improved to 5 mm, but distal adding-on has occurred (arrow).

None of the patients required revision surgery, but PSI has been associated with unfavorable sequelae such as DA, progression of new lumbar curves, and trunk shift [10–12]. Furthermore, the PSI group exhibited significantly greater coronal imbalance, measured by C7–CSVL deviation, at the final follow-up, suggesting that PSI may contribute to long-term malalignment. These findings support the view that proactive prevention of PSI is essential to optimize surgical outcomes in patients with AIS.

Effect on patient-reported outcomes

Previous studies have linked PSI with dissatisfaction among AIS patients and their families [13,14]. In contrast, our analysis demonstrated no statistically significant differences in SRS-22 domain scores, including self-image, between the PSI and non-PSI groups at the final follow-up. Nevertheless, self-perception is multifactorial, and given the predominance of female patients in AIS populations, surgeons should remain attentive to potential cosmetic concerns even when PSI appears mild on radiographs.

Spontaneous improvement of shoulder balance

Our findings support previous reports that shoulder balance often improves over time as a result of muscle adaptation and relief of postoperative discomfort [15]. In particular, RSH significantly decreased between the early postoperative period and the final follow-up. Prior studies have shown that spontaneous improvement in clavicle height difference and CA occurs in approximately 90% of cases within 7–12 months, with continued improvement up to 24 months [16]. In our cohort, similar recovery patterns likely reflected muscular adjustments around the scapula or compensatory curve changes, such as DA or the development of new lumbar curve.

T1 Tilt as a modifiable intraoperative factor

Our prior research identified postoperative ΔT1 tilt as an independent predictor of PSI [8]. Building on that observation, the present study demonstrated that intraoperative changes in T1 tilt (ΔiT1) were strongly correlated with changes in both RSH and CA, highlighting ΔiT1 as a modifiable intraoperative factor with predictive value. Minimizing ΔiT1 during surgery may therefore reduce the risk of developing PSI.

Previous studies have linked T1 tilt to medial shoulder imbalance, characterized by trapezial asymmetry and upper rib elevation [17], whereas lateral shoulder imbalance (LSI) is more closely related to RSH and CA. However, as noted by Ono et al. [18], the correlation between T1 tilt and LSI is relatively weak. These observations imply that PSI is multifactorial, yet our results emphasize that real-time intraoperative control of T1 tilt is a practical strategy for prevention.

Kwan et al. [19] proposed the crossbar technique as a reliable method for intraoperative assessment of UIV tilt, enabling prediction and control of T1 tilt intraoperatively. Consistent with their findings, our study supports the integration of intraoperative T1 tilt monitoring into surgical workflows as a preventive measure against PSI in Lenke type 2 AIS.

Clinical significance and surgical strategies

Our findings highlight the clinical significance of intraoperative monitoring and adjustment of T1 tilt. In our surgical approach to Lenke type 2 AIS, we aimed for maximum correction of the PT curve, particularly in relatively rigid cases, through the use of a grade 2 osteotomy. When intraoperative radiographs indicated a positive ΔT1 tilt, in-situ rod bending combined with convex-side compression and concave-side distraction was applied. This approach may have contributed to improved PT curve correction, which could explain why the PT curve/MT curve ratio was not identified as a significant factor in this study.

Practical strategies when ΔT1 tilt cannot be controlled

When ΔiT1 could not be adequately minimized intraoperatively and PSI was anticipated, our surgical strategy prioritized achieving shoulder symmetry rather than maximal coronal correction. To this end, we intensified PT curve correction (including grade 2 osteotomy when necessary), fine-tuned UIV tilt through in-situ bending with targeted compression/distraction, and selectively moderated MT correction to avoid relative overcorrection. Although this approach may result in slightly reduced curve correction, it can help prevent persistent PSI and improve long-term balance. Notably, we do not extend fusion distally solely to address anticipated PSI, as PSI is primarily influenced by proximal alignment. Distal extension, which increases fusion length and surgical morbidity, is reserved only for unequivocal indications such as structural DA, progressive LIV+1 disc wedging, or global decompensation.

Limitations

Some limitations of this study should be acknowledged. As a retrospective single-center study, the potential for selection bias cannot be excluded. The relatively small number of patients in the PSI group may have reduced the statistical power of the analyses, and the findings in this group should therefore be interpreted with caution.

Variability in LIV selection, ranging from T11 to L3, may also have influenced the incidence of PSI through differences in lumbar modifier classification. In addition, cases with a “directional mismatch” between T1 tilt and elevated shoulder (e.g., left T1 tilt with right shoulder elevation) were not encountered in this cohort and thus were not evaluated.