Postoperative proximal junctional kyphosis in Lenke 5C adolescent idiopathic scoliosis following posterior surgery: extending fusion versus thoracolumbar/lumbar fusion

Background: Proximal junctional kyphosis is a common complication after posterior fusion in adolescent idiopathic scoliosis patients, and was correlated with postoperative change of thoracic kyphosis. However, no previous literature has evaluated the correlation of PJK with thoracervical parameters in the AIS patients. Methods: Data from 61 patients who underwent posterior fusion for Lenke 5C AIS and had 2 years of follow-up were retrospectively reviewed. Patients were fused higher than two levels above the upper-end-vertebra in the extended fusion group (n=32), whereas they were fused below or at UEV+2 in the TL/L fusion group (n=29). Results: During an average follow-up of 32.5 months, 14/61 patients had PJK. The extended fusion group had a greater incidence of PJK than the TL/L fusion group did (11/32 vs. 3/29,P = 0.03) and a significant more decrease of TK than the TL/L group(p<0.01). Patients with PJK had a significantly larger Thoracic inlet angle(TIA) than those without PJK (P <0.01). Multivariate analysis revealed a greater preoperative TIA to be a positive predictor for PJK (p = 0.015). TIA >72° with extended fusion was a risk factor for PJK at final follow-up. The SRS-22 score did not significantly differ between the extended fusion and TL/L fusion groups. Conclusions: TIA could be a predictor of PJK. In Lenke 5C AIS patients, those with TIA >72° were more likely to develop PJK. And extended fusion in patients with Lenke 5C adolescent idiopathic scoliosis may be at an increased risk of proximal junctional kyphosis.

ere more likely to develop PJK. And extended fusion in patients with Lenke 5C adolescent idiopathic scoliosis may be at an increased risk of proximal junctional kyphosis. Background PJK is a common complication after posterior fusion in patients with Lenke 5C adolescent idiopathic scoliosis (AIS) [1][2][3][4]. Multiple studies have reported that the risk factors for PJK are a large lumbar lordosis (LL) angle, large sagittal vertical axis (SVA), and low pelvic 3 incidence (PI) [1][2][3][4]. Other factors strongly associated with PJK are a large preoperative TK and large immediately postoperative TK in AIS [2,4,5]. Previous study indicated that the thoracic inlet angle (TIA) was strongly associated with TK, global TK, and cervical kyphosis in AIS patients [6][7][8], and that the TIA is a constant morphological parameter that is not changed by the patient's position or any other conditions. And these correlations are maintained at 2 years after corrective surgery in patients with Lenke type 1 AIS patients [8]. Hence, the TIA is an ideal parameter that can be used to evaluate the sagittal balance of the proximal thoracic area and may could be a predictor of PJK .
The choose of upper instrumented vertebra may also be associated with the incidence of PJK in Lenke 5c patients [4]. Posterior thoracolumbar/lumbar (TL/L) fusion is commonly used for Lenke 5C (AIS) [9,10]. To avoid the progression of the unfused thoracic curve after thoracolumbar/lumbar (TL/L) fusion, experienced surgeons may choose to perform extended fusion in patients with Lenke 5C AIS with a bending thoracic Cobb angle of more than 20° [11.12.13.15].Furthermore, Kwan et al. [14] suggested extended fusion with a bending thoracic curve of more than 15°. However, the extended fusion may strongly reduce the thoracic kyphosis (TK) in the sagittal profile [14,15], and the reduction of TK is a proven risk factor for proximal junctional kyphosis (PJK) and may also result in great change in thoracervical parameters in AIS patients [2.4.5.16-20].
The present study focused on analysis of the radiological features of Lenke 5C patients to identify the correlations among TIA, PJK and other sagittal parameters. Our study also aimed to evaluate the relationship between the choice of upper instrumented vertebra and PJK in patients with Lenke 5C AIS.

Methods
The present retrospective study included 61 patients with Lenke 5 AIS who underwent posterior-only surgery in our department between January 2010 and January 2016. All 4 included patients had follow-up for more than 2 years with complete radiographic data.
The present study was approved by the local ethics committee. Patients were fused higher than two levels above the upper-end-vertebra in the extended fusion group, whereas they were fused below or at UEV + 2 in the TL/L fusion group. Patients with a Cobb angle of the thoracic curve of > 40° or a bending thoracic curve of > 15° generally underwent extended fusion; however, a few patients with a rigid thoracic curve chose to undergo TL/L fusion to preserve the flexibility of the thoracic spine after acknowledging complete understanding of the benefits and risks of these two procedures. The Legacy or CD Horizon M8 screw-rod system (Medtronic Sofamor Danek, Inc., Memphis, TN) was used for fixation.
Standing full-length posteroanterior and lateral radiographs were routinely taken using the multi-purpose Digital R/F System (Sonial Vision Safire 17, Shimadzu Corporation, Kyoto, Japan) preoperatively, at 3 months postoperatively, and at final follow-up. Radiologic parameters were independently measured and documented on an electronic system by two surgeons who were not involved in the surgeries, and the average values were calculated to increase the accuracy of the measurement of the upper thoracic region [19].
The coronal parameters measured were the main TL/L Cobb angle, thoracic Cobb angle, reduced bending angle, and correction rate of the two curves. The sagittal parameters measured were the PI, sacral slope (SS), pelvic tilt (PT), LL, PI-LL, TK, GTK, T1 tilt, CL, TIA, proximal junctional angle (PJA), and SVA. TK, GTK, TK, LL, PI, SS and PT were measured using previously described standard methods (Fig. 1). The TIA was defined as the angle between the vertical line of the T1 superior endplate and the line passing through the midpoint of T1. CL was measured as the Cobb angle between the lower endplates of C2 and C7. CK was defined as a CL angle of less than 0°.

Result
The age of 61 Lenke 5 AIS patients in this study was 15.1 ± 2.0 years. There were 45 females and 16 males, and the mean Risser sign was 2.8 ± 1.5. The mean follow-up of these patients was 33.2 months (range from 22 to 81 months). The main TL/L curve Cobb angle on average was 53.0 ± 8.4° before surgery and 6.8 ± 6.7° at final follow-up.
Extended fusion versus TL/L fusion 6 Twenty-nine patients had TL/L fusion, and 32 patients chose extended fusion. Both groups had an excellent correction rate for the main TL/L curve at final follow-up time (85.4% in the extended fusion group and 86.9% in the TL/L fusion group). The correction rate of the thoracic curve in the extended fusion group (70.2%) was significantly greater than that in the TL/L fusion group (57.3%, p < 0.01) ( Table 1).
PI and TIA kept stable after surgery and through the follow-up time in both the extended fusion and TL/L fusion groups. A significant difference was found in the immediate change of TK (-3.7 ± 9.7 vs. 3.8 ± 6.0, p < 0.01) and increasing PJK angle (7.1 ± 8.6 vs. 3.6 ± 4.8, p = 0.04) between the extended fusion and TL/L fusion groups (Table 1). No significant difference was found between extended fusion and TL/L fusion in the preoperative or final SRS-22 scores for function/activity, pain, self-image, mental health, and satisfaction with treatment. (Table 4a) Proximal junctional kyphosis According to our PJK criteria, 14/61 (23.0%) patients were considered PJK at final follow-up time. We found no significant difference exists in PI, PT, SS, LL, PI-LL, or even SVA distance preoperatively, at three months, and final follow-up time. In addition, PJA did not differ between the two groups preoperatively. However, the mean preoperative TIA, GTK, and CL value in PJK patients were significantly larger than that in non-PJK patients ( Table 2). After surgery, the extended fusion group exhibited much more PJK than the TL/L fusion one (11/32 vs. 3/29, p < 0.01). At final follow-up time, lower TK, GTK, and CL were found in the PJK group. The preoperative TIA of patients with PJK at the latest follow-up was larger than that of patients with non-PJK (73.2 ± 5.6 vs. 66.6 ± 7.4, p < 0.01), and the ROC curve showed that the cutoff value was 72°. None of the 27 CK patients had PJK, which is significantly lower than 14/34 in the CL group (p < 0.01). No significant difference was found between post-op PJK and non-PJK patients in the preoperative or final SRS-22 7 scores (Table 4B).

Bivariate correlation and curve estimations for PJK and TIA
The correlation analysis was done between the sagittal parameters with the increasing PJK angle (Fig. 2). Bivariate correlation tests of the following radiographic parameters showed

Multivariate analysis
Multivariate logistic regression results for predictors of PJK is presented in Table 3. It revealed a greater preoperative TIA to be a positive predictor for PJK (p = 0.015, log odds 0.212). Greater preoperative PJA was a negative predictor for PJK (p = 0.045, log odds − 0.355).

Discussion
Proximal junctional kyphosis and extendned fusion Kim[1]and Wang [2] defined PJK in AIS patients as both an increase of > 10° in kyphosis between the UIV and the UIV + 2 and the final PJA greater than or equal to 10°, which was widely used in AIS patients [1,2,4,5] and was chosen in our study. Previous studies have 8 reported incidences of PJK after surgery for Lenke 5C AIS of 8.49% [4]. Sun et al. [3] reported that an increase of PJK larger than 10° occurred after selective fusion in 17.1% of patients with Lenke 5C AIS at 2 years postoperatively, which is comparable to the PJK incidence found in the present study (23.0%). To reduce the incidence of PJK, it is critical to restore and maintain normal TK [1,2,3,5]. Important risk factors for PJK in AIS include reduction of TK and large preoperative TK [2,4,5].
Patients with PJK had a greater GTK than those without PJK in our study. The immediately postoperative change in TK is reportedly a risk factor for PJK [2,4,5]. Similarly, the present study revealed a correlation between the immediately postoperative change in TK and the increasing PJK angle. However, PJK resulted from either a large immediately postoperative decrease or increase in TK. Patients with an immediately postoperative decrease in TK need to restore the GTK to attain sagittal balance; the kyphosis would increase at the proximal part of the fusion level and thus lead to PJK. In patients with an immediately postoperative increase in TK, the greater risk of PJK may be due to the overcorrection of LL [3]. Hence, the larger postoperative LL could lead to an immediate increase in TK, and the spontaneous change to attain sagittal balance may lead to greater PJK to obtain much larger TK to match the overcorrected LL. However, the present study did not find a correlation between PJK and LL.
Extended fusion had more correction in the coronal plane with less restoration of TK [14,15]. In our study, the extended fusion group had a significantly immediate, larger reduction in TK. This immediate decrease in TK meant that the extended fusion group had a significantly greater incidence of PJK than the TL/L fusion group did. A previous study showed that placement of the UIV cephalad to the upper-end vertebra was associated with an increased risk of PJK in Lenke 5 curves [4]. Similarly, extending the fusion to the thoracic curve in the present study resulted in higher rate of PJK. 9 Thoracic inlet angle and Proximal junctional kyphosis Preoperative TK, LL and global TK are alternate parameters that could potentially be used to predict postoperative change of the Sagittal balance of the thoracic spine. However, these parameters are easily influenced by posture, because it is difficult to ensure that every patient has a horizontal vision line and is standing perfectly straight during radiography. So they are not ideal predictors of the increasing PJK angle.In contrast, the TIA is a constant morphological parameter that is not influenced by the posture under any conditions, similarly to the PI of the spinopelvic unit [6][7][8]. Recent studies found significant correlations between preoperative TIA and TK and Global TK angle [6][7][8].
Preoperative global kyphosis is a key parameter influencing the postoperative thoracic and thoracervical spinal sagittal balance [20]; In our study, the TIA could predict the GTK at final follow-up, which was also strongly correlated with the final increasing PJK angles.
The TIA did not change after surgery or during two years of follow-up and could be an appropriate predictor for proximal junctional kyphosis.
In patients with a reduction in TK, which is a reported risk factor for PJK [2.5], the occurrence of the proximal junctional kyphosis is mainly determined by the variation in the TIA (Fig. 5A-B).This may explain the role of the TIA in predicting PJK. A larger preoperative TIA, which predicts larger final GTK, makes it easier for the proximal junction to become more kyphotic during follow-up. In the patients who underwent extended fusion of the main thoracic curve, this increase in sagittal kyphosis could only be obtained at the proximal junctional area, so the PJK angle increased. However, patients with a lower TIA, which indicated lower final GTK and T1 tilt, were much more likely to have final CK [8].
In the extended fusion group, the TK decreased immediately after surgery. However, the GTK needed to stay stable from the preoperative one, which made it easier for PJK to occur in the proximal thoracic region. According to the ROC curve, a TIA of > 72° was the cutoff point for the occurrence of PJK. Hence, extended fusion should be avoided in patients with Lenke 5C AIS with a TIA of > 72°. These patients with large preoperative TIA and GTK should be carefully evaluated prior to extended fusion, especially when the surgeon tried to correct the preoperative hyperkyphosis.
However some patients with large TIA had relatively small TK and GTK because of the rotation of the thoracic curve, these patients developed an immediate increase of TK after surgery, and the large TIA also meant that they needed a large GTK at the final follow-up.
The increasing TK and GTK immediately after surgery were not enough for these patients; the TK and GTK continued to increase during follow-up and led to PJK at the final followup. This mechanism could explain why some of the PJK patients had an immediate increase of TK.
Patients who develop PJK following surgical correction of adult spinal deformity reportedly tend to have increased pre-and postoperative T1 slopes and C2-7 angles [21.22].
Increased progression of CL was associated with the development of PJK in these studies [21.22]. In present study, none of the 27 CK patients had Proximal junctional kyphosis at the final follow-up. This result suggests the incidence of PJK could be correlated with the cervicothoracic parameters.

Clinical outcome
The mean SRS-22 score was satisfactory in both the extended fusion and TL/L fusion groups. Furthermore, the domain values for pain, self-image, function, mental health, and satisfaction were good or excellent in both groups. No significant difference was found in the clinical outcome between the PJK and non-PJK groups with at least a 2-year follow-up time. This result was similar to the recent study [5]. We also believe a longer follow-up time is needed to evaluate the degeneration and back pain from the PJK.
The present study had several limitations. First, the number of patients was relatively 11 small with at least 2 years of follow-up, and there might not have been enough patients with preoperative abnormal sagittal profile to detect whether this made a difference in postoperative PJK. Second, the TL/L fusion may less accurately reflect the thoracic kyphosis than the extended fusion might, due to the unfused thoracic curve. However, the final Cobb angle of thoracic curve was not significantly different between the two groups.
Third, the accuracy of parameter measurement in PJK patients is still a large problem [19].
Although we had already measured and documented on an electronic system the patients' Author Contributions: BH and XY conceptualized and designed the study, drafted the initial manuscript. BH and CZ carried out the initial analyses, reviewed and revised the manuscript. XH and LW analyzed and interpreted the data, reviewed and revised the manuscript. LL and YS coordinated and supervised data collection, critically reviewed and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.