The Spine Journal 13 (2013) 1485–1492

Clinical Study

Effectiveness of cross-linking posterior segmental instrumentation inadolescent idiopathic scoliosis: a 2-year follow-up comparative study

Arjun A. Dhawale, MD, Suken A. Shah, MD*, Petya Yorgova, MS, Geraldine Neiss, PhD,Douglas J. Layer, Jr., BS, Kenneth J. Rogers, PhD, ATC, Peter G. Gabos, MD,

Laurens Holmes, Jr., PhD, DrPHDepartment of Orthopedics, Nemours/Alfred I. duPont Hospital for Children, 1600 Rockland Rd, PO Box 269, Wilmington, DE 19899, USA

Received 8 March 2012; revised 8 February 2013; accepted 4 May 2013

Abstract BACKGROUND CONTEXT: Surgeons conti

FDA device/drug

Author disclosure

Scoliosis Straight Fo

Royalties: Depuy Syn

Synthes Spine, Inc (D

ing/Teaching Arrange

port (Staff/Materials)

directly to institution

Board of Directors: S

Straight Foundation (n

close. DJL: Nothing

1529-9430/$ - see fro

http://dx.doi.org/10.10

nue to debate the need for a cross-link (CL) in pos-terior spinal instrumentation constructs with segmental pedicle screws in adolescent idiopathic sco-liosis (AIS). Advantage of CLs is increased stiffness of the construct, and disadvantages includeadded expense and risk of late operative-site pain and pseudarthrosis.PURPOSE: To compare the effectiveness of using CLs versus using no cross-links (NCLs) in pos-terior segmental instrumentation in AIS.STUDY DESIGN: Retrospective comparative study, level of evidence 3.PATIENT SAMPLE: Seventy-five AIS patients less than 21 years of age, who underwent poste-rior spinal instrumentation with segmental pedicle screws (25 with CLs and 50 with NCLs) at a sin-gle institution with 2-year follow-up, are described.OUTCOME MEASURES: Physiologic measures include imaging: thoracic and lumbar Cobb an-gles, correction rate, apical vertebral translation (AVT), and apical vertebral rotation (AVR); self-report measures include Scoliosis Research Society (SRS) domain outcome scores.METHODS: Preoperative (pre-op) and postoperative first erect, 1-year, and 2-year follow-up ra-diographs were measured. Instrumentation-related complications and normalized SRS scores wererecorded. Independent sample t test, c2 test, and repeated-measures analysis of variance were usedfor analyses.RESULTS: The average age at surgery was 14 years, the mean pre-op Cobb angle was 57�, and themean number of levels fused was 10.9. The groups were similar preoperatively with respect to age,sex, Lenke curve, Cobb angle, AVT, and Risser grade and were similar intraoperatively for levelsfused and anchor density. There was no difference in AVR, Cobb angle, correction rate, or AVTbetween the groups (pO.05). Complications included one wound infection in the CL group andone painful scar in the NCL group. There were no differences in SRS domain scores.CONCLUSION: We observed no differences in maintenance of correction, SRS scores, and com-plications with or without cross-linking posterior segmental instrumentation in AIS patients over 2-year follow-up. Further follow-up is necessary. � 2013 Elsevier Inc. All rights reserved.

Keywords: Cross-links; Transverse connectors; Pedicle screw; Posterior instrumentation; AIS

status: Approved (cross-link).

s: AAD: Nothing to disclose. SAS: Grant: Setting

undation (E, Paid directly to institution/employer);

thes Spine, Inc (E), Arthrex (B); Consulting: DePuy

), Ethicon endosurgery (B), K Spine (none); Speak-

ments: DePuy Synthes Spine, Inc (B); Research Sup-

: Setting Scoliosis Straight Foundation (E, Paid

/employer); Stock options: Globus Medical (D);

coliosis Research Society (none), Setting Scoliosis

one). PY: Nothing to disclose. GN: Nothing to dis-

to disclose. KJR: Nothing to disclose. PGG:

Consulting: DePuy Synthes Spine (B); Speaking/Teaching Arrangements:

DePuy Synthes Spine (B). LH: Nothing to disclose.

The disclosure key can be found on the Table of Contents and at www.

TheSpineJournalOnline.com.

Institutional Review Board approval was obtained from the Nemours

Institutional Review Board.

* Corresponding author. Department of Orthopedics, Nemours/Alfred

I. duPont Hospital for Children, 1600 Rockland Rd, PO Box 269, Wil-

mington, DE 19899, USA. Tel.: (302) 651-5904; fax: (302) 651-5951.

E-mail address: sshah@nemours.org (S.A. Shah)

nt matter � 2013 Elsevier Inc. All rights reserved.

16/j.spinee.2013.05.022

ContextWhile cross-linking dual rods in posterior instrumenta-

tion improved measured stability of constructs, whether

this improved clinical outcomes is unclear.

ContributionIn this retrospective cohort study in patients undergoing

fusion for AIS with and without cross-linking of the in-

strumentation, the authors found no differences in clini-

cal or radiographic outcomes.

ImplicationsThe primary implication of these data are in terms of

cost, which might be lowered by the avoidance of

cross-linking. That said, the numbers in this study are

small; differences between the groups may be present

but were undetected by the current analysis.

1486 A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

Introduction

Surgeons continue to debate the need for a cross-link (CL)in posterior spinal instrumentation constructs with segmentalpedicle screws in adolescent idiopathic scoliosis (AIS).Biomechanical testing of hook/rod constructs has shown in-creased stiffness and torsional rigidity of the construct whentransverse CLs were used compared with unlinked rods [1].With the advent of pedicle screw constructs, biomechanicalstudies were performed to determine if transverse CLs pro-vided any additional benefit [2–10]. A significant numberof these studies demonstrated that CLs provided additionalstiffness and torsional rigidity [2–6,10]. This additional stiff-ness could be beneficial in maintaining correction and pro-ducing a stiffer fusion mass [7,11].

In contrast, Wood et al. [7] biomechanically tested an in-dustrially fabricated spine model and showed that with theuse of distal pedicle screw anchors in scoliosis constructs,CLs provided no significant additional rotational stiffness.Valdevit et al. [8] performed a biomechanical analysis ofdifferent CL configurations with rod screw constructs onthoracic porcine spines (T4–T10) and found no significantdifferences in torsional stiffness of the transverse CL con-struct compared with the unlinked rod constructs; although,they did find that diagonal fixation provided a more stableconstruct. Burney et al. [9] found no differences in rota-tional stiffness with segmental pedicle screw fixation of10-level cadaver spine model constructs with or withoutthe use of CLs. Disadvantages of using CLs include addedexpense, bursa formation, and risk of late operative-sitepain [12,13] and pseudoarthrosis [14].

We are not aware of any clinical studies comparing the ef-fectiveness of posterior screw rod constructs with CLs versusconstructs with no cross-links (NCLs) in AIS patients. Our

primary aim was to compare the maintenance of radiologicalcorrection in AIS patients treated with and without CLs ata single institution over a 2-year follow-up. Our secondaryaims were to compare the self-reported Scoliosis ResearchSociety (SRS) outcome domain scores and assess theinstrumentation-related complications.

Study design

Retrospective cohort comparative study, level ofevidence 3.

Patient sample

After institutional review board approval, we retrospec-tively reviewed a database of AIS patients who underwentprimary posterior spinal instrumentation and fusion withsegmental pedicle screws, by either of two surgeons at a sin-gle institution between December 1, 2005, and April 30,2009. Two groups of patients were identified: a CL groupin which the posterior instrumentation was cross-linkedwith one or two transverse connectors and an NCL groupin which the posterior instrumentation was not cross-linked. The use of CLs was based on surgeon preference.Complete 2-year follow-up included a preoperative (pre-op) visit and postoperative (post-op) first erect (FE), 1-year,and 2-year visits. Seventy-five patients who had complete2-year follow-up details were included in the study. Ofthese, there were 25 patients in the CL group and 50 pa-tients in the NCL group.

Outcome measures

Physiologic measures

ImagingThoracic and lumbar Cobb angle, correction rate, apical

vertebral translation (AVT), and apical vertebral rotation(AVR). Full-length standing posteroanterior and lateralroentgenograms were used for all radiological measure-ments. At pre-op, the measurements recorded were thoracicCobb angle, lumbar Cobb angle, lateral Cobb angle (fromT5–T12), curve flexibility on bending films, Lenke classifi-cation [15], Risser grade [16], thoracic AVT, and thoracicAVR using the method of Nash and Moe [17]. Measure-ments on FE, 1-year, and 2-year radiographs included tho-racic Cobb angle, lumbar Cobb angle, lateral Cobb angle,thoracic AVT, and Risser grade. Percentage correctionwas calculated as the difference between the pre-op andpost-op Cobb angles at each visit; this difference was thendivided by pre-op Cobb angle and multiplied by 100. Post-operative residual thoracic AVR was recorded at each visitusing a radiographic grading system, described by Upasaniet al. [18] based on the location of the apical vertebralscrew tip with respect to the rod. The grading scale wasas follows: grade 0, approximately 0� to 8� rotation; grade1, 9� to 12� rotation; and grade 2, more than 13� rotation.

1487A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

Quality-of-life self-reported measures

We evaluated the normalized self-reported SRS out-comes [19] for the five domains (pain, general function,mental health, satisfaction, and self-image) and the totalscores at pre-op, 1-year, and 2-year follow-ups where theywere available.

Methods

Demographic data

Preoperative visit details, operation and instrumentationdetails, complications, and other outcome measures (as de-scribed previously) were recorded over a 2-year follow-up.

Surgical details

All patients had posterior spinal instrumentation and fu-sion using the Expedium 5.5 system (DePuy Spine, Rayn-ham, MA, USA). Segmental pedicle screw fixation wasperformed at every level on the concave side and at inter-rupted levels on the convex side. Two precontoured 5.5-mm longitudinal rods (predominantly stainless steel [SS])were used in all cases. The number of pedicle screws andinstrumented (fused) vertebral levels were recorded. Uni-planar pedicle screws were predominantly used (O80%)in all cases. The anchor density was calculated by dividingthe number of instrumented (fused) vertebral levels by thenumber of pedicle screws. The use of CL connectors(SFX; Depuy Spine) was documented along with the num-ber and location. One CL was used in 21 constructs, and 2were used in 4 constructs.

Complications

Possible instrumentation-related complications were re-corded. These included late wound infections (O6 monthspost-op), implant prominence, bursa formation, lateoperative-site pain (LOSP), characterized bymidline or peri-scapular pain that was made worse by direct palpation of theincision [12], implant failure, and pseudarthrosis. Radio-graphic criteria for assessment of pseudarthrosis were lossof correction more than 10� and instrumentation failure.

Power and sample size estimations

To estimate the power for the comparison of indepen-dent samples of thoracic Cobb angle between CL andNCL groups, we performed an a priori power analysis usingthe following parameters: a50.05 (type 1 error tolerance);sample size n1 (NCL)550, n2 (CL)525, n2/n150.5; mean1(NCL)520; mean2 (CL)525; standard deviation153.0; andstandard deviation256.0, with mean difference set at 5�.We obtained an adequate power of 0.9 (97%).

To determine whether our study would be able to detectthe minimum difference in AVR between the CL and NCL

groups, if one really existed, we performed power analysisusing the following parameters: a50.05 (type 1 error toler-ance); sample size n1 (NCL)550, n2 (CL)525; and effectsize50.35. We obtained a sufficient power of 0.80 (80%).

Statistical analysis

Our hypothesis was that there was no difference in main-tenance of curve correction, AVT, and AVR between the CLand NCL groups. The continuous variables (age, coronalthoracic Cobb angle, lumbar Cobb angle, lateral Cobb an-gle, percent correction, AVT, number of levels fused, num-ber of anchors, and anchor density) were tested fornormality with the Shapiro-Francia test. After assessingnormality, the CL and NCL groups were compared. In-dependent samples t test (parametric test) was used tocompare the baseline pre-op and operative measures.Repeated-measures analysis of variance (parametric test)was used to compare the pre-op and FE, 1-year, and 2-year follow-up measurements to assess the maintenanceof coronal Cobb angle, curve correction, lateral Cobb an-gle, and AVT. The chi-square (c2) test (nonparametric)was used to compare the ordinal measures, gender, Rissergrade, Lenke curve type (pre-op), and AVR grade (pre-opand FE, 1 year, and 2 year) between the CL and NCLgroups. The normalized SRS outcome scores for the do-mains were compared with the median c2 test. All testswere two-tailed, and the significance level was 0.05. STA-TA version 11 (StataCorp, College Station, TX, USA) andSPSS version 17 (SPSS, Chicago, IL, USA) were usedfor analyses.

Results

There were 14 males and 61 females with a mean ageof 14.062.3 years and mean thoracic Cobb angle of57�611.3�. The mean number of levels fused was 10.961.7,the mean number of anchors was 19.263.1, and the mean an-chordensitywas1.7660.1.Thecomparisonof themeanage atsurgery, pre-op thoracic and lumbar Cobb angles, kyphosis,AVT, number of levels fused, number of anchors, and anchordensity between the CL and NCL groups is shown inTable1.The twogroupswere similarwith respect to thesevari-ables (pO.05).

Sex, Risser grade, and Lenke curve distribution in theCL and NCL groups with c2 test values are shown inTable 2. There were no differences between the groups(pO.05).

Of the 75 constructs, there were 63 SS (84%), 11 tita-nium (Ti) (14.7%), and 1 cobalt-chrome implants (1.3%).In the CL group, there were 18 SS (72%), 6 Ti (24%),and 1 cobalt-chrome (4%) implants. In the NCL group,there were 45 SS (90%) and 5 Ti (10%) implants. Therewas no significant difference between the groups with re-spect to implant material with Pearson c2 test (p5.08).

Table 1

Patient characteristics of the CL and NCL groups describing baseline pre-

op and intraoperative variables

Variables

CL group,

N525

NCL group,

N550

t p Value*Mean6SD Mean6SD

Age (y) 13.862.2 14.162.4 0.5 .64y

Thoracic Cobb angle ( �) 58.6611.2 56.2611.5 �0.87 .38y

Lumbar Cobb angle ( �) 37.8614.1 36.5611.5 �0.42 .67z

Kyphosis (T5–T12) ( �) 23.4612.6 19.9614.2 �1.10 .28z

AVT (cm) 4.961.9 4.063.7 �1.45 .15z

Levels fused (n) 11.361.6 10.861.8 �1.3 .19y

Anchors (n) 19.963.0 18.963.1 �1.2 .22y

Anchor density 1.7660.1 1.7660.1 0.11 .91

CL, cross-link; NCL, no cross-link; pre-op, preoperative; SD, standard

deviation; t, two-sample t test; AVT, apical vertebral translation.

* Significance level, p!.05.y t with equal variance.z t with unequal variance.

1488 A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

Preoperative AVR grade percentages in the CL and NCLgroups were similar (c2, p5.74). First erect, 1-year, and2-year follow-up AVR grade percentages (Fig. 1) showedno differences with c2 test (pO.05, Table 3). In comparingthe FE and 2-year grades, we found that one patient in theCL group and three patients in the NCL group had progres-sion of AVR grade; however, this was not statisticallysignificant.

No differences in maintenance of correction of meanthoracic Cobb angle (Fig. 2), lumbar Cobb angle, lateralCobb angle, AVT, or percent correction were found be-tween the CL and NCL groups on repeated-measures anal-ysis of variance (Table 4).

Table 2

Patient characteristics of the CL and NCL groups describing sex, Rissser

grade, and Lenke curve type

Variables

CL group, N525 NCL group, N550

c2 (df) p Value*n % n %

Sex 0.70 (1) .40y

Females 19 76 42 84

Males 6 24 8 16

Risser grade 4.06 (5) .54y

0 5 20 10 20

1 3 12 6 12

2 1 4 6 12

3 5 20 4 8

4 6 24 17 34

5 5 20 7 14

Lenke type 8.14 (5) .11z

1 14 56 35 70

2 4 16 10 20

3 3 12 6 6

4 3 12 0 0

5 0 0 1 2

6 1 4 1 2

CL, cross-link; NCL, no cross-link; df, degree of freedom.

* Significance level, p!.05.y Pearson c2 test.z Fisher exact test.

There were no discernable differences in the individualSRS domains of pain, general function, mental health, sat-isfaction, and self-image nor in the total scores (median-based c2 test, pO.05, Table 5).

Complications

Deep wound infectionOne patient in the CL group had a deep wound infection

that required exploration. The patient presented 6 monthsafter surgery with increasing pain and some swellingaround incision area. Deep lumbar spinal abscess wasfound at exploration, and drainage and debridement witha vacuum-assisted closure were performed twice. Culturesgrew Pseudomonas aeruginosa sensitive to cefepime. Thepatient was doing well on subsequent follow-up.

Painful scarOne patient in the NCL group had a painful operative

scar at 6 months and was referred to a plastic surgeon forscar revision.

There were no reports of LOSP, implant failure, pseu-darthrosis, symptomatic implant prominence, or bursaformation.

Discussion

Although biomechanical studies have evaluated the roleof CLs, there is some variability in the results [2–10]. Thiscould be because of differences in the CL designs of differ-ent instrumentation systems, in the model used for testing(ie, biological models [cadaver, porcine, calf], industrialspine models, or wood-block models), and in the lengthof the constructs. In some studies, biomechanical flexibil-ity testing in lateral bending (z axis) did not show signif-icant differences in stiffness with use of CLs [3,6,7,10]. Incontrast, others found increased lateral bending stiffnesswith use of CLs [2,4,10]. Our findings showed no dif-ferences in maintenance of coronal Cobb angle correctionand AVT in the groups. No significant progression ofthe major curve was seen in any of the patients. Biome-chanical flexibility testing in flexion-extension (x-axis)has not shown differences in stability with use of trans-verse CLs [3,6,7,10]. Lim et al. [20] have shown that di-agonal transfixation provided more rigid fixation inflexion and extension. In our study, there were no differ-ences in maintenance of lateral Cobb angle correction inthe groups.

Biomechanical testing in axial rotation (y-axis) has dem-onstrated significant increase in torsional stiffness with theuse of CLs in a significant number of studies [1–6,10]. Incontrast, Wood et al. [7] have found that with the use ofpedicle screws as distal anchors, the CLs provide very littleadditional rotational stiffness. Valdevit et al. [8] found nodifferences in torsional stiffness of the transverse CL

Fig. 1. Comparison of the percentages of apical vertebral rotation (AVR) grades between the cross-link (CL) and no cross-link (NCL) groups at first erect, 1-

year, and 2-year follow-ups showed no significant differences. The postoperative (post-op) 2-year AVR shows the lines are superimposed, implying absolutely

no differences in the AVR between the CL and NCL groups.

1489A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

construct compared with the unlinked rod constructs, al-though they did find that diagonal fixation provided a morestable construct. Burney et al. [9] found no differences inrotational stiffness with segmental pedicle screw fixationwith or without the use of CLs, although they did find

Table 3

Comparison of AVR between CL and NCL groups

AVR

CL group,

N525

NCL group,

N550

c2 p Value*n % n %

Pre-opy 1.25 .74z

0 1 4 3 6

1 9 36 14 28

2 13 52 31 62

3 2 8 2 4

FEx .97k

0 6 24 12 24 0.04

1 13 52 27 54

2 6 24 11 22

Post-op (1 year)x 0.47 .78k

0 5 21 11 22

1 15 60 26 52

2 5 20 13 26

Post-op (2 years)x 0 .99k

0 5 20 10 20

1 13 52 26 52

2 7 28 14 28

AVR, apical vertebral rotation; FE, first effect; CL, cross-link; NCL, no

cross-link; pre-op, preoperative; post-op, postoperative.

* Significance level, p!.05.y Pre-op AVR grading (Nash and Moe [17]).z Fisher exact test.x Post-op AVR grading (Upasani et al. [18]).k Pearson c2 test.

a difference with wood-block models. We found no signif-icant differences in AVR grade percentages between the CLand NCL groups, either pre- or post-op. The method ofNash and Moe [17] has fair-to-good reliability for thepre-op AVR measurement [21,22]. We did not do computedtomography (CT) scans for assessment of rotation becauseof the radiation risk involved. The method of post-op AVRmeasurement described by Upasani et al. [18] has agood correlation with CT measurements (r50.92, p!.001),with an average absolute difference of 1.9�62.0�. Inaccur-acies with this measurement technique can occur with sur-geons who use only polyaxial screws or those who do notuse intrapedicular screw placement techniques [18]. Therewere only two surgeons; so, there is less variability in themethod of screw insertion. The surgeons predominantlyused uniplanar screws and always used intrapedicular screwplacement techniques. This is a single-center study; so, themethod of obtaining radiographs for all patients wasstandardized.

The advantage of the use of CLs is the additional stiff-ness provided, which could be beneficial in maintainingcorrection and producing a stiffer fusion mass [7,11]. Dis-advantages are increased expense, bursa formation, risk ofLOSP [12,13], and risk of pseudoarthrosis [14]. Cook et al.[12] and Asher et al. [13] identified that transverse connec-tor design was an important factor influencing the reopera-tion need in the long-term survivorship after primaryposterior spinal instrumentation and fusion. They identifieda group of patients with severe LOSP of unknown causewarranting implant removal. The average interval to reoper-ation was 46 months, and implant corrosion at the rod andtransverse connection sites was frequently observed in the

Fig. 2. Comparison of mean thoracic Cobb angles between the cross-link (CL) and no cross-link (NCL) groups showed no significant difference at preop-

erative, first erect, 1-year, and 2-year follow-ups.

1490 A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

reoperated patients. The LOSP was thought to be caused bylocal soft-tissue reaction. The threaded transverse connectordesign was changed to a stronger, closed drop-entry trans-verse connector. The stronger transverse connector designalong with the most stable lower instrumented vertebra

Table 4

Comparison of Cobb angle, AVR, and correction rate between the CL and NCL

Variables

CL group, N525

Mean6SD

Thoracic Cobb angle

Pre-op 58.6611.2

Post-op FE 16.166.2

1 year 17.765.2

2 years 20.566.3

Lumbar Cobb angle

Pre-op 37.8614.1

Post-op FE 11.667.0

1 year 12.566.1

2 years 12.967.0

Thoracic Cobb correction

Post-op FE 71.5612.3

1 year 68.9610.6

2 years 64.0612.2

Lumbar Cobb correction

Post-op FE 69.3617.2

1 year 65.1613.8

2 years 66.5615.4

AVT (cm)

Pre-op 4.961.9

Post-op FE 0.661.3

1 year 1.161.4

2 years 1.461.1

Lateral Cobb angle (T5–T12)

Pre-op 23.4612.6

Post-op FE 19.466.1

1 year 18.366.5

2 years 19.565.4

F value, obtained with F test for repeated-measures analysis of variance; AVR

NCL, no cross-link; FE, first effect; SD, standard deviation; pre-op, preoperative

* Significance level (between-subject effects) on repeated-measures analysis

anchor with pedicle screws provided the best survival with-out reoperation with the system (Isola; Depuy Spine, Inc.).We did not have any cases of LOSP in our series. One ofthe reasons could be that our follow-up is not long enough.Kim et al. [14] studied the risk factors for pseudarthrosis in

groups

NCL group, N550

F value p Value*Mean6SD

0.05 .81

56.2611.5

15.665.8

19.666.9

20.467.2

0.009 .92

36.5611.5

12.067.0

12.967.4

12.967.7

0.43 .52

72.069.8

64.6612.6

62.8613.4

0.34 .56

66.3617.9

63.5618.6

64.4619.5

2.8 .09

4.063.7

�0.161.5

0.761.4

0.961.5

0.01 .92

19.9614.2

19.466.1

20.965.9

19.567.1

, apical vertebral rotation; AVT, apical vertebral translation; CL, cross-link;

; post-op, postoperative.

of variance, p!.05.

Table 5

Comparison of self-reported SRS domain outcomes between CL and NCL

groups

SRS domain

CL group NCL group

c2 value p Value*Median IQR Median IQR

Pre-op

Pain 4.0 0.6 4.2 0.8 0.45 .50

Self-image 3.4 0.6 3.45 0.7 0.04 .84

General function 4.5 0.7 4.75 0.5 1.30 .25

Mental health 4.4 0.4 4.2 1.0 0.71 .39

Satisfaction 3.5 1.0 4.0 1.5 2.37 .12

Total 4.0 0.4 4.0 0.7 0.04 .84

Post-op (1 year)

Pain 4.6 0.4 4.6 0.8 3.07 .08

Self-image 4.6 0.4 4.6 0.8 0.13 .71

General function 4.7 0.5 5.0 0.3 1.50 .11

Mental health 4.6 0.2 4.6 0.9 0.12 .72

Satisfaction 5.0 0.5 5.0 0.5 0.56 .57

Total 4.6 0.2 4.6 0.5 0.81 .36

Post-op (2 years)

Pain 4.4 1.0 4.6 0.8 1.07 .30

Self-image 4.8 0.6 4.8 0.6 0.55 .45

General function 4.8 0.2 5.0 0.3 0.36 .71

Mental health 4.6 0.8 4.4 1.1 0.12 .72

Satisfaction 5.0 0.5 5.0 0 0.9 .32

Total 4.5 0.4 4.6 0.5 0.04 .83

CL, cross-link; NCL, no cross-link; pre-op, preoperative; post-op, post-

operative; IQR, interquartile range.

* Pearson c2 test; significance level, p!.05. Pre-op SRS scores: n517

patients (CL) and 44 patients (NCL); 1-year SRS scores: n517 patients

(CL) and 48 patients (NCL); 2-year SRS scores: n517 patients (CL)

and 48 patients (NCL).

1491A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

16 patients with adult idiopathic scoliosis and found that thesite of CLs correlatedwith the pseudarthrosis site in 69%.Wedid not have any reports of pseudarthrosis sofar, although further follow-up is necessary. Radiographiccriteria for assessment of pseudarthrosis were loss of cor-rection of more than 10� and instrumentation failure. WhenCLs were used, specific precautions were taken, includingmeticulous decortications and bone grafting before placingthe CL. The design of the CL and clearance underneath theCL is important so that it does not impede fusion.We encoun-tered one case of wound infection in the CL group; however,we were unable to determine the cause of this delayed infec-tion. One patient in the NCL group developed a painfuloperative-site scar, though this was not related to the implantprominence.

All patients with CLs during the study period with 2-year follow-up available were included to avoid a selectionbias and to increase the sample size. We had 21 patientswhere a single CL was used in the construct and 4 patientswhere 2 CLs were used. Biomechanical studies have shownthat after long-segment pedicle screw constructs, the addi-tion of a single CL decreases the axial rotation and T4–T10range of motion by 20%, with the location of the CL withinthe construct being irrelevant [6]. Addition of a second CLhas an additive effect.

It is important to state here that our study population ofAIS patients had an average age of 14 years, a mean pre-op

Cobb angle of 57�, and a mean number of levels fused of10.9. None of the patients underwent vertebral column resec-tion. The results should be interpreted in context of the studypopulation. Cross-linking may be desirable in certain situa-tions, such as decreased implant density, significant spinal in-stability, and decreased bone implant interface. The authorscurrently use CLs for long fusions to the pelvis (eg, neuro-muscular scoliosis), poor bone quality, osteopenia, con-structs with low implant density, after vertebral columnresection for severe scoliosis, and in some younger patients.For a typical AIS patient with sufficient segmental fixation,CLs are not routinely used.

Limitations

There are some limitations of this study. The use of CLswas not randomized but based on surgeon preference; how-ever, it is unlikely that this study is influenced by the non-randomized design, given the comparability of the baselinedata. We reviewed our radiographs and have not identifiedany cases of implant failure, substantial loss of correction,or symptomatic pseudarthrosis. At minimum follow-up of 2years, our follow-up period may not be long enough toevaluate the late complications, namely implant failureand pseudarthrosis. Also evaluation of pseudarthrosis wasdone on radiographs and not CT. Although CT evaluationwould be more accurate to evaluate the fusion, the radiationrisk did not warrant obtaining routine CT in AIS patients.We acknowledge that evaluation of pseudarthrosis at 2years is difficult and further follow-up is necessary. Apicalvertebral rotation measurement may be influenced by pa-tient positioning during the X-ray. This measurement isgraded on an ordinal scale and may not be sensitive in de-tecting smaller differences; although, this may not be clin-ically significant. SRS outcome scores were not available insome patients.

Conclusions

There were no differences in maintenance of radiologi-cal correction, self-reported SRS domain scores, and com-plications with or without cross-linking of posterior spinalinstrumentation with pedicle screws, in a similar and com-parable cohort of AIS patients followed over a period of 2years. Further prospective studies and extended follow-upare necessary to assess late complications like pseudarthro-sis, implant failure, and LOSP.

Acknowledgment

We would like to acknowledge the Harms Study Groupof the Setting Scoliosis Straight Foundation.

References

[1] Asher M, Carson W, Heinig C, et al. A modular spinal rod linkage

system to provide rotational stability. Spine 1988;13:272–7.

1492 A.A. Dhawale et al. / The Spine Journal 13 (2013) 1485–1492

[2] Brodke DS, Bachus KN, Mohr RA, Nguyen BK. Segmental pedicle

screw fixation or cross-links in multilevel lumbar constructs. A bio-

mechanical analysis. Spine J 2001;1:373–9.

[3] Dick JC, Zdeblick TA, Bartel BD, Kunz DN. Mechanical evaluation of

cross-link designs in rigid pedicle screw systems. Spine 1997;22:370–5.

[4] Lynn G, Mukherjee DP, Kruse RN, et al. Mechanical stability of thor-

acolumbar pedicle screw fixation. The effect of crosslinks. Spine

1997;22:1568–72.

[5] Pintar FA, Maiman DJ, Yoganandan N, et al. Rotational stability of

a spinal pedicle screw/rod system. J Spinal Disord 1995;8:49–55.

[6] Kuklo TR, Dmitriev AE, Cardoso MJ, et al. Biomechanical contribu-

tion of transverse connectors to segmental stability following long seg-

ment instrumentation with thoracic pedicle screws. Spine 2008;33:

E482–7.

[7] Wood KB, Wentorf FA, Ogilvie JW, Kim KT. Torsional rigidity of

scoliosis constructs. Spine 2000;25:1893–8.

[8] Valdevit A, Kambic HE, McLain RF. Torsional stability of cross-link

configurations: a biomechanical analysis. Spine J 2005;5:441–5.

[9] Burney MU, Mukherjee DP, Ogden AL, et al. A biomechanical study

of posterior spinal instrumentation using pedicle screws with and

without cross-links. J Spinal Disord Tech 2005;18:364–8.

[10] Lim TH, Eck JC, An HS, et al. Biomechanics of transfixation in ped-

icle screw instrumentation. Spine 1996;21:2224–9.

[11] Johnston CE II, Ashman RB, Baird AM, Allard RN. Effect of spinal

construct stiffness on early fusion mass incorporation. Experimental

study. Spine 1990;15:908–12.

[12] CookS,AsherM, Lai SM, Shobe J. Reoperation after primary posterior

instrumentation and fusion for idiopathic scoliosis. Toward defining

late operative site pain of unknown cause. Spine 2000;25:463–8.

[13] Asher M, Lai SM, Burton D, et al. Safety and efficacy of Isola instru-

mentation and arthrodesis for adolescent idiopathic scoliosis: two- to

12-year follow-up. Spine 2004;29:2013–23.

[14] Kim YJ, Bridwell KH, Lenke LG, et al. Pseudarthrosis in primary

fusions for adult idiopathic scoliosis: incidence, risk factors, and out-

come analysis. Spine 2005;30:468–74.

[15] Lenke LG, Betz RR, Harms J, et al. Adolescent idiopathic scoliosis:

a new classification to determine extent of spinal arthrodesis. J Bone

Joint Surg Am 2001;83-A:1169–81.

[16] Risser JC. The iliac apophysis; an invaluable sign in the management

of scoliosis. Clin Orthop 1958;11:111–9.

[17] Nash CL Jr, Moe JH. A study of vertebral rotation. J Bone Joint Surg

Am 1969;51:223–9.

[18] Upasani VV, Chambers RC, Dalal AH, et al. Grading apical vertebral

rotation without a computed tomography scan: a clinically relevant

system based on the radiographic appearance of bilateral pedicle

screws. Spine 2009;34:1855–62.

[19] Asher M, Min Lai S, Burton D, Manna B. The reliability

and concurrent validity of the scoliosis research society-22 patient

questionnaire for idiopathic scoliosis. Spine 2003;28:63–9.

[20] Lim TH, Kim JG, Fujiwara A, et al. Biomechanical evaluation of di-

agonal fixation in pedicle screw instrumentation. Spine 2001;26:

2498–503.

[21] Kuklo TR, Potter BK, Polly DW Jr, et al. Reliability analysis for

manual adolescent idiopathic scoliosis measurements. Spine 2005;

30:444–54.

[22] Kuklo TR, Potter BK, O’Brien MF, et al. Reliability analysis for dig-

ital adolescent idiopathic scoliosis measurements. J Spinal Disord

Tech 2005;18:152–9.