Abstract
Objective
This systematic review and meta-analysis evaluated the diagnostic accuracy of conventional computed tomography (CT) and weight-bearing computed tomography (WBCT) in assessing Lisfranc injuries, comparing their sensitivity, specificity, and reliability for detecting structural abnormalities between injured and uninjured cases.
Methods
A systematic search of PubMed, Scopus, WOS, and Medline databases identified 736 articles, of which 16 studies met the inclusion criteria. Seven studies were included in the meta-analysis. The analysis examined measurements such as M1-M2 and M2-C1 base distances, TMT1 and TMT2 dorsal step-offs, axial joint area and volume, and alignment parameters. Subgroup analyses compared conventional CT and WBCT methods.
Results
The injured group showed significantly greater M1-M2 and M2-C1 Base Distances than the uninjured group (M1-M2: MD = 0.88 mm; M2-C1: MD = 2.61 mm; p-value < 0.0001), with no significant differences between imaging modalities. The injured group had greater TMT2 dorsal step-offs (MD = 0.81 mm; p < 0.001), while TMT1 dorsal step-off differences varied between modalities. Injured joints demonstrated significantly larger Axial Joint Area and Volume (MD = 16.26 mm²) and volumes (MD = 189.1 mm³; p < 0.0001). Additionally, WBCT demonstrated high sensitivity and specificity for Lisfranc injuries, particularly for parameters like axial joint volume (AUC = 0.91). Conventional CT showed variable diagnostic performance, with sensitivity and specificity ranging from 69 % to 97.9 % and 39.3–100 %, respectively. The “Mercedes sign” (a three-pointed star appearance on axial CT indicating C1-M2 diastasis) and “Peeking cuneiform sign” (visible medial cuneiform between first and second metatarsals on AP view) showed over 92 % sensitivity and specificity.
Conclusions
Both conventional CT and WBCT effectively identify Lisfranc injuries when comparing injured and uninjured cases, with WBCT demonstrating superior sensitivity, specificity, and diagnostic accuracy in certain parameters. Conventional CT remains reliable but exhibits variability in specific assessments. Future research should focus on standardizing imaging protocols and refining diagnostic thresholds to enhance consistency and accuracy.
1
Introduction
Lisfranc injuries involving the tarsometatarsal (TMT) joints are frequently overlooked or misdiagnosed, presenting significant diagnostic challenges particularly when subtle instability is present . ,, The Lisfranc ligamentous complex (LLC), comprising the TMT, intertarsal, and intermetatarsal joints, provides crucial midfoot stability through its ligamentous and bony architecture . The strongest plantar ligament, connecting the medial cuneiform (C1) to the second metatarsal base (M2), serves as the keystone of this complex . When injured, LLC disruption can lead to subluxation, biomechanical alterations, and subsequent degenerative joint disease if left untreated ,, .
Current diagnostic methods face limitations in detecting subtle Lisfranc injuries. Plain radiographs may miss injuries without significant displacement, while conventional computed tomography (CT), despite offering excellent bone detail, may overlook unstable injuries when performed in non-weight-bearing conditions . Magnetic resonance imaging (MRI), though useful for identifying ligamentous damage, lacks the precise bone detail needed for instability assessment and cannot replicate physiological loading conditions . Traditional measurements, such as the 2–5 mm diastasis criteria on plain radiographs, provide too wide a range to adequately characterize subtle injuries.
Weight-bearing computed tomography (WBCT) has emerged as a promising diagnostic tool that combines the superior bone visualization of CT with physiological loading conditions ,, . This technology enables reliable three-dimensional evaluation of the Lisfranc joint under weight-bearing stress, potentially identifying subtle instabilities that conventional imaging might miss . However, systematic comparison of diagnostic accuracy between conventional CT and WBCT for Lisfranc injuries remains limited. This systematic review and meta-analysis evaluates and compares the diagnostic accuracy, sensitivity, and specificity of conventional CT and WBCT in detecting structural abnormalities between injured and uninjured cases with Lisfranc injuries.
2
Methods
2.1
Study protocol and registration
To conduct this systematic review and meta-analysis , we used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA). We followed all the steps mentioned in Cochrane’s Handbook of Systematic Reviews of Interventions .
2.2
Search strategy and data collection
We searched four databases (Scopus, PubMed, WoS, and Medline via WoS) for studies on CT and Lisfranc injuries from 1996 (when CBCT was first published) up to Oct 9, 2024. The review included studies using WBCT or conventional CT to assess Lisfranc injuries, comparing radiological parameters (e.g., M1-M2 base distance, TMT1 dorsal step-off, Lisfranc joint area) between injured and uninjured conditions, with subgroup analyses for each CT type. This study also analyzed data on the diagnostic accuracy and observer reliability of conventional and WBCT imaging for Lisfranc injuries. Included studies reported intra- and inter-observer reliability (e.g., kappa coefficients) and diagnostic accuracy (e.g., sensitivity, specificity, PPV, NPV, AUC). We focused on diagnosing Lisfranc injuries and postoperative assessments using these imaging methods. (Supp file 1)
2.3
We included only english studies
We removed duplicates with Endnote (X-9) and screened studies in two steps: title/abstract screening and full-text review. Two authors screened independently, with a third resolving conflicts.
2.4
Data extraction and outcome measurements
We extracted data using Excel in two sheets: a summary and baseline sheet, and an outcome sheet, by two separate authors, and the third author solved the conflict.
2.5
Quality assessment
The methodological index for non-randomized studies (MINORS) criteria was employed .
2.6
Data analysis
We conducted this meta-analysis using R Studio (version 2024.09.0 +375), employing the meta and metafor packages for computations and the forestplot and grid packages for creating detailed visualizations .
2.7
Double-arm outcomes
All outcomes compared injured and uninjured conditions and included subgroup analyses to assess differences between conventional and WBCT methods where data was available . Key outcomes included M1-M2 Base Distance, M2-C1 Base Distance, TMT2 Dorsal Step-off, Axial Lisfranc Joint Area, and Axial Lisfranc Joint Volume. Continuous variables were analyzed as mean differences (MD) with corresponding standard errors (SE). Forest plots were generated to display study-specific effects (represented by navy blue squares) and pooled estimates (depicted as navy diamonds). Subgroup analyses for all double-arm outcomes were conducted using fixed-effects and random-effects models based on the DerSimonian method, with subgroup differences between conventional CT and MRI assessed using chi-square (χ²) tests .
2.8
Diagnostic accuracy and observer reliability of conventional and WBCT
The results underwent a comprehensive stratification process based on several key factors. Diagnostic features, such as the “Mercedes sign” (a three-pointed star appearance on axial CT views indicating C1-M2 diastasis) and “Peeking cuneiform sign” (visualization of the medial cuneiform between the first and second metatarsal bases on anteroposterior views), were used as primary classification criteria to organize the findings. Additionally, different imaging planes, including axial and coronal views, served as secondary stratification parameters. Study-specific characteristics were also considered to ensure appropriate categorization of results. For the interpretation of reliability metrics, a structured framework was implemented. Results showing a Kappa value of 0.80 or greater were classified as having excellent agreement, representing the highest level of reliability. Those with Kappa values between 0.60 and 0.79 were determined to have moderate agreement, indicating good but not optimal reliability. Fair agreement was assigned to results with Kappa values ranging from 0.40 to 0.59, suggesting limited but acceptable reliability. Finally, any results with a Kappa value below 0.40 were categorized as having poor agreement, indicating minimal reliability that might not be suitable for clinical decision-making.
Diagnostic accuracy outcomes were presented as reported, with AUC values interpreted according to standard thresholds: ≥ 0.90 as excellent, 0.80–0.89 as good, 0.70–0.79 as fair, 0.60–0.69 as poor, and < 0.60 as failed. Sensitivity and specificity were summarized as percentages with 95 % confidence intervals when available.
2.9
Heterogeneity assessment
Heterogeneity was assessed using the I² statistic and categorized as follows: 0–40 % (unimportant), 30–60 % (moderate), 50–90 % (substantial), and 75–100 % (considerable). When significant heterogeneity was detected (I² > 50 % and p < 0.1), random-effect models were applied.
2.10
Sensitivity analyses and publication bias
Leave-one-out sensitivity analyses were conducted to assess the impact of individual studies on the overall effect size and identify outliers causing heterogeneity. Publication bias was evaluated using funnel plots with color-coded significance contours (e.g., p < 0.1, p < 0.05) and visual inspection of plot symmetry.
3
Results
3.1
Study selection
A literature search across PubMed, Scopus, WOS, and Medline identified 736 articles, with 305 duplicates. After screening 431 titles and abstracts, 127 articles underwent full-text review. Ultimately, 16 studies were included in the systematic review, and 7 in the meta-analysis. ( Fig. 1 ).
Study flow diagram illustrating the selection process for the systematic review and meta-analysis. A total of 736 articles were identified through database searches (PubMed, Scopus, WOS, and Medline), with 305 duplicates removed. Screening of titles and abstracts was conducted for 431 articles, and 127 full-text articles were reviewed. Sixteen studies were included in the systematic review, with seven studies proceeding to meta-analysis.
3.2
Baseline characteristics
Conventional CT studies were mostly retrospective single-center cohorts (1991–2022), primarily from China and the U.S., with patients aged 32–57 years and approximately 300 female participants. WBCT studies combined cadaveric and retrospective cohorts (2015–2024), all single-center from the U.S. and Switzerland, with predominantly male participants aged 32–76 years. Reported BMI ranged from 24 to 30 kg/m² ( Table 1 ).
Table 1
Summary of the included studies and baseline characteristics of the patients.
| Conventional CT Studies | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Study ID | Design | Centers | Country | Duration | Sample Size (Total/Injured/Control) | Inclusion Criteria | Conflict of Interest | Fund | Age (years) | Female N | BMI |
| Tang et al. (2024) | Retrospective cohort study | Single-center | China | 2013–2022 | 31 (31/0) | Minor foot injury, midfoot pain, Lisfranc injury confirmed by surgery | N/A | N/A | 32.26 ± 5.76 | 10 (Total) | N/A |
| Chen et al. (2023) | Retrospective cohort study | Single-center | China | 2017–2019 | 407 (307/100) | Age 18–80 years, acute foot injury, Lisfranc injury confirmed via CT scan | N/A | China’s National R&D Program, Shanghai grants | N/A (Total: 46.2 ± 14.9) | 160 (Total) | N/A |
| Tamir et al. (2023) | Retrospective case-control | Single-center | Israel | 2010–2020 | 102 (52/50) | Injured: Lisfranc injury confirmed intraoperatively; Control: normal CT of contralateral foot | N/A | N/A | 40 | 25 (Total) | N/A |
| Essa et al. (2022) | Retrospective cohort study | Single-center | Israel | 2020 | 44 (N/A) | Suspected Lisfranc injuries, CT performed in 2020, no prior foot/ankle surgery | N/A | N/A | 41.4 | 19 (Total) | N/A |
| Rikken et al. (2022) | Retrospective cohort study | Multicenter | United States | 1991–2018 | 72 (47/25) | Lisfranc: surgically confirmed injury; Hallux Valgus: surgically treated bilateral hallux valgus | Yes (Board memberships) | N/A | Lisfranc: 32.6 ± 15.0, Hallux: 50.8 ± 15.3 | 40 (Total) | Lisfranc: 27.4 ± 4.8, Hallux: 24.2 ± 4.6 |
| Shim et al. (2022) | Retrospective case-control | Single-center | South Korea | 2014–2020 | 115 (30/85) | Subtle Lisfranc injuries, treated operatively/non-operatively, bilateral CT scans | N/A | N/A | Surgical: 34.7 ± 10.9, Conservative: 36 ± 14.6 | 65 (Total) | N/A |
| Ponkilainen et al. (2020) | Retrospective cohort study | Multicenter | Finland | 2012–2016 | 100 (66/34) | Acute foot trauma, Lisfranc joint injury confirmed via CT/CBCT | N/A | Finnish Research Foundation | 40.9 | 45 (Total) | N/A |
| Clements et al. 2018 | Retrospective cross-sectional | Single-center | United States | 2014 | 30 (N/A) | CT scans showing tarsometatarsal joints, free of trauma or arthritis | N/A | N/A | N/A | N/A | N/A |
| Kunas et al. (2017) | Retrospective case-control | Single-center | United States | 2008–2014 (AAFD)/ 2006–2013 (Control) | 58 (38/20) | AAFD: Age > 40, flatfoot deformity stage II; Control: Lisfranc injuries | N/A | N/A | AAFD: 57 ± 12, Lisfranc: 36 ± 13 | 29 (Total) | N/A |
| Li et al. (2017) | Retrospective cohort study | Single-center | China | 2010–2016 | 79 (79/0) | Closed Lisfranc injuries, treated with ORIF, CT data available | N/A | National Natural Science Foundation of China | 35.6 | 36 (Total) | N/A |
| Summary | Total studies: 10 | Single: 8, Multi: 2 | 5 countries | Ranges from 2010 to 2022 | Total: 1038 (680 injured) | Mostly retrospective with Lisfranc injury confirmation via CT | Most have N/A COI | 6 studies with funding | Age range: 32–57 ± 12 | ∼300 + females | BMI: Mostly N/A |
| Weight-Bearing CT Studies | |||||||||||
| Study ID | Design | Centers | Country | Duration | Sample Size (Total/Injured/Control) | Inclusion Criteria | Conflict of Interest | Fund | Age (years) | Female N | BMI |
| Bhimani et al. (2024) | Cadaveric study | Single-center | United States | N/A | 10 (10/0) | Amputated feet with no pre-existing conditions, Lisfranc injuries induced during study | N/A | N/A | 67 ± 14 | 0 (All male) | N/A |
| Falcon et al. (2023) | Retrospective cohort study | Single-center | United States | 2018–2020 | 56 (N/A) | WBCT of bilateral lower extremities for Lisfranc or syndesmosis injury | N/A | N/A | Injured: 37.5 ± 18.8, Control: 38.02 ± 17.23 | 28 (Total) | 29.4 ± 7.4, 26.6 ± 4.3, 30.2 ± 8.0 |
| Bhimani et al. (2020) | Retrospective cohort study | Single-center | United States | 2015–2019 | 50 (14/36) | Injured: unilateral Lisfranc injury; Control: bilateral WBCT for ankle assessment, no foot injury | N/A | N/A | 32.2 ± 15.5, Control: 35.7 ± 16.5 | 27 (Total) | Injured: 26.3 ± 5.3, Control: 26.7 ± 4.6 |
| Penev et al. (2020) | Cadaveric study | Single-center | Switzerland | N/A | 8 (8/0) | Fresh-frozen lower limbs with no pre-existing damage | N/A | N/A | 76 ± 11 | 3 (Total) | N/A |
| Sripanich et al. (2020a) | Cadaveric study | Single-center | United States | N/A | 12 (12/0) | Male cadavers, 18–65 years, BMI ≤ 30, no prior foot/ankle issues | N/A | N/A | 48 ± 15.2 | 0 (All male) | 24.8 ± 4.8 |
| Sripanich et al. (2020b) | Cadaveric study | Single-center | United States | N/A | 12 (12/0) | Male cadavers, BMI ≤ 30, no prior foot/ankle surgery | N/A | N/A | 46 ± 14.8 | 0 (All male) | 25.3 ± 4.2 |
| Summary | Total studies: 6 | All single-center | 2 countries | Ranges from 2015 to 2024 | Total: 148 (66 injured) | Cadaveric studies and retrospective cohorts | No COI reported | No clear funding sources | Age range: 32–76 | Mostly male | BMI range: 24–30 |
3.3
Quality assessment
As shown in Table 2 , the assessment of the included studies using the MINORS scale demonstrated high methodological quality, with scores ranging from 16 to 23 out of a possible 24 points ( Table 2 ).
Table 2
Assessment of the quality of studies through methodological index for Non-Randomized studies (MINORS).
| Study ID | Clearly Stated Aim | Consecutive Patients | Prospective Collection of Data | Endpoints | Assessment of Endpoint | Follow-up Period | Loss < 5 % | Study Size | Adequate Control Group | Contemporary Group | Baseline Control | Statistical Analyses | MINORS Score |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Tang et al. (2024) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 0 | 0 | 0 | 2 | 16 |
| Chen et al. (2023) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 23 |
| Tamir et al. (2023) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 23 |
| Essa et al. (2022) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 22 |
| Rikken et al. (2022) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 22 |
| Shim et al. (2022) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 23 |
| Ponkilainen et al. (2020) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 23 |
| Clements et al. 2018 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 22 |
| Kunas et al. (2017) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 22 |
| Li et al. (2017) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 0 | 0 | 0 | 2 | 22 |
| WBCT | |||||||||||||
| Bhimani et al. (2024) | 2 | 2 | 2 | 2 | 2 | 1 | 2 | 1 | 0 | 0 | 0 | 2 | 16 |
| Falcon et al. (2023) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 22 |
| Bhimani et al. (2020) | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 22 |
| Penev et al. (2020) | 2 | 2 | 2 | 2 | 2 | 1 | 2 | 1 | 0 | 0 | 0 | 2 | 16 |
| Sripanich et al. (2020a) | 2 | 2 | 2 | 2 | 2 | 1 | 2 | 1 | 0 | 0 | 0 | 2 | 16 |
| Sripanich et al. (2020b) | 2 | 2 | 2 | 2 | 2 | 1 | 2 | 1 | 0 | 0 | 0 | 2 | 16 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree




