N E U R O I M A G I N GC L I N I C S

O F N O R T H A M E R I C A

Neuroimag Clin N Am 17 (2007) 105–115

105

ScoliosisJohan Van Goethem, MD, PhDa,b,*, A. Van Campenhout, MDc,Luc van den Hauwe, MDa, Paul M. Parizel, MD, PhDa

- ClassificationEtiologyCurve locationAge at onsetCurve types

- Prevalence- Clinical features- Natural history

Congenital scoliosisInfantile idiopathic scoliosisJuvenile idiopathic scoliosis

Adolescent idiopathic scoliosisAdult idiopathic scoliosis

- Imaging in scoliosisPlain film imaging techniqueImaging intervalMeasurementsSpecialized imaging

- TreatmentNon-operative treatment (braces)Surgical treatment

- References

In the frontal plane, the normal load-bearingspine is straight. Scoliosis is a structural lateralcurvature of the spine with a rotatory component.A small deviation (<10�) is sometimes called spinalasymmetry, whereas ‘‘true’’ scoliosis has a deviationof R10�. This deviation is accompanied by a rota-tion that is maximally at the apex of the curve. Inthe thoracic region, this rotation creates anasymmetry of the thoracic cage that produces thetypical chest wall prominence known as the Adamssign.

Imaging in scoliosis is important. Most cases ofscoliosis are idiopathic, and imaging is used rou-tinely in monitoring the changes of the deformitythat take place during growth. Imaging is alsocrucial in determining the underlying etiology innon-idiopathic cases of scoliosis and is used inpre- and postoperative monitoring.

Generally, scoliosis is treated by orthopedicsurgeons who have special training in spinal and

1052-5149/07/$ – see front matter ª 2007 Elsevier Inc. All righneuroimaging.theclinics.com

pediatric problems. Patients who have scoliosismay present directly to the radiology departmentthrough a primary health care physician or maybe referred from the pediatric, neurology, or neuro-surgery departments. Many physicians look towardthe radiologist as the spinal expert. Therefore,radiologists should know the basics of scoliosis,how to perform the radiologic examination, howto read these films correctly, and how to makea coherent and helpful interpretation.

Classification

Scoliosis can be classified according to etiology,curve location, age at onset, and curve type.

Etiology

Congenital scoliosisCongenital scoliosis is the most frequent congenitalspinal deformity. It is present at birth as the result of

a Department of Radiology, University of Antwerp, Wilrijkstraat 10, B2650 Edegem (Antwerp), Belgiumb Department of Radiology, AZ Nikolaas, Hospitaalstraat 8, B9100 Sint-Niklaas, Belgiumc Department of Orthopaedics, University Hospitals Leuven, Pellenberg, Weligerveld 1, 3212 Pellenberg,Belgium* Corresponding author. Department of Radiology, University of Antwerp, Wilrijkstraat 10, B2650 Edegem(Antwerp), Belgium.E-mail address: johan.vangoethem@ua.ac.be (J. Van Goethem).

ts reserved. doi:10.1016/j.nic.2006.12.001

Van Goethem et al106

embryologic or intrauterine maldevelopment ofvertebral elements. The term ‘‘congenital’’ is slightlymisleading because it implies that the curvature isapparent at birth, but this is not necessarily so.The vertebral anomalies are present at birth, andthe clinical deformity develops with spinal growthand may not become apparent until later child-hood [1]. These may be caused by failure of forma-tion or failure of segmentation. They are commonlyassociated with cardiac or urologic abnormalitiesthat develop during the same period (before 48days of gestation) (Fig. 1) [2]. Vertebral maldevel-opment can be classified as defects of segmentationor defects of formation. Curve progression isstrongly related to the type of vertebral abnormalitywith the poorest prognosis for unilateral unseg-mented bars with contralateral hemivertebrae (upto 10�/yr progression), a less severe progression in

Fig. 1. A 5-year-old boy who has Klippel-Feil syn-drome type II. Klippel-Feil syndrome occurs in a het-erogeneous group of patients unified only by thepresence of a congenital defect in the formation orsegmentation of the cervical spine [2]. Numerous as-sociated abnormalities of other organ systems maybe present. As a consequence of these fusion and seg-mentation anomalies, this patient has a congenitalcervical and thoracic scoliosis. Klippel-Feil type Ishows massive fusion of cervical and upper thoracicvertebrae. Type II shows fusion of a limited numberof vertebrae and hemivertebrae. Occipitoatlantal fu-sion and other lower thoracic anomalies arepresent. Type III shows cervical fusions and lower tho-racic or lumbar fusions. C2-C3 fusion, also present inthis case, is the most common form of congenitalfused cervical vertebrae and is probably dominantwith variable expression. This patient also showshemivertebrae and unilateral fused vertebrae.

cases of hemivertebrae or double hemivertebrae(1–2.5� and 2–5�/yr, respectively), and a leastsevere progression in patients who have block andwedge vertebrae (<1�/yr progression) (Fig. 2). Themost common anomaly is hemivertebrae, whichis seen in about 40% of cases.

Idiopathic scoliosisMost frequently, scoliosis is idiopathic (80% ofcases). Substantial research efforts have identifiedseveral factors contributing to the development ofidiopathic scoliosis.

Genetic factors are a potential etiologic compo-nent in the development of scoliosis. There isevidence for several different modes of inheritance,including multifactorial, autosomal dominant, andX-linked dominant, with variable phenotypicexpression. Several candidate regions have beenidentified, including on chromosomes 6p, distal10q, 17p11, 18q, and 19p13 [3]. Family membersof affected individuals have an increased incidenceof scoliosis. Studies in families with twins haveidentified 73% to 92% concordance in mono-zygotic twins and only 36% to 63% concordancein dizygotic twins. The prevalence of scoliosis in-creases seven times in individuals who have affectedsiblings and three times in those who have affectedparents.

Idiopathic scoliosis is apparently a result of inad-equate control of spinal growth. The deformityprogresses most rapidly during adolescent growth,and there is evidence that adolescents who haveidiopathic scoliosis have an earlier growth spurt[4], are taller and thinner, and have an increasedlevel of growth hormone.

Vertebral growth anomalies may be related toadolescent idiopathic scoliosis. Differentialgrowth rates between left and right sides of thespine may lead to asymmetry that could be ac-centuated by the Heuter-Volkmann effect (sup-pression of growth on the concave side of thecurve). When anterior spinal growth outpacesposterior growth in an adolescent patient, hypo-kyphosis is produced, with subsequent bucklingof the vertebral column. Scoliotic spines in girlsbetween 12 and 14 years of age have longer tho-racic vertebral bodies, shorter pedicles, anda larger interpedicular distance compared withthe spines of normal, aged-matched girls [5].The differential growth between the anteriorand posterior elements is not only significantlydifferent in scoliosis versus normal spines but isalso correlated to the severity of scoliosis. Thisovergrowth in length mainly occurs by enchon-dral ossification, whereas circumferential growthis slower and happens by membranousossification.

Scoliosis 107

Fig. 2. Postero-anterior plain film (A)and multiplanar reconstruction 3-Dreformation of a multi-row detectorCT in a patient who has congenitalthoracic scoliosis with wedge verte-brae and hemivertebrae (B). Curveprogression is strongly related tothe type of vertebral abnormality,with the poorest prognosis forunilateral unsegmented bars withcontralateral hemivertebrae (up to10�/yr progression), a less severe pro-gression in cases of hemivertebrae ordouble hemivertebrae (1–2.5�/yr and25�/yr, respectively), as in this case,and a least severe progression in pa-tients who have block and wedgevertebrae (<1�/yr progression). Asso-ciated rib anomalies are seenfrequently.

Scoliosis in generalized diseasesand syndromesNeural axis abnormalities have a prevalence be-tween 19.5% and 26% among infantile and juvenilescoliosis cases [6]. Several factors have been identi-fied that correlate with a higher incidence in scolio-sis, such as equilibrium and vestibular dysfunction,melatonin deficiency, syringomyelia (Fig. 3), Chiarimalformation, and spinal tumors. Neuromusculardisorders (eg, cerebral palsy and muscular dystro-phy) and some generalized diseases and syndromes(eg, Marfan, neurofibromatosis, rheumatoid dis-ease, or bone dysplasia) are associated with

scoliosis. Scoliosis is seen in 8.7% of patients whohave Down syndrome [7].

Traumatic scoliosisTraumatic scoliosis can be caused by bony lesions(eg, fractures and dislocations) or by soft tissuelesions (eg, burns and postempyema).

Degenerative scoliosisDegenerative lumbar scoliosis is a lateral deviationof the spine that typically develops after 50 yearsof age [8]. This is type 1 adult scoliosis, which isprimarily degenerative. It is caused most frequently

Fig. 3. Clinically suspected infan-tile scoliosis is confirmed by plainfilm (A). All cases of infantileand juvenile idiopathic scoliosisrequire further work-up with MRimaging because they are com-monly associated with a varietyof neural axis abnormalities andneuromuscular disorders, includ-ing syringomyelia (as in this case)(B).

Van Goethem et al108

by a disc or facet joint arthritis, affecting those struc-tures asymmetrically. Type 2 adult scoliosis is theprogression of adolescent scoliosis in adulthood.Type 3 adult scoliosis is a secondary scoliosismostly caused by osteoporosis [9]. Although theclinical presentation may vary, degenerative scolio-sis is usually associated with loss of lordosis, axialrotation, lateral listhesis, and spondylolisthesis. Itis associated with degenerative disk and facetdisease and hypertrophy of the ligamenta flava,typically leading to neurogenic claudication andback pain. Rarely, sagittal or coronal imbalancemay develop.

Curve location

Curve location is defined by its center, known as theapex, which is the most lateral disc or vertebra ofthe curve. Usually the apical vertebra is also themost horizontal. Scoliosis can be classified ascervical (apex between C2 and C6), cervicothoracic(C7–T1), thoracic (T2–T11), thoracolumbar(T12–L1), lumbar (L2–L4), or lumbosacral (L5and below).

Age at onset

Age at onset or diagnosis is used to classify scoliosisas the following types: type 1, infantile (0–3 years);type 2, juvenile (4–10 years, mean age of diagnosisof juvenile scoliosis is around 8 years); type 3,adolescent (11–17 years); and type 4, adult (R18years).

Curve types

Primary curves are the first to develop. Secondarycurves develop as a means to balance the headand trunk over the pelvis, not only in the frontalbut also in the sagittal plane. At the time of diagno-sis, it is not always possible to differentiate primarycurves from secondary curves (Fig. 4). Structuralcurves (as opposed to nonstructural curves) cannotbe corrected with side-bending or traction. Non-structural curves can be secondary curves or func-tional curves (postural, secondary to short leg,muscle spasm).

Several classifications according to different curvepatterns have been proposed as a preoperativeassessment. The use of these classification schemesallows scoliosis practitioners to compare varioustreatments of similar curve patterns and to recom-mend selective fusions of the spine when appropri-ate. The most widely used classification, developedby Moe and reported by King and colleagues [10](the King-Moe Classification), was designedprimarily to help the clinician to decide when toinstrument the thoracic curve alone and when toinstrument the thoracic and lumbar curves. Thisclassification is not comprehensive, and a morecomplete and reliable classification was proposedby Lenke and colleagues [11]. The Lenke Classifica-tion considers three components: curve type (types1–6), a lumbar spine modifier (A, B, or C), anda sagittal thoracic modifier (�, N, or 1). The sixcurve types have specific characteristics on frontal

Fig. 4. Follow-up of scoliosis in an adolescent girl. (A) 13 years. (B) 14 years. (C) 14.5 years with brace. (D) 15.5years. Imaging during bracing confirms that the thoracic curve is structural and that the thoracolumbar curveis secondary. The thoracic curve progresses 7� between ages 13 and 14, after which bracing is started. Duringbracing, the secondary thoracolumbar curve can be partially corrected. The thoracic curve only progresses 2�

over the following 1.5years during bracing.

Scoliosis 109

and sagittal radiographs that differentiate structuraland nonstructural curves in the proximal thoracic(PT), main thoracic (MT), and thoracolumbar/lum-bar regions (TL/L). The major curve is the one withthe largest Cobb measurement and is included infusion surgery for idiopathic adolescent scoliosis.The minor curves are all other nonmajor curves.One of the main debates in scoliosis surgery iswhether to include those minor curves in thefusion. Thus, six curve types are distinguished inthis system based on whether the PT, MT, and TL/Lregions are major, minor structural, or nonstruc-tural, including type 1, MT; type 2, double thoracic;type 3, double major; type 4, triple major; Type 5,TL/L; Type 6, TL/L-MT [12]. Many other classifica-tion types exist, such as the Scoliosis Research Soci-ety Classification for Adult Spinal Deformity, butthese usually differ mainly on surgical interpreta-tion points.

Prevalence

The prevalence of scoliosis (R10�) in the child-hood and adolescent population is between0.5% and 3.0%. Adolescent idiopathic scoliosisis present in 2% to 4% of children between 10and 16 years of age. Larger curves (>30�) are re-ported between 0.04% and 0.29%. In childhoodscoliosis, 0.5% is reported in the infantile group,10% in the juvenile group, and the remainderin the adolescent group.

Clinical features

There is no difference in the prevalence of back painor mortality between patients who have untreatedadolescent idiopathic scoliosis and the general pop-ulation. Patients who have mild idiopathic scoliosis(<25�) usually have little or no discomfort. Cardio-pulmonary complications are almost exclusivelyseen in early-onset scoliosis (<5 years of age).Patients presenting with severe pain, neurologicsymptoms, or rapidly progressing scoliosis requirethorough further examination. In adults, degenera-tive scoliosis may contribute significantly to facetjoint degeneration and spinal and foraminal steno-sis with subsequent clinical symptomatology.

Infantile idiopathic scoliosis presents as a leftthoracic curve in 90% of cases. The male/femaleratio is 3:2. In juvenile idiopathic scoliosis, themale/female ratio is 1:2 to 1:4; boys are moreaffected between 3 and 6 years of age (1:1), and girlsare more affected between 6 and 10 years of age(1:8). The number of right and left curves is equalin the younger group (<6 years at presentation),and right curves predominate in the older group(80%).

In North America and Europe, a screeningexamination in school often leads to a first referralfor scoliosis. The goal of school screening programsis to detect childhood scoliosis at a stage wheresurgical correction can be avoided.

Clinical tests are available to assess scoliosis. Theforward-bend test or Adams test is probably the bestknown. In this test, the patient bends forward withthe knees straight and the palms together. Duringthis test, the thoracic and lumbar regions shouldstay symmetric. An asymmetric rotational humpof 5� to 7� is associated with a scoliosis of 15� to20�. A referral and imaging is recommended whenthe angle of trunk rotation is greater than 7�. Thisis a sensitive, although not a specific, test (2–3%referral).

Natural history

Congenital scoliosis

Congenital scoliosis shows progression in 75% ofcases. The poorest prognosis is for thoracic curvesand for multiple hemivertebrae and a convex uni-lateral bar (failure of segmentation) opposite thehemivertebrae. Block and wedge vertebrae showprogression <1�/yr, hemivertebrae show a meanprogression of 1� to 2.5�/yr, double hemivertebraeincrease at a double rate, and unilateral unseg-mented bars with contralateral vertebrae may prog-ress up to 10�/yr. The management of congenitalscoliosis requires frequent clinical and radiographicfollow-up to detect progression. Curve progressionor severe vertebral anomalies known to cause curveprogression require prompt treatment to preventdeformity and morbidity, such as thoracic insuffi-ciency syndrome [13].

Infantile idiopathic scoliosis

The vast majority of these curves are self-limiting.The few that progress (usually double structuralcurves) can be difficult to manage. In cases wherethe rib vertebral angle difference is larger than20�, progression is likely. The rib vertebral angledifference is defined as the difference in angulationof the left and right ribs on the apical vertebra asmeasured on an anteroposterior (AP) radiograph.

Juvenile idiopathic scoliosis

Juvenile idiopathic scoliosis is often progressive(70%). The potential for trunk deformity with car-diac and pulmonary compromise exists especiallyin scoliosis with onset before 5 years of age. Curvesof greater than 30� are almost always progressive, ata rate of 1� to 3�/yr before 10 years of age and ata rate of 4.5� to 11�/yr after 10 years of age. If thescoliosis is in the thoracic region, surgery is requiredin more than 95% of these cases.

Van Goethem et al110

Adolescent idiopathic scoliosis

Roughly 2% of adolescents have a scoliosis (>10�),but only 5% of these have a progression of the curveto greater than 30� (Fig. 4). The progression ofscoliosis is dependent on the growth velocity andthe magnitude of the curve at the first visit. Progres-sion is most notable with a growth velocity ofgreater than 2 cm/yr, between 9 and 13 years ofage, at bone ages between 9 and 14 years, at Rissersigns 0 to 1, and between 0.5 and 2 years beforemenarche [14].

The key risk factors for curve progression are theremaining spinal growth (skeletal immaturity)combined with the curve magnitude at a giventime and, to a lesser degree, female gender [15].Skeletal immaturity or remaining skeletal growthis determined by age, menarchal status, and Rissersign (radiologic) or Tanner staging (clinical). Themain progression occurs at the time of most rapidskeletal growth. This occurs at 11 to 13 years of agefor girls and at 13 to 15 years of age for boys.Determining the peak height velocity (PHV)timing by accurate serial heights measurements isdifficult. Usually, patients present without accurateprior height measurements. In these cases, othermarkers have to be used. The Risser stage is still0 at the time of PHV, and this phase of acceleratedskeletal growth is split into halves by the closure ofthe triradiate cartilage [16]. Other markers can beused, including digital uncapped phalangealepiphyses, which are indicative of pre-PHV, andfused epiphyses, which are indicative of post-PHV. Capped but nonfused epiphyses are indeter-minant. Tanner stage 1 for breast strongly indicatespre-PHV. Stage 3 for breast and pubic hair occurs ator after the PHV, and stage 4 occurrs after PHV[17]. In the decelerating growth phase, a Risserless than 1 is associated with progression in 60%to 70% of patients, whereas Risser 3 has a riskfor progression in less than 10%. Curve patternhas also been identified as an important variablefor predicting the probability of progression.Primary thoracic curve scoliosis, especially Kingtypes II, III, and V, progress more than primarylumbar curve scoliosis.

Adult idiopathic scoliosis

Curves of less than 30� usually show no progres-sion. Curves measuring 30� to 50� at skeletalmaturity progress at an average of 10� to 15� duringa normal lifetime. Curves between 50� and 75�

show a continuing rate of nearly 1�/yr. In untreatedpatients, an increased mortality rate due to cardio-pulmonary disorders is seen most frequently inscoliosis of greater than 90�.

Imaging in scoliosis

Plain film imaging technique

The ideal imaging modality for screening in scolio-sis is the upright posteroanterior radiograph of theentire spine (Fig. 4). The head and pelvis should beon the same film. The patient must be standing,but in young patients or patients who have severeneuromuscular disorders, a sitting or supine radio-graph may be the only possibility. In general, noattempt should be made to equalize differences inleg length. A lateral film is not required as a partof the screening examination.

Radiographic techniques should be used tominimize radiation of sensitive organs (eg, breast,thyroid, ovaries, bone marrow, and lens). It isimperative that radiation-lowering techniques areused judiciously to minimize the radiationburden. In a recent study of 5,466 women whoreceived an average of 24.7 radiographs witha mean estimated cumulative radiation dose tothe breast of 10.8 cGy (range, 0–170 cGy), therisk of breast cancer was found to be 70% higherthan in women in the general population [18].There were 77 breast cancer deaths among thesepatients, compared with 46 expected deaths basedon United States mortality rates. Risk increased sig-nificantly with increasing number of radiograph ex-posures and with cumulative radiation dose. Aposterior-to-anterior technique reduces radiation tothe breast 3 to 7 fold compared with an anterior-to-posterior technique. If breast shielding is used,one should take care that automatic exposuresystems do not increase the radiation doseaccordingly. A further decrease in radiation dosecan be accomplished using the air-gap technique.This technique was first described in 1934 byLindblom [19]. In clinical practice, it was soonsuperseded by the use of anti-scatter grids. Airgaps are still used in lung examinations and com-pare favorably with the use of grids in pediatricradiology. The introduction of digital radiographyis likely to create a new interest in the air-gaptechnique [20]. With air-gap and computed radiol-ogy techniques, the mean effective dose can bereduced by a factor of 10 [21].

When surgical treatment is considered, lateralbend radiographs and a lateral film should beacquired. Bend films aid in deciding what levelsshould be included in the instrumentation(Fig. 5). Lateral bending is usually performed asa standing postero-anterior film, but in some insti-tutions supine AP films are used. The Stagnara obli-que view is taken perpendicularly to the ribprominence and shows a more accurate picture oflarge curves with a true magnitude of the scoliosis.

Scoliosis 111

Fig. 5. Follow-up of scoliosis in an adolescent girl. (A) 14 years. 14.5-years postero-anterior full spine (B) undertraction (C) and with lateral bending (D). Postoperative postero-anterior full spine at 15 years (E) and 16 years(F). Curve progression of 5� thoracic and 8� lumbar in 6 months warranted surgical stabilization. Before surgerytraction and lateral bend, films are used to determine which curves are structural (both were structural in thiscase) and the extent of the surgical fixation. Surgery reduced the curves with 22� and 25�, respectively. Bothcurves progress 4� in the year after surgery.

Several studies have used three-dimensional(3-D) techniques to evaluate idiopathic scoliosis.These have showed that although the deformity ofthe spine is 3-D, the regional deformity is almostalways two dimensional, but in a plane differentfrom the standard frontal or sagittal views.

Sagittal balance is an important concept innormal spinal stability. Thoracic kyphosis and lum-bosacral lordosis help to maintain a normal posturewith minimum energy expenditure and absorb theloads applied to the spinal column. Many studies

have shown the negative effects of a reduced lumbarlordosis with fixed sagittal imbalance after spinalinstrumentation [22].

Imaging interval

Follow-up is necessary in patients who have severecurves that are at risk for significant curve progres-sion or require some form of treatment. Idiopathiccurves should be monitored every 4 to 12 months,depending on the age and growth rate of the patient(see Fig. 4). After skeletal maturity, only curves

Van Goethem et al112

greater than 30� should be monitored for progres-sion and usually only every 5 years.

Measurements

Curve measurement is usually performed by theCobb method (Fig. 6). The caudal and cranialend vertebrae of a scoliosis are the most tilted ina frontal plane. A line parallel to the superior end-plate of the cranial end vertebra and a line parallelto the inferior endplate of the caudal end vertebraare drawn first (on film or digitally on a diagnosticworkstation). Then perpendiculars to these lines aredrawn, and the angle is measured where these linescross. A Cobb angle is measured for each curve thatis present. When comparing different radiographs,the endvertebrae usually remain the same, althoughcorrections can be needed over time. There is a wideinter- and intraobserver variation with this tech-nique, usually in the order of 5�.

Cobb angle measurements are done on antero-posterior radiographs. Because of the associated ver-tebral rotation, these are not true AP views of therotated spinal segment. Cobb angle measurementscan increase by more than 20% when measuredon these true AP views [23]. Therefore, in the fol-low-up of scoliosis, consistent patient positioningis of utmost importance.

For surgeons, it is important to recognize the im-portant decrease of curves in the frontal and sagittalplane due to prone positioning, anesthesia, andexposure during surgery. When patients resume theirstanding position, a ‘‘spring-back’’ effect is noted inthe sagittal plane with loss of correction [24].

The Ferguson Method is used much less fre-quently than the Cobb angle measurement. TheFerguson Method measures the angle between linesdrawn from the centers of the end vertebrae to thecenter of the apical vertebra/disc.

As a result of the increased appreciation of the3-D nature of scoliosis and modern spinal instru-mentation’s improved corrective capabilities, therehas been renewed interest in the correction andmeasurement of vertebral rotation [25]. Vertebralrotation is maximal at the apex of the curve andcan be quantified by different methods, all of whichare inaccurate. CT is limited in its clinical utilityowing to cost, radiation exposure, and the effectsof postural changes on scoliosis curves and conse-quently vertebral rotation. Therefore, and becauseof their simplicity, the Perdriolle [26] and Nash-Moe [27] techniques remain the standard mea-surements for providing a reasonable estimate ofpre- and postoperative vertebral rotation.

On lateral films, sagittal balance can be assessed.Normal sagittal balance is the alignment of C7 to

Fig. 6. Curve measurement is usually performed by the Cobb method. These measurements are repeated foreach curve present, in this case thoracic (apex T9) and lumbar (apex L3) (A). The caudal and cranial end vertebraeof each scoliosis are the most tilted in a frontal plane (T5 and T12 for the thoracic curve) (B). A line parallel to thesuperior endplate of the cranial end vertebra and a line parallel to the inferior endplate of the caudal end ver-tebra are drawn first (C). The Cobb angle is measured for each curve that is present (44� for the thoracic curve inthis case). When comparing different radiographs, the endvertebrae usually remain the same (see Figs. 4 and 5),although corrections can be needed over time.

Scoliosis 113

the posterior superior aspect of the sacrum. The sag-ittal plumb line, as drawn from center of C7, shouldbe �2 cm from the sacral promontory. Thoracickyphosis depends mostly on the spinal deformity,whereas lumbar lordosis is influenced mainly bythe pelvic configuration. The scoliotic curve typeis not associated with a specific pattern of sagittalpelvic morphology and balance [28]. Positive sagit-tal balance (ie, an anterior deviation of the C7plumb line) is more significantly associated withpain and disability than curve magnitude, curvelocation, or coronal imbalance [29].

Skeletal maturity is usually assessed using theRisser sign. The lateral to medial ossification ofthe iliac crest occurs in a predictable fashion overan 18- to 24-month period. Risser 0 is the absenceof ossification, Risser stages 1 to 4 correspond topartial ossification, and Risser 5 indicates the fusionof the fully ossified apophysis to the ilium. Anotheruseful landmark is the status of the triradiate carti-lage of the acetabulum. This usually closes beforeRisser 0, at the stage of maximal spinal growth.

Specialized imaging

CT, especially multi-row detector CT, is the bestmethod for visualization of complex scolioticdeformities. In general, it is used in cases ofcomputer-assisted surgery because the placementof pedicles screws on the concavity in the apical re-gion of thoracic curves can be critical because ofsmall endosteal pedicle width [30]. CT is also usefulto define abnormalities and to pick up previouslyunrecognized anomalies in patients who have con-genital scoliosis [31]. The excision of hemivertebrais a technically challenging procedure and can beperformed as an anterior–posterior procedure oran isolated posterior procedure, and the use of CTis helpful in the operative planning of these patients[32]. Most surgeons prefer multi-row detector CTwith 3-D reconstructions over planar CT in the pre-operative depiction of congenital scoliosis [33]. MRimaging is required in infantile and juvenileidiopathic scoliosis (see Fig. 3), congenital bonyanomalies, and scoliosis associated with specificneurologic or cutaneous abnormalities.

The prevalence of neural axis abnormalities in in-fantile and juvenile idiopathic scoliosis with a curveof more than 20� is approximately 20% [34]. Theseinclude Chiari malformations, syringomyelia, and,less frequently, spinal or brain tumors. In adoles-cent idiopathic scoliosis, MR imaging should beconsidered in cases where any of these red flags ispresent [35] (eg, severe pain, a left thoracic curve,or an abnormal neurologic examination).

A more recent study indicates that pain as a soleindicator is not reliable for detecting pathology[36]. An atypical curve pattern most frequently is

the only indicator of abnormal MR imaging find-ings. This includes left thoracic curve, short-seg-ment scoliosis (4–6 levels), decreased vertebralrotation, absence of thoracic apical segment lordo-sis, and rapid progression. Other curve patterns areassociated with an increased incidence of neuralaxis abnormalities, including left thoracic, doublethoracic, triple thoracic, and a long right thoraciccurve with end vertebra caudal to T12, and witha high or low apex or end vertebra, especially inmale patients and in patients who have a normalto hyperkyphotic thoracic spine [37]. Patientswho have severe curves despite skeletal immaturityand an abnormal neurologic examination havea significant probability of neurogenic lesions[38]. In patients who have juvenile idiopathic scoli-osis and back pain, preoperative MR imagingshould be performed to eliminate the risk of post-operative neurologic deficits, except in patientswho have Lenke type 1 idiopathic scoliosis if intra-operative neural monitoring is to be performed.

Spinal cord abnormalities are seen in 3% ofpatients who have adolescent idiopathic scoliosisand mainly include syringomyelia and less fre-quently Chiari malformations [39]. Whether preop-erative MR imaging in all patients who haveadolescent idiopathic scoliosis is routinely indi-cated remains controversial [40]. The role of special-ized imaging in extremely severe scoliosis remainsunclear. MR screening of all patients who havescoliosis is not indicated.

Treatment

Non-operative treatment (braces)

In most cases, the aim of orthotic treatment (braces)is to avoid spinal surgery. In growing children,a spinal orthosis (brace) is indicated when a curveprogresses to 25� to 30� (see Fig. 4). Lesser curveswith an annual growth of more than 5� are an indi-cation for bracing. Braces are used only in patientswho have substantial remaining spinal growth(Risser 3 or less). The upper limit of curves manage-able with braces is 45�. Even in the most cooperativepatients, the final result of brace treatment is themaintenance of the curve degree at the level of thestart of bracing. Braces should be used 23 h/d,usually for several years, until the curve is stabilized.Generally, the brace should be worn at night untilskeletal maturity is reached (Risser 5 or no spinalgrowth for 18 months). Curve progression can belimited to less than 5� in 75% of patients, comparedwith 35% in a comparable nontreated group.

Surgical treatment

In general, curves greater than 45� in patients whohave remaining spinal growth should be corrected

Van Goethem et al114

surgically (see Fig. 5). Curves greater than 30� at theonset of the pubertal growth spurt increase rapidlyand present a 100% prognosis for surgery. Evencurves between 20� and 30� have a high progressionrisk and need careful follow-up. Timing of spinalsurgery is of utmost importance and depends onexpected curve progression. In congenital scoliosis,the curves tend to be short with little flexibility anddo not show substantial response to brace treat-ment. Progressive congenital scoliosis is thereforegenerally treated with surgery. The operative treat-ment of adult lumbar degenerative scoliosis isa challenge, and major complication rates rangefrom 56% to 75% [41]. Corrective instrumentation(rods) in combination with arthrodesis (strength)is the best method for achieving long-term results.

The typical posterior spinal approach uses theHarrington instrumentation. It consists of a distrac-tion rod with hooks at either end and a threadedcompression rod attached to the transverse pro-cesses on the convex side of the curve. This originalconcept corrected scoliosis at the cost of a decreasedthoracic kyphosis. This system was subsequentlymodified with different systems.

The Cotrel-Dubousset system is more recent(1980s) and uses a multihook concept that allowsdistraction and compression on the same rod.Many of these systems can be attached with hooks,wires, or pedicle screws.

Anterior spinal instrumentation is a newer tech-nique with several systems on the market. Initially,it was primarily used for the correction of lumbar orthoracolumbar scoliosis, but it is now also used inthe thoracic region. It can also be helpful in a com-bination anterior and posterior approach, espe-cially for curves larger than 75�, and in youngerpatients.

The correct surgical technique depends on thecurve pattern. For example, in the idiopathic rightthoracic curve pattern (Lenke type 1), posteriorspinal instrumentation and fusion of the thoraciccurve are common. The segment to be fused shouldbe as short as possible but long enough to mini-mize residual imbalance or progression. The lowesthook is attached above the level where the centralvertical sacral line bisects the spine. Shorter fusionsare possible with anterior instrumentation, includ-ing all vertebrae in the measured Cobb angle. Ina double thoracic curve (Lenke type 2), instrumen-tation is often extended up to T1 or T2. Differentschemes exist for other patterns.

Although instrumentation generally achievesgood to excellent improvement of the Cobb angle,there are conflicting reports on the long-term func-tional results. Complications range from blood lossover hardware failures to neurologic injury. Urinarytract infections are the most common medical

complication associated with adult spinal defor-mity surgery. Pulmonary complications, includingpneumonia and pulmonary embolism, are amongthe most frequently seen life-threatening complica-tions with deformity procedures [42].

References

[1] McMaster M. Spinal growth and congenital de-formity of the spine. Spine 2006;31(20):2284–7.

[2] Tracy MR, Dormans JP, Kusumi K. Klippel-Feilsyndrome: clinical features and current under-standing of etiology. Clin Orthop Relat Res2004;424:183–90.

[3] Alden KJ, Marosy B, Nzegwu N, et al. Idiopathicscoliosis: identification of candidate regions onchromosome 19p13. Spine 2006;31(16):1815–9.

[4] Stokes IA, Windisch L. Vertebral height growthpredominates over intervertebral disc heightgrowth in adolescents with scoliosis. Spine2006;31(14):1600–4.

[5] Guo X, Chau WW, Chan YL, et al. Relative ante-rior spinal overgrowth in adolescent idiopathicscoliosis: Results of disproportionate endochon-dral-membranous bone growth. J Bone JointSurg Br 2003;85(7):1026–31.

[6] Benli IT, Uzumcugil O, Aydin E, et al. Magneticresonance imaging abnormalities of neural axisin Lenke type 1 idiopathic scoliosis. Spine 2006;31(16):1828–33.

[7] Milbrandt TA, Johnston CE II. Down syndromeand scoliosis: a review of a 50-year experienceat one institution. Spine 2005;30(18):2051–5.

[8] Tribus CB. Degenerative lumbar scoliosis: evalu-ation and management. J Am Acad Orthop Surg2003;11(3):174–83.

[9] Aebi M. The adult scoliosis. Eur Spine J 2005;14(10):925–48.

[10] King HA, Moe JH, Bradford DS, et al. The selec-tion of fusion levels in thoracic idiopathic scoli-osis. J Bone Joint Surg Am 1983;65(9):1302–13.

[11] Lenke L, Betz R, Harms J, et al. Adolescent idio-pathic scoliosis: a new classification to determineextent of spinal arthrodesis. J Bone Joint Surg Am2001;83A:1169–81.

[12] Lenke LG, Edwards CC II, Bridwell KH. TheLenke classification of adolescent idiopathic sco-liosis: how it organizes curve patterns as a tem-plate to perform selective fusions of the spine.Spine 2003;28(20):S199–207.

[13] Kose N, Campbell RM. Congenital scoliosis.Med Sci Monit 2004;10(5):104–10.

[14] Ylikoski M. Growth and progression of adoles-cent idiopathic scoliosis in girls. J Pediatr OrthopB 2005;14(5):320–4.

[15] Lonstein JE, Carlosn JM. The prediction of curveprogression in untreated idiopathic scoliosisduring growth. J Bone Joint Surg 1984;66:1061–71.

[16] Charles YP, Daures JP, de Rosa V, et al. Progres-sion risk of idiopathic juvenile scoliosis duringpubertal growth. Spine 2006;31(17):1933–42.

Scoliosis 115

[17] Sanders J, Browne R, Cooney T, et al. Correlatesof the peak height velocity in girls with idio-pathic scoliosis. Spine 2006;31(20):2289–95.

[18] Morin Doody M, Lonstein JE, Stovall M, et al.Breast cancer mortality after diagnostic radiogra-phy: findings from the U.S. Scoliosis CohortStudy. Spine 2002;25(16):2052–63.

[19] Lindblom K. Secondary screening by means offiltering. Acta Radiol 1934;15:620–7.

[20] Persliden J, Carlsson GA. Scatter rejection by airgaps in diagnostic radiology: Calculations usinga Monte Carlo collision density method andconsideration of molecular interference in coher-ent scattering. Phys Med Biol 1997;42:155–75.

[21] Hansen J, Jurik AG, Fiirgaard B, et al. Optimisa-tion of scoliosis examinations in children. PediatrRadiol 2003;33:752–65.

[22] Kim Y, Bridwell K, Lenke L, et al. An analysis ofsagittal spinal alignment following long adultlumbar instrumentation and fusion to L5 orS1: can we predict ideal lumbar lordosis? Spine2006;31(20):2343–52.

[23] Gocen S, Havitcioglu H. Effect of rotation onfrontal plane deformity in idiopathic scoliosis.Orthopedics 2001;24(3):265–8.

[24] Delorme S, Labelle H, Poitras B, et al. Pre-, intra-,and postoperative three-dimensional evaluationof adolescent idiopathic scoliosis. J Spinal Dis-ord 2000;13(2):93–101.

[25] Kuklo TR, Potter BK, Lenke LG. Vertebral rotationand thoracic torsion in adolescent idiopathic sco-liosis: what is the best radiographic correlate?J Spinal Disord Tech 2005;18(2):139–47.

[26] Perdriolle R. La scoliose. Maloine, SA; Paris: 1979.[27] Nash C, Moe J. A study of vertebral rotation.

J Bone Joint Surg 1969;51:223–9.[28] Mac-Thiong JM, Labelle H, Charlebois M, et al.

Sagittal plane analysis of the spine and pelvisin adolescent idiopathic scoliosis according tothe coronal curve type. Spine 2003;28(13):1404–9.

[29] Glassman SD, Bridwell K, Dimar JR. The impactof positive sagittal balance in adult spinal defor-mity. Spine 2005;30(18):2024–9.

[30] Liljenqvist UR, Link TM, Halm HF. Morphomet-ric analysis of thoracic and lumbar vertebraein idiopathic scoliosis. Spine 2000;25(10):1247–53.

[31] Newton PO, Hahn GW, Fricka KB, et al. Utility ofthree-dimensional and multiplanar reformattedcomputed tomography for evaluation of pediat-ric congenital spine abnormalities. Spine 2002;27(8):844–50.

[32] Hedequist DJ, Emans JB. The correlation of pre-operative three-dimensional computed tomogra-phy reconstructions with operative findings incongenital scoliosis. Spine 2003;28(22):2531–4.

[33] Bush CH, Kalen V. Three-dimensional computedtomography in the assessment of congenitalscoliosis. Skeletal Radiol 1999;28(11):632–7.

[34] Dobbs MB, Lenke LG, Szymanski DA, et al. Prev-alence of neural axis abnormalities in patientswith infantile idiopathic scoliosis. J Bone JointSurg Am 2002;84(12):2230–4.

[35] Reamy BV, Slakey JB. Adolescent idiopathicscoliosis: review and current concepts. Am FamPhysician 2001;64(1):111–6.

[36] Davids JR, Chamberlin E, Blackhurst DW. Indi-cations for magnetic resonance imaging in pre-sumed adolescent idiopathic scoliosis. J BoneJoint Surg Am 2004;86-A(10):2187–95.

[37] Spiegel DA, Flynn JM, Stasikelis PJ, et al. Scoli-otic curve patterns in patients with Chiari I mal-formation and/or syringomyelia. Spine 2003;28(18):2139–46.

[38] Morcuende JA, Dolan LA, Vazquez JD, et al. Aprognostic model for the presence of neurogeniclesions in atypical idiopathic scoliosis. Spine2004;29(1):51–8.

[39] Hausmann ON, Boni T, Pfirrmann CW, et al.Preoperative radiological and electrophysiologi-cal evaluation in 100 adolescent idiopathic scoli-osis patients. Eur Spine J 2003;2(5):501–6.

[40] Do T, Fras C, Burke S, et al. Clinical value of rou-tine preoperative magnetic resonance imaging inadolescent idiopathic scoliosis: A prospectivestudy of three hundred and twenty-sevenpatients. J Bone Joint Surg Am 2001;83-A(4):577–9.

[41] Akbarnia BA, Ogilvie JW, Hammerberg KW.Debate: degenerative scoliosis: to operate ornot to operate. Spine 2006;31(19):S195–201.

[42] Baron EM, Albert TJ. Medical complications ofsurgical treatment of adult spinal deformityand how to avoid them. Spine 2006;31(19):S106–18.