Reliability and validity of different ankle MRI scanning planes for the anterior talofibular ligament injury diagnosis: a cadaveric study

An ankle sprain is a common injury in sports and daily living. After an ankle sprain, anterior talofibular ligament (ATFL) is the most frequently injured ligament. If injured ATFL is untreated or treated inappropriately, some patients will develop residual symptoms like pain, giving way, feeling of instability, or early-onset osteoarthritis [1]. Some cohort studies estimate more than 20% ankle sprain patients eventually developed persistent symptoms [2]. The patients’ history and clinical manual tests are important for the diagnosis of acute or chronic ATFL injury. However, a systemic review investigated the accuracy of clinical manual tests and suggested that ligament injury cannot be ruled out even when clinical manual tests are negative [3]. When the results of clinical tests are ambiguous, imaging techniques like magnetic resonance imaging (MRI), stress radiography, and ultrasonography could be helpful. MRI is widely utilized in diagnosing ankle ligament injury [4, 5] and could also provide information about concomitant lesions and influence the precise operative technique for a certain patient [6].

ATFL courses anteriorly, medially, and inferiorly from fibula to talus and sometimes cannot be delineated clearly on a routine axial plane in MRI [7]. To better delineate ATFL in ankle MRI, two approaches were utilized. One is to reposition the tested foot with tape or custom-made device [7, 8]. The other is to utilize oblique axial plane scanning or multiplanar reconstruction of MR images, parallel to ATFL course [9, 10]. The second method was more convenient in clinical practice. ATFL was reported to be directed at approximately 45° medially from the sagittal plane and at approximately 25° inferiorly from the horizontal plane [11, 12]. The oblique axial-coronal plane and oblique axial-sagittal plane were parallel to ATFL course in different planes. Kim et al. reported the oblique axial-coronal plane (ATF view) from the center of talus to the center of navicular delineated full-length view of ATFL better than routine axial view [13]. Previous studies reported grades of injury, subjective scales of optimal visualization and the rates of full-length view in an oblique axial plane compared with the routine axial plane. However, reliability and diagnostic accuracy of the oblique axial plane have not been investigated. This limited the clinical application of oblique MRI scanning planes. The objective of the current study is to compare reliability, accuracy, sensitivity, and specificity in MRI evaluation of cadaveric ATFL among routine axial plane, oblique axial-coronal plane, and oblique axial-sagittal plane. Our hypothesis is that both oblique axial-coronal MRI scanning and oblique axial-sagittal MRI scanning will deliver better reliability and diagnostic performance compared to routine axial MRI scanning.

The interobserver agreement of three scanning methods was listed in Table 1. The interobserver agreement regarding the injury of ATFL was fair to good (kappa, 0.55 to 0.65) in routine axial planes, fair to good (kappa, 0.557 to 0.75) in oblique axial-sagittal planes, and excellent (kappa, 0.85 to 0.95) in oblique axial-coronal planes.

The diagnostic performance of each scanning plane was listed in Table 2. For reader 1, the sensitivities of routine axial, oblique axial-coronal, and oblique axial-sagittal MRI scanning were 0.65, 0.85, and 0.6 respectively. The specificities of routine axial, oblique axial-coronal, and oblique axial-sagittal MRI scanning were 0.65, 0.9, and 0.75 respectively. The accuracy of reader 1 was significantly higher (p = .018) when utilizing oblique axial-coronal MRI scanning (accuracy = 0.875) than routine axial MRI scanning (accuracy = 0.65).

For reader 2, the sensitivity of reader 2 was also significantly higher (p = .008) when utilizing oblique axial-coronal MRI scanning (sensitivity = 0.95) than routine axial MRI scanning (sensitivity = 0.6). The specificities of routine axial, oblique axial-coronal, and oblique axial-sagittal MRI scanning were 0.75, 0.9, and 0.75 respectively. The accuracy of reader 2 was also significantly higher (p = .005) when utilizing oblique axial-coronal MRI scanning (accuracy = 0.925) than routine axial MRI scanning (accuracy = 0.675).

For reader 3, the sensitivities of routine axial, oblique axial-coronal, and oblique axial-sagittal MRI scanning were 0.85, 0.95, and 0.75 respectively. The specificities of routine axial, oblique axial-coronal, and oblique axial-sagittal MRI scanning were 0.85, 0.95, and 0.85 respectively. The accuracies of routine axial, oblique axial-coronal, and oblique axial-sagittal MRI scanning were 0.85, 0.95, and 0.8 respectively.

The oblique axial plane, which is parallel to the course of ATFL, was reported to delineate ATFL more clearly in MRI scanning [13]. However, the application of oblique scanning planes is uncommon in clinical practice, because the reliability and validity of the oblique axial plane were still unknown. The current cadaveric study investigated the interobserver agreement and diagnostic performance of two kinds of oblique axial plane. According to our results, the diagnostic performance of the routine axial plane and oblique axial-sagittal plane were comparable. The diagnostic accuracy of the oblique axial-coronal plane was higher than the routine axial plane.

MRI is helpful when the diagnosis of ATFL injury is uncertain after careful inquiry of the patient’s history and physical tests. MRI is also frequently performed to confirm or exclude the presence of concomitant lesions of ligamentous injury and influence the precise operative technique for a certain patient. Ultrasonography and stress radiograph were also utilized in the diagnosis of ATFL injury. However, ultrasonography is of limited value in assessing bone or cartilage lesions and may be much less accurate in less experienced hands. Stress radiograph was reported to have a high rate of false negative results [15].

The previously reported diagnostic efficacy of MRI was listed in Table 3. Due to variable MRI parameters, gold standards, and injury types in individual study, the diagnostic efficacy of MRI varied largely. The sensitivity, specificity, and accuracy of MRI diagnosing ATFL injury were reported to be 0.5 to 1, 0.5 to 1, and 0.588 to 1 respectively, and the interobserver agreement was reported to be 0.4 to 0.939 (kappa value) [4‐6, 10, 16‐27]. The sensitivity, specificity, and accuracy of routine axial MRI scanning diagnosing ATFL injury in the current study were 0.65 to 0.85, 0.65 to 0.85, and 0.65 to 0.85. These results were comparable with other studies.

To better delineate a ligament, the MRI plane needs to be parallel to the long axis of the ligament. Previous anatomic study of ATFL delineates the course of ATFL from fibula to talus. The fibular origin of ATFL was reported to be 10 to 13.8 mm proximal to the tip of the fibula [28], which is approximately the midpoint between the inferior tip and anterior tubercle of the fibula. The talar insertion of ATFL was reported to be 14.2 to 18.1 mm to the subtalar joint or 11.3 to 14.8 mm to the anterolateral corner of the talar body [28], which is approximately the midpoint between the lateral talar process and the anterolateral corner of the trochlea. ATFL was reported to be directed at approximately 45° medially from the sagittal plane and at approximately 25° inferiorly from the horizontal plane [11, 12]. The oblique axial-coronal plane and oblique axial-sagittal plane utilized in the current study were parallel to the ATFL course in different planes.

The oblique axial-coronal plane was also referred to as the ATF view in previous studies [13]. Kim et al. reported the full length of 97.4% ATFL was delineated utilizing the ATF view [13]. In the current study, the accuracy of the oblique axial-coronal plane was significantly higher compared to the routine axial plane for readers 1 and 2 (reader 1: p = .018; reader 2: p = .005). The accuracy for reader 3 was also raised utilizing the oblique axial-coronal plane but with the numbers available, no significant difference could be detected. The sensitivity and specificity were also raised utilizing the oblique axial-coronal plane, but only sensitivity for reader 2 showed statistical significance (p = .008). The interobserver agreement of the oblique axial-coronal plane was excellent. In contrast, the interobserver agreement of the routine axial plane was fair to good.

Schneck et al. [7] reported ATFL has a slightly descending course and was best delineated in 10 to 20° dorsiflexion of the tested foot. This angle is slightly less than 25° oblique axial-coronal plane utilized in the current studies. The methods of repositioning the tested foot were also reported to be effective by Farooki et al. [8] However, the optimal position of the foot was not identical among studies [7, 8]. The discrepancy in the optimal angle might be due to anatomic variance. Compared to the methods of repositioning the tested foot and multiplanar reconstruction of images, oblique plane MRI scanning was more convenient in clinical practice. It only took a couple of minutes for the additional oblique plane MRI scanning of one specimen in the current study.

The oblique axial-sagittal plane was firstly proposed in the current study. The oblique axial-sagittal plane was parallel to the course of ATFL in the coronal plane. However, this method did not show enough improvement in reliability or validity from the routine axial plane. This might be due to the large variance for angles in the coronal plane. Moreover, ATFL was a flat quadrilateral ligament [28]. In planes 45° degree from the sagittal plane, ATFL might demonstrate the full length on fewer slices than 25° oblique planes.

The oblique axial-coronal plane could be added to routine MRI scanning protocol for better diagnosis of ATFL injury. However, the routine axial plane is still irreplaceable in assessing lesions of other structures. Kim et al. [13] reported anterior inferior tibiofibular ligament injury, posterior inferior tibiofibular ligament injury, and posterior tibialis tendinitis were better delineated on the routine axial plane compared to ATF view. Park et al. [29] reported better delineation of the calcaneofibular ligament on an oblique coronal plane.

One of the limitations of the current study is that we utilized cadaveric feet rather than clinical patients to investigate the diagnostic efficacy. Tear of ATFL in vivo is more irregular, including partial tear and tear with an avulsion fracture of fibular. We simulate the tear of ATFL by a cut in its middle part. This is the most common situation during the surgery. Exposing cadaveric ligament (ATFL) by surgically removing fat and connective tissue creates the condition not quite comparable to clinical cases. Removal of fat could be a clue that surgery was performed. We also did sham operations in the intact groups to decrease the influence of exposing the ligament. The MRI signal intensities of cadaveric ATFLs were not identical as the ATFLs in a physiological environment. We were not able to assess the conditions of ATFLs through the altered MRI signal intensities and through the contrast of synovial fluid. T2-weighted images were more common in clinical practice, because they delineate the ligaments better through the contrast of synovial fluid. However, in the current cadaveric study, T1-weighted images delineate the ligaments well (Fig. 2). Compared to acute ligament injury, the proper diagnosis of chronic ligament injury is more difficult. But it is difficult to study the MRI diagnosis of chronic ligament injury in the current cadaveric study.

However, we utilized cadaveric feet in the current study for two reasons. One is that the anatomic findings can be used as a gold standard when investigating diagnostic accuracy. This was more reliable than clinical diagnosis or arthroscopic findings as other studies mentioned. Some studies utilizing clinical findings as gold standards might have difficulties in determining whether ATFLs of “healthy volunteers” were intact or not, because they could only assess “healthy volunteers” clinically rather than through direct observation of the ligament. The other reason is that the tested feet in the intact group and injured group could be strictly matched. We scanned 20 intact feet and then scanned 20 identical artificially injured feet. The images were randomly numbered, and the orders were blind to the observers. Previous studies utilizing arthroscopic findings as gold standards might have difficulties in collecting adequate true negative patients, because most patients who underwent arthroscopic procedure were injured. This situation may cast bias on the results.

Another limitation is that this experiment included only 20 specimens. The number of specimens was relatively small. However, significantly higher accuracies were still observed for oblique axial-coronal MRI scanning (reader 1: p = .018; reader 2: p = .005), and the results of the three observers showed fair to excellent interobserver agreement.

Moreover, MRI is time-consuming and expensive. So, it is difficult to send all patients for MRI. Ultrasonography and stress radiograph were also utilized in the diagnosis of ATFL injury. Studies comparing diagnosing efficacy of different MRI scanning planes with ultrasonography and stress radiograph are warranted. However, MRI is irreplaceable when clinicians need to confirm or exclude the presence of concomitant osteochondral lesions with ligamentous injury. According to our results, the oblique axial-coronal plane could be added to the MRI scanning protocol during clinical practices to improve the diagnostic accuracy of ATFL injury.

The diagnostic accuracy of the oblique axial-coronal plane MRI was higher than the routine axial plane concerning ATFL injury, and the interobserver agreement was excellent. The oblique axial-coronal plane could be added to the MRI scanning protocol during clinical practices to improve the diagnostic accuracy of ATFL injury.