Introduction

Degenerative lumbar scoliosis (DLS) is highly prevalent in individuals aged > 65 years, affecting between 8.3 and 68% of the population [1,2,3,4]. DLS is a chronic condition that results from degenerative bone and soft tissue changes. Degeneration ultimately leads to pathological changes in load-bearing at the intervertebral and facet joints, resulting in spinal deformity [5]. DLS often leads to radiculopathy or ultimately instability (via spondylolisthesis or rotatory subluxation) that results in spinal stenosis and low back pain [6]. Posterior instrumentation and fusion for DLS have become the mainstay of operative treatment [7]. One of the most prevalent complications following surgical correction of DLS is Proximal junctional kyphosis (PJK) [8]. The prevalence rates of PJK after DLS surgery vary widely, which have been reported to range between 20 and 40% [9, 10]. The current literature evaluating sagittal alignment in the management of spinal deformity and the risk radiological factors and outcomes associated with proximal PJK have been largely based on the Schwab criteria [11] and GAP score [12]. Although the etiology of PJK is multifactorial, some radiological risk factors are still not fully understood.

An influential and widely employed classification system for the lumbo-pelvic profile in the sagittal plane has been proposed by Roussouly et al. [13]. Taking into consideration the apex of the lumbar curve and the Sacral slope (SS) in the standing position, the method allows defining four different lumbo-pelvic types, which have recently been extended to 5 by accounting for the presence of pelvic anteversion [14]. Under the Roussouly classification, two concepts deserve attention. First, the Lordosis tilt (LT) is an important positional parameter. LT is defined as the angle between the line of the anterior upper edge of L1 and S1 vertebrae and the plumb line, which indicates the inclines of the total lordosis. Second, as the lower arc of lordosis is the main portion, extending to the concept of Lordosis distribution index (LDI), defined as L4–S1 lordosis/L1–S1 lordosis \(\times\) 100%, determines the lower-arc lordosis [12]. LDI is a Pelvic incidence (PI)-based proportional radiographic paralordosis, which is important because it alters the distribution of the load [15, 16]. With the same amount of Lumbar lordosis (LL), there may be different forms of LT and LDI. In the past, we paid more attention to the amount of lumbar lordosis correction. The significance of LT and LDI in DLS correction has not been fully studied.

This study aimed to investigate the potential value of LT and LDI in the occurrence of PJK following DLS correction surgery, and provide practical guidance on surgical strategies to improve clinical outcomes.

Methods

This was a single-center, retrospective study that included consecutive patients with DLS who underwent posterior instrumentation and fusion treatment at our institution between 2009 and 2018. Institutional review board approval was obtained from all the participants. The inclusion criteria for patients were as follows: (1) Cobb angle of lumbar curves ≥ 10°; (2) age > 45 years at the time of surgery; (3) long instrumentation from thoracolumbar to L5 or the sacrum with a minimum of four vertebral segments of fusion; (4) complete preoperative and postoperative radiographic data and functional evaluation forms; and (5) at least a 2‐year follow‐up. The exclusion criteria were as follows: (1) history of previous lumbar spine surgery and (2) other types of scoliosis.

Radiographic measurements and clinical outcomes.

Radiological evaluations were performed using the entire standing spine. X-ray and all radiological measurements were performed by two experienced surgeons who were not involved in the operations. Radiographic data collection consisted of full-length lateral radiographs obtained preoperatively, postoperatively, and at the latest follow-up. The following radiographic parameters were measured: (1) pelvic parameters: PI, pelvic tilt, and SS. (2) Local curvature: Cobb angle; LL; LDI was defined as L4–S1 lordosis/L1–S1 lordosis \(\times\) 100%; thoracic kyphosis (TK). (3) Global parameters: sagittal vertical axis; LT, the angle formed by the vertical line and the line through the anterior superior edge of S1 and the first lumbar (L1) vertebra. LT is expressed as a negative value if L1 is posterior to the anterior aspect of S1 and positive if it is anterior to S1 (Fig. 1).

Fig. 1
figure 1

Schematic diagram of measured parameters

Full size image

PJK was defined as ≥ 10° increase in kyphosis between the Upper instrumented vertebra (UIV) and the UIV + 2 between early postoperative and follow-up radiographs. Proximal junctional failure was defined as fracture of the UIV or UIV + 1, pullout of instrumentation at the UIV, and/or presence of sagittal subluxation.

Clinical outcomes were evaluated using the Japanese orthopedic association (JOA) score, Oswestry disability index (ODI), and Visual analog scale (VAS) scores at the final follow-up. These are valid and rigorous functional measures used to assess spinal disorders.

Preoperative clinical function questionnaires, including the VAS of low back pain, JOA, and ODI scores for patients were completed on admission for surgery without assistance. All enrolled patients were followed up for at least 2 years from the date of surgery. Questionnaires were completed in the outpatient department at the final follow-up.

Statistical analysis

Statistical analysis was performed using SPSS version 23.0. All continuous data are expressed as the mean and standard deviation. The normality of continuous data was tested using the Kolmogorov–Smirnov test. Normally distributed values were compared using Student’s t-test, while values with skewed distribution were compared using the Kruskal–Wallis test. Categorical data were presented as numbers, and statistical significance was performed using the χ2‐test. Pearson correlation analysis was performed to investigate the relationship between postoperative radiographic parameters. ROC curve estimation was performed to predict PJK using postoperative parameters. Statistical significance was set at p < 0.05.

Results

Patient population

A total of 122 patients were enrolled in this study; 24 were men and 98 were women, with an average age of 63.5 ± 6.3 years. The patient demographics were shown in Table 1. They underwent 6.1 ± 1.6 levels of fusions. Postoperatively, the spinopelvic alignment of patients improved significantly, and the mean preoperative and postoperative alignments were summarized in Table 1. LDI, LT, ODI, JOA, VAS in the preoperative periods were 107.4 ± 107.2%, 6.0 ± 11.7°, 58.3 ± 15.4, 11.4 ± 3.8, and 6.4 ± 1.7, respectively. The postoperative LDI and LT were 83.7% ± 20.3% and − 2.8 ± 7.4°, respectively. LDI, LT, ODI, JOA, and VAS at final follow-up were 85.9 ± 36.3%, -4.2 ± 8.0°, 27.4 ± 18.9, 20.2 ± 5.5, and 3.5 ± 2.4, respectively.

Table 1 Patient demographics
Full size table

Relationship between distal fusion level and clinical outcomes

All patients were divided into floating and lumbosacral groups. We compared the clinical outcomes between the two groups and found no significant differences. The results revealed no significant difference in Cobb angle of the curves at the final follow-up time (p = 0.766), ODI at final follow-up (p = 0.963), JOA at final follow-up time (p = 0.379), final VAS (p = 0.103), and rate of PJK (p = 0.9), which were shown in Table 2. These results confirmed that the distal fusion level does not influence the next steps of the analysis in this study.

Table 2 Comparison of therapeutic effects or complications between groups with long instrumentation and fusion with lumbar vertebra or S1
Full size table

Incidence and clinical importance of PJK at final follow-up

Thirty patients (24.6%) had PJK at the final follow-up. Eighteen patients had PJK grade A (PJA < 20°), 10 patients had grade B (PJA increase from 20° to 29°), and two patients had grade C (PJA increase > 30°). Besides, there were 7 patients (23.3%) got PJF in the patients with PJK. As shown in Table 3, ODI, JOA, and VAS scores at the final follow-up were worse in the PJK (–) group than in the PJK ( +) group.

Table 3 Comparison of clinical outcome in non-PJK group and different PJK group
Full size table

Comparison of PJK and Non-PJK

PJK patients were significantly older and had larger preoperative LT and Cobb angles. Comparison of PJK (–) and PJK ( +) patients did not reveal any other differences in preoperative sagittal parameters. No significant differences in sex or body mass index were observed between PJK ( +) and PJK (–) patients (Table 4). Postoperative sagittal spinopelvic alignment in patients with and without PJK is summarized in Table 5. The PJK (–) patients had a significantly smaller angle LT (− 1.6 ± 7.2 vs. − 6.2 ± 6.6°; p = 0.002) and Cobb angle of the curves (10.9 ± 6.1 vs. 14.3 ± 8.0; p = 0.017). At final follow-up, the PJK (–) patients had a significantly smaller LT, TK and Cobb angle of the curves and significantly larger LDI and SS, which were shown in Table 6.

Table 4 Preoperative characteristics of patients in non-PJK group and PJK group
Full size table
Table 5 Summary of postoperative parameters comparing patients in non-PJK group and PJK group
Full size table
Table 6 Parameters of patients in non-PJK group and PJK group at last follow-up
Full size table

Postoperative reciprocal change (RC) of LT, LL and LDI

LL was significantly increased postoperatively in patients with DLS (p < 0.001). LT was also significantly increased in patients with dynamic light scattering (p = 0.001). LDI was also significantly increased in patients with DLS (p = 0.013). In addition, there was a significant positive correlation between increased LL angles and increased LT angles (p = 0.006) (Fig. 2). These results indicate that postoperative RC in LT occurred with LL corrective surgery. In contrast, there were no significant correlations between the increased LT angles and increased LDI angles.

Fig. 2
figure 2

Relationship between the changed LT and changed LL of preoperative and postoperative

Full size image

ROC curve analysis of postoperative parameters to predict PJK

We established an ROC curve to predict PJK by including factors that were < 0.05, such as age, postoperative LT, postoperative LDI, postoperative Cobb angle, and PI-LL, which have been proven to be related to PJK [17]. This is shown in Table 7. We concluded that the Cobb angle of the curve and LT were independent risk factors for PJK in patients with DLS. According to the You Den Index of ROC analysis, the threshold (integer) for identifying patients at high risk of PJK was set at 8.0, and patients with postoperative LT < − 8° were more likely to develop PJK (43.8% vs. 18.0%, p < 0.05).

Table 7 ROC curve of postoperative parameters to predict PJK
Full size table

Discussion

This study showed a correlation between postoperative LT, LDI, and clinical outcomes. There was also a significant difference in the Cobb angle of the curves in the postoperative time and last follow-up time among the groups, which indicated the great effectiveness in coronal realignment.

In the present study, the prevalence of PJK was 24.6%. Consistent with previous reports [18, 19], the clinical outcomes of patients with PJK were significantly worse than those of patients without PJK. The relationship between spinopelvic parameters and quality of life remains controversial. A previous study showed that flat-back syndrome was an indication of unsatisfactory clinical outcomes after lumbar fusion surgery [20]. Mac-Thiong [21] demonstrated that sagittal spinal and global balance was strongly associated with ODI in adults with scoliosis. In our study, there was no significant difference in clinical outcomes between the two groups when fused to S1. Therefore, the position of distal vertebral fixation was not different in these patients. We found that patients with PJK had worse clinical outcomes, which indicated the important role of sagittal balance in clinical outcomes.

Preoperative risk factors for PJK were older age, larger LT, and larger Cobb angle of the curves. The postoperative risk factors for PJK were LT and Cobb angle of the curves (Fig. 3). Previous studies indicated that patients with worse preoperative spinopelvic alignment and a large correction in LL are more likely to develop PJK [19, 22,23,24]. The distribution of lordosis is also important [16]. In our study, we used the age, postoperative LT, postoperative LDI, postoperative Cobb angle of the curves, and PI-LL, which have been proven to be related to PJK [17], in ROC analysis. According to the ROC curves, we found a strong correlation between postoperative LT and PJK without a relationship with LDI. LT is a spinopelvic parameter for analyzing the relationship between the relative positions of the spine and pelvis, which indicates the inclines of the total lordosis. In total, it reflects the total LL position. In addition, we also found a positive relationship between LL and LT. Thus, attention should be paid to both LT and LL. LT had no effect on LDI. The distribution of the lordosis was considered to be aligned if the LDI was within the range 50–80%. Almost all the patients met the criteria in this study. Therefore, we believe that the relationship between LDI and PJK remains uncertain. It is still controversial whether small LDI is essential in severe deformity as a risk factor for PJK in patients with DLS using the method that was concluded by other studies [25,26,27]. According to the Yuden index of the ROC curve, we concluded that patients with LT < − 8° were at a higher risk of PJK.

Fig. 3
figure 3

The relationship between the postoperative risk factors and PJK. In the pictures, patient without PJK was shown in a, b, c and d. LDI, LT and LL in the preoperative period were 849.0%, − 26.2° and 5.1° (a and b). LDI, LT and LL in the postoperative period were 84.3%, − 3.9° and 40° (c). LDI, LT and LL in the preoperative period were 115.2%, − 6.9° and 24.3°(d). Patient with PJK was shown in e, f, g and h. LDI, LT and LL in the preoperative period were 146.6%, − 19.1° and 17.8° (e and f). LDI, LT and LL in the postoperative period were 65.1%, − 20.4° and 31.5°(g). LDI, LT and LL in the preoperative period were 54.4%, − 18° and 37.3° (h).

Full size image

To our knowledge, this is the first report showing a critical correlation between postoperative LT and lordosis distribution and consequent PJK. Importantly, LT is a modifiable factor depending on the surgical technique, and we can apply this knowledge to surgical strategies. In contrast with previous reports, a smaller postoperative LT was not found to be a risk factor for PJK. This is because in the previous study, they did not pay attention to this parameter. Recent reports have focused on the impact of UIV orientation on the development of PJK [28, 29]. We can control LT by adjusting the rod curvature of the position between L1 and S1 in our operation.

This study has some limitations. First, this was a retrospective study that may have caused selection bias. Second, the sample size was limited in a single-center study. Because of these limitations, a multicenter, prospective randomized study is needed to determine the ideal LT in Chinese patients.

Conclusions

Preoperative risk factors for PJK were older age, larger LT, and larger Cobb angle of the curves. Postoperative risk factors for PJK included postoperative LT and postoperative Cobb angle of the curves, which were smaller than those preoperatively. LT is a spinopelvic parameter for analyzing the relationship between the relative positions of the spine and pelvis, which indicates the incline of the total lordosis. Control of postoperative LT is crucial for preventing PJK.