Introduction

Scoliosis, which affects six to nine million individuals in the United States [1–4], is defined by the Cobb angle of spine curvature in the coronal plane, and is often accompanied by vertebral rotation in the transverse plane and hypokyphosis in the sagittal plane. Although the coronal plane deformity is the main concern in the diagnosis of scoliosis, this complex three-dimensional thoracic deformity affects both the spinal column and the rib cage due to secondary anatomical interconnections [2–8]. These abnormalities in the spine, costal-vertebral joints, and the rib cage produce a ‘convex’ and ‘concave’ hemithorax [9, 10], and sternal deviation relative to the apical vertebrae helps define the transverse plane rotational deformity [11]. If thoracic deformity progresses to the point that respiration, lung growth or biomechanical motions are compromised, the condition is called thoracic insufficiency syndrome (TIS) [12].

Historically, the treatment for progressive adolescent idiopathic scoliosis AIS has been spinal fusion with instrumentation, which primarily aims to restore a balanced spine position [13]. With more severe cases of scoliosis, thoracoplasty and rib resection are performed to help reconstruct the thoracic cage to correct the thoracic distortion [14–16]. To support this correction and to preserve spine growth, the vertical expandable prosthetic titanium rib (VEPTR) has been used to preserve spinal and thoracic growth in children too young to be candidates for fusion [17, 18]. While physical therapy and bracing are used to treat milder forms of scoliosis to maintain cosmesis and avoid surgery [19], the intended goals of acute surgical interventions are to address not only the skeletal deformities, but also the functional outcomes. Determining the optimal time to intervene, however, requires a broad understanding of both the underlying thoracospinal disorder, its progression, and how it impacts pulmonary function. There is a paucity of information on the inter-relationship between spine deformities, thoracic cage shape and pulmonary function.

Despite extensive literature emphasizing the 3D complexity of AIS and subsequent rib cage deformity [11, 20–32], current pre-surgical planning uses 2D coronal and sagittal plane radiographs to assess deformity. In an effort to characterize vertebral, spinal, and rib cage deformity in the transverse plane, several thorax deformity parameters have been proposed in the literature [11, 20–32]. A large discontinuity exists between the scoliosis research society (SRS) glossary of spine deformity and the published thorax deformity quantification parameter studies. While the SRS glossary is an extensive compendium on spine deformity evaluation methods, it does not include thorax deformity measures and their respective correlations with spine deformity [33]. A comprehensive collection of both spine and thorax deformity parameters along with their associated correlations is required to better describe these 3D deformities. This review serves as a collection of radiographic and CT-based parameters developed to assess thoracic deformity organized by similar features.

Methods

A comprehensive search was performed using the PUBMED search engine for publications on radiographic and CT-based parameters developed in the assessment of the skeletal deformities associated with scoliosis. The PUBMED search engine was mined using the following key words in various combinations: scoliosis, thorax deformity, thoracic insufficiency syndrome, computer tomography parameters, and radiographic parameters. Articles not written in English were excluded.

Results

Thorax deformity in the anterior-posterior and medial—lateral planes is primarily assessed for clinical purposes using planar radiographs [11, 26, 28, 31]. CT imaging, the gold standard for transverse plane thorax deformity characterization, is also used to complement AP and lateral radiographs [6, 20, 22–29, 31]. A comprehensive literature review on the quantification of thorax deformity was performed, and a total of 25 thorax deformity parameters were compiled into eight independent categories based on their similarities of deformity assessment (Table 1). The categories of thorax deformity (in alphabetical order) are: (1) Anterior chest angulation, (2) Area enclosed by rib cage, (3) Coronal asymmetry, (4) Hemithorax depth asymmetry, (5) Hemithorax width asymmetry, (6) Posterior rib asymmetry, (7) Sagittal depth, and (8) Sternum deviation. Figure 1 describes the anatomical landmarks used to quantify thorax deformity using radiographic and CT-based measures.

Table 1 Radiographic and CT-based thorax deformity parameters
Full size table
Fig. 1
figure 1

Radiographic and CT-Based landmark data

Full size image

Relationship of thorax deformity parameters with Cobb angle and vertebral rotation

Scoliosis-induced lateral spine curvature in the coronal plane and vertebral rotation are commonly used for clinical assessment and their inter-relationship has been widely documented [6, 26, 27, 30, 31, 34]. Furthermore, additional evidence supports the subsequent contribution of spine distortion towards progression of thoracic cage deformity [2–5]. It is important to understand the associations that may exist between primary spine deformity and secondary thorax deformity parameters to validate and establish clinical relevance. Therefore, a comprehensive literature review was performed to study the established relationships of thorax deformity parameters with spine curvature and vertebral rotation. The results are organized by thorax deformity category and are summarized below.

Anterior chest angulation

Anterior chest wall angle was shown to significantly correlate with Cobb angle (r = 0.377, p < 0.001), but not with vertebral rotation [27]. It also noted that the most severe anterior chest wall deformity occurred in patients with the apical level at T9. Sternal Tilt was developed by Hong et al. [24] to measure the angulation of the sternum in PE patients with AIS. However, Sternal Tilt did not correlate to Cobb angle and its relationship with vertebral rotation has not been studied. Angle of Sternum Relative to Apical Vertebrae was found to significantly correlate with Cobb angle and vertebral rotation (r = −0.401, p < 0.001 and r = −0.757, p < 0.001, respectively) [27]. Despite coupling that may exist between apical vertebral rotation, rib head deformity, and anterior chest angulation in AIS, the aforementioned literature data may be influenced by the causal nature of vertebral rotation on the angle of sternum relative to apical vertebrae.

Area enclosed by rib cage

Kyphosis-lordosis index from T6–T12 have been shown to significantly correlate with Cobb angle (r = −0.25 to −0.405, p < 0.01), but was not studied to correlate with vertebral rotation [31]. However, no significant correlation was observed between pectus index [23] and Cobb angle or vertebral rotation [24]. Haller index [25], Frontosagittal index [25], and transverse diameter [20] were not studied to correlate with either Cobb angle or vertebral rotation.

Coronal asymmetry

Apical rib vertebral angle difference did not significantly correlate with Cobb angle or vertebral rotation [26]. The relationship between space available for the lung and spine deformity measures have not yet been studied.

Hemithorax depth asymmetry

CT-based measures are used to clinically assess hemithorax depth asymmetry in pectus excavatum (PE) patients. PE is a congenital anterior thoracic cage deformity that has been shown to affect cardiopulmonary function due to the significant reduction in chest volume. While there exists a 22.58 % increase (38.46 % for female) in incidence of adolescent idiopathic scoliosis (AIS) within the PE population [24], hemithorax depth asymmetry measurements are not assessed in AIS subjects. So, there is a lack of correlative analysis associating asymmetry index, chest asymmetry index, and chest flatness index with lateral curvature or vertebral rotation.

Hemithorax width asymmetry

Sternum-rib ratio was shown to significantly correlate with both Cobb angle and vertebral rotation (r = 0.514, p < 0.001 and r = 0.213, p < 0.05, respectively) [27]. Apical vertebral body-rib ratio was also found to correlate with Cobb and vertebral rotation (r = 0.57, p < 0.005 and r = 0.49, p < 0.005, respectively) [26]. Posterior hemithoracic symmetry ratio [11] has not been assessed for its relationship with Cobb angle or vertebral rotation.

Posterior rib rotation

The rib hump index originally defined by Aaro et al. [20] but later modified by Takahashi et al., both consider latero-lateral axis width measurements and showed significant correlation with lateral spine curvature (Aaro: r = 0.601, p < 0.001, Takahashi: T6–T12, r = 0.306–0.507, p < 0.001) and vertebral rotation (Aaro: r = 0.36, p < 0.02, Takahashi: not studied) [20, 30, 31]. Rib hump, based on lateral radiographic measurements was shown to correlate with Cobb angle (r = 0.65, p < 0.0001) and vertebral rotation (r = 0.53, p < 0.002) [26]. If vertebral rotation leads to rib deformity, the most deformed ribs would be coupled with the apical vertebrae. However, there exists a systematic flaw in the use of CT-based measurements to assess posterior rib asymmetry, in that the posterior ribs observed in the axial ‘slices’ through the apical vertebrae connect to vertebrae superior to the apical vertebrae due to the natural droop of the costals.

Posterior rib rotation and back surface rotation demonstrated significant correlation with Cobb angle (r = 0.63 and 0.88 respectively, p < 0.05) and vertebral rotation (r = 0.63 and 0.77 respectively, p < 0.05) [30]. While Mao et al. reported a significant correlation between Cobb angle and Angle of Trunk rotation (r = 0.517, p < 0.05); Erukla et al. did not observe any relationship. More recently, Carlson et al. [21] used a Bunnell scoliometer to measure Angle of Trunk Inclination and found significant correlations with Cobb angle (r = 0.711, p < 0.004) and vertebral rotation (r = 0.53, p < 0.02).

Sagittal depth

Takahashi et al. [31] assessed the sagittal diameter at every vertebral level, and reported significant negative correlations between sagittal diameter and lateral curvature at the apical vertebra (T9: r = −0.218, p < 0.022) and the adjacent vertebral levels. Although not quantitatively assessed, it was also noted that vertebral rotation was most severe at the same vertebral levels that corresponded with sagittal diameter [31]. Sternovertebral distance [25] has never been correlated with Cobb or vertebral rotation.

Sternum deviation

Midline deviation was shown to significantly correlate with Cobb angle (r = 0.76, p < 0.01) [20, 34] and was not studied further to determine relationship with vertebral rotation. Vertebral translation has been shown to correlate significantly with both Cobb angle and vertebral rotation (r = 0.657, p < 0.001 and r = 0.317, p < 0.006, respectively) [6]. The relationships between thoracic rotation [11] and either Cobb angle or vertebral rotation have not been studied.

Discussion

Three-dimensional imaging methods and deformity characterization are essential in guiding surgical restoration of abnormal spine curvature, thoracic volume, symmetry, and function [35]. CT and MRI techniques may provide a level of detail needed to assess out-of-plane spine rotation and thoracic volume changes that often accompany scoliosis. Currently, AP radiography is the gold standard to assess scoliosis, and it is a favored imaging method due to low radiation exposure, making it very practical for longitudinal observations. Although sufficient in assessing spine curvature in the coronal and sagittal planes, AP radiographs are inadequate for describing the vertebral rotation and rib cage distortions; which can be better assessed using transverse plane measurements [20].

The current literature on scoliosis-induced thorax deformity characterization is mainly focused on CT-derived transverse plane measurements. Due to the more pronounced caudal rotation of the ribs in scoliosis, a transverse image alone may not capture the true thoracic distortion of the deformity [34]. With a limited number of studies evaluating the relationship of transverse measurements with AP and lateral radiographs, a comprehensive understanding of the 3D thoracic deformity does not exist [6]. Such knowledge is especially critical, and timely, since we routinely use space available for the lung (SAL), a 2D assessment as a preoperative indicator for VEPTR implantation [11]. Although the VEPTR corrects hemithorax height asymmetry in the coronal plane, the impact of such correction on thoracic measurements in the transverse plane has not been well-studied.

With these limitations, AP radiographs are still used to assess and describe scoliosis-induced thorax deformity with parameters such as rib hump, apical vertebral body-rib ratio, rib-vertebral angle difference and space available for the lung [11, 26, 28]; measures used as proxy for 3D thorax deformity characterization without sufficient validation in the literature. While rib hump and apical vertebral body-rib ratio were shown to correlate significantly with Cobb angle and vertebral rotation. There was no correlation between Cobb angle and there were no comparisons made between Cobb angle and Space Available for the Lung [26, 28]. These findings indicate the inherent limitations of 2D imaging-based measures in describing a 3D deformity.

The current review of thorax deformity quantification presents only 2D measures derived from 3D imaging, and subsequently does capture the true 3D thorax deformity. To our knowledge, crude thoracic volume index (CTVI) is the only parameter that assesses 3D volume related changes in the thorax [36], and was shown to correlate well with functional pulmonary output. Future studies could combine planar data derived from any imaging modality to comprehensively describe changes in thoracic volume.

While surgical interventions for scoliosis primarily focus on correcting abnormal spine curvature, the ultimate goal is to restore thoracic function, assessed by improvements in thorax geometry, rib cage motion and pulmonary function [37, 38]. The impact of 2D geometry measurements on rib cage motion and thoracic biomechanics is incomplete. Although the effect of 2D deformity measures such as Cobb angle and vertebral rotation on pulmonary function has been extensively studied, these efforts do not take into consideration the interplay between spine deformity and subsequent thorax distortion [38–41]. Upadhyay et al. [40] reported that the patients with rotational flexibility less than 55 %, thoracic kyphosis less than 15°, and rib-vertebral angle asymmetry greater than 25° had relatively poor pulmonary function; with 85 % of patients meeting two of the three measurements having a vital capacity of 70 % or less of their predicted values. While these data could significantly contribute towards clinical prognosis, they provide limited information to further our understanding of changes in pulmonary function due to 3D thorax deformity.

Although radiographic measurements are clinically the most feasible, the development of 3D geometric characterization based on imaging modalities such as CT and dynamic MRI may be valuable. However, with CT-based methods, there is a growing concern related to an increased risk of radiation in children whose conditions may require longitudinal imaging studies. More recently, optimization of CT radiation dosage and scan speed have successfully shown to reduce radiation exposure of the whole spine from an average effective dose of 6 to 0.3 mSv without compromising effective clinical evaluation of scoliotic deformity [42].

Despite these advances in CT protocol, radiation exposure still remains 2–3 times greater than that of a plain radiograph for the whole spine. Additionally, there may be measurement errors due to subject position–i.e. standing versus supine [20]. In comparison with standing radiographic measurements of Cobb angle and vertebral rotation, Yazici et al. [43] observed a 29.78 % and 24.39 % decrease when the patients were imaged using a supine CT. More recently, Lee et al. [44] also reported that a supine MRI underestimate Cobb angle measured using radiographs by an average of 10°. However, these results may be confounded by differences in imaging modalities used. Alleviating radiation and subject-position related limitations, newer imaging methods such as EOS® (EOS Imaging, Paris, France) offer high-quality biplanar radiographs of the entire body in the standing position with an ultra-low radiation dose [45, 46]. Whole-body, weight bearing, erect radiographs provide accurate relationships of scoliotic deformities, compensatory curves, global sagittal balance, trunk and rib cage relationships, and describe the complex interplay of the spine and pelvic (pelvic incidence) [47–50]. While current dynamic MRI techniques enable position- and time-variant functional assessment of the spine and thorax with no radiation exposure, future advancements in medical image processing methods may also help render real-time biplanar radiograph-based deformity characterization [51–53].

Conclusion

Practitioners and researchers involved in the treatment of children with scoliosis need to have a thorough understanding of the three-dimensional nature of scoliosis and other spinal deformities and relative changes in the rib cage and thorax. This review serves as the first comprehensive summary of thorax deformity quantification measures. Continued development of advanced imaging and further work in uncovering the complex relationships of spinal deformity, rib cage development and pulmonary function will optimize the treatment of these children and improve their quality of life.