Introduction

Adolescent idiopathic scoliosis (AIS) is a common three-dimensional spinal deformity affecting teenagers [1]. Females are 1.4–7.2 times more likely to have AIS than males [2]. While the etiology and risk factors for the development or progression of scoliosis remain inconclusive [3], AIS and/or related treatment may cause back pain, insomnia [4], psychological distress, poor body image [5], and suboptimal quality of life [4].

Patients with AIS are not uncommon to demonstrate compromised pulmonary function [6]. Multiple studies have substantiated the presence of suboptimal pulmonary function (e.g., decreased vital capacity) in these patients [7,8,9]. While up to two-thirds of AIS patients with large scoliotic curves demonstrate restrictive respiratory abnormalities [10], recent research suggests that suboptimal ventilatory function may occur in patients with mild or moderate AIS [7, 11]. Despite the controversy [12, 13], the suboptimal pulmonary function in these patients may be related to the distortion/restriction of the spine and/or thoracic cage [14], locations of the deformity, reduced chest wall mobility [15], and/or obstructive lung disorders secondary to intrathoracic airway compression [16]. Further, patients’ pulmonary function can be compromised by the curve progression [9], and/or bracing [8, 9, 16, 17]. Conversely, these changes may be reversed by aerobic exercises.

Although a recent systematic review with meta-regression analysis revealed that the pulmonary function in patients with idiopathic scoliosis was inversely related to the curve severity [18], this review was limited by the summary of associations between coronal Cobb angles and various pulmonary parameters among patients of different types of idiopathic scoliosis. Because AIS is a three-dimensional spinal deformity, examining these associations based on coronal Cobb angles alone is incomprehensive [19]. Additionally, findings from a mixed cohort of patients with various idiopathic scoliosis cannot be generalized to patients with AIS. The current systematic review and meta-analysis addressed these limitations and summarized the evidence regarding the associations between various pulmonary functions and spinal parameters in patients with AIS, which may help clinicians identify patients at risk of having pulmonary impairment. Therefore, this review aimed to summarize the evidence regarding the: (1) associations between various pulmonary parameters and the severity of scoliosis in AIS patients; and (2) temporal relations between changes in the spinal curve due to progression/conservative treatments and the corresponding changes in pulmonary function.

Methods

This review protocol was registered with PROSPERO (CRD42016043599) and followed the Preferred Reporting Items of Systematic Reviews and Meta-analysis guidelines [20].

Search strategy

Seven databases: PubMed, Embase, PEDro, SPORTDiscus, CINAHL, Cochrane Library, and PsycINFO were searched for potential articles from inception to March 14, 2022. There were no restrictions on languages, but only English, Chinese, and Italian publications were screened. Search terms included keywords related to pulmonary function, spinal deformity, and AIS. Specifically, the Boolean search strings included (“cardiac*” OR “pulmonary” OR “lung” OR “thoracic” OR “cardiopulmonary”) AND (“test*” OR “exam*”) AND (“adolescent*” OR “teen*” OR “puberty” OR “youth”) AND (“AIS” OR “adolescent idiopathic scoliosis”). The detailed search strategy is included in Supplementary Material eTable 1.

Eligibility criteria

Articles were included had they reported an association between pulmonary function and the severity of spinal curve in patients with AIS aged between 10 and 18 years [21]. Longitudinal, cross-sectional, and case–control studies were eligible. Randomized controlled trials were included if they reported the targeted associations in AIS patients preoperatively, or before and/or after conservative treatments. Studies were excluded had they examined patients with scoliosis other than AIS, cognitive impairment, Marfan syndrome, or pectus deformity.

Screening

Three independent reviewers (MK, JY, and RC) paired up to screen titles and abstracts of all identified citations for eligibility. Studies deemed to be eligible by either reviewer were included for full-text screening. Reviewers repeated the same procedure for full-text screening. If disagreements in inclusion could not be resolved by discussion, a senior reviewer (AW) arbitrated the disagreement. The reference lists of all included articles were screened and forward citation tracing was conducted on Scopus to identify additional articles. The corresponding authors of all included studies were contacted by emails to identify omitted studies, or to seek raw data for our meta-analyses.

Data extraction

Two reviewers (MK and RC) independently extracted data from each included paper. Any disagreements were resolved with the third reviewer (AW). The collected data included: (1) study characteristics (e.g., year of publication, study design); (2) participants’ characteristics (e.g., age and gender); (3) absolute values and/or percentage predicted values of pulmonary parameters; (4) spinal/thoracic deformity parameters; and (5) statistical analyses of the associations between (3) and (4). If the included studies conducted subgroup analyses, relevant data were extracted. Missing data was marked as “not reported.” A list of definitions of pulmonary and spinal parameters is shown in eTable 2. This includes common terminology such as Lenke and King's classifications of scoliotic curve [23], angle of trunk rotation [22], surface spinal penetration index [24] and endothoracic hump ratio [25].

Risk of bias assessments

Two independent reviewers (RC and AW) assessed the methodological quality of prospective studies using the Quality in Prognostic Studies (QUIPS) [26], cross-sectional studies using Appraisal tool for Cross-Sectional Studies (AXIS) [27], and case–control studies using Newcastle–Ottawa Scale (NOS) [28]. Studies that retrospectively analyzed data or collected data at a single time point were assessed using AXIS. Any disagreements in the assessment results were resolved by consensus.

Data syntheses

Meta-analyses

The primary measure was the associations (e.g., Spearman’s/Pearson coefficients) between various spinal parameters and pulmonary functions in patients with AIS. The correlation coefficients were interpreted as weak, moderate, and strong if their values were 0.3, 0.5, and 0.7, respectively [29]. If three or more clinically homogenous studies investigated the same association, data were pooled for meta-analysis using random-effects model using the Comprehensive Meta-analysis version 3.0 software (Biostat, NJ, USA). Separate meta-analyses were conducted for studies involving multiple regression models. The significance level was set at 0.05. Statistical heterogeneity of the included studies in meta-analyses was graded as low, moderate, and high if the I2 statistics were ≤ 25%, between 26 and 74%, and ≥ 75%, respectively [30]. If meta-analyses were not conducted, the correlations were summarized narratively.

Subgroup analyses

Subgroup analyses were planned to examine the cross-sectional/longitudinal associations between spinal parameters and pulmonary functions based on: (1) genders; (2) severity of the pulmonary impairment; (3) severity of scoliosis; (3) thoracic or lumbar scoliosis; and (4) before and after conservative treatments or curve progression.

Levels of evidence

Levels of evidence were rated as strong, moderate, limited, and very limited based on established criteria (eTable 3) [31, 32].

Results

Database searches yielded 3723 non-duplicated titles and abstracts for screening. Twenty-one out of 278 full-text articles were included. Further, manual searches of reference lists and forward citation of the included articles yielded six additional included articles (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Study characteristics

Table 1 summarizes characteristics of the 22 included cross-sectional studies [8, 9, 11, 16, 17, 33,34,35,36,37,38,39,40,41,42,43,44,49] and five case–control studies [7, 50,51,52,53] that involved 3,162 participants. All included studies used spirometry and some also used plethysmography [9, 16, 33, 41, 49, 53] to evaluate lung function. The reported pulmonary function parameters included the absolute values of forced vital capacity (FVC), forced expiratory volume in the first second (FEV1), FEV1/FVC ratio, forced expiratory flow at 25% and 75% of FVC (FEF25–75%), FEF25%, FEF50%, peak expiratory flow (PEF), vital capacity (VC), residual volume (RV), total lung capacity (TLC), RV/TLC ratio, functional residual capacity, etc. Some studies reported these parameters as the percentage of predicted values (e.g., %FVC, %FEV1, and %TLC). The predicted values were determined according to age-, gender-, and height-matched normative data [54]. Some included studies used equations to estimate patients’ “actual height” from the arm span [7, 8, 16, 17, 35,36,37, 39, 47, 48, 53] or Cobb angles [11, 42, 45] to predict participants’ pulmonary functions. Pulmonary functions were considered normal if their measured values exceeded 80% of the predicted values [9, 49].

Table 1 Characteristics of the included studies (in alphabetical order of the first author’s surname)
Full size table

Spinal parameters were measured by X-rays or computed tomography scans. Five studies used biplanar X-rays with three-dimensional reconstruction of the spine and/or rib cage [17, 33, 43, 49, 55]. The reported spinal parameters included proximal thoracic Cobb angles [9, 35, 43, 55] main thoracic Cobb angles on anteroposterior radiographs in a standing [7, 8, 16, 17, 29,30,35, 34,35,36,37,38,43, 41,42,43,44,49, 53, 55] or supine bending position [39], thoracic kyphosis angles [9, 35, 41, 43, 45], and apical vertebral rotation angles [17, 39, 43, 49, 50, 55] (Table 1). Most included studies used univariate analyses to determine the associations of interest. Nine studies used multiple regression to evaluate such associations [9, 16, 38, 39, 37,38,39,44, 51] (Table 1).

Risk of bias assessments

All 22 included cross-sectional studies did not justify their sample sizes, nor report the response rate or non-responders’ characteristics [8, 9, 11, 16, 17, 29,30,35, 33,34,35,36,37,38,39,40,41,42,43,44,49, 55] (eTable 4). Five cross-sectional studies [17, 37, 38, 40, 55] did not describe participants’ demographics (e.g., gender distribution) [17, 38, 40], while seven studies [11, 34, 37, 39, 46, 48, 49] did not discuss their limitations. Nine included studies did not mention the ethical approval or the informed consent process [9, 11, 34, 33,34,39, 43, 44, 46]. Similarly, all included case–control studies [7, 46,47,48,53] did not describe the non-response rate, while four of them [7, 47,48,53] did not describe the recruitment process of controls (eTable 5).

Associations between spinal parameters and lung function

Univariate correlations between 43 spinal parameters and 32 pulmonary function parameters were reported (Table 2 and eTable 6). Twenty-seven meta-analyses were conducted to reveal 22 significant correlations. Of them, 20 showed significant but weak correlations. Further, 11 included studies used multivariate analyses to identify independent spinal parameters that predicted pulmonary function [9, 11, 16, 38, 39, 41,42,43,44, 51, 53]. Given the numerous investigated correlations, only significant correlations with at least limited-quality evidence were reported and discussed in this review.

Table 2 Summary of significant correlation between various spinal deformities and pulmonary function parameters
Full size table

No included studies investigated the gender-related correlations between spinal parameters and lung function. No included studies reported the temporal relations between changes in spinal structure and the corresponding changes in pulmonary function in conservatively treated patients with AIS. Although Lin et al. [35] reported that lung function parameters did not significantly differ between AIS patients with and without a history of brace usage, some studies [44, 51] found that compared to “non-brace” patients, those with bracing had poorer lung function. Furthermore, some studies compared patients’ lung function based on different Lenke classification types [47, 48], a Cobb angle cutoff [52], a kyphosis angle cutoff [17, 37], and BMI [38, 42]. Two included studies compared spinal parameters based on the severity of lung impairment [41, 49].

Significant negative univariate correlations

Meta-analyses showed that proximal thoracic Cobb angles were negatively related to %FVC (r =  − 0.194; 95% confidence interval [95% Cl]: − 0.253 to − 0.134) and %FEV1 (r =  − 0.234; 95% Cl: − 0.291 to − 0.175) (Supplementary Material eFigure 1), while main thoracic Cobb angles were negatively associated with %FVC (r =  − 0.302), FVC (r =  − 0.245), %FEV1 (r =  − 0.348), FEV1 (r =  − 0.232), FEV1/FVC ratio (r =  − 0.166), %TLC (r =  − 0.183), TLC (r =  − 0.302), and %VC (r =  − 0.272) (Fig. 2). Similarly, significant negative correlations were noted between the number of involved thoracic vertebrae and %FVC (r =  − 0.262; 95% Cl: − 0.524 to − 0.215) and %FEV1 (r =  − 0.255; 95% Cl: − 0.346 to − 0.159), between main thoracic apical vertebral rotation and %FVC (r =  − 0.215; 95% Cl: − 0.314 to − 0.112) and %TLC (r =  − 0.126; 95% Cl: − 0.240 to − 0.009), between maximum rib hump and FVC (r =  − 0.225; 95% Cl: − 0.367 to − 0.072), as well as between lumbar lordosis and %FVC (r =  − 0.099; 95% Cl: − 0.165 to − 0.032) and %FEV1 (r =  − 0.116; 95% Cl: − 0.182 to − 0.049) (Table 2; eFigures 2–5).

Fig. 2
figure 2

Forest plots of univariate meta-analysis of main thoracic Cobb angles and pulmonary parameters

Full size image

Significant positive univariate correlations

Thoracic kyphosis angles were positively related to %FVC (r = 0.180; 95% Cl, 0.151 to 0.432) and %FEV1(r = 0.193; 95% Cl: 0.007 to 0.365) (Table 2; eFigure 6). Other meta-analyses revealed that higher rib cage thickness (r = 0.377; 95% Cl: 0.155 to 0.562), width (r = 0.635; 95% Cl: 0.486 to 0.748), and volume (r = 0.784; 95% Cl: 0.716 to 0.838) were significantly associated with higher FVC (Table 2; eFigures 7–9).

Reported multivariate analyses

Three meta-analyses showed significant correlations between main thoracic Cobb angles and %FVC (r =  − 0.309), as well as between main thoracic kyphosis angles and FEV1 (r = 0.318) or %FVC (r = 0.226) after adjusting for confounders (Fig. 3). These results were similar to the corresponding meta-analyses of univariate analysis (Fig. 2; eFigure 6).

Fig. 3
figure 3

Forest plots of multivariate meta-analysis

Full size image

Discussion

This is the first systematic review and meta-analysis to summarize the associations between various spinal parameters and pulmonary function parameters in patients with AIS. Limited-quality evidence supports that increased thoracic Cobb angles, number of involved thoracic vertebrae, apical vertebral rotation, rib hump, and lumbar lordotic angles are related to decreased %FVC, whereas increased thoracic kyphosis angles are associated with larger %FVC and %FEV1. Rib cage parameters are positively correlated with FVC.

Scoliosis involves three-dimensional spinal deformity and thoracic cage distortion that may affect each other and worsen lung function [56]. Notably, thoracic cage deformity may alter spinal curvature, and causes rotation and shortening of the thoracic spine, leading to compromised chest wall compliance [57], decreased lung volume under the rib hump, and lung impingement on the concave side. The compressed lung tissues may reduce lung compliance, causing restrictive lung diseases [58]. Similarly, the rotational vertebral deformity may cause thoracic asymmetry [11], which increases the chest wall stiffness [59], reduces the efficiency of respiratory muscles and the diaphragm [8]. The vertebral rotation and rib hump can also cause imbalance in bilateral paraspinal and respiratory muscles [60], limiting the elevation of ribs and reducing lateral and anteroposterior movements of the thoracic cage [11]. These altered chest wall and respiratory muscle mechanics may decrease TLC [61], and increase the risk of hypercapnia, hypoxemia, and alveolar hypoventilation, causing irreversible lung atrophy [57].

While it is well known that patients with thoracic Cobb angles > 50° display clinically significant pulmonary impairments [62], our findings suggest that pulmonary impairment exists even in patients with mild-to-moderate idiopathic scoliosis [53]. However, some patients with severe spinal curvature may not show pulmonary decline if they have good apical vertebral rotational flexibility [39]. Notably, AIS patients with a flexible spine (rotational flexibility > 55%) have normal lung function [56]. Therefore, thoracic curve flexibility should be considered in evaluating the associations.

The decreased %FVC and %VC but a normal FEV1/FVC ratio among patients with AIS in the included studies indicate that they show restrictive lung characteristics [7, 55]. However, there are conflicting findings regarding the relation between spinal deformity and obstructive lung disease in patients with AIS. While one included study reported no significant relation between main thoracic Cobb angles and FEV1/FVC ratio [34], another included study found that 68 out of 176 AIS patients with thoracic Cobb angle > 40° had obstructive lung diseases although no significant correlation between Cobb angles and FEV1/FVC ratio was noted [16]. The latter study also showed that 73% of these 68 patients had irreversible obstructive lung disease that could not be improved by bronchodilator [16]. Although multiple factors (e.g., lower airway malacia, asthma) may lead to obstructive lung characteristics [16], rib cage deformity-related intrathoracic airway compression or respiratory muscle weakness may contribute to such findings [8]. Given the high prevalence of irreversible obstructive lung diseases in patients with moderate to severe AIS, endoscopy or chest imaging may be indicated for this airway obstruction [16].

The consistent findings of significant but weak correlations between various structural characteristics and pulmonary function may be ascribed to no adjustment for confounders (e.g., BMI and duration of bracing). Abnormal mechanical loading of respiratory muscles and altered muscle length-tension relationship can affect respiratory muscle contraction and lung function [7, 63], especially in patients with mild AIS [60]. Higher BMI is associated with better %FVC in teenagers with Cobb angle > 40° [38, 42]. Research found that the association between BMI and %FVC was stronger than those between thoracic Cobb angles or kyphosis angles and %FVC [38]. Heavier teens tend to have greater thoracic kyphosis, which yields better %FVC than hypokyphotic peers [38]. Additionally, one study found that brace wearing temporarily compromised %FEV1 in AIS patients after accounting for thoracic kyphosis [51], although it was unclear whether participants took off the brace during spirometry. Compared to AIS patients without bracing, patients with a thoracic curve and bracing displayed significantly poorer %FVC and %FEV1 [44, 51]. However, the pulmonary function/compliance restores to previous conditions once the brace is removed [64]. Likewise, the negative association between lumbar lordosis and %FVC or %FEV1 might have disappeared if confounders were considered.

This review had some limitations. Because many included studies performed pulmonary function tests on AIS patients preoperatively, their findings may represent patients with more severe curves. Further, most included studies did not define the vertebral levels for classifying the proximal and main thoracic curves, which might introduce discrepancies in our pooled results.

Implications

Most included studies measured anteroposterior and lateral spinal features on radiographs. Future studies should adopt low-dose biplanar X-ray imaging for three-dimensional thoracic cage and spinal structure reconstruction [19], which could capture the three-dimensional impacts of spinal/thoracic deformities on patients with AIS. This allows comprehensive evaluation of the relations between spinal deformities and lung function, which may guide clinical management and research.

While patients with mild AIS may not show respiratory dysfunction at rest, they may display reduced functional capacity [65], or maximum oxygen uptake during exercise tolerance tests [50]. Spirometry may not detect subtle deterioration or dyspnea on exertion, which may indicate scoliosis-related respiratory decline. Clinicians should conduct progressive exercise tests on patients with suspected respiratory impairments to detect early respiratory dysfunction. If the curve progresses, regular progressive exercise tests are recommended [57].

Scoliosis can directly (spinal deformity) or indirectly (respiratory muscle weakness/ inefficiency) affect respiratory function. Although patients with mild-to-moderate scoliosis may not experience dramatic pulmonary impairments during daily activities, it is important to use bracing or physiotherapy scoliosis-specific exercises to prevent or delay curve progression in these patients [64]. However, bracing should be worn for at least 16 h per day to prevent curve progression [57, 66]. Therefore, aerobic training should be prescribed to patients with bracing to optimize their lung functions.

Since pulmonary deficits in AIS patients may worsen with curve progression, patients indicated for surgical correction may experience pulmonary impairment secondary to severe scoliosis [67]. While AIS patients with moderate lung volume are less likely to require postoperative ventilatory support [37], those with moderate or severe defects (< 60% of predicted VC) may indicate high-risk surgical fusion. The latter should undergo full spirometry before surgery. VC can be used as a screening indicator for all patients before spinal surgery [37] because such surgery may adversely affect pulmonary function/compliance [68].

Because there was no included prospective study, the causal relations between changes in spinal/thoracic deformity and changes in lung function remain unclear. Future prospective studies should investigate such relations after adjusting for confounders. Further, as prior research involving AIS patients aged > 18 years revealed that patients had worsening pulmonary function (e.g., FVC) as they aged [67, 69], future prospective research with long-term follow-ups should determine whether AIS patients with near-normal, mild, or moderate lung dysfunction would experience declined lung function and body’s functionality in later life [70].

Conclusions

This systematic review highlights that larger proximal and main thoracic Cobb angles, smaller kyphosis angles, greater lumbar lordotic angles, a longer thoracic curve, a larger rib hump, increased apical vertebral rotation angles and smaller rib cages are associated with poorer pulmonary functions. Other factors can also affect the lung function in these patients. Nevertheless, the clinical impact of scoliosis on lung function is mainly subclinical except for those with severe structural deformity. We definitely need more research to strengthen the quality of evidence. Future prospective studies should evaluate the temporal relations between changes in spinal/thoracic parameters and changes in pulmonary function in order to inform the clinical management of AIS patients with potential respiratory decline.