Usefulness of Unenhanced MRI and MR Arthrography of the Shoulder in Detection of Unstable Labral Tears

Transcription

1 Musculoskeletal Imaging Original Research Unenhanced MRI and MR rthrography for Unstable Labral Tears Musculoskeletal Imaging Original Research Thomas 1,2 T Keywords: labral tear, MRI, shoulder DOI: /JR Received December 8, 2014; accepted after revision May 6, NeuroSkeletal Imaging Institute (NSI), 255 North Sykes Creek Pkwy, Merritt Island, FL ddress correspondence to T. (tmageerad@cfl.rr.com). 2 Department of Radiology, University of Central Florida School of Medicine, Orlando, FL. JR 2015; 205: X/15/ merican Roentgen Ray Society Usefulness of Unenhanced MRI and MR rthrography of the Shoulder in Detection of Unstable Labral Tears OJECTIVE. Shoulder surgeons commonly intervene to repair unstable labral tears (tears that displace with patient movement). Surgeons can detect unstable tears at surgery. It is difficult to be certain if a tear is unstable from a static MR image. This study reports the comparative benefits of using unenhanced MRI and MR arthrography together to detect unstable labral tears. MTERILS ND METHODS. One hundred fifty consecutive unenhanced shoulder MRI and MR arthrography examinations performed on the same patients were reviewed retrospectively by consensus reading of two musculoskeletal radiologists. oth unenhanced MRI and MR arthrography were performed on each patient on the same day. Labral tears were assessed. It was also determined if there was any difference in position of the labral tear between unenhanced MR images and MR arthrograms. change in position of 4 mm or more between unenhanced MR images and MR arthrograms was considered indicative of an unstable tear. ll patients proceeded to arthroscopy. RESULTS. Of these 150 patients, 94 had superior labral anterior-to-posterior (SLP) tears, 53 had posterior labral tears, and 42 had anterior labral tears on MRI. ll lesions described on MRI were described on arthroscopy. Twenty-three SLP tears, 16 posterior labral tears, and 17 anterior labral tears showed a change in the position of the labral tear when comparing unenhanced MR images versus MR arthrograms of 4 mm or more. ll of these labral tears were considered unstable by the surgeon, and all of these patients had surgical tacking performed. The other labral tears showed motion on unenhanced MR images and MR arthrograms of less than 4 mm. ll of these tears except for three were considered stable on arthroscopic examination. There were five SLP tears, three anterior labral tears, and four posterior labral tears seen on arthroscopy that were not seen on unenhanced MRI or MR arthrography. The tears not seen on unenhanced MRI or MR arthrography were considered stable at arthroscopy. CONCLUSION. In this study, unenhanced MRI and MR arthrography of the shoulder was useful in diagnosing unstable labral tears in 23 patients with SLP tears, 16 patients with posterior labral tears, and 17 patients with anterior labral tears. This information was useful in surgical planning. Of 133 tears that moved less than 4 mm on unenhanced MR images and MR arthrograms, 130 were considered stable on arthroscopy. L abral tears are common injuries that often require surgery. In our practice, labral tears are commonly seen on MRI examination. High-field-strength MRI as well as increased awareness of labral tears has resulted in labral tears being commonly reported on MRI examinations. Very high sensitivity for detection of labral tears has been reported on both 1.5- and 3-T MRI [1, 2]. Labral tears reported on MRI examinations are often found to be stable at arthroscopy [3, 4]. Some of these stable tears do not require surgical intervention. Unstable labral tears are more likely to require surgical in- tervention than are stable tears [3, 4]. It is difficult to determine if a labral tear is unstable on the basis of a static MR image. Preoperative determination of whether a labral tear is stable or unstable would be of great benefit to the surgeon. For this reason, I sought to determine if a change in position of a labral tear between an unenhanced MR image compared and an MR arthrogram would be useful in determining if a labral tear was stable or unstable preoperatively. Materials and Methods One hundred fifty consecutive unenhanced MRI and MR arthrography examinations of the 1056 JR:205, November 2015

2 Unenhanced MRI and MR rthrography for Unstable Labral Tears C Fig year-old man with shoulder pain., T2-weighted axial MR image (TR/TE, 3250/55) shows anterior labral tear (arrow)., T1-weighted fat-saturated axial MR arthrogram (TR/TE, 677/12) shows anterior labral tear (arrow). There is motion of this anterior labral tear by more than 4 mm as compared with unenhanced MR image. This was found to be unstable tear at arthroscopy and was surgically tacked. C, Line on T1-weighted fat-saturated axial MR arthrogram (TR/TE, 677/12) shows how labral displacement, here due to anterior labral tear, was measured on PCS. shoulder performed between June 2013 and January 2014 on patients presenting with shoulder pain who subsequently underwent arthroscopy were retrospectively read in consensus by two musculoskeletal radiologists. ll patients were referred from one orthopedic group. The age range of the 150 patients was years (mean, 33 years). Postoperative shoulder patients were excluded from this study. One hundred thirty-one male patients and 19 female patients were included. ll patients underwent surgery within 51 days of the MRI examination (range, 1 51 days; mean time between MRI examination and surgery, 21 days). ll patients had prospective MR image and MR arthrogram readings performed by one of three musculoskeletal radiologists with more than 10 years experience. Institutional review board approval was obtained for this retrospective review and correlation with surgical records [5]. ll patients underwent shoulder MRI in oblique coronal, oblique sagittal, and axial planes on a 3.0-T scanner (Signa, GE Healthcare). Oblique coronal and sagittal fast spin-echo T1-weighted (TR/TE, 550/10; number of excitations, 2), oblique coronal and sagittal fast spinecho fat-saturated T2-weighted (TR/TE, 3850/55; number of excitations, 4), and fast spin-echo fatsaturated proton density weighted axial (TR/TE, 3250/55; number of excitations, 3) sequences were used, with an FOV of 12 cm on all images. Slice thickness was 4 mm with a 10% interslice gap on all sequences except for the fast spin-echo proton density weighted axial sequence, which had a 3-mm slice thickness. The echo-train length was 10 on all T2-weighted and proton density weighted sequences and 3 on the T1-weighted sequences. The bandwidth was khz on all sequences. The imaging time was 4 minutes 43 seconds for the oblique coronal and sagittal T2-weighted sequences; 3 minutes 26 seconds for the proton density axial sequences; and 2 minutes 28 seconds for the T1-weighted sequences. The matrix was for all T2-weighted sequences and for all T1-weighted sequences. 3-channel phased-array shoulder coil was used [5]. ll 150 patients also underwent MR arthrography immediately after the unenhanced MRI examination. t our institution, my colleagues and I routinely perform both unenhanced MRI and MR arthrography. MR arthrography was performed with approximately 15 ml of a mixture of Mag nevist (ayer HealthCare) and saline, diluted to a concentration of 0.15 ml of Magnevist per 20 ml of normal saline. 22-gauge needle was placed in the glenohumeral joint via an anterior approach under fluoroscopic guidance. The injection was performed by one of three musculoskeletal radiologists. The shoulder was exercised after injection, and T1-weighted (TR/TE, 677/12) fat-saturated oblique coronal, oblique sagittal, and axial images were obtained. T1-weighted (TR/TE, 677/12) fat-saturated oblique coronal images were obtained before MR arthrography for direct comparison with images after MR arthrography. The imaging time for the T1-weighted sequences was 2 minutes 28 seconds. ll MR arthrograms were acquired with a 12-cm FOV and a 4-mm slice thickness (with a 10% interslice gap) [5]. ll 150 MRI examinations were retrospectively reviewed by consensus reading of two musculoskel- Fig year-old man with shoulder pain., T2-weighted coronal MR image (TR/TE, 3250/55) shows superior labral anterior-to-posterior (SLP) tear (arrow)., T1-weighted fat-saturated coronal MR arthrogram (TR/TE, 677/12) shows SLP tear (arrow). There is motion of this SLP tear by more than 4 mm as compared with unenhanced MR image. This was found to be unstable tear at arthroscopy and was surgically tacked. JR:205, November

3 Fig year-old woman with shoulder pain., T2-weighted axial MR image (TR/TE, 3250/55) shows posterior labral tear (arrow)., T1-weighted fat-saturated axial MR arthrogram (TR/TE, 677/12) shows posterior labral tear (arrow). There is motion of this posterior labral tear by more than 4 mm as compared with unenhanced MR image. This was found to be unstable tear at arthroscopy and was surgically tacked. etal radiologists, each with more than 10 years of experience reading musculoskeletal MR images. The readers were blinded to arthroscopic findings at the time of consensus review. Retrospective MR image readings were then correlated with arthroscopic results (n = 150). The surgeons were aware of prospective MR image interpretations before the patients underwent arthroscopy. There was no reporting of labral stability or instability on prospective MR image readings. The unenhanced MR images and MR arthrograms were reviewed together. The examinations were read in two separate sessions with unenhanced MR images and MR arthrograms for each patient compared directly. Reviewers were blinded to names of patients, prospective MRI reports, and arthroscopy results. Scans were assessed for superior labral anterior-to-posterior (SLP) tears and anterior or posterior labral tears. MRI criterion used for diagnosis of labral tear was a morphologic abnormality of the glenoid labrum, increased signal intensity, or both. Labral tears were diagnosed on MRI examinations when there was high signal intensity on T2-weighted images in the labrum, an irregular or detached labrum, or both. SLP tear was defined as superior labral irregularity, high signal intensity on T2-weighted images within the superior labrum, or both [6]. SLP tears that extended anteriorly were defined as SLP tears, not as anterior labral tears. n anterior or posterior labral tear was defined as an area of abnormal increased signal intensity on T2-weighted images in the labrum, an irregular or detached labrum, or both. MR arthrography criterion for a labral tear also included high signal intensity within the labrum owing to gadolinium extending into the labrum on T1 fat-saturated images. The images were reviewed for the presence or absence of anterior labral, posterior labral, or SLP tears and were not graded. That is, on retrospective review, images were not graded for highor low-grade SLP tears or detached or nondetached labral tears, but rather for whether the lesions were present or absent. It was also determined whether there was any difference in position of the labral tear between unenhanced MR images and MR arthrograms. change in position of 4 mm or more between unenhanced MR images and MR arthrograms was considered indicative of an unstable tear. The 4-mm cutoff was determined by a previous analysis of 20 patients who subsequently underwent surgery. Unenhanced MRI and MR arthrography examinations were correlated with arthroscopy results. In this pilot study, all patients who had labral displacement of 4 mm or more between unenhanced MR images and MR arthrograms were found to have unstable labral tears at arthroscopy. The 4-mm measurement was acquired in either the axial or oblique coronal plane on a PCS workstation. The axial plane was used for anterior and posterior labral tears, and the oblique coronal plane was used for SLP tears. ll patients proceeded to arthroscopy. MR arthrograms alone were assessed in a separate reading session to determine if unenhanced MR images were needed to assess if a labral tear was unstable. MR arthrograms were assessed for displacement of 4 mm or more. Tears with displacement of 4 mm or more on MR arthrograms were then correlated with arthroscopy to determine if this criterion was useful in detection of unstable labral tears. Results Of these 150 patients, 94 had SLP tears, 53 had posterior labral tears, and 42 had anterior labral tears on MRI. ll lesions seen at MRI were seen at arthroscopy. Twentythree SLP tears, 16 posterior labral tears, and 17 anterior labral tears showed a change in the position of the labral tear when comparing unenhanced MR images versus MR arthrograms of 4 mm or more. ll of these labral tears were considered unstable by the surgeon, and all of these patients had surgical tacking performed. The other labral tears showed motion between unenhanced MR images and MR arthrograms of less than 4 mm. ll of these tears except for three were Fig year-old man with shoulder pain., T2-weighted coronal MR image (TR/TE, 3250/55) shows superior labral anterior-to-posterior (SLP) tear (arrow)., T1-weighted fat-saturated coronal MR arthrogram (TR/TE, 677/12) shows SLP tear (arrow). There is no motion of this SLP tear when comparing unenhanced MR images versus MR arthrograms. This was found to be stable tear at arthroscopy and was débrided JR:205, November 2015

4 TLE 1: Sensitivity and Specificity of Unenhanced MRI and MR rthrography for Detection of Unstable Labral Tears Versus MR rthrography Unenhanced Only MRI and MR rthrography for Unstable Labral Tears Examination(s) True-Positives False-Positives True-Negatives False-Negatives Sensitivity (%) Specificity (%) oth unenhanced MRI and MR arthrography MR arthrography only a a Displaced by 4 mm. considered stable at arthroscopy. Of 133 tears that moved less than 4 mm between unenhanced MRI and MR arthrography, 130 were considered stable at arthroscopy (see Figs. 1 5). There were five SLP tears, three anterior labral tears, and four posterior labral tears seen at arthroscopy that were not seen on unenhanced MR images or MR arthrograms. The tears not seen on unenhanced MR images or MR arthrograms were considered stable at arthroscopy. The sensitivity and specificity for detection of an unstable labral tear by using the criterion of displacement of 4 mm or more comparing unenhanced MR images and MR arthrograms were 95% and 100%, respectively (Table 1). n additional review of MR arthrography only was performed to assess whether displacement of the labrum by 4 mm or more was accurate in determining if a labral tear was unstable. The 23 SLP tears, 16 posterior labral tears, and 17 anterior labral tears that showed displacement by 4 mm or more compared with unenhanced MRI examination were all found to be unstable at arthroscopy. n additional eight SLP tears, six posterior labral tears, and five anterior labral tears were displaced by 4 mm or more on MR arthrograms that showed minimal or no difference in position compared with the unenhanced MR images. One of these SLP tears and one of these anterior labral tears were found to be unstable at arthroscopy. The sensitivity and specificity for detection of an unstable labral tear by using the criterion of displacement of 4 mm or more on MR arthrography only were 98% and 88%, respectively (Table 1). Discussion There were a large number of labral tears in this study. Most of the labral tears were seen on both unenhanced MRI and MR arthrography. The main use of MR arthrography in this study was to determine if a tear was stable or unstable. Unstable tears are more likely to result in clinical symptoms for the patient and often require surgical tacking. Stable tears are less likely to cause clinical symptoms, usually do not require surgical tacking, and often warrant no surgical intervention at all [3, 4]. It is difficult for the surgeon to determine if a labral tear is stable or unstable just on physical examination. t arthroscopy, the surgeon can determine if the tear is unstable by probing the tear to see if it moves. Movement with probing is indicative of an unstable tear [3, 4]. In this study, very accurate prediction of unstable labral tears was achieved by using the criterion of a change in the position of the labral tear when comparing unenhanced MR images versus MR arthrograms of 4 mm or more. ll 23 SLP tears, 16 posterior labral tears, and 17 anterior labral tears that showed a change in the position of the labral tear when comparing unenhanced MR images versus MR arthrograms of 4 mm or more were considered unstable by the surgeon, and all of these patients had surgical tacking performed. Of the 133 labral tears that showed motion on unenhanced MRI and MR arthrography of less than 4 mm, 130 were considered stable at arthroscopy. The tears were either débrided or no surgical intervention was performed on them. Three of these 133 labral tears with motion of less than 4 mm when comparing unenhanced MR images versus MR arthrograms were considered unstable at arthroscopy, and surgical tacking was subsequently performed. previous study by Modi et al. [7] showed that abducted and externally rotated (ER) positioning after arthrography improved accuracy in diagnosis of labral tears. In Modi et al. s study, it was suggested that ER positioning be used routinely when ordering MR arthrography examinations to increase diagnostic accuracy for labral tears. This study did not address the use of ER positioning to determine if a tear was stable or unstable [7]. (ER positioning has elsewhere been shown to be useful in detection of some partial thickness supraspinatus tendon tears and some anterior labral tears [8, 9].) nother study, by Marcus et al. [10], showed that MRI examinations performed with the arm in internal rotation increased accuracy in diagnosis of anterior labral tears. Their study did not assess whether the tears were stable or unstable [10]. third study, by Chiavaras et al. [11], assessed use of flexion, adduction, and internal rotation positioning as an adjunct to unenhanced MR arthrography. Chiavaras et al. found that flexion, adduction, and internal rotation positioning was useful in detecting additional posteroinferior labral abnormalities as compared with unenhanced MR arthrographic sequences. Their study did not address whether the labral tears were stable or unstable [11]. The aforementioned three studies [7, 10, 11] made use of changing arm position to highlight labral tears. In the present study, I compared unenhanced MR versus MR arthrography to determine if labral tears would change position when the joint was distend- Fig year-old man with shoulder pain., T2-weighted axial MR image (TR/TE, 3250/55) shows posterior labral tear (arrow)., T1-weighted fat-saturated axial MR arthrogram (TR/TE, 677/12) shows posterior labral tear (arrow). There is motion of this posterior labral tear by less than 4 mm as compared with unenhanced MR image. This was found to be unstable tear at arthroscopy and was surgically tacked. JR:205, November

5 ed with intraarticular contrast. In our practice, my colleagues and I do not routinely use ER positioning because many of our patients do not feel comfortable in that position. In a previous surgical study by Hurley and nderson [3], one hundred shoulder arthroscopies were reviewed. In Hurley and nderson s study, most labral lesions found at arthroscopy were not associated with instability; many of these lesions were not treated surgically. In another surgical study by Glasgow et al. [4], 72% of labral injuries noted at arthroscopy were not associated with instability; again, many of these lesions were not treated surgically. The aforementioned studies [3, 4] in the surgical literature describe a large number of labral lesions seen at arthroscopy that were not felt to be associated with instability. In our practice, labral tears are commonly diagnosed at MRI. High-field-strength MRI allows very sensitive detection of such tears. Many of these tears are stable and may not be considered clinically significant to the surgeon [12]. In most cases, anterior labral tears were tacked surgically. nterior labral tears often occur secondary to anterior dislocations in a younger population. nterior labral tears are often unstable at arthroscopy [12]. Posterior labral tears and SLP tears are often degenerative in nature and stable at arthroscopy. These tears occur in all age groups. t our institution, the surgeons tack these lesions when they are found to be unstable at arthroscopy [12]. change in position of 4 mm or more of a labral tear on unenhanced MRI versus MR arthrography was used as a cutoff to show movement of the labral tear. The reason for using 4 mm as a cutoff is that this degree of motion shows a definite change in labral tear position. Technologists at our institution are instructed to position the patient exactly the same for unenhanced MRI and MR arthrography; nevertheless, there remains a possibility that the patient could be slightly differently positioned. This cutoff of 4 mm allows some variation in position on unenhanced MR images and MR arthrograms, therefore enabling a definitive diagnosis of motion. ll patients with motion of 4 mm or more between unenhanced MR images and MR arthrograms had unstable labral tears at arthroscopy that required surgical tacking. In the present study, there were three patients with motion of labral tears when comparing unenhanced MR images versus MR arthrograms of less than 4 mm who had unstable labral tears at arthroscopy. Of 133 tears that moved less than 4 mm between unenhanced MR images and MR arthrograms, 130 were considered stable at arthroscopy. The criterion of 4 mm or more displacement on MR arthrogram images only (i.e., without assessment of unenhanced MR images) yielded excellent sensitivity and good specificity for detection of unstable labral tears (sensitivity, 98%; specificity, 88%) (Table 1). The criterion of displacement of 4 mm or more when comparing unenhanced MR images versus MR arthrograms yielded excellent sensitivity and specificity for detection of unstable labral tears (sensitivity, 95%; specificity, 100%) (Table 1). Imaging times for performing both unenhanced MRI and MR arthrography versus MR arthrography only are the same because the same sequences are used. The total imaging time is approximately 21 minutes. The additional time needed for both unenhanced MRI and MR arthrography versus MR arthrography only is the time required in order to move the patient on and off the table. There is a slight additional cost to perform both unenhanced MRI and MR arthrography versus MR arthrography only. Current Medicare reimbursement rates in the United States are $ for only shoulder MR arthrography versus $ for both unenhanced MRI and MR arthrography of the shoulder. In our practice, my colleagues and I routinely perform both unenhanced MRI and MR arthrography because we believe that doing so allows better analysis of joint effusion and better analysis of labral tears. The unenhanced portion of the examination provides important diagnostic information for those patients who cannot complete the arthrogram portion of the examination. The unenhanced portion of the examination is also useful in postoperative patients to determine if there is significant metallic artifact on MRI. This can preclude MR arthrography in some cases. In those cases, we proceed to CT arthrography instead. There are limitations to this study. This study included only those patients who subsequently underwent surgery. This could have introduced bias with respect to the patient population. The criterion of 4 mm of labral displacement for instability has not been previously scientifically proved as a cutoff number in the literature. The use of 4 mm is based on a pilot study of 20 patients. The study design involved consensus reading rather than independent reading. In conclusion, comparison of unenhanced MR images and MR arthrograms can be useful in determining if a labral tear is stable or unstable preoperatively. Labral tear motion of 4 mm or more at MR arthrography allowed detection of all but three unstable labral tears. There were no stable labral tears that showed motion of 4 mm or more when comparing unenhanced MRI versus MR arthrography. Use of unenhanced MR images and MR arthrograms together to determine if there is labral tear motion at MR arthrography helps the surgeon determine if a labral tear is stable or unstable preoperatively. This may aid presurgical planning. References 1. TH, Williams D. Sensitivity and specificity in detection of labral tears with 3.0 T MRI of the shoulder. JR 2006; 187: Gusmer P, Potter HG, Schatz J, et al. Labral injuries: accuracy of detection with unenhanced MR imaging of the shoulder. Radiology 1996; 200: Hurley J, nderson TE. Shoulder arthroscopy: its role in evaluating shoulder disorders in the athlete. m J Sports Med 1990; 18: Glasgow SG, ruce R, Yacibyccu GN, Torq JS. rthroscopic resection of glenoid labral tears in the athlete: a report of 29 cases. rthroscopy 1992; 8: T. 3-T MRI of the shoulder: is MR arthrography necessary? JR 2009; 192: Tuite MJ, Cirillo RL, DeSmet, et al. Superior labrum anterior-posterior (SLP) tears: evaluation of three MR signs on T2-weighted images. Radiology 2000; 215: Modi CS, Karthikeyan S, Marks, et al. ccuracy of abduction-external rotation MR versus standard MR in the diagnosis of intra-articular shoulder pathology. Orthopedics 2013; 36:e337 e Saleem M, Lee JK, Novak LM. Usefulness of the abduction and external rotation views in shoulder MR arthrography. JR 2008; 191: Cvitanic O, Tirman PF, Feller JF, et al. Using abduction and external rotation of the shoulder to increase the sensitivity of MR arthrography in revealing tears of the anterior glenoid labrum. JR 1997; 169: Marcus M, Malhotra, Peri M, et al. Kinematic shoulder MRI: the diagnostic value in acute shoulder dislocations. Eur Radiol 2013; 23: Chiavaras MM, Harish S, urr J. MR arthrographic assessment of suspected posteroinferior labral lesions using flexion, adduction, and internal rotation positioning of the arm: preliminary experience. Skeletal Radiol 2010; 39: T. How often do surgeons intervene on shoulder labral lesions detected at MR examination? retrospective review of MR examinations correlated with arthroscopy. r J Radiol 2014; 87: JR:205, November 2015