Risk Factors for Multiple Implant-Related Complications (MIRC) with Growing-Rod for Early-Onset Scoliosis

Study design: A retrospective single center study. Objective(cid:0)To identify risk factors for multiple implant-related complications with growing-rod for early-onset scoliosis. Background: High incidence of implant-related complications in the treatment of early-onset scoliosis with traditional growing rod. The risk factors for multiple implant-related complications (MIRC) have not been adequately studied. Methods: Data of 59 early-onset scoliosis patients who had been underwent growing rod surgery at Beijing Chao-yang Hospital from September 2007 to December 2017 were reviewed. All patients had complete clinical and radiographic data. Patients were divided into groups with or without MIRC. The univariate and multivariate logistic regression analysis were performed to identify the risk factors associated MIRC. Results: The average age of insertion was 8.9 years and mean follow-up was 51.91 months. 234 implantation or expansion surgeries were performed and the average operation interval was 11.4 months. A total of 60 implant-related complications occurred. Ultimately, MIRC developed in 20 (33.9%) of 59 patients. Number of surgery procedure > 3 times , follow-up time ³50 months , preoperative thoracic kyphosis > 50°, postoperative thoracic kyphosis >50°, postoperative lumbar lordosis >50°, postoperative sagittal vertical axial >40mm are potential risk factors for MIRC (P<0.1). Multivariate logistic regression analysis showed that Number of surgery procedure > 3 times, postoperative thoracic kyphosis >50°, postoperative lumbar lordosis >50°are independent risk factors for MIFRC (P<0.05), Among them, patients with postoperative thoracic kyphosis >50°had an 18.647 times higher risk of MIFRC than postoperative thoracic kyphosis angle <50°. Conclusions: Traditional growing rod in the setting of EOS has excellent clinical and radiographic outcomes but a high multiple implant-related complications. Number of surgery procedure > 3 times, postoperative thoracic kyphosis >50°, postoperative lumbar lordosis >50° are independent risk factors for MIRC.


Introduction
Early-onset scoliosis (EOS) refers to spinal deformities that occur before the age of 10 and are caused by a variety of pathological factors [1]. Because EOS patients are in the rapid growth period of the spine and the critical development period of cardiopulmonary function, early detection and timely treatment are essential [2]. EOS patients develop rapidly, and spinal deformity maybe worsens rapidly in a short period of time, and the initial treatment may not be effective all the time, therefore, close observation is also a very important measure. Surgical intervention is mainly used to treat scoliosis that cannot be controlled by bracing or plaster treatment, or the child is too young to insist on wearing it [3].
The traditional growing rod (TGR) is the most widely used technology in the treatment of EOS, not only preserves the growth ability of the spine, but also maintains the development of lungs and thorax [4]. To our knowledge, some authors have described outcomes of EOS patients treated with growing rods, which achieving satisfactory results in correcting deformity, spinal height, and thoracic volume [5][6][7][8][9][10]. The dual growing rod technique can obtain better correction of coronal deformities and provide su cient support for spine growth, and the incidence of complications is lower than the single growing rod, but the incidence of implant-related complications is higher than the single growing rod. Previous studies have shown total complication rates for TGR procedures to be as high as 48%-79% [5,[11][12]. Thus the incidence of implant-related complications were 21.7%-59.3% [5,9,[11][12]. Implant-related complications include implant dislodgement, implant loose, rod fracture, proximal junctional kyphosis and so on. Thoracic kyphosis (TK) is an important parameter for assessing the severity of scoliosis, increased kyphosis may cause high loading in the middle of the growing rod and increase the risk of local fatigue fracture of the implant, the proximal and distal foundation sites may bear more stress, which leads to implant-related complications [13]. In a multicenter retrospective study, thoracic scoliosis >50 degrees and Cobb angle >50 degrees are risk factors for unplanned reoperations after growing-rod surgery [14]. In addition, some researchers have observed that EOS patients with hyperkyphosis tend to suffer more complications [13]. The risk factors for treating EOS with implant-related complications were well evaluated in previous studies, however the risk factors for multiple implant-related complications have not been adequately studied. The study was conducted to identify the rate and risk factors of more than once implant-related complications with growing rod surgery for EOS.

Patient Selection
We reviewed the records of patients with EOS who received traditional growing rod surgical for scoliosis.
Inclusion criteria were: 1) unsuccessful conservative management; 2) Cobb angle ≥50 degrees or curve progression; 3) a minimum of 2-years follow-up; 4) lengthening surgery more than once. Patients with insu cient follow-up time, incomplete data and revision surgeries were excluded. We evaluated patients who were suffered from more than once implant-related complications (MIRC) due to EOS with TGR or less than once implanted-related complications (N-MIRC). All the radiographs were calibrated to achieve accurate distance measurements. Clinical records were reviewed for data including age at the initial surgery, gender, height, weight, etiology, body mass index (BMI), living altitude. The clinical data and the surgical data were all collected and shown in Table I. Complications were also recorded, details were noted.

Radiographic Evaluation
Radiographic measurements were performed independently by two of the authors using a picture archiving and communication system (PACS). All upright standing or sitting posteroanterior and lateral images preoperatively, after surgery, at the last follow-up with growing rods were measured. The radiological measurements included Cobb angle, thoracic kyphosis(TK), lumbar lordosis(LL), T1-S1 range(mm), T1-T12 range(mm), sagittal balance, coronal balance, distance between the C7PL and sagittal vertical axis, pelvic incidence(PI), sacral slope(SS), and pelvic tilt(PT).
Institutional review board approval was obtained from our institution before initiation of the study. All methods were performed in accordance with the relevant guidelines and regulations, and a consent form was signed by parents of each child.

Surgical Technique
Both the initial surgery and lengthening surgery were all performed under general anesthesia. Neurophysiological monitoring was essential. Single growing rods were used for well-balanced trucks or thin patients, while dual growing rods were mainly used for patients with unbalanced trucks.. The skull to the upper thoracic curve and the stable vertebrae and were selected as the lower instrumental vertebra (LIV) and the upper instrumental vertebra (UIV), respectively. When stripped under the periosteum in the upper and lower xation areas, the integrity of the joint capsule should be protected. Pedicle screws, hook or hybrid xations were used. The selection of the anchor sites were depending on severity of the curve, etiology. The proximal and distal anchor points can be limited fusion locally. The connecting rod should be bent for adapting with the curvature of scoliosis. Proper lengthening procedure should be carried out initially. Lengthening surgery was performed for generally 6-12 months. When lesser distraction can be obtained, a risser sign >1°, or menstruation in female patients, nal fusion should be adopted.

Statistical Analysis
SPSS Statistics for windows (version 24.0; IBM, Armonk, NY, USA) was used for statistical analysis. Chisquare test or Fisher exact test was utilized for evaluate differences in categorical data. Student-t test was used to examine continuous variables. The risk factors with a p-value <0.1 on univariate analysis were identi ed as potential of multiple implant-related complications. Multivariate logistic regression was performed to determine the independent risk factors for MIRC.

Results
Fifty-ve patients (24 males, 35 females) with EOS from our department treated with TGR between September 2007 and December 2017 met the inclusion criteria. The average age was 8.9±2.4 years at the time of initial surgery. Median follow-up was 51.91±25.23 months (range 24-132). In terms of etiology, 31 patients were idiopathic, 1 patient was syndromic, 18 patients were congenital, 10 patients were neuromuscular. The average number of lengthening was 3.97±1.72 (range, 2-10). An average interval of 11.4±3.0 months (rage from 4 to 24) was performed. Thirty-one of the 59 EOS patients had underwent de nitive fusion.
Basic information of MIRC group and N-MIRC group are shown in Table 1. The MIRC group consisted of 10 males and 10 females, with the average initial surgery age of 8.6±2.46 years(range, [5][6][7][8][9][10][11][12][13][14], The N-MIRC group consisted of 14 males and 25 females, with the average initial surgery age of 9.05±2.40 years range,5-14 . Etiologies included 11 idiopathic, 6 congenital, 2 neuromuscular, 1 syndromic in MIRC group, and 19 idiopathic, 12 congenital, 8 neuromuscular in N-MIRC group. In MIRC group, 6 underwent dual growing rod surgery and 14 received single growing rod treatment, thus 22 underwent dual growing rod surgery and 17 received single growing rod treatment in N-MIRC group. There was no signi cant difference between groups for age at initial surgery, curve etiology, sex, height, weight, BMI, single or dual rods, subcutaneous or submuscular, average interval of operations, patients with nal fusion. In MIRC group, the mean follow-up time was 65.2±27.98 months (range,24-132 and the mean follow-up time was 44±20.59 months (range, 24-107 in group N-MIRC, which showed a statistically signi cant difference(P<0.05). Distribution of UIV and LIV were showed in Table 2         patients with early-onset scoliosis treatment with dual growing rods and demonstrated curve correction, but the incidence of implant-related complications was 54.5%, which were the main reasons for unplanned reoperation [9]. A retrospective study indicated 110 patients with growing rod for early-onset scoliosis, a total of 87 patients (79%) had 263 complications and 84 unplanned reoperations, of which implant-related complications were 49%, surgical site infections were 23%, the primary operation age less than 7.6 years, the thoracic kyphosis angle greater than 38° and the main curve Cobb angle greater than 84° were risk factors for complications [11]. Zhang et al [17] identi ed growing rod treatment of 55 early- for IRCs include: female patients, younger age and smaller pelvic incidence angle, for patients with low pelvic incidence angle, more attention should be paid to the sagittal balance. In a multi-center study by Akbarnia [19], 138 growing rods were used to treat patients with EOS and 56 patients (40.6%) had at least one growing rod fracture, the study found that growing rod fracture was associated with pathology, BMI, and mean interval lengthening are irrelevant. Stainless steel growing rods are more prone to fracture than titanium alloy growth rods, and growing rod fracture often occurs in a diameter of less than 4mm. A study on IRCs in scoliosis patients with neuro bromatosis type, including 59 patients, 17 cases (28.8%) of IRCs occurred, of which scoliosis aggravated (7 cases), screws Loosening (3 cases), adding-on (3 cases), rod fracture (2 cases), PJK (2 cases), nail cap loosening (1 case), pedicle screw extraction (1 case), univariate logistic regression analysis showed that age <9 years, thoracic kyphosis angle < 50° and growing rod application are risk factors for IRCs, multivariate logistic regression analysis showed that thoracic kyphosis angle >50° and the application of growth rods are independent risk factors for IRCs, the author believes that the kyphosis angle is too large, and there may be malnutrition for the kyphotic vertebral body, the apical vertebral area around the internal xation bears more pressure, which will cause the connecting rod to break or the screw to pull out, younger patients need more growth of the spine, and the diameter of the internal xation implanted for the rst time is smaller, so IRCs are more likely to occur, the authors suggest that the pathological classi cation of dural expansion and EOS should be considered when selecting distal and proximal xation segments [20].
In our study, there was no statistical difference in gender, height, weight, and BMI between the MIRC group and the N-MIRC group (P=0.297, 0.168, 0.317, 0.621), and there was no statistical difference in the diagnostic classi cation of the other two groups (P=0.405). The MIRC group was younger than the N-MIRC group preopreatively, but the difference was not statistically signi cant (P=0.5). In previous study, younger patients before the initial surgery was prone to occur IRCs [21], but in our study, although patients in the MIRC group were younger, it was not statistically different.
The number of single and dual growing rods in the MIRC group and N-MIRC group were 14/6 and 22/17 respectively. There was no signi cant difference in statistical analysis (P=0.311). previous studies have shown that single growing rod is more prone to occur IRCs complications, thus our study is different from previous [10,13,17,22]. The interval lenthening in the MIRC group and N-MIRC group was 11.31 to 3.33 months and 11.48 to 2.71 months, respectively. There was no signi cant difference in statistical analysis (P=0.71). This conclusion is consistent with the results of Akbarnia [19]. The rod interval had no effect on the occurrence of multiple implant-related complications.
The pelvic incidence of patients in the MIRC group changed from 39.95°∼11.89° preopreatively to 38.4°∼10.25° at the last follow-up, there was no signi cant difference in statistical analysis (P>0.05). The pelvic incidence angle of the N-MIRC group was changed from preoperatively 39.16°∼11.04° became 40.74°∼10.98° at the last follow-up. There was no signi cant difference in statistical analysis (P>0.05).
In the past, it was reported that with age increasing, the PI angle continued to increase until maturity, thus the increase of PI was limited due to the curvature of the spine or application of implant in EOS patients [21,[23][24], which was con rmed in our literature. There was no signi cant difference of PT between the MIRC and N-MIRC group preoperatively, postoperatively and at the last follow-up (P=0.801, 0.648, 0.43).
The preoperative PT of the two groups were 11.90°∼13.5° and 5.52°∼9.76°, respectively (P=0.042). There was no signi cant difference in the PT between postoperative and the last follow-up (P=0.891, 0.525).
There was no signi cant difference between the two groups of SS preoperatively and postoperatively (P=0.105, 0.409). There was signi cantly difference of SS at the last follow-up between two groups (P=0.048), due to the PT and SS are two variety parameters as age increasing, body position changes or other pathological factors exist [18].
Here major Cobb angle in the MIRC group was 97.42°∼16.14° preoperatively, while 89.1°∼23.2° in N-MIRC group (P=0.158), Previous studies [20] indicated that the Cobb angle of the major curve greater than 84° was a risk factor for complications. The complications involved in implant-related complications, surgery site infection and other complications, our study mainly research implant-related complications.
The correction of major Cobb angle in two groups were all satisfactory, but MIRC group was better than N-MIRC group (P<0.05). There was no signi cant difference in TK between the two groups preoperatively and at the last follow-up (P=0.378, 0063), while the postoperative TK of the two groups were 44.42°∼23.3° and 32.02°∼15.68°, respectively (P=0.018). There was no signi cant difference in T1-S1 comparison between the two groups preoperatively, postoperatively, and the last follow-up (P=0.249, 0.088, 0.984). The results of the study showed that the occurrence of multiple implant-related complications did not affect T1-S1 growth. There was statistically signi cant difference in the length of T1-12 between the MIRC group and the N-MIRC group preoperatively and postoperatively (P=0.039, 0.045) but there was no signi cant difference in the T1-12 between the two groups at the last follow-up (P=0.465), and the correction rate of T1-12 in the two groups was also not statistically signi cant (P=0.301). The results show that despite multiple implant-related complications, the growth of T1-12 was not affected.  [11, 13-14, 20, 25].
This purpose of our study mainly explored the occurrence the risk factors for more than twice implantrelated complications. Postoperatively TK>50° should be pay more attention, the main considerations as follow: rstly, the abnormal postoperatively TK lead to more stress on proximal and distal anchors, EOS patients were young and had poor bone mineral density, pedicle screw pullout, loosen and kyphosis were Easier to occur. Secondly, Abnormal TK required larger curvature of the growing rod to match it and the initial implantation would bring more growth, thus the shape of the growing rod might decrease stiffness of the growing rod. Thirdly, patients with abnormal TK might strip more in the distal and proximal anchor points during the initial operation, the articular capsule of adjacent segments was damaged, which might cause the stability of the anchor points to decrease. Fourthly, EOS patients with postoperatively TK>50°, the balance of spinal sequence and growing rod might be changed after the initial surgery, which resulted in abnormal stress between spinal sequence and implant, that might increase the incidence rate of MIRC.
In the previous study of our team [26], the results showed that postoperative upper thoracic scoliosis >50°w as a risk factor for PJK. This study also emphasizes the importance of keeping the proximal posterior joint capsule and posterior ligament intact.
Multivariate logistic regression analysis showed that the number of operations >3 times was the risk factor for MIRC. Previous studies reported that longer follow-up time and more operations were the risk factors of postoperative complications [17].  [13] reported that patients with single growing rod were more prone to growing rod fracture and repeated rod fracture than patients with dual growing rod. In this paper, the proportion of patients with single growing rod in the MIRC group was 60.9% and 56.4% in the N-MIRC group (P = 0.311). The purpose of this paper was to study the causes of MIRC rather than rod fracture or multiple rod fracture. Previous study indicated that the follow-up time is a risk factor for complications with growing rod for EOS, the longer follow-up time, the greater possibility of complications [17]. In our study, one-way ANOVA showed that the follow-up time of MIRC group and N-MIRC group were 65.2 ± 27.98 months and 44 ± 20.59 months respectively (P = 0.002), which was similar to that reported in the literature, but multivariate logistic regression analysis showed that the follow-up time ≥50 months was not a risk factor for MIRC.
In this study, 1 patient had 4 times of IRCs (1 implant dislodgement and 3 screw loosening), 3 patients had 3 times of IRCs (1 screw loosening and 1 rod fracture and 1 proximal dislodgement, 1 PJK and 1 dislodgement and 1 screw loosening, 1 PJK and 2 rod fracture), 16 patients had more than once implant-related complications. 15 patients with 1 IRC during treatment were divided into N-MIRC group. When EOS patients had one of the three factors during operation as follow: number of operations >3 times or postoperative TK >50° or postoperative LL >50°, the patient should be treated individually and followed up closely in combination with the patient's initial operation age, severity of scoliosis, etiological classi cation, radiographic parameters, reduce the incidence of IRCs. Patients have MIRC during the treatment process, which increases di culties in the treatment process, brings more pain to patients, and even serious complications may occur, resulting in the impact of the nal treatment effect. Therefore, more research are needed for MIRC.
Our study has several limitations. Firstly, this was a retrospective study, and no randomization was performed between the MIRC and N-MIRC group. Secondly, the sample size was not big enough, and all cases were from a single institution, which can cause selection bias, thirdly, the incidence of MIRC with growing rods might be underestimated since not all of the patients were followed until nal fusion, nally.

Conclusions
Traditional growing rod in the setting of EOS has excellent clinical and radiographic outcomes but a high multiple implant-related complications. Number of surgery procedure > 3 times, postoperative thoracic kyphosis >50°, postoperative lumbar lordosis >50° are independent risk factors for MIRC.