A meta-analysis based on 26 studies showed that the pooled incidence of PSI was 25% (95% confidence interval: 20–31%). The study also summarized the risk factors for PSI, such as the Risser sign, correction rate of PTC at follow-up, and correction rate of MTC at follow-up 7. The included studies only used RSH greater than 10 mm or 20 mm as the standard for shoulder imbalance. With the definition of shoulder imbalance by Ono et al., the understanding of shoulder imbalance has been developed1, which meant that lateral and medial shoulder imbalance were two different patterns. RSH represents only lateral shoulder imbalance, and there are few studies on medial shoulder imbalance. Although the lateral shoulder balance were achieved after Lenke 2 AIS corrective surgery, this study found that up to 69.2% of patients suffered from MSI. A stepwise multiple regression analysis showed that postoperative UIVt, PTC, PTC/MTC and T2-VR were risk factors for MSI. The results of the SRS-30 questionnaire were similar with those of earlier research8, and the MSI group were dissatisfied with their appearances. The results of this study provide a wake-up call to surgeons to consider the common occurrence of MSI, and surgeons should make adequate preoperative planning to reduce the occurrence of this phenomenon.
There is no consensus on the definition of shoulder imbalance. Lateral shoulder imbalance corresponded to RSH, Cla-A, coracoid height difference, and clavicle-rib intersection difference9, and RSH ≥ 10 or 20 mm was usually used as the criterion of imbalance. Medial shoulder imbalance correlated well with the radiographical measurements of T1 tilt, first rib angle, and upper thoracic curve size, and a T1 tilt value greater than 3° or 4° was defined as imbalance. However, these two phenomena were independent of each other10. Chung et al. studied the incidence and patterns of medial and lateral shoulder discordance among 151 Lenke 1 and Lenke 2 AIS patients and found that 46.4% of AIS patients had a shoulder discordant pattern, among which Lenke 2 was as high as 67.5%4. In this study, we defined lateral and medial shoulder balance as RSH < 10 mm and T1 tilt < 3°, respectively. Our results with a high incidence of MSI seems to be contrary to Ono and Amir’s view that surgeons had more control correcting medial shoulder balance than clinical shoulder balance after spinal fusion and instrumentation for AIS1,11. We speculated that there were three reasons for such a high incidence. First, the upper thoracic curve of Lenke 2 AIS was structural and hard to correct, making it difficult to eliminate risk factors of MSI. Second, the movement of the shoulder and sternoclavicular joints, as well as the surrounding soft tissues, might constitute a compensating mechanism for the imbalance of the lateral shoulder, which is absent in the medial shoulder. Third and most importantly, surgeons had an inadequate understanding of MSI. Through this research, we advocate that PTC should be sufficiently corrected, UIV should ensure good tilt and derotation, and PTC and MTC should be a good match to achieve postoperative medial shoulder balance. This statement needs to be confirmed in prospective studies. Typical cases are shown in Fig. 3 and Fig. 4.
Optimization of UIV tilt for postoperative medial shoulder imbalance has been widely discussed. Kwan et al. studied the relationship between postoperative UIV tilt and shoulder balance in 60 patients with Lenke1 and Lenke2 AIS and found that patients whose UIV tilt angle deviated away from the reserve motion of the UIV would have an increased risk of clinical medial shoulder imbalance postoperatively12. They further defined the magnitude of the value of the UIV and concluded that patients with a positive value of postoperative UIV tilt had 14.9 times increased odds of developing positive medial shoulder imbalance13. The results of this study also supported the results of these previous studies in which there was a significant correlation between postoperative UIV tilt and postoperative medial shoulder and neck imbalance12–14. In the univariate linear regression relationship between UIV and T1 tilt (T1 tilt = 0.82*UIVt + 3.05), if T1 tilt was < 3, the value of UIV was negative, which coincided with the research of Kwan et al. The understanding behind these results was that the UIV tilt tends to be positive in most patients with left upper thoracic curvature, and as a base for T1, UIV is level or negative to correct the presence of malformations.
Sufficient correction of the PTC could be associated with improved medial shoulder balance postoperatively11. Our results are consistent with previous studies. In the study, the preoperative PTC of the MSB group was smaller than that of the MSI group, and there was no significant difference in the flexibility of PTC and surgical factors (such as the position relationship between UEV of PTC and UIV and implant density of PTC). However, the MSB group achieved a greater PTC correction rate and a smaller postoperative PTC. This may be due to artificial surgical factors caused by different surgeons or a lack of guidance from a surgeon causing the side bending of the PTC to not be complete during the radiograph. Furthermore, several studies have reported a coordinating correction of both the PTC and the MTC consistent with postoperative LSB15–16. The postoperative PTC/MTC ratio correlated with LSB17. In this study, we found that the postoperative PTC/MTC ratio might also be an important factor for the onset of MSI. In the report of Li et al., the shoulder height difference greater than 10 mm was divided into a PSI group and a non-PSI group, where the mean values of PTC/MTC were 1.81 and 1.56, respectively17. In this study, the overall mean PTC/MTC was 1.45, with 0.79 and 1.72 in the MSB and MSI groups, respectively. This meant that the balance of the medial shoulder was more rigorous for PTC/MTC than for the lateral shoulder.
Vertebral rotational deformity and decreased rotational stability may explain the pathogenesis of scoliosis and its progression18–19. The pattern of vertebral rotation generally determines the curve pattern20, and vertebral derotation in AIS can achieve promising corrections in the coronal and sagittal planes21–23. There are many radiographic and CT methods for vertebral rotation measurement24. One of these methods, called the Cobb method, divides the vertebral body into six parts, with the area where the spinous process is aligned determining the assigned grading6. According to this method, we further quantified it, dividing the displacement of the spinous process by half of the transverse diameter of the vertebral body to evaluate the rotation of the vertebra. Then, the T1, T2, and upper thoracic apex vertebrae were measured, and we found that their values were lower in the MSB group than in the MSI group. Multiple regression analysis showed that T2 rotation was a risk factor for T1 tilt. Therefore, we think that derotation of the upper thoracic spine can eliminate part of the cause of T1 tilt and achieve medial shoulder balance.
The strength of this study was the analysis of unique cases of LSB with MSI. At the same time, surgeons can gain a new and deep understanding of the occurrence of MSI. Nevertheless, this study has some limitations. Firstly, this is a retrospective study involving a relatively small sample size. Secondly, this study is based on radiological indicators, and radiological indicators cannot completely substitute for clinical parameters. Thirdly, only 60% of patients responded to the questionnaire survey. The answers of the preoperative questionnaire were incomplete and could not be included.