Summary
Our understanding of adolescent idiopathic scoliosis (AIS) has improved dramatically over time, and with that increasing understanding, the importance of the sagittal plane has become more evident. Our early understanding of idiopathic scoliosis identified the lordotic component of the thoracic spine but failed to appreciate the true severity of the hyperlordosis and its representation on two-dimensional radiographs. With three-dimensional (3D) studies, we can better understand the true 3D curvature of the spine and see how we have been underestimating the degree of thoracic lordosis present. The sagittal profile is important as we see the impact it has on patients in adulthood with issues such as quality of life, proximal junctional kyphosis, and cervical alignment. This chapter discusses techniques and factors that influence surgeons’ ability to restore thoracic kyphosis. It will broach patient-specific properties, surgical techniques, and implant properties that allow us to optimize kyphosis restoration in patients with AIS.
Key words
thoracic kyphosis – sagittal plane – adolescent idiopathic scoliosis – surgical restoration – postoperative21 Kyphosis Restoration in Adolescent Idiopathic Scoliosis
21.1 Introduction
As our understanding of scoliosis has improved, the importance of the sagittal plane in adolescent idiopathic scoliosis (AIS) has become increasingly evident. One of the hypotheses linking an etiologic explanation for scoliosis and the three-dimensional (3D) deformity is based on anterior overgrowth causing lordosis of the spine. Dickson et al 1 , 2 proposed that thoracic lordosis was the primary problem of AIS and hypothesized that anterior overgrowth contributed to its development. Several others have published radiographic analyses supporting that anterior vertebral body and disk height is larger than the posterior vertebral body in AIS, 3 , 4 but whether the differences are causative or secondary remains unclear.
Anterior vertebral body overgrowth can lead to rotation, causing the 3D deformity seen in AIS. Stagnara 5 described an oblique view of the apex on radiography to visualize the maximal deformity, and Perdriolle and Vidal 6 similarly reported lordosis of the thoracic spine on a true lateral X-ray. These imaging techniques helped us better understand the hypokyphosis associated with AIS. However, with 3D, we can now better understand the true deformity and better visualize thoracic lordosis. Newton et al 3 compared 3D measures of sagittal profiles and found excessive lordosis of the thoracic spine in Lenke type 1 to 4 curves and excessive lordosis of the lumbar spine in Lenke type 5 and 6 curves when compared to a normal thoracic spine. Similarly, Hayashi et al 7 have reported on how routine pre-op lateral X-rays overestimate the kyphosis of a curve (most curves have less kyphosis than measured on two-dimensional [2D] X-rays) and the increasing margin of error with a larger curve size. Understanding the 3D anatomy is important to comprehend the underlying disease process and achieve optimal surgical correction.
Initial instrumented methods to correct scoliosis were focused on the coronal plane with Harrington rod distraction and have improved to address sagittal plane and axial plane deformity, thus addressing the 3D deformity of scoliosis. However, the etiology and anatomic morphology affect our surgical management. With increasingly powerful tools and techniques to correct spinal deformity, it is imperative to adequately define our surgical goals to achieve our desired outcomes. Increasing evidence is highlighting the importance of restoring physiologic sagittal profiles in our patients. 8 , 9 , 10 , 11
21.2 Background
With the advent of instrumented techniques to correct scoliosis, emphasis was placed on the coronal plane. Harrington rods achieved correction through the distraction of the concave side with improvement of the curves; however, some patients return with the flat back syndrome due to a loss of physiologic lumbar lordosis. 12 Instrumentation design and surgical techniques have improved enough to allow better sagittal contouring and axial derotation, thus targeting all three planes of deformity. However, now with 3D imaging, we can better understand the deformity in AIS and we are starting to realize that we have underestimated the severity of the thoracic lordosis present.
Even though the Lenke classification has been widely adopted in categorizing AIS curves and helping standardize surgical management, it likely still underemphasizes the importance of the sagittal plane in AIS. 13 , 14 Abelin-Genevois et al 15 have proposed a classification based on the sagittal profile alone and divides patients into four categories (normal kyphosis, hypokyphosis, thoracolumbar kyphosis, or cervicothoracic kyphosis), but their classification system was not based on 3D imaging. Hayashi et al 7 have shown that 2D imaging underestimates the true lordosis of the spine and that variation increases with larger magnitude curves (Fig. 21‑1). Parvaresh et al 16 developed a prediction formula for estimating 3D T5–T12 sagittal measurement from standard 2D measurements:
The average model error between predicted and measured 3D T5–T12 kyphosis was ±7 degrees. Ultimately, as our understanding of 3D scoliotic anatomy improves, we will likely need a new classification system to better emphasize the importance of the sagittal plane.
Use of 3D imaging has helped clarify some confounding results in the literature regarding the impact of pedicle screws on the sagittal profile. Hayashi et al 7 showed that pre-op lateral X-rays do not accurately reflect the true lateral alignment of a scoliotic spine. They reported that X-rays had to be rotated a mean of 13 ± 3 degrees to obtain a true lateral view of the apex and were on average 10 degrees less kyphotic. In some cases, the thoracic apex was frankly lordotic in a true lateral radiograph as opposed to kyphotic on a standing lateral view. Using 3D imaging, Newton, et al 17 also showed that the mean difference in preoperative thoracic kyphosis between 2D and 3D imaging was 11 ± 7 degrees (range –1 to 40 degrees). Postoperatively, once the curve was partially corrected and the radiographs were more in plane, the difference from 2D to 3D was only 1 ± 1 degrees (range –2 to 5 degrees; Fig. 21‑2). Therefore, much of our existing literature likely underestimates the preoperative relative hypokyphosis of the spine, and this likely explains why thoracic pedicle screws were previously thought to create more hypokyphosis.
Also, since many studies group all patients together, differences in kyphosis can be averaged and mask any notable differences. Several studies have shown that hyperkyphotic curves tend to decrease postoperatively and, if not separated from the larger cohort, the analysis may confound the results. 18 , 19 Many studies including only hypokyphotic curves or frankly lordotic curves have more evidently reported changes in the sagittal plane and highlighted the increase in kyphosis imparted with pedicle screw constructs. 20 , 21 Due to these limitations, surgeons should be cautious in interpreting the breadth of literature available.
21.3 Importance of Sagittal Plane Correction
Restoring an appropriate sagittal profile surgically is not only important to correct the thoracic segment but also has repercussions throughout the spine. In particular, changes in the thoracic spine may impact adjacent lumbar or cervical alignment. The relationship between pelvic parameters and lumbar lordosis, as well as their impact on health-related quality-of-life (HRQOL) measures, has been well established in the adult literature and to some extent in the pediatric population. 8 , 22 However, the link between thoracic profile and lumbar lordosis in the pediatric population has been more variable, with publications both supporting and refuting a relationship. 23 , 24 , 25 , 26 Sudo et al 24 correlated thoracic kyphosis with lumbar lordosis and lordosis with pelvic parameters. Similarly, Newton et al 27 found a correlation between thoracic kyphosis restoration and lumbar lordosis postoperatively that persisted at 2 years. However, others have not established a relationship between thoracic kyphosis and lumbar lordosis. 25 , 26 Some of the conflicting data may be due to reliance on 2D X-rays and do not truly depict a 3D relationship between the thoracic and lumbar spine. Clément et al 28 , 29 may have also partially explained the differences by linking the thoracic kyphosis to the upper lumbar lordosis (r = 0.69 pre-op) and similarly the pelvic parameters to the lower lumbar lordosis. They proposed that the lumbar segment has such significant motion and adaptability that it can buffer changes and mask a correlation between the pelvis and thoracic spine in children. They also reported a correlation between changes in thoracic kyphosis restoration and lumbar lordosis. 29
The thoracic profile in AIS is also linked to the cervical spine. 9 , 10 , 30 Several papers have supported the relationship preoperatively (Fig. 21‑3) and postoperatively between thoracic kyphosis and cervical lordosis or even frank kyphosis. A few papers have found a correlation between thoracic kyphosis and the C7 slope and cervical lordosis as well as postoperative associations. 10 , 15 The rostral sagittal profile may influence cervical alignment and even contribute to junctional problems as the upper instrumented vertebra (UIV) is often near the cervicothoracic junction.
Radiographic proximal junctional kyphosis (PJK) has also been reported ranging from 7 to 46% in AIS cases postoperatively. 11 , 31 , 32 Some authors have suggested that PJK develops as a compensatory mechanism to realign the head after surgical decompensation, and risk factors reported have included pelvic incidence, lumbar lordosis, and sagittal C7 slope. 32 , 33 In a meta-analysis, Zhong et al 34 reported a PJK incidence of 14% in AIS and found risk factors to include use of screws rostrally, entire screw constructs, larger pre-op thoracic kyphosis, larger pre-op lumbar lordosis, and greater post-op change in either kyphosis or lordosis. Although the reoperation rate for PJK is low in pediatric AIS, it is more common in adults, and attention must now be paid to minimizing the potential risk of creating iatrogenic problems in the future as our patients age. 35 , 36 , 37
Ultimately, our surgical goal is to achieve a balanced spine with sustained optimal clinical outcomes. We understand that in adult patients the sagittal profile is one of the most important factors contributing to disability and pain. 38 The relationship between the pelvis and lumbar spine primarily dictates the sagittal profile in adults and influences HRQOL measures. 22 Therefore, emphasis on optimizing the sagittal profile in our AIS patients to last them a lifetime, or at least minimizing iatrogenic problems as adults, is critical.
To be able to create an optimal 3D profile postoperatively, we not only need to understand the pre-op 3D anatomy but also have an understanding of our radiographic goal. Chapter 4 discusses the range of normal radiographic parameters, but, in brief, Dickson et al 2 approximated normal kyphosis in adolescents in the range of 20 to 40 degrees. These approximations have been consistently reproduced in subsequent studies 25 , 39 and, although some publications have suggested that normal range may extend from 10 to 50 degrees, 20 to 40 degrees is largely accepted as the normal range. 40 , 41 , 42 Therefore, our surgical goal is defined as thoracic kyphosis between 20 and 40 degrees.
21.4 Surgical Restoration of Kyphosis
21.4.1 General
In the surgical restoration of thoracic kyphosis, many variables likely contribute to a successful outcome. Several studies have tried to isolate individual factors, but even with single-center, single-surgeon series where a technique is modified over time, both unknown and known variables may confound results. Monazzam et al 43 reviewed a multicenter series of 275 patients who had improvement in their thoracic kyphosis postoperatively and analyzed five variables that have previously been associated with kyphosis restoration in the literature: pre-op kyphosis, rod metal type, implant density, use of Ponte osteotomies, and surgeon. Only the surgeon was associated with the amount of thoracic kyphosis restoration. Another study by the same senior author, Newton et al, 44 showed that it is possible to create thoracic kyphosis in frankly lordotic spines but concluded that variation was likely attributed to surgeon-specific factors and was multifactorial in nature. Using 3D imaging, 134 patients were analyzed, and factors associated with 2-year increase in 3D T5–T12 kyphosis were identified. The continuous variables of pre-op 3D T5–T12 kyphosis as well as first erect and 2-year thoracic percent correction were significant as well as the categorical variables of the female gender, use of Ponte osteotomies, use of stainless steel (SS) metal (vs. cobalt chromium), and the surgeon. In a Classification and Regression Tree (CART) regression analysis, only the surgeon remained significant, but the surgeon group that achieved more post-op kyphosis tended to more often perform osteotomies and use segmental pedicle screw fixation and SS rods. 44
In combination with the variability of 2D lateral imaging to accurately measure true kyphosis in larger curves and lack of subdivision by pre-op hypokyphosis, determining which individual factors most influence the iatrogenic restoration of thoracic kyphosis can be difficult. Nonetheless, existing literature can provide some framework for guidance and help direct future research ideas.
21.4.2 Curve Attributes
Although the degree of kyphosis restoration may be largely due to surgeon technique, other variables may be intrinsic to the patients and their specific curve. Intuitively, many studies have correlated greater preoperative kyphosis with greater post-op kyphosis. 45 , 46 , 47 It makes sense that preexisting kyphosis is easier to maintain rather than taking a more hypokyphotic spine and creating kyphosis. Interestingly, in a multicenter study, Lonner et al 47 found a larger coronal curve magnitude at 2 years post-op to be associated with greater kyphosis, whereas Luo et al 45 found the flexibility of the main thoracic curve to be significant. Having a larger residual curve may translate into more measured kyphosis due to the difference in 2D and 3D X-rays as less coronal correction likely means less “in plane”/accurate measures and therefore may exaggerate difference from the true kyphosis. Alternatively, greater correction in the coronal plane could rotate the longer anterior column into the sagittal plane and have a hypokyphotic effect. Each study had slightly differing parameters for inclusion, such as thoracic bands in Kokabu et al, 46 and surgical techniques that may explain the variability in findings as well. Intuitively, rigid, lordotic curves would be more challenging to make kyphotic than flexible, kyphotic ones. However, surgeons can attempt to create more flexibility intraoperatively using facetectomies or osteotomies.
21.4.3 Release of Spine: Posterior
Although the inherent flexibility of the spine facilitates surgical correction of deformity, several procedures can be utilized to increase our desired surgical goals. Often, inferior facetectomies are performed to localize anatomy for pedicle screw placement, which may increase the flexibility of the spine, and Ponte osteotomies can be performed to achieve even more iatrogenic flexibility. It is important to distinguish Ponte osteotomies used in this setting to create increasing kyphosis through distraction and lengthening of the posterior elements rather than the more common application of posterior element compression in the correction of Scheuermann kyphosis (Fig. 21‑4).
Several authors have reported improved kyphosis with the use of multiple facetectomies but mostly using bands. 21 , 46 , 48 Kokabu et al 46 reported a correlation in thoracic kyphosis restoration with the number of facetectomies, whereas Sudo et al 48 reported a mean increase of 9.1 degrees with a correlation of the concave implant density. However, as discussed previously, most of these studies have multiple variables that could account for the change in kyphosis, and that could confound results. All of these studies are also limited by their 2D analysis of kyphosis using standard X-rays, as mentioned previously, and may not truly depict a 3D anatomy.
With respect to Ponte osteotomies, most studies include pedicle screw–only constructs. Demura et al 20 used ultra-strength SS rods with uniplanar screws and noted stable thoracic kyphosis post-op but an increase from 13.5 ± 5.0 to 20.0 ± 4.0 degrees when grouped by pre-op hypokyphotic spines. Shah et al 18 similarly reported an improvement in hypokyphotic spines from 8.1 to 18.3 degrees. Samdani et al 49 compared patients having had Ponte osteotomies to a control group in 191 patients. Although curves in the Ponte group were less flexible pre-op (47 ± 22 vs. 55 ± 23%), the addition of Ponte osteotomies led to greater improvement in all three planes of correction. Specifically, in the sagittal plane they achieved 3 ± 11.6 degrees of kyphosis in the Ponte group versus –0.4 ± 9.9 degrees in the group without Ponte osteotomies. All three studies have some degree of patient overlap, as they are the primary contributing sites to the Harms Study Group database. In hypokyphotic patients, Feng et al 19 noted an increased creation of kyphosis with Ponte osteotomies (7.3 ± 1.9 to 18.4 ± 3.2 degrees) versus controls (8 ± 1.4 to >11.5 ± 2.4 degrees).
In contrast, Halanski et al 50 reviewed a small cohort of 35 patients and over time stopped performing Ponte osteotomies. Comparing to their historic cohort of Ponte osteotomies, they reported the same sagittal kyphosis post-op using only multiple facetectomies. However, they recognized that curves were larger in the Ponte group (59 ± 11 vs. 52 ± 8 degrees), and they did not stratify patients by their pre-op profile. Overall, the use of Ponte osteotomies does appear to help create additional kyphosis, but significant variability exists between studies, making direct comparisons difficult.
21.4.4 Anterior Surgery/Video-Assisted Thoracoscopic Surgery
Although anterior fusion for AIS has become less common since the advent of pedicle screw fixation, it remains a powerful tool to restore kyphosis. Anterior surgery can be divided into anterior fusion (either open or thoracoscopic—VATS) or anterior releases followed by posterior fusions. Chapter 14 discusses the technique, nuances, and rationale for anterior surgery in more detail.
Betz et al 51 showed that in patients with less than 20 degrees of pre-op kyphosis, 81% with an anterior fusion had thoracic kyphosis in the normal range post-op, whereas 60% of patients with posterior fusion remained hypokyphotic. In patients who had pre-op kyphosis greater than 20 degrees, 40% of patients with anterior fusion ended up hyperkyphotic, and 29% of patients with posterior fusion ended up hypokyphotic. Lonner et al 47 used the same Harms registry and also corroborated that anterior surgery was associated with the greatest kyphosis creation, with 7.7 degrees more than the posterior approaches. Sucato et al 52 compared anterior surgery versus hybrid constructs versus hook only constructs and noted that anterior surgery created the most kyphosis, with 22.4% of patients hyperkyphotic at 2 years.
With VATS, Newton 53 showed an increase in thoracic kyphosis from T2 to T12 of 5 degrees, or 19%. Kim et al 54 showed a mean increase of kyphosis of 2.2 degrees initially, increasing to 4.2 degrees over time (18.2 ± 7.7 to 22.4 ± 7.2 degrees), whereas Lonner et al 55 demonstrated a mean increase of 8.7 degrees (17.4 ± 8.7 to 26.1 ± 8.1 degrees). Both anterior techniques shorten the anterior column and allow significant creation of kyphosis (Fig. 21‑5).
Anterior releases followed by posterior instrumentation likely include a selection bias toward larger or more rigid curves but do still shorten the anterior column. Lonner et al 47 did not find a significance of kyphosis creation using anterior releases prior to posterior fusions (p = 0.16), but the study may have been limited by a small sample size of 11 patients. Ferrero et al 32 just included hypokyphotic patients and noted better thoracic kyphosis post-op after anterior releases (18.3 ± 13.6 degrees) compared to without (15.2 ± 9.1 degrees).
Other studies have corroborated that excessive progression of kyphosis from anterior approaches is possible, especially with continued growth, and some have advocated creating less kyphosis in patients who are skeletally immature and kyphotic (>30 degrees). 56 , 57 Undoubtedly, the anterior approach is a powerful tool to shorten the anterior spine and create kyphosis, but overall this has fallen out of favor for multiple reasons that will be discussed in Chapter 12.
21.5 Posterior-Based Surgery
21.5.1 Type of Fixation
Bands
Although AIS deformity correction is mostly performed posteriorly using hooks or screws, the advent of synthetic bands has gained some attention as well. Authors using sublaminar bands or clamps have claimed equivalent, if not better, results, often using bands in the thoracic spine and screws in the lumbar region. However, most studies are smaller retrospective series. 58 , 59 Ilharreborde et al 60 reported a series of 35 patients with pre-op hypokyphosis (<15 degrees) and with 3D imaging observed a mean increase of 8 ± 7 degrees in T4–T12 kyphosis. Imagama et al 61 reported a similar correction in the sagittal profile with a change from 15.2 to 23 degrees, and Blondel et al 23 observed a mean of 6.5 degrees of increased kyphosis. Other variables such as rod metal and size as well as rod contouring likely have a significant effect on kyphosis creation but have not been studied with this type of fixation to date.
Pedicle Screws or Hooks
Several studies have raised concern that the use of pedicle screw constructs can induce greater hypokyphosis in AIS correction. 47 , 52 , 62 Clements et al 63 suggested greater reduction of kyphosis with more screws and a mean change in kyphosis of –3.8 ± 12 degrees with all-screw constructs. Similarly, Fletcher et al 64 reported that all-screw constructs led to less kyphosis (17.3 ± 10.9 degrees with screws vs. 24.4 ± 11.4 degrees) in hybrid constructs. Cao et al 65 performed a meta-analysis of 24 articles and concluded that both screw-only constructs and hybrid constructs (rostral hooks and distal screws) restore kyphosis but that hooks “create more kyphosis.” As discussed previously in the chapter, Newton et al 17 have shown that thoracic pedicle screws do not reduce kyphosis but rather the initial degree of hypokyphosis is underestimated (on average by 11 ± 7 degrees), whereas post-op values are more accurate (1 ± 1 degrees). Therefore, pedicle screws create kyphosis, but pre-op X-rays underestimate the initial severity in the sagittal plane. In addition, it is possible that derotation techniques can influence the sagittal profile due to the anterior vertebral length. With increasing direct vertebral body derotation, shifting the longer anterior column into the sagittal plane could lead to less kyphosis.
Suk et al 66 reported greater kyphosis restoration when only looking at patients starting with hypokyphosis. They reportedly used 7-mm SS rods, which may have been so powerful that a larger degree of kyphosis was created, thus nullifying the previously hypokyphosing effect reported from screws. Again, the recurrent theme of variability by surgeon-dependent factors may have the greatest impact on the successful creation of kyphosis.
Screw Properties
Several reports have emphasized the utility of uniplanar screws in the restoration of kyphosis, whereas other authors advocate using polyaxial screws in their correction; however, few studies directly compare screw types with sagittal profiles. 46 , 67 , 68 Badve et al 67 compared the sagittal plane correction using monoaxial versus uniplanar screws (Fig. 21‑6) and reported greater post-op kyphosis (23.6 vs. 18.0 degrees, p = 0.02) with uniplanar screws and 6.35-mm SS rods. Presumably, the uniplanar screw allows the screw to shift along the plane of the contoured rod using distraction, thus lengthening the posterior column and overcoming the anterior overgrowth. 18 This technique of kyphosis restoration may again illustrate the multifactorial nature of kyphosis restoration between the screw type, rod contouring, metal properties, and surgeon.
21.5.2 Implant Density
In the series by Clements et al 63 reporting increased hypokyphosis using screws versus hybrid constructs, increasing implant density was also associated with greater hypokyphosis. Also from the Harms Study Group database, Lonner et al 47 similarly found greater implant density and greater correction in the coronal plane to be associated with less kyphosis. It is possible that increased coronal correction places the relatively longer anterior part of the vertebral body in the sagittal plane, thus creating less kyphosis. However, as mentioned previously, most of these studies are based on 2D radiographic measures that underestimate the true lordosis of the thoracic spine.
In contrast, Liu and Hai 68 compared four groups of Lenke type 1 patients based on implant density of screws (<0.6 vs. ≥0.6) and rod stiffness (5.5- vs. 6.35-mm titanium) and felt that the most thoracic kyphosis was achieved in patients with a higher implant density and stiffer rods. Luo et al 45 similarly reported a series of patients with pre-op hypokyphosis in Lenke type 1 to 4 curves and found that increased concave screw density, pre-op kyphosis, and flexibility of the main thoracic curve correlate with the creation of kyphosis. They noted an increase from 4.75 ± 3.45 to 17.30 ± 5.13 degrees. The subdivision of patients into a hypokyphotic group may help eliminate some of the confounding analyses discussed earlier and highlight the effect of implant density in this case.
Related posts:
13 Halo Traction in Large Idiopathic Scoliotic Curves
14 Indications and Techniques for Anterior Release and Fusion
12 Posterior Correction Techniques in Adolescent Idiopathic Scoliosis
17 Assessment and Management of Shoulder Balance
19 The Surgical Treatment of Double and Triple Curves (Lenke Types 3, 4, and 6)
18 Surgical Treatment of Lumbar and Thoracolumbar Curve Patterns (Lenke V)
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
