Cross-links in posterior pedicle screw-rod instrumentation of the spine: a systematic review on mechanical, biomechanical, numerical and clinical studies

Dorsal screw-rod instrumentations are used for a variety of spinal disorders. Cross-links (CL) can be added to such constructs, however, no clear recommendations exist. This study aims to provide an overview of the available evidence on the effectiveness of CL, potentially allowing to formulate recommendations on their use. A systematic literature review was performed on PubMed and 37 original articles were included and grouped into mechanical, biomechanical, finite element and clinical studies. The change in range of motion (ROM) was analyzed in mechanical and biomechanical studies, ROM, stiffness and stress distribution were evaluated in finite element studies and clinical outcome parameters were analyzed in clinical studies. A relative consistent reduction in ROM in axial rotation with CL-augmentation was reported, while minor and less consistent effects were observed in flexion–extension and lateral bending. The use of CLs was clinical beneficial in C1/2 fusion, while the limited clinical studies on other anatomic regions show no significant benefit for CL-augmentation. While CL provides some additional axial rotation stability in most situations, lateral bending and flexion–extension are less affected. Based on clinical data, CL-augmentation can only be recommended for C1/2 instrumentations, while for other cases, further clinical studies are needed to allow for evidence-based recommendations.


Introduction
Dorsal instrumentation is an effective tool in the surgical treatment of various spinal disorders.It is generally based on a screw-rod construct composed of pedicle or lateral mass screws combined with vertically oriented connecting rods.It has been shown that the rigidity of instrumentations correlates with fusion rate [1] and consequently, high rigidity in all loading directions is aspired.A large plethora of implants like plates, hooks, cerclages, cross-links and interbody devices were developed to be added to a dorsal screwrod construct or to be used as a substitute for certain parts of such constructs.Finally, the configuration chosen must be safe and effective, and therefore, the optimal compromise between construct rigidity, surgical exposure, surgical time and implant cost must be aspired.
Cross-links (CL), connecting the contralateral rods or screw heads, can be added to virtually all configurations of dorsal instrumentation.Despite a large amount of the literature on CL, its application remains controversial in the clinical practice, since the biomechanical benefit must be 1 3 weighed against the additional implant, larger exposure, increased surgical time and implant cost.So far, guidelines on the use of CL-augmentation in dorsal screw-rod constructs are lacking.
The aim of this systematic review was to provide an overview of the mechanical, biomechanical, numerical and clinical effect of CL-augmentation on dorsal screw-rod constructs.This information could help develop recommendations for the use of CLs in specific clinical situations.

Materials and methods
A systematic review in accordance with the PRISMA statement was conducted on PubMed (https ://www.ncbi.nlm.nih.gov/pubme d/) with the search terms "crosslink spine", cross-link spine", "transfixation spine" and "transverse connector spine" on July 2019.Excluding duplicates, 195 studies were identified and evaluated in accordance with the process shown in Fig. 1.Abstracts, case reports and conference presentations as well as studies on instrumentations other than dorsal screw-rod constructs and studies not analyzing the effect of CL were excluded.Studies in languages other than English or German, and studies published before 1990 were excluded as well.To be included into the review, studies had to compare dorsal screw-rod constructs with the same construct with one or more CLs within a mechanical or biomechanical setting, or they had to compare clinical outcome parameters of patients treated with dorsal instrumentations with or without CLs.Studies found through reference lists were evaluated with the same process.The included studies were grouped into mechanical experiments on synthetic bone models, biomechanical experiments on animal cadavers, biomechanical experiments on human cadavers, finite element simulations and in vivo studies with human patients.

Biomechanical studies on synthetic bone models
Four studies analyzed the effect of CL using synthetic bone models [2,[19][20][21] (Table 1).All four studies added one and two hR-R CL to synthetic models instrumented with a dorsal pedicle screw-rod construct.The results of three studies were converted from stiffness values to relative change in ROM (Fig. 3) [19][20][21] A significant decrease in AR-ROM of up to 31% for one and up to 44% for two regular hR-R CL was reported.In FE, ROM was measured to be reduced less than 8% with one and less than 12% with two regular R-R CL.Minimal changes in LB-ROM were reported with one and two regular hR-R CL.In two of the studies, the effect of vertical rod length was investigated [2,20].Both studies demonstrated that ARrigidity decreases with longer construct, while the addition of a hR-R CL had a stronger effect on such longer and more unstable constructs [2,20].Custom built cross-links with very large cross-sectional areas were observed to be more effective in AR and FE compared to regular cross-links [20].
The application of one hR-R CL led to a reduction in AR-ROM ranging from 5 to 59%, while the addition of two hR-R CL reduced AR-ROM between 10 and 30%.In FE, one hR-R CL led to a reduction in ROM from − 1 to 18%, while two hR-R CL reduced ROM from 1 to 6%.In LB, one hR-R CL reduced ROM from 1 to 16%, while two hR-R CL reduced ROM from 8 to 14% (Fig. 4).
Comparing the effectiveness of hR-R CL on single-level to multi-level instrumentations, here, the included animal cadaver studies did not show a consistent difference.The effect of hR-R CL was more pronounced in segmental than in non-segmental instrumentations in LB [24].This trend Fig. 3 The change in range of motion (ROM) due to the addition of one and two horizontal rod-rod cross-links (1 CL) are plotted for axial rotation, flexion-extension and lateral bending for the mechani-cal studies performed on synthetic bone models.Changes reported as statistically significant are marked with an asterisk (*) Fig. 4 The change in range of motion (ROM) due to the addition of one and two horizontal cross-links (1 CL) are plotted for axial rotation, flexion-extension and lateral bending for all biomechanical stud-ies on animal cadavers.Changes reported as statistically significant in the studies are marked with an asterisk (*) was however not seen to the same extent in AR, FE and AC [24,27].
One study compared different CL-configurations: double hR-R CL, double hS-S CL and double dS-S CL.The greatest increase in AR-stability was observed with double dS-S CL, followed by double hS-S CL and double hR-R CL [22].In another study, however, single hS-S CL was observed to be more effective in AR and LB compared to single dS-S CL, while the opposite relation was observed for FE [12].

Occipitocervical
Two studies analyzed the effect of CL-augmentation on dorsal occipitocervical instrumentations [13,14].In both studies, the effect of one hR-R CL on the ROM was reported to be non-significant.Specific values were however not provided in the manuscripts.

Cervical
Of the four biomechanical studies on human cervical spines, two analyzed the effect of CL on the atlantoaxial region with one study using C1 and C3 lateral mass screws [15] and the other using a special implant configuration at C1-C2 [28].In both studies, the addition of one hR-R CL reduced ROM in AR in the presence of an odontoidectomy significantly (−54% and −37%, respectively).The effect of CL in the intact situation (without odontoidectomy) was not quantified in the first and described as non-significant in the second study (− 15%).LB-ROM was reduced significantly in the second study only in the odontoidectomy situation (− 43%), while the effect was non-significant in the first study (− 13%) and in the intact situation of the second study (−24%).The effects in FE were non-significant in both studies with values below − 21%.
The effect of different CL locations and types of CL on the ROM of C3-C7 lateral mass screw instrumentations was analyzed in another study [17].Without laminectomy, no CL-configuration changed the ROM in a statistically significant way.After laminectomy, however, the addition of two CL in three of the tested configurations led to a significant reduction of ROM in AR.The two types of CL used in this study (hS-S CL, hR-R CL) showed no statistically significant difference in effectiveness [17].
Analyzing the cervical region with C3-C6 lateral mass screws and pedicle screws at C7, a significant reduction in AR-ROM with one (− 29%) and two hS-S CL (− 44%) as well as a significant reduction of LB-ROM with two hS-S CL (− 13%) was observed in both, the intact and the destabilized situation [18].The relative effect of the CL remained relatively constant with an increasing extent of decompression and facetectomy [18].

Cervicothoracic
In dorsal instrumentations of C6-T2, the addition of 2 h-R CL led to a significant reduction in AR-ROM in the 3-column injury model at C7/T1 (− 27%).A similar trend was observed in the 2-column-injury model at C7/T1.Thoracic hR-R CL showed a trend of being more effective than the cervical hR-R CL [4].

Thoracic
In thoracic instrumentations of T4-T10, the addition of one and two hR-R CL reduced ROM significantly (average 21% for one and 34% for two CL) [5].In LB and FE, the addition of one and two hR-R CL did not reduce the ROM significantly [5].Analogously, another study analyzing the same anatomical region with an additional pedicle subtraction osteotomy at T7 found a reduction of AR-ROM with one (26%, ns.) and two hR-R CL (48%, *), while no relevant effect was observed in FE and LB [16].

Thoracolumbar
The thoracolumbar region has been investigated in four biomechanical studies performed on human cadaveric spines [6][7][8]26].In testing of embalmed specimens, completely destabilized at L1 and instrumented at T12 and L2, the addition of one and two CL decreased ROM in AR up to 38% and in LB up to 89%, whereas only a subgroup of the differences were statistically significant [26].In another study on specimens destabilized and dorsally instrumented at T12/L1, the addition of one hR-R CL reduced ROM significantly in AR (21%), while LB and FE were not affected to the same extent [7].In a study on cadavers with corpectomy and spacer instrumentation at T12, the addition of one hR-R CL to two different types of dorsal instrumentations led to a reduction in AR-ROM of 44% and 51%, while the effect in LB and FE was below 18% and 12%, respectively [6].ROM after T12 corpectomy and instrumentations at T11 and L1 was significantly reduced in AR by the addition of one and two CL (16% and 31%, respectively), while the effects in FE and LB were non-significant [8].

Lumbar
The biomechanical effects on the lumbar spine have been investigated by two groups [9,10].In destabilized specimens at L3/4 and L4/5, the addition of one hR-R CL to a dorsal instrumentation at L3 and L5 reduced ROM in AR significantly (21%) only with CoCr vertical rods, while the effect with Titanium rods was similar (18%), but statistically non-significant.The non-significant effects in FE and LB were below 1% and 4%, respectively [9].In specimens with corpectomy L3 and dorsal instrumentation at L2 and L4, the change in ROM due to one hR-R CL was not altered significantly in AR, LB and FE, while the observed effect was most prominent in LB with 16% [10].
One study focused on the thoracolumbar region [29], four studies on the lumbar region [11,12,30,31], and one study evaluated a long instrumentation from T1-sacrum [32].The values reported for change in ROM are illustrated in Fig. 6.
One study analyzed the stiffness of constructs with one hR-R CL, one dR-R-CL and two dR-R CL (X-shaped) effects of CL on longer constructs, while the largest effect was seen in AR and LB when simulating the augmentation with an X-shaped CL configuration.[29].
Bone loads (von Mises) were observed to decrease with the use of a CL [12,29].Similarly, loads (von Mises) on the pedicle screw neck decreased [29].This is opposed by another study in which a general increase in screw loads (strain energy) in AR and LB was observed [12].Peak loads (peak von Mises) on the vertical rod were seen to increase in one study [31], while another study calculated better stress dispersion on the rod in situations with CL-implantation [32].
Regarding optimal CL-positions, larger effects on the rigidity were computed for placement in the proximal part of the construct [11,33] and at an osteotomy-site [33].

In vivo studies
Six in vivo studies on human patients were identified [33][34][35][36][37][38] (Table 5).Two studies were performed on atlantoaxial instrumentations and found earlier bony fusion in the hS-S CL group [34] and in the hR-R CL group [33] compared to the no-CL groups.Furthermore, favorable neurological scores were observed in the hR-R CL group compared to the no-CL group [33].
In the thoracolumbar region, two studies on patients with adolescent idiopathic scoliosis treated with dorsal instrumentations found no clinical, radiologic or surgical differences for the groups with and without hR-R CL [36,37].Another study suggests to omit CL-augmentation since no rotational instability was observed without the use of CL [38].Comparing the effectiveness of different types of CL, lower reoperation rates were observed in closed drop entry CL compared to other types of CL [35].

Discussion
The stability of dorsal instrumentation is of great importance for bony fusion and consequently for a good clinical outcome [1].To achieve the required stiffness, a plethora of augmenting components like interbody cages, pedicle hooks, cerclages, end-to-end and side-to-side connectors as well as CLs have been proposed for the use in certain clinical settings.For the application of CLs, no clear recommendations or guidelines exist, and the use of these components is based on the subjective evaluation of the surgeon.The addition of elements to a biomechanical construct can be argued to generally increase stability, while also cost, surgical exposure, surgical time and additional foreign body implantation have to be considered and weighed against the beneficial effects.The aim of this systematic review was to create an overview of the effect of CL-augmentation to contribute to a more informed decision on whether or not to include CL in dorsal instrumentations.

Synthetic models
Synthetic models, having the advantage of eliminating anatomical variability, show a clear effect of horizontal R-R CL with up to 30% reduction in ROM in axial rotation.The effects in FE and LB were found to be much smaller with values below 12% and 3%, respectively.

Animal cadaver studies
Animal cadaver studies show very similar results to the synthetic models with the largest effect of CL-augmentation in AR.While these studies integrate factors like the screwbone interface, morphology and bone density can vary from humans.

Human cadaver studies
The biomechanical studies on the different parts of the human spine showed similar results with the largest effects of CL-augmentation being observed in AR, while only minimal effects were recorded in FE and more irregular results were observed in LB.

Numeric simulations
The general stress in the vertebral bodies and the mean stress in the instrumentations were observed to be reduced by CLaugmentation.However, peak loads and stress accumulation in the vertical rods were increased, which might be problematic in situations where material failure poses a problem.Several generally applicable biomechanical observations can be stated: • CL-augmentation adds stability to the construct mainly in AR. • Two CLs are more effective than a single CL [2,4,5,8,11,16,17,[19][20][21]26].• A CL with a larger cross section is more effective than a smaller one [19,20].• Diagonal CL shows higher effectiveness in FE and LB, while horizontal CL are more effective in AR [12,29].
Based on the mechanical, biomechanical and numerical studies, we evaluate the effectiveness of CL to be based on the prevention of shear deformation of the screw-rod construct, also described as the "windshield wiper" effect.As illustrated in Fig. 7, such deformation primarily occurs during AR, less during LB and only minimally during FE.In conclusion, the largest effect of CL-augmentation can be expected in AR, less in LB and only minimal effects in FE, which agrees with the literature.
Although in vitro experiments are crucial to understand and predict the biomechanical effect of CL-augmentation, this information is not sufficient to predict the effect on clinical outcome.To answer this question, in vivo studies are indispensable, however, only a limited number of studies is available.In the setting of C1/2-instrumentation, CL-augmentation has been observed to result in earlier bony fusion, superior clinical outcome and in no additional complications.The reduced number of clinical and radiologic follow-up controls and the improved patient satisfaction was stated to possibly compensate for the additional cost of about 1000 dollars per CL [33,34].In retrospective analysis, CL-augmented instrumentations in idiopathic scoliosis patients were not associated with a better clinical or radiologic outcome [36,37].While biomechanically, CLs provide some benefits in such constructs, CL-augmentation was also observed to be associated with soft tissue irritation and delayed infection [40] and with an increased risk of pseudarthrosis due to the reduced bone graft volume at the CL position [41].
In conclusion, based on clinical data, the general use of CL-augmentation can only be recommended for C1/2 instrumentations.In the surgical treatment of idiopathic scoliosis, CL-augmentation cannot be recommended as a standard measure.Based on biomechanical studies, an increase in AR-stability can be expected for many other clinical situations, however, no clinical data is available.The increase in AR-stability could be especially beneficial in situation of relevant AR-instability (e.g., after bilateral facetectomy), however, due to the fact that CL-augmentations provides potential drawbacks, such as the risk of irritation, delayed infection, pseudoarthrosis, increased costs, surgery time and exposure, clear clinical evidence is required to justify the usage of CL.Therefore, further clinical research on this topic is strongly suggested to further assess the benefit of CL in distinct clinical situations and to legitimate its usage.

Fig. 5
Fig. 5 The change in range of motion (ROM) due to the addition of one and two horizontal rod-rod cross-links (1 CL) are plotted for axial rotation, flexion-extension and lateral bending for all biome-

Fig. 6
Fig.6 The change in range of motion (ROM) due to the addition of one, two and three horizontal cross-links (Nr.CL) are plotted for axial rotation, flexion-extension and lateral bending for the finite element simulations

Table 1
(5)mechanical studies on synthetic bone models (*) Difference described as statistically significant, (n.s.) difference described as non-significant, (1) values converted from stiffness to range of motion, (2) values for different CL-configurations Difference described as statistically significant, (n.s.) difference described as non-significant, (1) values converted from stiffness to range of motion, (2) values for different CL-configurations, (3) values for different instrumentations, (4) values for flexion and extension combined to FE(5)values for regular CL-placement and placement through spinal process, respectively, (6) calculated difference in ROM in neutral zone

Table 4
Biomechanical studies with numerical simulations (*) Difference described as statistically significant, (1) values for different CL-configurations, (2) values for different instrumentations, (3) values for flexion and extension combined to FE, (4) values for regular and weak bone, respectively

Table 5
In vivo studies