Adult scoliosis (AS) is defined as a three-dimensional deformity of the spine in a skeletally mature patient. According to previous epidemiological studies, the incidence of AS has been 17.0%–29.4% in the past decade (1, 2). As the ageing of populations in modern society accelerates, AS is becoming increasingly burdening (3). Currently, correction surgery is the only effective treatment for AS patients with a large magnitude curve (4, 5). Postoperative shoulder imbalance (PSI) is a common complications of AS correction surgery, which considerably impacts the postoperative satisfaction of patients (6). However, achieving postoperative shoulder balance remains challenging, with the total incidence of PSI ranging from 25% to 57% (6, 7). Identifying the independent risk factors of PSI can enhance our understanding of this phenomenon and aid in reducing its incidence.

Previous studies regarding the risk factors of PSI mainly focused on adolescent idiopathic scoliosis (AIS). The selection of upper-instrumented vertebra (UIV) is considered one of the main factors responsible for postoperative shoulder balance (8). Previous investigation found that a proximal UIV can avoid the occurrence of PSI (9). However, recent studies have indicated that PSI is not affected by the UIV level (10, 11). Andy et al. reported that a higher preoperative Cobb angle and increased surgical correction lead to an increased risk of PSI (10). In a retrospective review of 145 patients with AIS, John et al. indicated that overcorrection of the main thoracic (MT) curve (>54%) with less correction (<52%) of the proximal thoracic (PT) curve lead to a higher incidence of PSI, regardless of the UIV(11). However, the study of risk factors of PSI in AS correction surgery has not been reported.

AS is a progressive spine deformity, which has a more severe and rigid curve. Correction surgeries for AS always require longer fusion segments, which means that achieving postoperative shoulder balance is more difficult (12–14). The purpose of this study was to evaluate the radiographic risk factors of PSI after AS correction surgery.

In this study, 79 AS patients who underwent posterior instrumentation correction surgery were included (Table 1). Among them, 58 were female and 21 were male. The average age was 35.9 ± 12.7 years (ranging from 21 to 62 years). The mean follow-uptime was 28 months (ranging from 12 to 60 months). Overall, 48 patients had shoulder balance and 31 had shoulder imbalance at follow-up.

As shown in Table 2, there was no significant difference between the balanced and unbalanced groups regarding whether the UIV was T1–2, T3–4, or below T4 (P = 0.512). The pre- and postoperative scoliosis parameters were shown in Table 3. The ICC for preoperative and postoperative RSH was 0.991 (95% CI, 0.987–0.994) and 0.998 (95% CI, 0.997–0.999), respectively. Univariate analysis indicated that the unbalanced group had significantly higher postoperative RSH, lower percentage PT correction, and greater percentage MT correction. The classification and regression tree analysis demonstrated that when the correction percentage of PT curve was more than 55.3%, 84.4% of the patients had balanced shoulder (Figure 1). In addition, when the correction percentage of PT curve was less than 55.3% and the correction percentage of MT curve was less than 56%, 75% of the patients achieved postoperative shoulder balance. However, when the correction percentage of PT was less than 55.3%, and the correction percentage of MT curve was more than 56%, 65.7% of the patients developed PSI (P = 0.038).

The aim of this study was to analyze the factors that predict PSI after AS correction surgery. We found that a lower percentage correction of the PT curve and greater percentage correction of the MT curve were independent radiographic risk factors of PSI. Larger correction of the MT (>56%) with a relatively lower correction of the PT (<55.3%) lead to PSI in 65.7% of the patients. In contrast, when the correction percentage of PT curve was more than 55.3%, 84.4% of the patients had balanced shoulder. In addition, the incidence of PSI was independent of the UIV level.

Achieving postoperative shoulder balance is a significant but difficult goal in correction surgery of spine deformity. The choice of UIV level is considered to be one of the main factors potentially responsible for PSI, though this is still controversial. To date, there is no consensus regarding the UIV selection in correction surgery, which is a point of contention among many spine surgeons (8, 21, 22). According to a previous guidelines for AIS, a UIV of T2 was suggested for patients with a preoperative high left shoulder, T3 for those with a balanced shoulder, T4 or below for those with a high right shoulder (9). However, Jaysson et al. found that choosing T4 as UIV was more effective to avoid PSI than either T2 or T3, regardless of which shoulder was raised preoperatively (23). Recently, several studies reported that PSI is not affected by UIV levels (10, 24). The findings of these articles were consistent with our results that UIV did not affect the incidence of PSI. Our results further suggested that a proximal UIV may not be sufficient to achieve postoperative shoulder balance; rather, adequate percentage correction of the PT is paramount in avoiding the occurrence of PSI.

Another key finding of this study was that a lower percentage correction of the PT curve and greater percentage correction of the MT curve were independent risk factors of PSI in AS correction surgery. Representative cases of a patient with postoperative balanced shoulder (relative larger correction of the PT and lower correction of the MT) and of a patient with postoperative unbalanced shoulder (relative lower correction of PT curve and greater correction of the MT curve) were shown in Figures 2A–D, respectively. In a systematic review of risk factors for PSI after correction surgery for scoliosis, Zhang et al. indicated that adequate correction of the PT and moderate correction of the MT was suggested to avoid PSI (6). Other studies also reported that overcorrection of the MT curve leads to a high incidence of PSI (24, 25). In addition, John et al. reported that larger correction of the MT curve (>54%) with simultaneous less correction (<52%) of the PT curve resulted in PSI in 59% of patients in Lenke type 1 and 2 AIS. Similar results were also observed in patients with AS in our study. Therefore, the PT curve should be sufficiently corrected to achieve postoperative shoulder balance in AS correction surgery when the MT curve is overcorrected.

To our knowledge, this is the first study to evaluate the risk factors of PSI after correction surgery of adult spine deformity. Patients with AS tend to have a larger and rigid curve, which are more difficult to correct than that in AIS. John et al. reported that when the correction percentage of PT curve was more than 52%, 80% of the patients achieved shoulder balance in Lenke type 1 and 2 AIS. However, only when the correction percentage of PT curve was more than 56% in AS correction surgery, a higher proportion of shoulder balance can be achieved. This means that a greater correction percentage of PT curve is required to maintain shoulder balance in AS. The reason may be that the PT curve in AS patients is relatively stiff, while the PT curve of AIS patients is less rigid, thus possessing self-correction ability. Indeed, several studies have reported that a flexible PT will continue to correct automatically after the MT curve is corrected (26–28). Although the PT curve is rigid in AS, the correction can be achieved through compression across the convexity and distraction through the concavity of the PT curve. During the past several decades, the posterior column osteotomy techniques have advanced considerably, thereby enabling spine surgeons to significantly correct the MT curve (29, 30). However, if the PT curve is not also adequately corrected, a higher proportion of PSI will occur.

This study has several limitations. First, this was a single-center study with a small sample size, which may result in a selection bias. Second, several other factors such as T1 tilt, clavicle angle, and coracoid height difference, were not measured and discussed. Third, only RSH was used to estimate shoulder balance in our study, which may not be fully representative of clinical shoulder balance.

In conclusion, we found that a lower percentage correction of the PT curve and greater percentage correction of the MT curve were independent radiographic risk factors of PSI after AS correction surgery, regardless of the UIV level. Greater correction of the MT (>56%) with relative lower correction of the PT (<55.3%) lead to PSI in 65.7% of the patients. On the contrary, when the correction percentage of PT was more than 55.3%, 84.4% of the patients had a balanced shoulder. Therefore, the PT curve should be sufficiently corrected to achieve postoperative shoulder balance in AS correction surgery when the MT curve is overcorrected.