Impact of fusion for adolescent idiopathic scoliosis on lung volume measured with computed tomography

Impaired pulmonary function is a common complication of adolescent idiopathic scoliosis (AIS), due to deformity of the spine and ribcage [1, 2]. Several factors that can potentially affect pulmonary function in patients with AIS have been identified, including structural features, the proximal thoracic curve, apex displacement, sagittal hypokyphosis and even thoracic lordosis, and the number of vertebrae involved in the major curve [2, 3]. One of purposes of corrective surgery for AIS is to prevent the deterioration in pulmonary function caused by progressive spinal deformity. Particularly, posterior spinal fusion with instrumentation (PSF) has been shown to yield small or moderate increases in pulmonary function in AIS patients [4‐6]. However, the underlying mechanisms of pulmonary impairment in AIS patients, and of its postoperative improvement, is not fully understood.

Pulmonary function testing (PFT) is commonly used to evaluate the effect of obstructive and restrictive pulmonary diseases on respiratory function [7, 8]. However, AIS is often initially assessed in children who are too young to cooperate with the PFT protocol [9]. Computed tomography (CT) can accurately measure actual lung volume (LV) in this population [3, 10‐14].

Although studies have reported that corrective surgery for AIS improves pulmonary function, two studies found no significant postoperative improvements in total LV among patients with AIS who underwent PSF [11, 13], and Lee et al. reported a significant decline in total LV at 4 weeks post-PSF in this population [12]. However, given the relatively small number of subjects and short follow-up periods in these studies, they might not represent the ultimate outcomes of these patients.

Between 2007 and 2014, patients who were treated surgically for AIS at our institution underwent low-dose CT scans prior to surgery, to develop the surgical strategy, and 2 years after surgery to detect postoperative screw loosening and perforation especially in the upper and lower instrumented vertebrae [15]. Thus, we were able to review detailed LV data for a relatively large pool of AIS patients treated by PSF. For this analysis, we used an established technique for reconstructing pulmonary anatomy from low-dose CT scans [10, 16]. We first analyzed differences between the preoperative and postoperative LV in the AIS patients in this dataset. We then sought to identify the factors associated with a postoperative LV reduction in this population.

Preoperative and postoperative radiographic parameters, PFT data, and LV Table 2 shows the preoperative and 2-year postoperative radiographic parameters and PFT data used in this evaluation. From the preoperative assessment to the 2-year follow-up, the average Cobb angle decreased significantly for the proximal thoracic (27.8° ± 9.3° vs 13.9° ± 6.6°; p < 0.01), main thoracic (54.9° ± 10.4° vs 17.4° ± 7.5°; p < 0.01), and thoracolumbar/lumbar curves (34.0° ± 13.4° vs 10.2° ± 7.9°; p < 0.01). We also observed significant differences between the preoperative and 2-year postoperative coronal balance (0.6 ± 16.8 mm vs − 4.6 ± 10.7 mm, p = 0.01) and apical translation (43.8 ± 37.8 mm vs − 2.6 ± 18.6 mm; p < 0.01). For sagittal radiographic parameters, we observed an increase, although not significant, in the magnitude of upper thoracic kyphosis (T2–T5; p = 0.07) and main thoracic kyphosis (T5–T12; p = 0.10) at the 2-year follow-up. Meanwhile, the average value of SVA was significantly decreased from the preoperative assessment to the 2-year follow-up ( − 16.9 ± 22.4 mm vs − 25.9 ± 21.0 mm; p < 0.01). PFT showed a significant decrease in %VC (p < 0.01) and significant increases in VC (p = 0.02), FEV1 (p < 0.01), and FEV 1% (p < 0.01) at the 2-year follow-up. Table 3 shows the preoperative and postoperative LV values. TLV increased from the preoperative evaluation (2.75 ± 0.65 L) to the 2-year follow-up (2.82 ± 0.65 L); this increase was marginally significant (p = 0.06). LLV increased significantly from the preoperative (1.24 ± 0.31 L) to the postoperative evaluation (1.29 ± 0.32 L; p = 0.01); an increase in RLV (from 1.51 ± 0.36 L to 1.53 ± 0.34 L) was not significant (p = 0.25). The RLV/LLV ratio decreased significantly from the preoperative (1.22 ± 0.15) to the 2-year postoperative evaluation (1.20 ± 0.11; p = 0.02).

Several studies have analyzed PFT data in patients with AIS. However, few have used CT images to measure and analyze preoperative and postoperative LV values [11, 13, 19]. To the best of our knowledge, this retrospective study is the largest to evaluate LV values after PSF in a population of patients with AIS (n = 111) who were followed at least 2 years [11, 13]. At the 2-year follow-up, TLV had increased, although marginally significant. Importantly, there was a significant increase in LLV but not RLV. In addition, our study is the first to examine risk factors for postoperative TLV reduction. Although our findings should be validated with a larger sample, the present study indicates that a longer fusion length ( ≥ 11 levels) may be a risk factor for a postoperative reduction in TLV.

Our results demonstrated significant associations between TLV and PFT data, including VC and FEV1, verifying TLV as an indicator of pulmonary function in patients with AIS. Whereas PFT evaluates dynamic breathing parameters involving the movement of the chest and rib cage, such as deep expiration and inspiration, LV calculated by reconstructing CT scans represents the static component of the lungs. The CT scan method is particularly useful for measuring LV accurately in patients who are unable to follow instructions for PFT [3, 10‐14], and it has the added advantage of measuring the right and left sides separately. CT-based LV measurements can provide further our understanding of anatomical characteristics associated with scoliosis and marked left–right asymmetry. Sarwahi et al. [13] assessed LV in 29 AIS patients and reported that some LV values were not altered after PSF. Similarly, our study indicated that PSF did not significantly increase the TLV. However, a more detailed analysis showed a significant postoperative increase in the LV at the concave side of the major curve, but not at the convex side. One possible explanation for this observation is that vertebral de-rotation for correction expands the LV at the concave side and contracts that of the convex side.

Karol et al. [20] reported that the proximal extent of the fusion increases the risk of developing restrictive pulmonary disease in patients younger than 9 years who were treated by thoracic spine fusion for non-neuromuscular scoliosis. In contrast, we did not find a significant correlation between UIV level and postoperative LV reduction in AIS patients. Demura et al. [21] reported that the presence of an upper thoracic curve in AIS was associated with decreased PFT values at the 2-year follow-up. However, in the present study, the presence of an upper thoracic curve (Lenke type 2) was not significantly associated with a postoperative LV reduction. Although these discrepancies between studies might be due to differences in object variables and analysis methods, these issues deserve further investigation. In the present study, univariable logistic regression analysis identified a fusion length of ≥ 11 vertebrae as a potential factor in postoperative LV reduction. Because cases with a construct ≥ 11 levels have more severe deformity compared with those with a construct < 11 levels (Supplementary Table 1), multivariable analysis was conducted by adjusting the factors including preoperative Cobb angle of main thoracic curves. Breathing is partially affected by spinal movement [22], which becomes more restricted as the fusion length increases, and this condition might be a factor in reducing the postoperative LV in AIS patients. Thus, our results suggest that using as short a fusion area as possible may help maintain LV after PSF.

This study had several limitations. This was a single-center study, with potential selection bias. Second limitation was a lack of direct comparison with anticipated changes in LV of a normal age-matched population or of a matched AIS group that did not have PSF. The normative reference data of Gollogly et al. [16] can be used for comparison. However, their study lacks the data of AIS patients aged 10–20 years. Without benchmarks, it is difficult to assess exactly whether PSF had any impact on LV. Third, despite being the largest study of LV values after PSF [11‐13], the sample size might not be large enough to draw solid conclusions from our findings. Finally, we do not recommend the routine use of CT to assess LV due to concerns about increasing the radiation burden, especially given recent reports of risks of induced malignancy [23]. Nevertheless, the present study successfully demonstrated the postoperative change in LV values after PSF in a relatively large subject sample.

In conclusion, we retrospectively analyzed LV prior to and 2 years after PSF in patients with AIS and conducted the first-ever multivariable analysis of risk factors for postoperative LV reduction. Our results indicate that a longer fusion area may lead to a postoperative LV reduction. Therefore, avoidance of a longer fusion area of ≥ 11 will be preferable for preserving lung volume.