Sagittal malalignment causes pain and diminished health-related quality of life (HRQOL) in spinal deformity patients [10,14,21,45]. Historically, deformity correction has focused on correcting scoliosis and preventing scoliotic curve progression. Spinopelvic and lower extremity parameters have been relatively in frequently evaluated when treating spinal deformities. Compensatory mechanisms are recruited not only in the spinal segment, but also in the pelvis and lower extremities to maintain a horizontal gaze with an erect spine [4,25]. To evaluate sagittal plane alignment, regional/global spinal-, spinopelvic alignment, and compensated knee and ankle f lexion should be evaluated [19]. The pelvis is important as an interpolated unit between the spine and the lower extremities. Failure to account for pelvic alignment when treating a spinal deformity increases the risk of spinal misalignment, decompensation, and treatment failure [2]. The reciprocal change of cervical lordosis (CL) following deformity progression in the thoracolumbar spine is significantly correlated with axial pain and accelerated disc degeneration [16]. Therefore, these changes of the mobile cervical spine in patients with a thoracolumbar deformity should be considered in order to prevent secondary cervical spine disorders.

Some realignment goals have been established for deformity correction surgery, including C7 sagittal vertical axis (SVA) (sagittal distance between the C7 plumb line (PL) and the vertical line through the postero-superior corner of the S1 endplate) <50 mm, PT <22°, and pelvic incidence (PI) – LL <±9° [15]. However, since there are no clear goals to be achieved in terms of CL, and it has become a generally accepted rule to correct cervical kyphosis to be as close to neutral as possible. A trend has emerged to define cervical sagittal alignment parameters similar to the accepted C7 SVA, which is traditionally used to measure sagittal alignment of the thoracolumbar spine [39]. T1 slope (TS), thoracic inlet angle, the C2 PL, and chin-brow to vertical angle (CBVA) are also being used with increasing frequency [38,42]. The CBVA is defined as the angle subtended between a line drawn from the patient’s chin to brow and a vertical line. Evaluating the CBVA requires a clinical whole skeletal photograph of the patient standing with hips and knees extended while the neck is in the neutral position [23,24].

EOS is a system that uses a gas particle detector using a multiwire proportional chamber, for which the Nobel Prize was awarded in 1992. The chamber, which generates a secondary flow of electrons, is placed between the X-ray and the distal detector. This in turn stimulates the distal detector that gives rise to the digital image. This secondarily increase in photons explains how a low-dose primary X-ray beam is sufficient to display a high-quality digital radiograph [46]. Conventional radiography causes distortion between the center and the edges of the radiograph, which increases the risk of size measurement error for structures located far from the central region [8]. The major advantage of the EOS system is that it enables single-shot X-ray imaging of the spine, pelvis, and lower extremities under weight-bearing condition without image distortion. The image distortion and standard error of measurement of the EOS only accounted for 0.5% to 1.5% of the total length measurement. Thus this variability would likely have minimal clinical impact in comparison to 15% in conventional X-ray [40]. This system takes simultaneous anteroposterior and lateral two-dimensional images of the entire body and makes it possible to obtain a three-dimensional reconstruction of the bones with a relatively low dose of radiation (50–80% less than that of conventional X-rays) [31].

The “cone of economy” concept refer to a stable region in the standing posture that requires a minimum of muscle activity in normal conditions [11]. Hasegawa et al. [17] reported the standing sagittal alignment of whole axial skeletons using EOS in 158 healthy human volunteers (mean age, 39.7 years; range, 20–69 years). They defined the center of the head as the center of acoustic meatus (CAM) (Fig. 1). The gravity line (GL), defined as a line through the center of vertical pressure, was measured by a force plate (ANIMA^®, Tokyo, Japan). The mean offset of the CAM was 0. The mean offset of the cervical and thoracic vertebrae was posterior to the GL with the apex of thoracic kyphosis at T7. Thoracic kyphosis changed to LL at the level of L2. The alignment changed to posterior through the apex of the LL at L4, and the base of the sacrum was just posterior to the GL. The hip axis was 1.4 cm anterior to the GL, the knee was 2.4 cm posterior to the GL, and the ankle was 4.8 cm posterior to the GL.

Posterior shifting of the lower lumbar vertebrae and sacrum increases with age, inducing loss of lordosis and an increase in PT. The C7 SVA increases with age. No correlation has been found between CAM–GL and age. Aging induces trunk stooping, but the global alignment compensates for by an increase in CL, PT, PI and knee flexion. This suggests that the horizontal gaze remains constant with age in healthy humans despite trunk stooping by compensatory changes in the cervical region, pelvis, and lower extremities.

Previously, PI has been believed to remain consistent with age. PI reflects the anatomical configuration of the pelvis and is an important parameter of sagittal balance that provides a target for optimal LL in thoracolumbar reconstruction surgery. Since the sacroiliac joint is thought not to move, PI has been considered to stable with age. However, some recent studies found that PI tended to increase with age [17,44]. It is thought that movement of the sacroiliac joint might increase due to osteoarthritic changes, but further study is needed.

Patients with thoracolumbar malalignment exhibit compensatory changes, including cervical hyperlordosis, pelvic shift, knee flexion, hip extension and pelvic retroversion [29]. Ha et al. [16] studied 49 patients undergoing thoracolumbar spine surgery results in reciprocal changes in GSA. They suggested that cervical alignment is affected by thoracic regional alignment, and that GSA and reciprocal changes are correlated with the magnitude of sagittal deformities. TS is correlated with the CL and the C7 SVA. Changes in the C7 SVA after surgery induce changes in TS, which may serve as a driving force for inducing reciprocal changes in cervical alignment. TS has been demonstrated to be the radiologic parameter most closely correlated with the C7 SVA in other studies [22]. Patients with a low preoperative C7 SVA group (C7 SVA ≤6 cm) tended to have hypolordotic cervical curvatures (low TS and CL) and to have increased CL after surgery, whereas those with a high C7 SVA group (C7 SVA ≥9 cm) had a high TS and CL preoperatively and decreased CL after surgery. CL increases after larger iatrogenic reductions in thoracolumbar spine surgery.

There is a reciprocal relationship between LL and thoracic kyphosis. Thoracic kyphosis and pelvic retroversion can corrected after restoration of LL [5]. Reciprocal reduction occurs in PT and C7 SVA in response to increases in LL. Posterior pelvic shift, knee flexion and hip extension also decrease when sagittal malalignment is corrected. The magnitude of the T1 pelvis angle is strongly correlated with reciprocal changes in the knee angle, hip extension, and posterior pelvic shift (Fig. 2) [7].

Several studies have investigated spontaneous improvements in cervical alignment after lumbar pedicle subtraction osteotomy (PSO) [6,35,36,41]. Lumbar PSO significantly improves LL. LL restoration induces a decreased need for compensation at the thoracic spine and pelvis. After lumbar PSO, thoracic kyphosis and sacral slope increase, and PT, PI–LL, knee flexion, and SVA decrease. In the cervical area, reciprocal changes lead to reductions in CL and TS following surgery. There is a significant decrease in distal CL because there is no more need for compensation within the distal cervical spine to maintain horizontal gaze, whereas there is no significant decrease in proximal CL following surgery.

Ames and colleagues [1] made a meaningful attempt to create a comprehensive classification of CSD using a modified version of the Delphi approach. The proposed classification includes a basic deformity descriptor and modifiers based on clinical relevance and impact on patient HRQOL. This classification provides a mechanism to assess CSD with GSA and clinically relevant parameters. The authors defined C2–C7 SVA, horizontal gaze, TS–CL, myelopathy and the Scoliosis Research Society (SRS)-schwab classification as five modifiers. First, regarding C2–C7 SVA, Tang and colleagues [43] reported that C2–C7 SVA was negatively correlated with SF-36 physical component scores and positively correlated with Neck disability index (NDI) scores. Second, for horizontal gaze, significant correlations were found between C2–C7 SVA and C1–C2 lordosis [43]. C1–C2 alignment is the terminal link between the cranium and the cervical spine to regulate the angle of the gaze. A CBVA of 10° has been described as an optimal target [39]. Third, for TS–CL, the relationship between TS and CL is similar to the relationship between PI and LL, in that a greater TS requires a greater CL for harmonious alignment [27]. Hyun et al. [18] reported that TS–CL was strongly correlated with disability. Fourth, myelopathy may occur in patients with CSD due to direct compression resulting from spondylotic changes and the deformity itself. The development of myelopathy has been associated with progressive cervical kyphosis and positive sagittal cervical malalignment (C2–C7 SVA) [39]. Fifth, regarding the SRS-Schwab classification, cervical deformities contribute to thoracolumbar deformities. Thus, it is apparent that surgeons should not only focus on cervical alignment, but also on GSA. Adults with positive sagittal spinopelvic malalignment tend to compensate with abnormally increased CL in an effort to maintain horizontal gaze. The assessment and classification of CSD should not occur in isolation, which is why the alignment of a thoracolumbar spinal deformity was selected as a modifier for the CSD classification. Therefore, a full-length lateral radiograph is needed even when evaluating a cervical deformity.

Recently, Hyun et al. [18] suggested revised modifiers (Fig. 3). The authors investigated the validity of a CSD classification system using long-term follow-up data. They studied 30 consecutive patients with average follow-up period for 7.3 years. They demonstrated that C2–C7 SVA values of 4 cm and 7 cm were correlated with moderate and severe disability based on NDI scores. The authors found that C2–C7 SVA values of 4 cm and 7 cm corresponded to TS–CL values of 20° and 25°, respectively. TS–CL mismatch was significantly associated with disability and cervical malalignment. Therefore, a TS–CL less than 20° might be used for preoperative surgical planning, similar to a PI–LL less than 9°.

Only a few authors have analyzed the effect of GSA changes after cervical reconstruction surgery [26,33]. Cervical sagittal imbalance causes significant impairment due to the inability to gaze horizontally and is associated with poorer HRQOL [20,28].

Mizutani et al. [33] recently reported on the effects of cervical reconstruction surgery on thoracolumbar alignment changes. The C7 PL and GL are usually concordant, and GL is measured as the center of gravity of the head plumb line (COG PL). A distance between COG PL and C7 PL over 30 mm is defined as occiput-trunk (OT)-discordance. Compensation in primary cervical kyphosis occurs via posterior shifting of C7 SVA, a small TS, and large LL. The goal of cervical reconstruction surgery is therefore to achieve OT-concordance and cervical sagittal balance. COG PL and C7 PL move onto the femoral head as needed to achieve OT-concordance [34]. Subsequent thoracolumbar alignment changes occur as needed to harmonize GSA.

Reciprocal changes after cervical reconstruction surgery depend on whether patients have head-balanced (compensated) or trunk-balanced (decompensated) kyphosis (Fig. 4). Headbalanced kyphosis is characterized by a head position on the femoral head with posterior shifting of the C7 PL, hypokyphotic thoracic curve, straightened thoracolumbar junction, and lumbar hyperlordosis. The preoperative posterior shifted C7 PL moves anteriorly after surgery, and subsequently TS and the thoracic curve increase while LL decreases. However, pelvic and lower extremity parameters do not change after kyphosis correction (Fig. 5). In the trunk-balanced kyphosis, the COG PL is anteriorly shifted, which causes the C7 PL to remain anterior (located on the pelvis) compared to head-balanced kyphosis. These decompensated patients have a large preoperative TS and low LL. The LL is at the upper limit of PI–LL harmonization, which is the result of lumbar extension with maximum effort or lumbar degeneration. There are no significant changes in thoracolumbar alignment in trunk-balanced patients after kyphosis correction (Fig. 6). While TS and thoracic kyphosis decrease, spinopelvic and lower extremity parameters do not change after kyphosis correction. The COG PL moves posteriorly to achieve better GSA after surgery.

We reviewed 11 articles to synthesize the available information on GSA and reciprocal changes in healthy humans and patients with deformities. Only recently has there been widespread recognition of the importance of global imaging of the spine in the setting of deformities, and only a few studies have sought to define reciprocal changes in terms of GSA. In some studies, surgical procedures were not uniform and the follow-up was short-term. The CSD classification was derived from a modified Delphi approach and expert opinion. Thus, although initial steps have been taken provide a mechanism to assess CSD within the framework of GSA, further cohort studies are needed to validate the classification.

Sagittal malalignment after surgery is a major source of pain and disability. Positive sagittal malalignment was identified as the radiographic parameter most closely correlated with adverse health status outcomes [15]. There is a significant correlation between C7 SVA and TS, as well as between LL and thoracic kyphosis [5,16]. Furthermore, reciprocal changes are affected by preoperative sagittal alignment. Therefore, when patients undergo corrective surgery for a deformity, it is important to evaluate whole-body images both preoperatively and postoperatively.

Radiographs produced by conventional X-ray, computed tomography and the EOS can provide reliable and accurate Cobb angle measurements for scoliosis assessments. None of the methods systemically underestimates or overestimates the Cobb angle measurement. However, the static global balance of a patient cannot be evaluated properly on a conventional radiograph or computed tomography in which all the elements of the chain of balance (lower limbs, cervical spine and head) are not taken into account [13]. EOS evaluation for the correction of a deformity would make a big difference in the preoperative planning and postoperative evaluation.