1 Introduction
Scapholunate interosseous ligament (SLIL) injury is a relatively common [
1,
2] wrist ligament condition which if not treated successfully may lead to carpal instability and degenerative osteoarthrosis [
3]. SLIL injury occurs most frequently with the wrist positioned in extension, ulnar deviation and carpal supination. Treatment of scapholunate instability depends upon the severity of the injury which can vary widely [
4]. For subjects presenting with dynamic scapholunate or static reducible instabilities, ligamentous reconstruction is a consideration [
5].
In flexion-extension of the wrist, the lunate rotates over the radius in the dorsal direction during flexion and in the volar direction during extension [
6]; therefore, reduction and stabilisation of the dorsal gap is important in flexion, whereas for extension the volar gap is significant. Until recently, SLIL reconstruction techniques, including the Brunelli tenodesis method and derivations [
7‐
9], have concentrated on reconstructing only the dorsal portion of the SLIL; thus, volar opening and sagittal plan rotation remains a potential complication, leading to altered kinematics [
3,
5].
More recently, techniques including Corella [
5] and scapholunate axis (SLAM) [
3] involving either a multiplanar scaphoid-lunate tether or volar reconstruction in addition to dorsal have been proposed in order to overcome this. Although preliminary studies suggest that multiple junction point techniques are better able to correct SL gap and angle compared to conventional techniques, further data, analysis and long-term follow-up studies are required to confirm this [
3,
10,
11].
In this study, the finite element method together with in vitro cadaveric tests were used to investigate the performance of the modified Brunelli tenodesis (MBT), Corella and SLAM reconstruction methods in regard to their ability to restore wrist stability [
6] following a simulated complete tear of the SLIL. A total of 30 3-D finite element models were created for the investigation. Neutral and ulnar-deviated clenched fist wrist positions were used to validate the models. In the latter position, virtual surgery of the MBT SLAM and Corella was performed in addition to the SLIL sectioning and non-sectioning (intact ligament) scenarios. For the neutral position, the intact (ligament) was only considered. The validation of the models was carried out through a comparison of the predicted SL dorsal gap and angle against the results obtained from the in vitro cadaveric tests.
Once the models had been validated, an investigation of the performance of the three reconstruction methods (MBT, SLAM and Corella), the intact (ligament) and the SLIL sectioning cases was undertaken with the wrist positioned at 20° flexion and 20° extension, at 15° ulnar deviation and 15° radial deviation. The predicted values of SL angle and SL gap at both dorsal and volar sides obtained from these models were used for comparison purposes between the reconstruction techniques.
The finite element method is widely employed for undertaking analyses in biomechanics offering a number of well-documented advantages compared to cadaveric studies including repeatability of analyses, ease of study parameter modification and lack of associated ethical issues. A particular advantage in utilising the finite element method in our study was that it facilitated calculation and comparison of both dorsal and volar angles for all wrist positions analysed, which is not currently feasible with the radiograph-based techniques currently employed for cadaveric/clinical studies.
4 Discussion
A variety of treatments currently exist for treating chronic SL instability. Ligamentous reconstruction techniques including capsulodesis, bone-ligament-bone and tenodesis are an option where patients present with non-repairable SLIL injury but a reducible SL dissociation [
5]. Until recently, tenodesis procedures have concentrated on reconstructing the dorsal component of SLIL thus volar opening and sagittal plan rotation leading to altered kinematics remains a potential complication [
3,
5]. More recently, techniques including Corella and SLAM involving either a multiplanar scaphoid-lunate tether or volar reconstruction in addition to dorsal have been proposed in order to overcome this.
Using a finite element model and cadaveric study data, we investigated the performance of the Corella, SLAM and MBT techniques.
SL gap and angle predictions from our model were in good agreement with those from the cadaveric study for the six scenarios considered, including the three ligament reconstruction techniques MBT, SLAM and Corella. SL gap predictions were all within 0.4 mm and SL angles within 7.7° of the corresponding experimental mean data values. The cadaveric data and predictions from our model showed that SLIL sectioning had a much greater effect on SL gap compared to angle, with experimentally obtained mean SL gap increasing by 50% or 1.2 mm compared to the intact ligament case for the ulnar-deviated clench fist position. In this case, SL gap values met the criteria for SL dissociation [
4] and were in good agreement with another study that measured SL gap following ligamentous sectioning in cadaver wrists loaded to produce an ulna-deviated clench fist posture [
23]. In contrast, the effect of sectioning on SL angle was relatively minor, causing a change of no more 6%, with SL angle remaining within the reportedly normal range, 30° to 60° [
25]. These results concur with those of other researchers who determined that solely dividing the SLIL does not have a significant effect on the rotational motion of the scaphoid and lunate for radial-ulnar deviation [
26].
For the ulnar-deviated clenched fist posture investigated, the tendon reconstruction techniques reduced dorsal SL gap to within 10% of the values obtained for the original intact loaded ligament and volar SL gap to within 26% in all cases whilst maintaining SL angle in the normal range, demonstrating the techniques’ abilities to restore dorsal SL gap following non-repairable SLIL injury. Of the three tendon reconstruction techniques, Corella was more effective in restoring SL gap, restoring dorsal gap back to the original intact loaded value and volar gap to within 10.5%.
In flexion-extension, the lunate rotates over the radius “in the dorsal direction” during flexion and “in the volar direction” during extension [
6]; therefore, for any tendon reconstruction technique, reduction and stabilisation of the dorsal gap is of significant importance in flexion, whereas for extension the volar gap is particularly important. Our model demonstrated that all three reconstruction techniques, Corella, MBT and SLAM, were able to restore dorsal SL gap to within 0.4 mm of the intact ligament during flexion and 0.1 mm during extension following simulated SLIL sectioning. This is as expected, as all three techniques involve reconstruction of the dorsal portion of the SLIL. Of the three techniques, Corella was able to restore dorsal SL gap and angle closer to that of the intact ligament, followed by SLAM then MBT. However, greater variation was found between the techniques in terms of their ability to restore volar SL gap, with the techniques involving either a multiplanar scaphoid-lunate tether (SLAM) or reconstruction of the volar portion of the SLIL in addition to the dorsal (Corella), performing better. In flexion and extension, the Corella technique was able to restore volar SL gap to the same as that for the intact ligament. In flexion, SLAM restored dorsal SL gap to within 5.8% of the intact ligament, but fared less well in extension, where volar opening is more significant, only being able to restore volar SL gap to within 16% of the intact. The MBT technique, which reconstructs just the dorsal portion of the SLIL, was not able to reduce volar SL gap at all compared to the SLIL-sectioned case in extension, and only by 11% in flexion.
The predictions from our FE models indicated that for radial and ulnar deviation, all three reconstruction techniques simulated were able to restore dorsal SL gap to within 0.3 mm and SL angle to within 1.8° of the intact ligament following SLIL sectioning. Again, this is not unexpected as all techniques involve dorsal SLIL portion reconstruction. However, more variation was found in the ability of the techniques to restore dorsal SL gap. The Corella technique, which involves reconstruction of the volar portion of the SLIL, restored volar SL gap back to that of the intact ligament for both radial and ulnar deviation. SLAM, which involves a multiplanar scaphoid-lunate tether, was able to restore volar SL gap to within 4.5% of the intact ligament in radial deviation and 21% in ulnar deviation, whereas the corresponding values for MBT, which reconstructs only the dorsal portion of the SLIL, were 9% and 32%, respectively.
The relative significance of the various portions of the SLIL is still undecided, with conflicting data available in the literature [
27]. However, the results from our study indicate that unless ligamentous reconstruction techniques involve multiple junction points between scaphoid and lunate, volar gap widening and sagittal plane rotation is likely to occur which may consequently lead to altered kinematics. Of the three reconstruction techniques considered, overall, we found Corella was better able to restore both dorsal and volar SL gap and SL angle following SLIL injury; however, further analysis and long-term clinical follow-up studies are required to confirm outcomes and evaluate potential creep and elongation with the reconstruction.
Limitations and assumptions apply to our study which are typical of complex numerical analyses in the field of biomechanics. A number of assumptions and simplifications were inevitably required including geometrical representations and material properties. In terms of soft tissue representations, a hyper-elastic material model was employed for cartilage which is considered to provide a more accurate representation of behaviour [
12,
13].
The majority of the ligaments included in the model were represented using spring elements. Whilst the use of one-dimensional representations of ligament geometries is commonplace in biomechanical joint models and has been shown to be valuable particularly for investigating kinematics where external loading is present, a number of limitations have been identified, including the inability to accurately capture non-uniform 3-D stress and strain, non-uniform deformations and joint orientation effects [
28]. Three-dimensional FE modelling approaches have been highlighted as being required for more accurate ligament representation; however, it is recognised that this is not straightforward and can be massively time-consuming [
28]. Linear elastic material properties were employed for the ligaments. In reality, ligaments typically exhibit non-linear viscoelastic behaviour so if ligament strain was low, then behaviour would fall within the non-linear region and ligament stiffness would be overestimated as a result which would affect joint motion prediction. However, accurate data for the large number of parameters required to describe non-linear viscoelastic ligament behaviour is not readily available [
28]. Furthermore, it has been determined that ligaments tend to operate at or close to the linear region, so an assumption of linear elastic behaviour should not introduce significant error [
29].
We validated our finite element wrist model by comparing predicted SL gap and SL angle with data obtained from an in vitro cadaveric study conducted on intact, SLIL-sectioning specimens and cadaveric wrists following simulation of the three reconstruction techniques with the hand in the neutral position and under ulnar-deviated clenched fist posture. The good agreement between predicted and experimentally obtained SL gap and angle data suggests that our model is able to model and represent behaviour to a good degree of accuracy and the assumptions used in the model do not introduce significant error.
We used CT scans from the wrist of a single volunteer for creating our FE models; therefore, caution should be taken from drawing extensive conclusions from the results. That said, model predictions compared well with mean results from simulation of the three reconstruction techniques undertaken on 15 cadaveric wrists; therefore, our wrist model appears to be a good representation and a certain degree of confidence can be placed in the results from it.
Although SLIL sectioning only partly replicates scapholunate instability, it is recognised that the SLIL is the primary stabiliser of the joint [
30]. To better simulate scapholunate instability, our model could be revised to take into account stretching of the supporting ligaments, in particular the radioscaphocapitate and dorsal intercarpal ligaments; however, the soft tissue reconstruction techniques we model are generally employed for early chronic or subacute SLIL disruptions before secondary constraints have been excessively compromised and irreducible SL subluxation has occurred [
3].
Fixation of tendon reconstructions to bone is treated similarly in every reconstructive option we modelled [
20], which may not be the clinical scenario. Our models focus on the relative position of the bones after the reconstruction scenarios. The simulations assume that the tendon graft is attached to a point which would not change if it is a suture, screw or tunnel anchor. Suture to a soft tissue could influence the results in that the tissue has some flexibility which allows relative motion of the tendon graft when it starts to deform.
SL gap and angle measurement of the cadaveric wrist specimens was based on posteroanterior (SL) and lateral (angle) plain radiographs and therefore any rotational error in X-ray positioning could potentially affect accuracy.