Influence of Paraspinal Growth-Friendly Spinal Implants in Children with Spinal Muscular Atrophy on Parasol Deformity, Rib-Vertebral Angles, Thoracic, and Lung Volumes

Introduction: Children with spinal muscular atrophy (SMA) and progressive neuromuscular scoliosis often require early growth-friendly spinal implant (GFSI) treatment for deformity correction with implant fixation either through pedicle screws or bilateral to the spine using ribto pelvis fixation. It has been proposed that the latter fixation may change the collapsing parasol deformity via changes in the rib-vertebral angle (RVA) with a positive effect on thoracic and lung volume. The purpose of this study was to analyze the effect of paraspinal GFSI with bilateral rib-to-pelvis fixation on the parasol deformity, RVA, thoracic, and lung volumes. Methods: SMA children with (n = 19) and without (n = 18) GFSI treatment were included. Last follow-up was before definite spinal fusion at puberty. Scoliosis and kyphosis angles, parasol deformity, and index, as well as convex and concave RVA, were measured on radiographs, whereas computed tomography images were used to reconstruct thoracic and lung volumes. Results: In all SMA children (n = 37; with or without GFSI), convex RVA was smaller than concave values at all times. GFSI did not crucially influence the RVA over the 4.6-year follow-up period. Comparing age- and disease-matched adolescents with and without prior GFSI, no effect of GFSI treatment could be detected on either RVA, thoracic, or lung volumes. Parasol deformity progressed over time despite GFSI. Conclusion: Despite different expectations, implantation of GFSI with bilateral rib-to-pelvis fixation did not positively influence parasol deformity, RVA and/or thoracic, and lung volumes in SMA children with spinal deformity directly and over time.


Introduction
Spinal muscular atrophy (SMA) is a rare, but severe, neuromuscular disorder with a loss of alpha motor neurons leading to progressive muscle weakness and atrophy.This is primarily caused by a deficiency of the survival motor neuron (SMN) protein due to mutation or deletion of the SMN1 gene [1,2].The severity of the disease ranges from very severe (SMA type I) to mild cases (SMA type IV) [2,3].The number of genomic copies of SMN2 is inversely associated with disease severity [4].
Children with severe SMA often develop progressive spinal deformity in combination with collapsing ribs, the so-called collapsing parasol deformity, and a pulmonary function decline [5,6].To control spinal deformity and supposedly support thoracic development, lung growth, and respiratory function, growth-friendly spinal implants (GFSI) have been used with a rib-to-pelvis fixation [7].The rationale behind this technique for the treatment of collapsing parasol deformity is possible horizontalization of the ribs due to implant-induced pressure, while obtaining these effects due to repetitive surgical distractions or outpatient lengthening procedures using an external remote controller over time [7][8][9][10].The therapeutic success of GFSI is usually measured by the reduction of the scoliosis and/or kyphosis angle on standardized radiographs of the whole spine [11], whereas evaluation of the torsion of the spine, 3D analysis and thoracic volume has gained importance just recently [12][13][14].Another radiographically measurable deformity parameter, the rib-vertebral angle (RVA) was established by Mehta [15] in 1972.The difference of the RVA can be used to predict the progression in infantile idiopathic scoliosis [16] and has been proven to have a correlation to the scoliosis angle measurement [17].Livingston et al. [6] and Johnston et al. [5] used different methods to evaluate parasol deformity in SMA children.
In the present study, the effects of paraspinal GFSI with rib-to-pelvis fixation in SMA children with neuromuscular spinal deformity were analyzed to predict influences on the collapsing parasol deformity via measurements of the parasol score, the RVA, thoracic, and lung volumes.Because of progressive pulmonary function decline in SMA, a larger thoracic cage may be beneficial to the overall lung function in these patients.

Material and Methods
After Ethics Committee approval, a prospective, nonrandomized cohort study was performed on 37 pediatric subjects with SMA and spinal deformity.Of these, 19 children were treated with bilateral GFSI (MAGEC ® , NuVasive, USA, or VEPTR ® , DePuy Synthes, USA) (Fig. 1) inserted parallel to the spine using a rib cradle fixation with an outrigger and pelvic hook fixation [7], whereas 18 SMA children had no prior spinal treatment (Table 1).The indication for GFSI treatment was a progressive spinal deformity in SMA, which could not be controlled by conservative treatment methods.Rib fixation was always performed in the area of rib three to seven.
The group with GFSI-treatment was evaluated prior to GFSI treatment (T0), directly after implantation-surgery (T1), after an average follow-up of 4.6 years (range 2.4-8.8years) with GFSI treatment (T2), and upon removal of GFSI before definite spinal fusion (T3) (Fig. 2a).GFSI are removed some weeks before definite spinal fusion to minimize implant-transmitted infection from GFSI.In the group without GFSI pretreatment, definite spinal fusion was planned and the late presentation without pretreatment with GFSI was either because of late deformity, fear of surgical treatment, repeated respiratory infection, cachexia or due to a refugee status.
Clinical data relative to gender, age, body mass index (BMI), body height, and weight were collected (Table 1).Radiographs of the children were obtained in anterior-posterior and lateral standardized sitting position at all time points.On radiographs, the scoliosis angle, kyphosis, and Mehta RVA [15] were measured using Centricity Enterprise Web version 3.0 (GE Healthcare Medical Systems, USA).Using the original Mehta RVA measurement [15], a "perpendicular (line) is drawn to the middle of either the upper or lower border of a selected thoracic vertebra."This is the datum line for that vertebra.Another line is drawn from the mid-point of the head of the rib to the mid-point of the neck of the rib, just medial to the region where the neck widens into the shaft of the rib.This rib line is extended medially to intersect the vertebral line to make the RVA" (Fig. 2b).The RVA on the convex and the concave side of the curve were measured at the level of all twelve ribs, even though some were twisted due to spinal torsion.
Parasol deformity was determined clinically by grading each hemithorax as "normal" or "parasol" at all examined time points.The parasol score was determined according to Livingston et al. [6].
Computed tomography (CT) images were taken directly before definite spinal fusion for planning of the surgery.CT images were subjected to three-dimensional reconstruction and all reconstructions with full representation of thoraces and lungs were used for measurements of thoracic and lung volumes using the software program Horos (The Horos Project, Horos version 3.3.6)(Fig. 2c, d).The apex of the lungs and the upper thoracic aperture formed the upper limit of the reconstruction, whereas it was limited by the ribs and the intercostal muscles laterally and by the diaphragm caudally.The ventral boundary was formed by the sternum and the dorsal boundary by the twelve thoracic vertebrae.The thoracic boundaries were manually drawn into the CT image, and subsequently, all regions of interest were combined and assembled into a reconstructed three-dimensional model.The measurements of the lung volumes were done using the same method with individual measures of the left and right lung, respectively.
Statistical analysis of all data was performed using Graph Pad Prism version 6 (GraphPad Software Inc., USA).Displayed in the figures are the mean and standard deviation as error bars.Some RVA could not be measured, for example, due to extreme torsion; therefore, the exact n-numbers analyzed are depicted in the corresponding bars of each graph.An unpaired t test was performed to compare RVA at each individual rib, differences in RVA (RVAD), thoracic volume, and lung volume between patients with and without GFSI.A paired t test was performed to compare the concave and convex side of each individual rib pair.Statistical significance was defined with levels as p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***).

Results
Data of 19 scoliotic children with GFSI treatment (9 female, 10 male) and 18 scoliotic children without GFSI treatment (10 female, 8 male) were analyzed (Table 1).At the time of GFSI implantation, children were on average In all SMA patients and at all times, there was a difference in the RVA between the convex and concave side with lower values of the convex side (Fig. 3).While at an average age of 8.3 years, the RVA difference between the convex and concave side was significant only in the lower thoracic region (Fig. 3a), a significant difference also became apparent in the upper thoracic region over time (Fig. 3b-d).At T0, the largest average difference between the concave and convex side was observed within rib pair 11, with the RVA on the convex side being on average 37°smaller (range 72°s maller -21°larger) than on the concave side.At T1, the largest average difference was observed within rib pair 11 as well, with the RVA on the convex side being on average 28°s maller (range 72°smaller -17°larger) than on the concave side.At T2, the largest average difference was observed within rib pair 8, with the RVA on the convex side being on average 34°smaller (range 93°smaller -34°larger) than on the concave side.At T3, the largest average difference was observed within rib pair 10, with the RVA on the convex side being on average 39°smaller (range 80°smaller -60°l arger) than on the concave side.
Significant differences of the RVA of the convex and concave side were also found in the SMA group without prior treatment (Fig. 3e).Here, the largest average difference was observed within rib pair 6, with the RVA on the convex side being on average 38°smaller (range 100°s maller -7°larger) than on the concave side.
To analyze the effect of GFSI on the RVA, the SMA treatment group was analyzed for an average of 4.6 (range 2.4-8.8)years.The RVA of all ribs was measured, but for a better overview we chose to display ribs 3 to 8 equaling the implant fixation points.Because of the fixation of rib implants, it was hypothesized that these were the ribs with the possibly most severe effect on rib horizontalization (Fig. 4).There was no difference in the RVA before (T0) and after (T1) GFSI implantation, neither on the concave (Fig. 4a) nor on the convex (Fig. 4a') side.Comparison of the RVA before GFSI implantation (T0) and after an average of 4.6 (range 2.4-8.8)years (T2) revealed significantly higher RVA on the concave (Fig. 4b) but not the convex (Fig. 4b') side.Thus, carrying GFSI over the course of several years significantly increased RVA in the area of implant-fixation on the concave but not the convex side.Comparing RVA directly before (T2) and after (T3) GFSI removal, there were unchanged values of the concave side and significantly reduced  values for rib 4-6 on the convex side of the curve (Fig. 4c, c').RVA of rib 4 was on average reduced by 7°(range reduction by 46°-increase by 7°), RVA of rib 5 was on average reduced by 20°(range reduction by 58°-increase by 7°), RVA of rib 6 was on average reduced by 15°(range reduction by 48°-increase by 25°).
To analyze the influence of GFSI on the collapsing parasol deformity, age-matched SMA adolescents with (n = 19) and without (n = 18) prior GFSI treatment were analyzed before spinal fusion (Fig. 2a; T3).Average kyphotic values did not differ significantly between these groups at that time point, average scoliotic values were significantly larger in patients without prior GFSItreatment (Table 1).Comparing both groups, there was no significant difference of RVA both on the concave (Fig. 5a) and on the convex side (Fig. 5b).
To evaluate the influence of GFSI treatment on thoracic and lung volume, available CTs with full thorax images were analyzed of 9 patients with and 16 patients without GFSI treatment.In these subgroups, no significant differences were observed when comparing age, weight, height, BMI, and kyphosis.Scoliosis was significantly larger in patients without prior GFSI-treatment.There were no significant differences in the thoracic (Fig. 5c) and total lung volumes (Fig. 5d).If at all, there was a very slight trend toward reduced thoracic and lung volumes in patients with GFSI treatment.Volumes of the left lung were significantly smaller than volumes of the right lung, which is explained by the anatomy.
Collapsed hemithoraces did not change due to implantation of GFSI (Fig. 6).However, there was an increase of parasol deformity over time (between T1 and T2).Analyzing the parasol score, again, there was no difference between T0 and T1 and between both T3 groups, but an increase of parasol deformity over time (between T1 and T2) despite GFSI treatment (Fig. 6).

Discussion
Most children with SMA type I or II develop spinal deformity in early childhood, which may require early surgical intervention [18].In the past years, implantation of GFSI with bilateral paraspinal rib-to-pelvis fixation was postulated [7] and deformity control in SMA children was shown both in short and long-term follow-up using this surgical technique [19,20].Additional to spinal deformity, thoracic deformity in SMA children is a common sequela.Weak intercostal muscles are unable to oppose the relatively stronger diaphragm resulting in a collapse of the rib cage, a constricted thorax, and hypoplastic lungs [6].This condition is known as the thoracic insufficiency syndrome [21].It was assumed that SMA children with progressive thoracic collapsing parasol deformity and declining lung function [5] may benefit from a paraspinal rib-to-pelvis implant fixation with lateral outriggers due to horizontalization of the ribs, e.g., elevation of the RVAs, caused by implant pressure on the upper rib portion.In general, RVAD strongly correlates with scoliosis angles and RVAD is recommended to guide treatment decisions in idiopathic scoliosis in infantile and juvenile types [15-17, 22, 23].In this paper, RVA in SMA children before, during, and after GFSI treatment was assessed, and data from the last analyzed time point (after GFSI-treatment and before definite spinal fusion) were compared to age-and diseasematched controls without prior GFSI-treatment.
In all SMA children with spinal deformity, an RVA asymmetry between the convex and the convex side was found with smaller values on the convex side.This is consistent with studies on idiopathic scoliosis, where concave RVA were larger than convex values [22,23].For the idiopathic scoliosis patient population, RVA linearly increased with larger scoliotic angles [23].The latter finding could not be reproduced in SMA children with spinal deformity indicating that in the SMA population RVA changes, known as the collapsing parasol deformity of the thorax, are rather related to the disease and muscle weakness itself than to scoliotic deformity.Our findings support the data of Livingston et al. [6] for hypotonic neuromuscular patients.Livingston et al. [6] found no correlation between the parasol deformity and the degree of scoliosis in their patient population.
In our study, bilateral GFSI treatment with significantly positive effects on spinal deformity correction was not able to positively influence the collapsing parasol deformity in SMA children.There was no increase in RVA especially of the more affected convex side directly after surgical intervention, after nearly 5 years of GFSI treatment and after implant removal.Also, in age-and disease-matched groups with and without prior treatment, no differences were seen.Again, these findings are in line with Livingston et al. [6], who did not find any thoracic deformity changes or differences analyzing ribbased in comparison to spine-based GFSI.When directly comparing treatment of SMA patients either by only GFSI or by combined treatment with GFSI and lateral chest wall support using transverse bars and rib cradles, however, Swarup et al. [24] demonstrated that GFSI combined with lateral chest wall support provided significantly better outcomes in hemithorax width compared to patients without chest wall support.According to our data, the hypothesis that GFSI with a cranial rib anchor raises the ribs to a more normal position and thus RVA, Thoracic, and Lung Volumes in SMA Patients with GFSI enlarge the thorax could not be validated.Consequently, thoracic and lung volumes were not influenced or changed by GFSI treatment.One limitation of this study is a small sample size and a further reduction of n-numbers due the limited availability of CT images displaying full thoracic and lung volumes and due to extreme rotation of some vertebrae that did not allow to measure corresponding RVA.Measurements of RVA and RVAD are known to come with a certain error due to inter-and intra-observer variability [25,26].Inaccuracy of radiographic RVA measurements is especially an obstacle in severe scoliosis with spinal rotation.Several studies focus on this problem [27,28] and alternative measurements for parasol rib deformity have been postulated [6].A further limitation of this study is that thoracic and lung volume data cannot be directly related to lung function.This study lacks reliable lung function data to clearly identify the effect of GFSI on lung function in SMA patients.

Fig. 2 . 3 (
Fig. 2. Timeline of patient evaluation and methods used in this investigation.a Schematic representation of analyzed patientgroups and time points.Patients with GFSI-treatment (n = 19) were analyzed at T0, T1, T2, and T3.Patients without GFSI treatment (n = 18) were analyzed at the age-matched time point corresponding to T3, directly before definite spinal fusion.Measurements of RVA were done at all time points, measurements of thoracic and lung volumes were done before spinal fusion in both groups.Numbers in brackets display the average age of patients in years.b Schematic representation of measuring the Mehta angle, e.g., rib-vertebral angles (RVA) for the concave and convex side on radiographs.c Marking of the thorax on a 2D-CT-plane with reconstruction of thoracic volume in the a.p. (c') and lateral view (c'').Left lung on a 2D-CT-plane (d) with reconstruction of lung volume, displayed from different angles (d', d'').a.p., anteriorposterior.

Fig. 5 .
Fig. 5. Comparison of age-mated SMA patients with and without prior GFSI-treatment before definite spinal fusion.Analysis of RVA on the concave (a) and convex (b) side and analysis of thoracic (c) and lung volumes (d).Note that there was no significant difference in any of the analyses between both groups.An unpaired t test was performed to compare the RAV between the two groups at each individual rib and to compare thoracic and lung volume between the two groups.For (a, b), the n-numbers are displayed within the bars.For (c, d) n = 16 without prior GFSItreatment, n = 9 with prior GFSI-treatment.

Fig. 6 .
Fig. 6.Parasol score and collapsed hemithoraces.a, b Measurement and calculation of the parasol score.c Parasol score in GFSItreated group at T0, 1, 2, and 3 and in GFSI untreated group at T3.A paired t test was performed to compare parasol score at T0 with T1, at T1 with T2, and at T2 with T3.An unpaired t test was performed to compare parasol score at T3 between GFSI-treated and GFSI-untreated groups.Note that a significant reduction of the parasol score was observed over time from T1 to T2.There was no difference, neither upon implantation (T0-T1) or explantation (T2-T3) of GFSI, nor between GFSI-treated or GFSI-untreated children at T3. d Percentage of children with collapsed thoraces on both sides or in total (on one or both sides).