Experimental Setup
Fourteen fresh-frozen, paired cadaveric feet from donors who had been sixty-three to ninety-three years old (median, 80.5 years old) at the time of death were used in the present study. The specimens were obtained from three female and five male donors; one pair of specimens was formed with opposing feet from two male donors. Each specimen was truncated by severing the tibia and fibula at a height of approximately 30 cm. The tibial medullary canal was filled with polymethylmethacrylate, and a 0.25-in (0.64-cm) threaded rod was driven through the canal and the talus and into the calcaneus in order to fix the ankle at approximately 30° of plantar flexion with the subtalar joint in neutral and to provide a mounting point for loading (Fig. 1).
The dorsal skin above the first and second cuneiforms and the proximal metatarsals was removed, and a triad of registration screws comprising two 1-in (2.54-cm) number-1 and one 3/16-in (0.48-cm) number-0 Phillips-head machine screws was inserted into three predrilled holes in the first cuneiform and three predrilled holes in the proximal second metatarsal in order to create approximately equilateral triangles between the screw heads while allowing space for the subsequent sectioning of the Lisfranc ligament. The 3/16-in (0.48-cm) screws were fixed flush to the bone surface, whereas the 1-in (2.54-cm) screws were partially inserted in order to allow most of their length to extend above the surface and to create larger triads for better positional accuracy. These triads formed the basis for tracking the three-dimensional rigid body positions of the associated bones.
Fluoroscopy was used to verify that the registration screws did not interfere with any articulating surfaces. The specimen was then fixed in the loading frame with the distal metatarsals resting on a wooden block padded by a ~1/4-in (0.64-cm)-thick rubber pad to minimize slippage during loading. All measurements were made with a simulated weight-bearing load of 35 kg (343 N) applied through the tibia as described by Coss et al.17. Three minutes of viscoelastic dissipation was allowed following the application of load prior to the recording of the positions of the triad screw heads.
A digitizer (MicroScribe-3D Digitizer; Immersion, San Jose, California) with a manufacturer-specified accuracy of 0.23 mm was utilized to record the three-dimensional positions of the heads of the triad registration screws in each condition. Digitizer calibration was validated prior to the recording of data for each specimen by digitizing the vertices of a reference triad in two widely separated positions within the workspace and ensuring that the edge lengths remained constant and within the stated accuracy. All digitizer measurements were made with the head of the digitizer parallel to the shaft of the screw to ensure position consistency. Original software for the three-dimensional reconstruction18 and analysis of absolute orientation19 facilitated the recording of data, the mapping of ligament attachment sites, and the determination of three-dimensional displacement between the attachment sites from the registration triads fixed to each bone.
Experimental Procedures
The intact condition was recorded, and then isolated sectioning of the Lisfranc ligament was achieved from the dorsal aspect of the foot. A number-11 blade was inserted deep between the bases of the first and second metatarsals, and then all intervening structures were incised proximally up to the middle of the intercuneiform joint between the medial and middle cuneiform bones, with care being taken to keep the remaining dorsal portion of the proximal intercuneiform ligament intact. Subsequently, the blade was reversed and the same cut was retraced to ensure that no attachments remained in the deep plantar layers.
Stabilization of the Lisfranc joint proceeded as follows. First a 5-mm incision was made over the medial border of the foot, overlying the medial cuneiform. Another 5-mm incision was made over the dorsum of the midfoot, at the level of the lateral border and approximately 1 cm distal to the base of the second metatarsal. A guidewire was inserted from the base of the second metatarsal into the medial cuneiform, and the position was checked under fluoroscopy. A cannulated 3.0-mm drill was passed over the guidewire. This drilled hole was used for subsequent alternating internal fixation with either the suture button or the cannulated screw.
Fixation with use of the suture button (TightRope repair kit; Arthrex, Naples, Florida) was performed in the following sequence (Fig. 2). The 1.6-mm guidewire with pull-through suture was passed into the drill-hole in a medial-to-lateral direction. The leading button was flipped on exiting the hole and was arranged so that it lay on the lateral aspect and just distal to the base of the second metatarsal. The pull-through suture was cut and removed. A bone clamp was used to reduce any displacement and to hold the medial cuneiform reduced to the base of the second metatarsal. The trailing medial button was then tightened down over the medial cuneiform by pulling on the free ends of the button suture and was secured with a knot. Screw fixation was performed with use of a 3.5-mm-diameter cannulated lag screw of appropriate length, with the screw being inserted over the guidewire into the predrilled hole in a lateral-to-medial direction.
Loading and Evaluations
The positions of the triad of screws on each foot were recorded in the following conditions, with the specimen unloaded and loaded between each condition. The intact condition was initially recorded. The Lisfranc ligament was sectioned, and the cut condition was recorded. One foot from each pair was randomly selected and was fixed first with the suture button; the fixation was then removed, and the foot was fixed with the cannulated screw. Each condition was recorded. Subsequently, the contralateral foot of each pair was subjected to the same procedure; however, the sequence of the types of fixation was reversed.
On completion of the experimental loading procedure, the first and second cuneiforms and the first and second metatarsals were removed as a single block and were evaluated to verify that the Lisfranc ligament had been successfully transected. The Lisfranc ligament attachment sites were identified, and the digitizer was used to record the positions of the centers of the attachment sites with reference to the registration triads. Displacements in each condition were determined as the three-dimensional distances between the centers of the attachment sites.
Hypotheses and Statistical Methods
Our primary hypothesis was that suture-button fixation provides stability comparable with that provided by the Lisfranc ligament or a screw. Our additional hypotheses were that significant displacement is created by transection of the Lisfranc ligament; that there is no significant difference between fixation of the left foot as opposed to the right foot; and that the sequence of fixation methods used to stabilize the Lisfranc joint (screw before suture button, or the reverse) does not significantly affect stability after fixation. The displacement in the intact condition was used as the reference displacement, and displacements for all other conditions were normalized as the difference between the displacement in each condition and the displacement in the intact condition.
The minimum sample size needed to achieve the desired level of significance (alpha = 0.05) was determined through an a priori power analysis on the basis of the means and standard deviations of the displacements from a pilot study. Two-way analysis of variance with use of the general linear model in SAS 9 (SAS Institute, Cary, North Carolina) was used to determine the significance (alpha = 0.05) of displacement differences. The 95% confidence intervals for the means were determined as the mean plus or minus twice the standard error of the mean.
Source of Funding
Funding for the study was provided to the department by Arthrex, Naples, Florida, and was used to acquire materials. The funding source did not play any other role or influence the investigation. None of the authors or their families received any monies personally.
The objective of the present study was to compare the stability provided by a suture button with that provided by a screw when used to stabilize the diastasis associated with a Lisfranc ligament injury with use of a three-dimensional analysis technique and a minimally disruptive model21. Our results demonstrated that, in these cadaveric specimens with a simulated static weight-bearing load, reduction and stabilization of the Lisfranc joint with the suture button achieved fixation equivalent to that achieved with a cannulated screw following isolated transection of the Lisfranc ligament. Additionally, our results verified that a significant displacement was created by severing the Lisfranc ligament, that there was no significant difference between the fixation of a left as opposed to a right foot, and that the fixation sequence yielded a significant difference in the resulting displacement.
Some previous Lisfranc injury models have utilized 100 N for biomechanical testing in a whole-foot setting5,22. Failure loads for the dorsal ligament and for the Lisfranc ligament have been reported, with applied axial tension in a bone-ligament-bone fashion, as 150.7 N (34 lb) and 368.8 N (83 lb), respectively23. Considering that one leg can easily bear the full body weight in static loading, we adopted a simulated weight-bearing value of 35 kg (343 N) as our maximum load17 as a compromise between compatibility with previous studies and the practical limitations of achieving ankle fixation under a larger load in the absence of physiological muscle loading. At the ankle joint, 30° of plantar flexion was chosen because loading the forefoot with the ankle in this position was found in our pilot study to most reliably produce displacement at the tarsometatarsal joints after a transection that is limited to the Lisfranc ligament and because it approximates the ankle position at the end of the stance phase of the gait cycle.
The present study had some limitations. The testing conditions were not physiological as this was a cadaveric study and the specimens were not subject to the physiological forces generated by the musculotendinous units and modulated by the ligament restraints. However, standardization of the test conditions allowed for comparison between the screw and the suture button in an identical situation. The fixation provided by the suture button depends on the knot-tying technique. The use of a bone-holding clamp to hold the reduction between the medial cuneiform and the base of the second metatarsal before tying the knot facilitates tensioning and securely tying the knot holding the reduction. To limit variability in the knots, the same investigator tied the knots with use of four locking throws and the same technique was used in all of the specimens. In vivo, the suture may become slack because of erosion or tissue remodeling due to the constant pressure underneath the buttons at both ends of the suture. Also, creep of the suture-button fixation under constant or cyclic loading during weight-bearing and walking may occur. However, in current clinical practice, weight-bearing is allowed only after a period of immobilization and limitation of activities for about eight to twelve weeks. This period of immobilization may be sufficient for the Lisfranc ligament to heal, in which case creep of the implanted suture may be irrelevant. The present study focused only on the initial fixation condition and was unable to address the issues of creep under prolonged or cyclic loading.
In the clinical setting, if screws were used to reduce the diastasis after an isolated Lisfranc ligament injury, the diastasis could recur after removal of the screws if the Lisfranc ligament has not healed adequately. Suture-button fixation may have a place in such a clinical setting or in an acute setting as a primary made of fixation24. Therefore, in our model, we compared the different sequences of fixation methods and found a small but significant difference in the resulting displacement. We speculate that this difference was due to permanent distortion of the bones following screw fixation. However, the results of this comparison should be viewed with caution because of the limiting factor of the inability to reproduce the effect of biological healing and other factors inherent in a cadaveric model as well as the lack of prolonged or cyclic loading.
The suture button is far more expensive than the screw is. Although the cost varies among institutions and regions and differs every time contracts are negotiated, at the authors' institution it was about twenty times more expensive. Additional research and clinical studies evaluating suture-button fixation compared with currently accepted methods for fixation at the site of Lisfranc injuries are required to determine whether the advantages of the suture-button approach outweigh the greater costs.
We conclude that fixation with a suture button may be an acceptable alternative to screw fixation in the treatment of isolated ligament injuries, providing initial reduction of diastasis equivalent to that provided by a screw. A suture provides a nonrigid anatomical reduction that may be more functionally similar to the natural Lisfranc ligament than a rigid screw is. It may continue to function in place of the Lisfranc ligament if the ligament fails to heal. In addition, subsequent surgery to remove hardware before weight-bearing, as well as the problems associated with screw breakage, can be avoided. 