Abstract
Background: The objective of the present study was to assess the utility of magnetic resonance imaging for the diagnosis of an injury to the Lisfranc and adjacent ligaments and to determine whether conventional magnetic resonance imaging is a reliable diagnostic tool, with manual stress radiographic evaluation with the patient under anesthesia and surgical findings being used as a reference standard.
Methods: Magnetic resonance images of twenty-one feet in twenty patients (ten women and ten men with a mean age of 33.6 years [range, twenty to fifty-six years]) were evaluated with regard to the integrity of the dorsal and plantar bundles of the Lisfranc ligament, the plantar tarsal-metatarsal ligaments, and the medial-middle cuneiform ligament. Furthermore, the presence of fluid along the first metatarsal base and the presence of fractures also were evaluated. Radiographic observations were compared with intraoperative findings with respect to the stability of the Lisfranc joint, and logistic regression was used to find the best predictors of Lisfranc joint instability.
Results: Intraoperatively, seventeen unstable and four stable Lisfranc joints were identified. The strongest predictor of instability was disruption of the plantar ligament between the first cuneiform and the bases of the second and third metatarsals (the pC1-M2M3 ligament), with a sensitivity, specificity, and positive predictive value of 94%, 75%, and 94%, respectively. Nineteen (90%) of the twenty-one Lisfranc joint complexes were correctly classified on magnetic resonance imaging; in one case an intraoperatively stable Lisfranc joint complex was interpreted as unstable on magnetic resonance imaging, and in another case an intraoperatively unstable Lisfranc joint complex was interpreted as stable on magnetic resonance imaging. The majority (eighteen) of the twenty-one feet demonstrated disruption of the second plantar tarsal-metatarsal ligament, which had little clinical correlation with instability.
Conclusions: Magnetic resonance imaging is accurate for detecting traumatic injury of the Lisfranc ligament and for predicting Lisfranc joint complex instability when the plantar Lisfranc ligament bundle is used as a predictor. Rupture or grade-2 sprain of the plantar ligament between the first cuneiform and the bases of the second and third metatarsals is highly suggestive of an unstable midfoot, for which surgical stabilization has been recommended. The appearance of a normal ligament is suggestive of a stable midfoot, and documentation of its integrity may obviate the need for a manual stress radiographic evaluation under anesthesia for a patient with equivocal clinical and radiographic examinations.
Level of Evidence: Diagnostic Level II. See Instructions to Authors for a complete description of levels of evidence.
Inadequate treatment of low-energy injuries of the Lisfranc articulation can result in substantial disability, deformity, and dysfunction1. These injuries often occur in young active individuals who are involved in sports activities, and they are frequently missed at the time of the initial presentation1-3. The mechanism of injury usually involves an axial load that is applied to the heel of a plantar flexed ankle (with the toes extended), resulting in a tensile force to the convexity of the tarsometatarsal articulations4. This forces the metatarsal bases dorsally and can disrupt the supporting ligaments, resulting in midfoot instability. Alternately, a twisting or bending force across the midfoot, with the foot in a fixed position and the individual's body rolling over the midfoot, can result in disruption of the supporting ligaments comprising the Lisfranc complex. The imaging definition of instability is diastasis of =2 mm between the second metatarsal base and the medial cuneiform on conventional plain radiographs5,6. Unfortunately, in cases of lower-energy injuries, the diastasis may not be evident unless an active stress is applied to the ligaments. This can be achieved with use of weight-bearing radiographs1, but this type of provocative examination is often technically limited by the pain experienced by the patient during weight-bearing and consequently may yield a false-negative result. The traditional alternative has been manual stress radiographs made with the patient under anesthesia1,7, which introduces an element of additional morbidity. Computerized tomography has been used as an alternative8, but while this modality is exquisitely sensitive for the assessment of fracture, it remains limited for the assessment of instability, and, without weight-bearing stress or clear diastasis, the status of joint stability can remain unclear. More recently, the role of magnetic resonance imaging in the detection of Lisfranc ligament injury has been described9. Previous studies have focused on the magnetic resonance imaging appearance of both normal and torn Lisfranc ligaments; however, they have not correlated results with surgical findings and have not included an analysis of the osseous and soft-tissue injuries that are often present10,11.
A recent biomechanical study in which cadaveric specimens underwent stress evaluation defined the patterns of ligament disruption that result in reproducible instability of the midfoot12. When these patterns have been seen, surgical stabilization has been recommended to prevent long-term disability13,14.
The objective of the present study was to assess the utility of magnetic resonance imaging in diagnosing injury of the Lisfranc and adjacent ligaments and to determine whether conventional noninvasive magnetic resonance imaging is a reliable tool for diagnosing Lisfranc joint complex instability, with manual stress radiographic evaluation with the patient under anesthesia and surgical findings being used as a reference standard.
The present study was undertaken after approval was obtained from the institutional review board of the Thomas Jefferson University Hospital, which allowed for all portions of the present retrospective review without obtaining retroactive informed consent.
Patients who presented for an orthopaedic foot and ankle evaluation between 2002 and 2007 with a mechanism of injury and clinical findings that were suggestive of a Lisfranc ligament injury and with equivocal findings on plain radiographs underwent magnetic resonance imaging of the foot. The present study included twenty patients (ten women and ten men) with a mean age of 33.6 years (range, twenty to fifty-six years). All patients had subsequently undergone manual stress radiographic studies under anesthesia and, when indicated, concomitant surgical stabilization of the midfoot. The magnetic resonance imaging findings were cross-referenced with the surgical reports and the findings of the examination with the patient under anesthesia. All of the examinations were performed by one orthopaedic surgeon (S.M.R.).
Anatomy
The articulations between the distal row of tarsal bones (the medial, middle, and lateral cuneiforms and the cuboid) and the second through fifth metatarsal bones are referred to as the Lisfranc joint. These individual joints are arranged in an arch configuration, which is maintained by the wedge-shaped configuration of the metatarsal bases and their associated ligaments. The second tarsal-metatarsal joint (or first Lisfranc joint) is recessed, which provides additional stability to the transverse arch of the midfoot, sometimes referred to as the "keystone" concept3,15.
Additional stabilization is provided by the complex of reinforcing ligaments across the Lisfranc joint complex. These are divided into dorsal and plantar ligaments, with the plantar ligaments being the primary stabilizing ligaments of the midfoot. Strong intermetatarsal ligaments connect the bases of the second, third, fourth, and fifth metatarsals but are absent between the first and second metatarsal bases. The base of the second metatarsal is primarily stabilized by the ligaments between the medial cuneiform and the bases of the second and third metatarsals, with the plantar ligament (pC1-M2M3) being significantly stronger than its corresponding dorsal counterpart (dC1-M2)16,17 (Figs. 1-A and 1-B). Disruption of the pC1-M2M3 ligament results in reproducible transverse midfoot instability, which has been demonstrated to be more reliably detected on manual stress radiographs than on weight-bearing radiographs12. Additional longitudinal midfoot stability is supplied by the plantar medial-middle cuneiform (pC1-C2) ligament, disruption of which is also best seen on manual stress radiographs. These ligaments have additionally been described on magnetic resonance images4.
Magnetic Resonance Imaging Protocols
Twenty-one magnetic resonance imaging examinations were performed for twenty sequential patients who had presented with a clinical suspicion of Lisfranc joint instability following a traumatic episode and who had equivocal findings on radiographs that had been made either in the radiology department or in the physician's office. Patients with radiographically overt fractures were excluded from the study. There were no other exclusion criteria, and no patient was excluded from the study or was lost to follow-up. Nineteen magnetic resonance imaging examinations were performed at 1.5 T (Signa LX; G.E. Healthcare, Milwaukee, Wisconsin) and two were performed at 0.7 T (Signa OpenSpeed 0.7T) with use of a standardized noncontrast imaging protocol. The magnetic resonance imaging evaluation included coronal T1-weighted images without fat saturation (field of view = 12 mm2, matrix = 256 × 256, number of excitations = 1, slice thickness = 3 mm, gap = 0.5 mm, repetition time = 400 to 800 ms, echo time = minimal, bandwidth = 16 kHz); coronal T2-weighted fast-spin-echo images with fat saturation (matrix = 256 × 192, repetition time = >2000 ms, echo time = 50 to 60 ms, number of excitations = 2, echo train length = 8, other parameters same as those for the T1-weighted coronal images); sagittal fast-spin-echo short tau inversion recovery images (field of view = 12 to 14 mm2, number of excitations = 3, echo time = 20 to 40 ms, time to inversion = 150 ms, other parameters same as those for the T2-weighted coronal images); sagittal T1-weighted images (field of view = 12 to 14 mm2, other parameters same as those for the T1-weighted coronal images); and axial T2-weighted fast-spin-echo images with fat saturation (slice thickness = 4 mm, gap = 1 mm, other parameters same as those for the T1-weighted coronal images).
Image Analysis
Two fellowship-trained musculoskeletal radiologists (S.D. and A.C.Z.) who were blinded to the clinical data reviewed the magnetic resonance images with regard to the integrity of the dorsal (dC1-M2) and plantar (pC1-M2M3) bundles of the Lisfranc ligament on short-axis (coronal) and long-axis (axial) images, the integrity of the first (pC1-M1) and second (pC2-M2) plantar tarsal-metatarsal ligaments on long-axis images, and the integrity of the medial-middle cuneiform (dC1-C2) ligament on long-axis images, and a diagnosis was determined by consensus. Furthermore, the presence and extent of fluid along the first and second metatarsals also were evaluated; in cases in which the interpretation of the long and short-axis images differed, the final reported diagnosis corresponded with the highest grade of injury. In all cases, the radiographic and intraoperative findings were compared with regard to the stability of the Lisfranc joint.
Statistical Analysis
Stepwise logistic regression was performed to identify the best indicators of Lisfranc joint complex instability. Linear regression, or the fitting of a line to a dataset, is a commonly used statistical technique for describing the relationship between variables. When the dependent variable can take on only two values (in this case, "stable" or "unstable"), then logistic regression must be used18.
Logistic regression is based on the logit function (the logarithm of the odds ratio): logit?(p)=p1-p, where p = the probability of a given outcome (in this case, instability). In logistic regression, the dependent variable is a logit and is set equal to the predictor. In the present study, we use the simple-linear-regression predictor, y = ß1x1ß2x2ß3x3ß4x4ß5x5ß6x6, where each of the xi is a different predictor:logit?(p)=p1-p
=ß0+ß1x1+ß2x2+ß3x3+ß4x4+ß5x5+ß6x6The solution to the model is pi=11+e-(ß0+ß1x1+ß2x2+ß3x3+ß4x4+ß5x5+ß6x6). Simply put, we can use the information from our study subjects to create an equation that will allow us to use predictors from the magnetic resonance imaging examinations to predict whether a patient will have Lisfranc joint complex instability. The predictors that were included in the model were the integrity of the dorsal (dC1-M2) and plantar (pC1-M2M3) bundles of the Lisfranc ligament, the first (pC1-M1) and second (pC2-M2) plantar tarsometatarsal ligaments, and the medial-middle cuneiform (dC1-C2) ligament. Data were reduced to dichotomous form by classifying the ligaments either as "intact" (comprising those that were determined on magnetic resonance imaging to be intact [Figs. 2-A and 2-B] or to have a grade-1 sprain) or as "ruptured" (comprising those that were determined on magnetic resonance imaging to have a grade-2 sprain [Fig. 3] or to be completely torn). Furthermore, the presence and extent of fluid along the first metatarsal and the presence of any fractures of the second metatarsal and other bones around the Lisfranc joint were also evaluated. Sensitivity, specificity, and positive predictive values were calculated for models including different predictors.
Source of Funding
No external funding or internal funding was used.
Of the twenty patients in our cohort, five patients (six feet) were professional American football players, one patient engaged in competitive aerobics, and one patient was a collegiate triple-jump athlete. Eleven injuries occurred during sports activities, four occurred after a fall, two occurred as the result of a motor-vehicle collision, two occurred as the result of a crush injury of the foot, one occurred when the foot was caught in an escalator, and one occurred as the result of a twisting injury of the foot when the patient was walking down a flight of stairs. All patients had tenderness and swelling over the first and second tarsometatarsal joints. Evaluation of the foot with use of weight-bearing radiographs was attempted in all cases, but all attempts were limited by pain, with the patients stating that they were unable to apply more than half of their weight to the foot. In each case, diastasis between the medial cuneiform and the second metatarsal base was found to be <2 mm. No fracture was identified radiographically in any case. A table summarizing the magnetic resonance imaging and intraoperative findings can be found in the Appendix.
Magnetic resonance imaging revealed a pC2-M2 ligament rupture in eighteen feet, a pC1-M2M3 ligament rupture in seventeen feet (Fig. 4), a dC1-M2 ligament rupture in thirteen feet, fluid along the first metatarsal in thirteen feet, and a pC1-M1 ligament rupture in three feet. None of the patients were found to have a dC1-C2 injury. Nine feet had a fracture, which involved the second metatarsal (seven), the medial cuneiform (one), or the lateral cuneiform (one). Five of the seven second metatarsal fractures were incomplete and involved the metaphyseal or diaphyseal region of the metatarsal (Figs. 2-A and 4). Two, however, were avulsion fractures involving the plantar base of the metatarsal at the attachment of the pC1-M2M3 ligament (Cases 4 and 14), and in both of those cases the midfoot was found to be unstable. None of these fractures were visible on plain radiographic evaluation.
Of the twenty-one Lisfranc joint complexes that were evaluated with magnetic resonance imaging, seventeen (81%) were confirmed to be unstable on manual stress radiographs made with the patient under anesthesia in the operating room (Fig. 5) and underwent surgical fixation. In the cases of seven (41%) of these seventeen injuries, the initial diagnosis was a simple sprain.
Stepwise logistic regression was performed to identify magnetic resonance imaging findings indicative of Lisfranc joint instability. The conditions of the dC1-M2, pC1-M2M3, pC1-M1, pC2-M2, and dC1-C2 ligaments ("intact" or "ruptured") as well as the presence or absence of fluid along the base of the first metatarsal were used to predict stability or instability. Results of the logistic regression indicated that the pC1-M2M3 ligament condition alone provided a model with a sensitivity of 94%, a specificity of 75%, and a positive predictive value of 94%. The coefficients of the model, regression equation, and predicted values for this regression are presented in Tables I and II.
A regression was performed to evaluate the predictive value of the status of the pC2-M2 ligament alone as it was the most commonly injured structure. The result for this finding was not significant; thus, the presence of a pC2-M2 ligament rupture was not useful for predicting Lisfranc joint complex stability. Further logistic regressions were performed with use of other combinations of possible predictors. None of the other factors (dC1-M2, pC1-M1, pC2-M2, or dC1-C2 ligament injury; fluid along the first metatarsal base; or fracture), alone or in combination, were found to be useful predictors of Lisfranc joint complex stability or instability.
Instability and incongruency of the articulation between the second metatarsal base and the medial cuneiform (Lisfranc joint) results in a high degree of long-term disability and dysfunction13,14,19. In most cases, diastasis of =2 mm between these two bones or the presence of an avulsion fracture at the base of the second metatarsal (at the plantar attachment of the C1-M2M3 ligament) as seen on standard radiographic evaluation is adequate to confirm a diagnosis of instability and to support the recommendation of treatment. Occasionally, despite an appropriate mechanism of injury and clinical findings suggestive of local tenderness over the articulation and inability to bear weight on the foot, plain radiographs are not diagnostic. In these cases, manual stress radiographic evaluation with the patient under anesthesia has been shown to be the most accurate determinant of midfoot stability12,13,20. Despite the suggestive clinical findings, a stable construct has been reported to be found at the time of examination with the patient under anesthesia in 10% to 20% of cases2, resulting in an unnecessary risk to the patient. Nevertheless, because of the potentially devastating effects of an untreated or misdiagnosed Lisfranc ligament injury13,20, examination with the patient under anesthesia was, and in many cases still is, considered acceptable.
In the present study, we were able to correctly classify nineteen (90%) of twenty-one Lisfranc joint complexes as either stable or unstable on the basis of magnetic resonance imaging of the plantar bundle of the Lisfranc ligament (pC1-M2M3) with use of logistic regression. This provided us with a sensitivity of 94% and a specificity of 75%. Only one foot in which instability was subsequently confirmed on stress radiographs had an intact pC1-M2M3 ligament. In this case, the patient (Case 4) had an avulsion fracture at the base of the second metatarsal plantarly, which may have had some protective effect on the appearance of the ligament on magnetic resonance images, on which a grade-1 sprain was initially diagnosed.
In addition, the majority (eighteen) of the twenty-one feet demonstrated disruption of the second plantar tarsal-metatarsal (pC2-M2) ligament. None of these disruptions were repaired at the time of surgery, and there was a poor correlation between this finding and instability demonstrated with the patient under anesthesia. At the present time, this injury is believed to be inconsequential in terms of treatment, planning, and outcome in cases of midfoot injuries. However, we believe that injury to the second plantar tarsal-metatarsal ligament, which in the present series was clearly seen in cases of a normal or sprained Lisfranc ligament, may confound the clinical picture during physical examination, resulting in unnecessary surgery. In this regard, it may be a clinical mimicker of a Lisfranc ligament tear. Magnetic resonance imaging has the unique ability to identify a torn plantar tarsal-metatarsal ligament and, in the setting of a normal Lisfranc ligament, to confirm the cause of pain and tenderness and to serve as a guide to nonoperative treatment.
While we believe that intraoperative correlation in 90% of the cases was a major strength of the present study, the study had several limitations. Most notable were the relatively small number of magnetic resonance imaging studies reviewed (n = 21) and the retrospective nature of the study. These limitations may have prevented us from finding more subtle relationships between the finding of a ligamentous injury on magnetic resonance imaging and Lisfranc joint instability. Another limitation was selection bias in that there was no attempt to evaluate magnetic resonance imaging studies of uninjured subjects. As most of the subjects in the present study had an unstable midfoot, it is difficult to completely assess the usefulness of this test. However, our anecdotal experience in reviewing magnetic resonance imaging examinations of the midfoot for reasons other than to evaluate a Lisfranc-type injury did reveal clearly visualized and intact plantar and dorsal ligament complexes as described in the normal anatomic situation, and thus we believe that our interpretation of findings is meaningful. We suggest that this work can be further explored by evaluating a larger series of cases prospectively in the future.
In conclusion, direct visualization of the Lisfranc ligament with magnetic resonance imaging is an accurate method for detecting Lisfranc joint complex instability when the plantar Lisfranc ligament bundle (pC1-M2M3) and a second metatarsal base fracture are used as predictors. The magnetic resonance imaging finding of a complete rupture or grade-2 sprain of the pC1-M2M3 ligament is highly suggestive of an unstable midfoot, for which surgical stabilization has been recommended to prevent long-term functional deficiency, arch collapse, and degenerative arthritis13,14,19. The appearance of a normal ligament, particularly in the absence of a metatarsal base avulsion fracture, is suggestive of a stable midfoot and may obviate the need for a manual stress radiographic evaluation to be performed with the patient under anesthesia when the clinical and radiographic findings are equivocal. To our knowledge, disruption of the second plantar tarsal-metatarsal ligament (pC2-M2) has not previously been reported in the setting of Lisfranc joint injury, but this finding was present in the majority of cases in the present study. Although it may be a common clinical confounder, this finding is not a useful predictor of instability.
Disruption of the dorsal C1-M2 ligament, the plantar C1-M1 ligament, the plantar C2-M2 ligament, or the dorsal C1-C2 ligament as well as the presence of fluid along the first metatarsal base are frequently seen in association with the Lisfranc injury mechanism but are not useful predictors of instability of the midfoot. Regional metatarsal and cuneiform fractures (except for avulsion fractures at the insertion of the pC1-M2M3 ligament) are commonly seen and do not, in isolation, indicate the presence of midfoot instability.
We propose a diagnostic and therapeutic algorithm for patients presenting with an injury mechanism and a clinical examination suggestive of a subtle Lisfranc injury and possible midfoot instability (Fig. 6).
A table presenting detailed clinical and magnetic resonance imaging information on all study subjects is available with the electronic versions of this article, on our web site at jbjs.org (go to the article citation and click on "Supplementary Material") and on our quarterly CD/DVD (call our subscription department, at 781-449-9780, to order the CD or DVD). 
Curtis MJ, Myerson M, Szura B. Tarsometatarsal joint injuries in the athlete. Am J Sports Med.1993;21:497-502.21497
1993
[PubMed][CrossRef]
Sherief TI, Mucci B, Greiss M. Lisfranc injury: how frequently does it get missed? And how can we improve? Injury.2007;38:856-60.38856
2007
[CrossRef]
Harwood MI, Raikin SM. A Lisfranc fracture-dislocation in a football player. J Am Board Fam Pract.2003;16:69-72.1669
2003
[CrossRef]
Preidler KW, Wang YC, Brossmann D, Trudell D, Daenen B, Resnick D. Tarsometatarsal joint: anatomic details on MR images. Radiology.1996;199:733-6.199733
1996
Foster SC, Foster RR. Lisfranc's tarsometatarsal fracture-dislocation. Radiology.1976;120:79-83.12079
1976
Norfray JF, Geline RA, Steinberg RI, Galinski AW, Gilula LA. Subtleties of Lisfranc fracture-dislocations. AJR Am J Roentgenol.1981;137:1151-6.1371151
1981
Coss HS, Manos RE, Buoncristiani A, Mills WJ. Abduction stress and AP weightbearing radiography of purely ligamentous injury in the tarsometatarsal joint. Foot Ankle Int.1998;19:537-41.19537
1998
Haapamaki V, Kiuru M, Koskinen S. Lisfranc fracture-dislocation in patients with multiple trauma: diagnosis with multidetector computed tomography. Foot Ankle Int.2004;25:614-9.25614
2004
Kavanagh EC, Zoga AC. MRI of trauma to the foot and ankle. Semin Musculoskelet Radiol.2006;10:308-27.10308
2006
[CrossRef]
Potter HG, Deland JT, Gusmer PB, Carson E, Warren RF. Magnetic resonance imaging of the Lisfranc ligament of the foot. Foot Ankle Int.1998;19:438-46.19438
1998
Preidler KW, Brossmann J, Daenen B, Goodwin D, Schweitzer M, Resnick D. MR imaging of the tarsometatarsal joint: analysis of injuries in 11 patients. AJR Am J Roentgenol.1996;167:1217-22.1671217
1996
Kaar S, Femino J, Morag Y. Lisfranc joint displacement following sequential ligament sectioning. J Bone Joint Surg Am.2007;89:2225-32.892225
2007
[CrossRef]
Faciszewski T, Burks RT, Manaster BJ. Subtle injuries of the Lisfranc joint. J Bone Joint Surg Am.1990;72:1519-22.721519
1990
Myerson MS, Fisher RT, Burgess AR, Kenzora JE. Fracture dislocations of the tarsometatarsal joints: end results correlated with pathology and treatment. Foot Ankle.1986;6:225-42.6225
1986
Peicha G, Labovitz J, Seibert FJ, Grechenig W, Weiglein A, Preidler KW, Quehenberger F. The anatomy of the joint as a risk factor for Lisfranc dislocation and fracture-dislocation. An anatomical and radiological case control study. J Bone Joint Surg Br.2002;84:981-5.84981
2002
[CrossRef]
de Palma L, Santucci A, Sabetta SP, Rapali S. Anatomy of the Lisfranc joint complex. Foot Ankle Int.1997;18:356-64.18356
1997
Johnson A, Hill K, Ward J, Ficke J. Anatomy of the Lisfranc ligament. Foot and Ankle Specialist.2008;1:19-23.119
2008
[CrossRef]
Portney LG, Watkins MP. Foundations of clinical research: application to practice. 3rd ed. Upper Saddle River, NJ: Pearson/Prentice Hall; 2008. p 696-700.
2008
Ly TV, Coetzee JC. Treatment of primarily ligamentous Lisfranc joint injuries: primary arthrodesis compared with open reduction and internal fixation. A prospective, randomized study. J Bone Joint Surg Am.2006;88:514-20.88514
2006
[CrossRef]
Myerson M. The diagnosis and treatment of injuries to the Lisfranc joint complex. Orthop Clin North Am.1989;20:655-64.20655
1989