Mismatch Between Proximal Rod Contour Angle and Proximal Junctional Angle: A Risk Factor Associated with Proximal Junctional Kyphosis After Growing Rods Treatment for Early-Onset Scoliosis

Study Design: A retrospective study Objectives: To investigate the impact of radiological and surgical factors as well as proximal rod contouring angle (PRCA) on the development of proximal junctional kyphosis (PJK) in early-onset scoliosis(EOS) patients after growing rod (GR) treatment. Methods: We reviewed a consecutive series of EOS patients who had undergone growing rod (GR) treatment between 2009 and 2018 (minimum follow-up of 2 years) at a single institution. Patients were divided into PJK and non-PJK groups according to the occurrence of PJK or not during the follow-up periods. The demographic data, surgical strategies, and radiographic parameters were recorded and compared between the PJK and non-PJK groups. PJK was dened as a PJA (cid:0) 10° at the last follow-up. PRCA was dened as the angle between the cephalad endplate of the UIV and the lower endplate of the second vertebra caudal to UIV and the PJA-PRCA was dened as the difference between the values of PJA and PRCA. Logistic regression was also performed to identify the risk factors for the occurrence of PJK. Results: This study nally included 95 patients. The mean age at the index surgery was 6.5±2.2 years. Mean follow-up lasted 4.4±1.9 years. Lengthening procedures averaged 4.0 ± 1.8 times. There were 20 patients who were observed with PJK (the incidence, 21.1%). In comparison with the non-PJK group, the PJK group showed a larger preoperative coronal cobb angle (81.8±20.6° vs 70.4±12.4°, P=0.041) and global kyphosis (GK) (56.0±15.3° vs. 45.9±12.9°, P=0.044), as well as a larger GK correction (40.4% ±10.0% vs. 30.0%±14.2%, P=0.035). In addition, the PJK group had signicantly larger postoperative PJA (10.8±3.1 vs. 5.3±3.1, P (cid:0) 0.001) and greater postoperative PJA-PRCA (5.3±3.0 vs. 3.66±2.9, P=0.031). The proportion of patients with a value of PJA-PRCA greater than 5° in PJK group was signicantly higher than that in the non-PJK group. Multiple logistic regression showed that preoperative GK (cid:0) 50°, postoperative PJA (cid:0) 10 and postoperative PJA-PRCA (cid:0) 5° were the risk factors in predicting PJK after GR treatment.


Introduction
Early-onset scoliosis (EOS) refers to a deformity that presents before 10 years old and progressed rapidly at early stage [1]. De nitive spinal fusion performed at young age may result in a short trunk, pulmonary dysfunction, and even iatrogenic thoracic insu ciency syndrome [2]. In past decades, growth friendly technique has become a mainstay of treatment for EOS patients due to its effectiveness in controlling spinal deformity while permitting spinal column growth and pulmonary system development [3]. However, it can't be ignored that this technique has a high incidence of complications, such as surgical site infection, implant-related and alignment-related complications [4][5][6].
Proximal junctional kyphosis (PJK) is a common alignment-related complications observed in adults and adolescents after surgery for scoliosis or kyphosis [7,8]. Also, a signi cant proportion of children under the age of 10 developed PJK. Chen [9] reported occurrence of PJK in 21 of 113 congenital scoliosis (CS) children who underwent posterior instrumented fusion. At the same time, PJK was reported in EOS patients who accepted growth-friendly treatment, such as vertical expandable prosthetic titanium rib technique(VEPTR) [10] and growing rods treatment(GR) [11], with a rate ranging from 12-56% [10][11][12].
Several risk factors, radiological or surgical, associated with PJK had been reported, including preoperative hyperkyphosis [13], pelvic incidence [14,15], locations of upper instrumented vertebra (UIV) [16] and lower instrumented vertebra (LIV) [11]. In addition, some studies have turned their attention to the proximal junctional area and found that the mismatch between proximal rod contour angle (PRCA) and proximal junctional angle (PJA) may be a possible risk factor of PJK in degenerative scoliosis [11].
To the best of our knowledge, the impact of the mismatch between PJA and PRCA on the occurrence of PJK in EOS has not been investigated. Thus, the current study aimed to investigate the incidence and risk factors of PJK and further to clarify the association between PRCA and PJK in EOS patients receiving GR treatment.

Study sample
A series of consecutive patients with EOS who had undergone GR treatment between January 2009 and May 2018 were identi ed from our spinal deformity database. Inclusion criteria to this study were as follows: (1) patients aged before 10 years old at index surgery; (2) had at least 2 lengthening procedures; and (3) a minimum follow-up of 2 years with complete radiographs. Exclusion criteria were as follows: (1) previous history of spinal surgery; (2) a distal anchor at the sacrum or pelvis; (3) VEPTR. Clinical demographic data including age at the index surgery, gender and etiology were recorded. Surgical data including type of GR (single or dual rods), location of UIV and LIV, anchor type (pedicle screw or hook or hybrid), and the times and intervals of lengthening procedures were also collected. This study was approved by the Institutional Review Board of A liated Drum Tower Hospital of Nanjing University Medical School, and informed consent have been exempt from requirement in the study.

Surgical strategy
All surgeries were performed by a single surgical team who were well experienced with the GR treatment.
The surgical strategy consisted of the index surgery and the following periodical lengthening procedures. In the index surgery, UIV and LIV for each patient were determined based on evaluations of age, scoliotic deformity and sagittal plane of the spine. Basically, the UIV and LIV were chosen in the stable zone from T1 to T5 and L1 to L5, respectively. During the index surgery, only upper and lower ends of the instrumentation area require subperiosteally exposure. Generally, polyaxial pedicle screws were used as anchors at the proximal and distal sites. If the pedicle was too small, a laminar hook was placed instead. Then, single/dual rods were contoured to normal sagittal alignments, placed with a tandem in a submuscular manner and attached to the screws at either end. Correction of the main curve was achieved once all set screws were tightened under traction. Lengthening procedure was scheduled every 8-12 months and was performed through a small midline incision with 1-2cm distraction. Somatosensory evoked potential and motor evoked potentials were applied in all surgical procedures.

Radiographic evaluation
All the radiographic measurements were performed on the Picture Archiving and Communications Systems (PACS) workstation. Radiographs were taken before and after the index surgery, as well as during each lengthening procedure. Standing postero-anterior radiographs were used for the measurements of the following parameters: coronal Cobb angle, thoracic kyphosis (TK), global kyphosis (GK), lumbar lordosis (LL), sacral slope (SS), pelvic tilt (PT), pelvic incidence(PI), sagittal vertical axis (SVA), PRCA and PJA. PJA was de ned as the angle between the caudal endplate of the UIV and the cephalad endplate of the second supradjacent vertebra above the UIV; PRCA was de ned as the angle between the cephalad endplate of the UIV and the caudal endplate of the second vertebra caudal to the UIV ( Fig. 1.). The PJA-PRCA was de ned as the difference between the values of PJA and PRCA.

Statistical Analyses
Analyses were performed by SPSS version 22.0 software (SPSS Inc., Chicago, IL). Patients were divided into PJK and non-PJK group based on whether PJK developed during the follow up or not. All data were reported as mean ± standard deviation. The signi cance of differences between the two groups was determined by independent-sample t test. The univariate analysis by Chi-square test was used to compare the differences of categorical variables between the PJK group and non-PJK group, and further served as a screening tool to select the possible candidates for the logistic regression analysis. Subsequently, logistic regression analysis was performed on predictors with a univariate P value of less than 0.10 to analyze the covariate effects of the possible indicators for the development of PJK. A P value of less than 0.05 was considered statistically signi cant.

Baseline Data
A total of 95 consecutive patients (45 boys and 50 girls) who accepted GR treatment were ultimately included in this study. The baseline patient characteristics are shown in Table 1. The mean age at the index surgery was 6.5±2.2 years. Among them, the etiologies were classi ed as: congenital in 61 patients, neuromuscular in 16, idiopathic in 11, and syndromic in 7, respectively. There were 73 patients treated with dual GR and 22 with single GR. Pedicle screws were placed as anchors proximally in 78 patients, while hooks in 9 and hybrid (screws in combination with hooks) in 8 respectively.

General Results After GR Treatment
A total of 383 lengthening procedures were performed, with an average of 4.0 ± 1.8 procedures per patient. The mean lengthening interval and follow-up duration were 11.0±1.1 months and 52.4±23.0 months, respectively. Fourteen patients who reached skeletal maturity underwent removal of the GR implants and de nitive fusion while one patient keep close follow-up without subsequent surgery after 5 times lengthening procedures.
As shown in Table 2, the major curve experienced satisfactory correction after the index surgery (73.0±18.8° VS. 36.8±11.4°, P<0.001) and increased slightly during subsequent lengthening surgery. The preoperative T1-S1 height increased from 24.7±3.3cm to 27.8±4.1cm after the index surgery (P<0.001), and increased to 32.9±4.3cm at nal follow up. On the sagittal plane, GK decreased signi cantly after the index surgery and increased slightly during the whole follow-up period.
Till the latest follow-up, PJK was identi ed in 20 patients. In addition, 13 patients developed rod fracture, 3 experienced screw loosening and 1 developed distal junctional kyphosis, which were treated in the next lengthening procedures. There were 3 patients with wound super cial infections who were treated successfully by dressing changes daily and antibiotic treatment. No neurological de cit occurred in this cohort either in the index surgery or in the subsequent lengthening procedures.

Incidence, classi cation and progression of PJK
The overall incidence of PJK in our cohort was 21.1%. There were 17 patients with type 1 PJK and 3 with type 3 PJK, respectively. None of our cohort had a type 2 PJK. Among the patients with PJK, 5 patients developed PJK within 6 months after the index surgery, 8 after the rst lengthening procedures, 5 after the second lengthening procedures, 1 after the third lengthening procedure and 1 after the forth lengthening procedure. In the PJK group, the average PJA increased from 5.9±4.1° to 10.8±3.1° after the index surgery, and continuously increased to 17.3±7.2° at nal follow-up. Six cases of PJK continued to progress while the other 14 cases were relatively stable. For these progressive PJK patients, brace treatment was prescribed. At the latest follow-up visit, only one patient received a revision surgery due to pulling out of the proximal hooks.

Comparisons between PJK and non-PJK groups
Comparisons between PJK and non-PJK groups revealed that younger age at the index surgery was associated with the occurrence of PJK. However, no signi cant difference was observed in terms of gender, type of GR and number of lengthening procedures or intervals between the two groups. (Table 3) The radiographic characteristics of both groups are listed in Table 4

Discussion
Incidence and mechanism of PJK after GR treatment According to previous studies, the incidence of PJK in EOS patients after GR treatment varied from 12% to 56%, which was mainly due to different inclusion criteria and surgical strategies [10][11][12]. The incidence of PJK in our study cohort was 21.1% over an average 52.8±22.2 months of follow-up. It's remarkable that PJK in this study mainly occurs after the rst distraction surgery. Li [19] revealed that all patients developed PJK within the rst year after VEPTR treatment. On the other hand, Wijdicks et al. [20] reported that the occurrence of PJK in patients treated with magnetically controlled growing rod increased as time progresses but had no correlation to the number of distractions.
Several studies tried to explain the patho-mechanism of PJK. A biomechanical study on adult scoliosis suggested that the dissection of posterior intervertebral elements might lead to the increase of the mechanical stresses within the junctional area. Also, the injury of the paraspinal musculature and the extensive dissected facet joints during the surgery may destroy the integrity and stability of the posterior structures of the spine [21]. According to our experience, the lengthening procedure itself is a local distraction surgery, which has a tendency to increase the kyphosis of the implant area or junctional area, eventually leading to the occurrence of PJK. In our study, the most common type of PJK was ligamentous failure which developed in 85% (17/20) patients. Our nding is in line with the study by Chen [9], but inconsistent with the study by Wang [22] who reported that implant/bone interface failure was the major type of PJK in congenital scoliosis patients receiving posterior short fusion.
Association of PJA-PRCA mismatch with PJK Currently, there are still no studies reporting the mismatch between PJA and PRCA as a risk factor for PJK in EOS patients receiving GR treatment. As a parameter to represent the proximal rod bending, the PRCA stays unchanged because the position of the screw and the rod is relatively xed after the index surgery and allows us to indirectly measure the contouring of the proximal part of the rod. Thus the PJA-PRCA was proposed to re ect the matching between the proximal rod contouring and supradjacent regional spinal alignment. In our study, there was no signi cant difference in PRCA and preoperative PJA-PRCA between the two groups. However, the PJK group has a signi cantly higher proportion of patients with a value of PJA-PRCA greater than 5° than that in the non-PJK group. We believe that PJK might be avoided if the postoperative PJA-PRCA matches well. But when the proximal area of the instrumentation is excessively corrected, resulting in mismatched postoperative PJA-PRCA, the risk of PJK will be greatly increased. Of note, similar ndings have been observed in previous studies. Yagi et al. [18] reported that the inadequate restoration of global sagittal alignment was a signi cant risk factor for PJK in adult scoliosis patients undergoing long instrumented spinal fusion. Yan [23] demonstrated that mismatch between postoperative PJA and PRCA led to a proximal compensation, namely PJK in degenerative scoliosis. After reviewing 84 adolescent idiopathic scoliosis patients, Wang et al. [24] found that larger postoperative PJA-PRCA, especially those with PJA-PRCA greater than 5° were risk factors for PJK.
In the previous literature, two methods of contouring angle measurements of the proximal part of the rod were reported. In Yan's [23] study, PRCA was de ned as the angle between the rst two proximal screws in sagittal plane after surgery in DS. The second method is that of Lonner used in AIS patients and the PRCA was de ned as the cobb angle between the UIV and once vertebra caudal to the UIV. These two methods have one thing in common, that is, they both use a contouring angle between 2 vertebra (UIV and UIV-1). In our study, we proposed a new measurement method to re ect the proximal rod contouring (PRCA was de ned as the angle between the cephalad endplate of the UIV and the lower endplate of UIV-2). Consequently, we propose some possible mechanisms. A large PJA-PRCA means the proximal rod is contoured insu ciently, resulting a greater biomechanical stress concentrated in the junctional area and leading to the occurrence of PJK subsequently. Moreover, to get satisfactory correction results, the proximal region may sustain excessive forces during the surgery. According to the Hueter-Volkman principle that compression forces inhibit growth and tensile forces stimulate growth, the posterior column of spine grows faster than the anterior column after the index surgery, further aggravating the kyphosis of the proximal junctional area. Therefore, even if the PJA is less than 10° before the lengthening procedures, it may progress to more than 10° after the subsequent distraction. Another cause of PJK is related to the particularity of GR treatment itself. GR treatment is different from other operations in that besides the index surgery, subsequent lengthening procedures are also needed. The stress generated during the lengthening procedures will be transferred to the junctional area, making the stress concentration, and nally lead to the occurrence of PJK. This is consistent with Bess et al. 12 ndings on the complications of GR treatment.

General risk factors of PJK
In this study, preoperative hyperkyphosis was also identi ed as an independent risk factor associated with PJK. This is also consistent with previous research. Chen et al. [10] reported that hyperkyphotic EOS (TK≥50°) tended to experience increased complications, such as rod fracture and PJK. Lee et al. [25] pointed out that correction surgery would impact the overall sagittal balance of the spine and PJK might be an early-stage adjustment of the overall balance compensation by the trunk. In severe kyphotic deformity, surgeons pay more attention to achieving satisfactory correction. Excessive forces were applied to the kyphosis area to the reconstruct the sagittal plane, leading to the stress concentration and eventually resulting in the occurrence of PJK.
We also found postoperative PJA 10° was a signi cant independent risk factor for PJK. Previously, several studies also detected larger postoperative PJA in the PJK group than the non-PJK group [11,16].
Patients with postoperative PJA greater than 10° are more likely to develop PJK, which is helpful for early prediction and intervention. We inferred that this might be related to the improper selection of UIV or the insu cient contouring rod during the index surgery.
Although we nd no signi cant differences between the location of UIV and LIV in PJK group and non-PJK group, some previous studies reported different opinions. Pan [16] reported UIV distal to T2 leads to an increasing stress concentration in the junctional area, which may accelerate the occurrence of PJK. Similarly, Watanabe [11] demonstrated that an LIV at or cranial to L3 may change the stress on the spine and then increase the risk of PJK.

Clinical relevance and limitations
The clinical relevance of our study lies in emphasizing the importance of appropriate rod contouring at the proximal junctional area in preventing the occurrence of PJK. During the index surgery, the curvature of the rods should be shaped into a slightly kyphotic pro le harmoniously matching the proximal spinal curve when contouring the rods. For patients with hyperkyphosis, the operation of excessive correction of kyphosis leading to the stress concentration on proximal junctional area should be avoided as much as possible. Additionally, for patients with postoperative PJA 10°, brace treatment will be prescribed.
Our study has several limitations. First, it was a retrospective study with a relatively short follow-up.
Second, we were unable to evaluate the impact of PJK on clinical outcomes because patients are too young to ll out questionnaire by themselves. Despite these limitations, all patients were treated by the same surgical strategy and protocol which reduced the impact of the variety of surgical procedures and ensured the homogeneity of surgical results.

Conclusion
In conclusion, the prevalence of PJK in patients with EOS undergoing growing rod treatment was 21.1%. PJK mainly occurred after the rst lengthening procedures. Preoperative GK 50°, postoperative PJA 10°a nd postoperative PJA-PRCA 5° were the risk factors of occurrence of PJK.