Following the Hospital Clinical Research Ethics Committee approval, this retrospective study was conducted on patients with DS-NF1 referred for corrective surgery at our institution from October 2011 to November 2018. The diagnosis of dystrophic scoliosis in NF-1 was made using established diagnostic criteria[15, 16]. Enrollment was limited to NF-1 patients with (1) dystrophic scoliosis and concomitant trunk shift causing convex coronal imbalance: coronal balance distance ≥ 3 cm; (2) intact neurological function before surgery; (3) minimum two-year follow-up with complete image data. The exclusion criteria were applied to those with (1) multiple sporadic dystrophic bone defects along the spine causing double or triple curves; (2) solitary dystrophic lesion in sacrum causing compensatory lumbar/thoracolumbar scoliosis and trunk shift; (3) presence of pelvic obliquity due to dystrophic bone defects in the lower limbs. A total 179 DS patients with NF-1 were operated during that time period, and finally, only 15 patients (age, 14.7 ± 4.4yrs; range, 10-26yrs; 7 males and 8 females; mean follow-up, 3.3± 1.5yrs) who fulfilled the inclusion and exclusion criteria were enrolled in this study. Their medical records, imaging scans, and operative reports were reviewed. The data collected include preoperative, postoperative and final main curve Cobb and kyphotic angles, patterns of convex CI, apex location, presence of vertebral rotatory subluxation (defined as a classic double-vertebrae sign on the axial computed tomography images)[17, 18], coronal balance distance (CBD), sagittal vertical axis (SVA), surgical strategies, fusion segments, implant density, ratio of laminar hook, postoperative neurological status, and surgical complications. Curve flexibility was not assessed for this special patient subgroup. This was attributable to the potential risk of neurological impairments if side bending movements were performed hinging the unstable apical region.
Classification of convex coronal imbalance
Antagonistic role of the upper and lower hemi-curve in the formation of coronal imbalance could be easily recognized: the lower hemi-curve played the role of imbalance driver causing the trunk shifting to the convex side, whereas the upper hemi-curve served the role of imbalance compensator. CI would occur if the upper compensation was distinctly insufficient. Based on the location of curve apex and the compensatory behavior of the upper hemi-curve, four types of CCI were determined:
Type 1: Thoracic convex coronal imbalance
For Type 1 CCI, the apex located above T12, and there were sufficient non-dystrophic vertebrae (≥3) locating distal to apex. The subtypes were identified according to the morphology of the upper hemi-curve.
Type 1A: The morphology of the upper hemi-curve was straight and vertical, indicating no compensation for CCI (Fig 1a).
Type 1B: The morphology of the upper hemi-curve was curved and inclined, moving the deviated torso partially back to midline (Fig 1b).
Type 2: Thoracolumbar/Lumbar convex coronal imbalance
For Type 2 CCI, the apex located between T12 and L4. There were limited (Thoracolumbar, apex between T12-L1) or insufficient (Lumbar, apex between L2-L4) non-dystrophic vertebrae (≤3) locating distal to apex. The subtypes were identified in a similar way.
Type 2A: The morphology of the upper hemi-curve was straight and vertical (Fig 1c).
Type 2B: The morphology of the upper hemi-curve was curved and inclined (Fig 1d).
All the recruited patients were stratified according to the above-mentioned classification. The concrete distribution of each CCI type was as follows: Type 1A: 1 case (6.7%); Type 1B: 3 cases (20.0%); Type 2A: 3 cases (20.0%); Type 2B: 8 cases (53.3%). Among them, 1 patient in thoracic CCI group (Type1B) and 3 patients in thoracolumbar/lumbar CCI group (Type2B) received staged surgery with combined posterior-anterior or anterior- posterior approach (Table 1), while the rest 11 patients (73.3%) underwent posterior-only spinal instrumentation and fusion. Supplementary anterior fusion utilizing structural fibular allograft (2 patients) (Fig 3) or autogenous rib grafts (1 patient) (Fig 4) was applied when the pedicle screw density in the apical region was distinctively low due to pedicle dystrophy. Stage 1 anterior release involving intervertebral disc resection and autogenous rib grafting was performed in 1 Type 2B patient, followed by skull-femoral traction for 2 weeks and subsequent stage 2 posterior spinal correction and fusion.
Spinal traction was indicated if Cobb angle > 90° or kyphosis > 80°. Aside from the aforementioned one Type 2 patient (9.1%), spinal traction was also applied for another 3 patients in Type 1 group (75%). Among them, two received halo-gravity traction (HGT) using a halo-wheelchair for one month, while the third one was applied with skull-femoral traction in bed for 2 weeks before posterior surgery. Posterior-only spinal instrumentation and fusion was performed with all pedicle screw or hybrid constructs, and was assisted with satellite rod technique for two patients being operated in late stage.
Radiographic assessments
The following parameters were also measured pre- and post-operatively for analysis of the separate contribution of the upper and lower hemi-curve to CI onset and prognosis: lower arc translation, upper arc translation, lower arc inclination, upper arc inclination, UIV (upper instrumented vertebra) tilt, LIV (lower instrumented vertebra) tilt, UIV translation and instrumentation mass inclination. All measurements were performed using the Surgimap spine software (Version 2.3.1.5; Spine Software, New York, NY). The definitions of the standard measuring techniques of the aforementioned parameters were defined as follows:
- Lower arc translation (LAT, Fig. 2a): defined as the distance from the center of apex to the center sacral vertical line (CSVL).
- Upper arc translation (UAT, Fig. 2b): defined as the distance from the center of UIV to the vertical line crossing the center of apex.
- Lower arc inclination (LAI, Fig. 2c): defined by the angle formed between the line drawn from the center of LIV to the center of apex and the vertical line.
- Upper arc inclination (UAI, Fig. 2d): defined by the angle formed between the line drawn from the center of UIV to the center of apex and the vertical line.
- UIV tilt: defined as the angle formed by the line drawn parallel to the superior end plate of the UIV and the horizontal line.
- LIV tilt: defined as the angle formed by the line drawn parallel to the superior end plate of the LIV and the horizontal line.
- UIV translation (Fig. 2e): defined as the distance from the center of UIV to the CSVL.
- Instrumentation mass inclination (Fig. 2f): defined by the angle formed between the line drawn from the center of UIV to the center of LIV and the vertical line.
The recruited patients were assigned into two groups according to whether or not the post-op residual CBD exceeded 3cm: the balanced group (post-op CBD <3cm) and the imbalanced group (post-op CBD ≥ 3cm). The operative changes of the translation and inclination for the upper and lower hemi-curves were recorded as △UAT, △LAT, △UAI and △LAI, respectively. Considering the antagonistic role of the upper and lower hemi-curve in the rebalance for coronal malalignment, the ratio of △UAT/△LAT and ratio of △UAI/△LAI were both calculated, which were indicative of whether the separate corrections of upper and lower hemi-curve were matched and cooperative or mismatched and antagonistic.
Statistical Analysis
Data analysis was conducted using statistical software (SPSS 20.0, SPSS Inc., Chicago, IL). Statistical data are presented as the mean ± standard deviation. Independent-sample t test was applied to compare the deformity parameters between the balanced and imbalanced groups. Comparisons between post-op and follow-up parameters were made by paired t test. Based on our clinical experience and meanwhile to avoid the pitfall of multicollinearity between parameters, △UAI/△LAI, △UAT/△LAT, △(UIV tilt + LIV tilt) and main curve correction rate were selected to explore whether they could contribute to the correction of CBD. A p value < 0.05 was considered statistically significant.