Role of Magnetic Resonance Arthrography (MRA) in the Diagnosis of Shoulder Injuries

Abstract

Background:

Imaging continues to have a crucial role in evaluating patients with shoulder pain, helping to make treatment choices. Magnetic resonance arthrography (MRA), rather than magnetic resonance imaging (MRI), is now routinely used to diagnose shoulder injuries. Against the gold standard investigation of arthroscopy, the study aimed to determine the accuracy of MRA in the evaluation of shoulder injuries.

Materials and Methods:

From June 2017 to May 2019, a cross-sectional study was conducted in a tertiary care health setup in North India. A total of 50 individuals were enrolled for diagnostic imaging with rotator cuff (RC) pathology. Patients were assessed using an MRA and their clinical history. In comparison with the gold standard for arthroscopy, the sensitivity (Sn), positive predictive value (PPV), and diagnostic accuracy (DA) of MRA for diagnosing shoulder injuries were evaluated. A P value under 0.05 was taken to be statistically significant.

Results:

The mean ± standard deviation (SD) age of the patients was 37.32 ± 12.78 years. There were seven women and 43 (86%) men. The different lesions were superior labral anterior–posterior (SLAP) tear (24 (48%)), Bankart lesions (19 (38%)), anterior labral periosteal sleeve avulsion (ALSPA) (4 (8%)), partial articular supraspinatus tendon avulsion (PASTA) lesion (2 (4%)), and Buford lesions (1 (2%)). The Sn, PPV, and DA of MRA were 98%, 100%, and 98%, respectively, for identifying the shoulder injury.

Conclusion:

It was concluded that MRA was reasonably accurate at diagnosing shoulder injuries. MRA demonstrated an excellent 98% Sn in identifying shoulder injuries and missed only one case of shoulder injury, suggesting that it can be an advantageous imaging tool for recognizing and evaluating shoulder injuries for immediate diagnosis and management.

INTRODUCTION

The anatomy of the shoulder joint is complex with different types of injuries involved. The most prevalent group of shoulder pathologies are rotator cuff (RC) tendon disorders that are multifactorial in origin and may cause progressive degeneration of the RC tendons and significant functional restrictions, necessitating urgent diagnosis and treatment.[1] The diagnosis and treatment can be difficult due to the shoulder joint’s complex anatomy.[2]

Imaging continues to have a crucial role in evaluating patients with shoulder pain, helping to make treatment choices.[1] Whether a patient with an RC tear may be managed conservatively or requires surgery depends on the extent of the tear, which includes whether it is full or partial thickness.[1,2] Once the proper diagnosis has been made, the surgical approach, open or arthroscopic, can be chosen.[3] There are many imaging tests, which have been used to evaluate the painful shoulder. Among them, unenhanced magnetic resonance imaging (MRI), indirect and direct magnetic resonance arthrography (MRA), and ultrasound have become the standards by which an RC tear is diagnosed. Selecting an open or arthroscopic surgical method is possible once the correct diagnosis has been determined. Numerous imaging tests have been employed to assess the painful shoulder. Among these, ultrasonography, direct and indirect MRA, and unenhanced MRI have emerged as the standards for diagnosing RC tears.[4,5,6]

MRA, rather than MRI, is now routinely used to diagnose shoulder injuries.[7] Additionally, MRA is regarded as the gold standard for determining instability and doing preoperative evaluations for shoulder injuries.[8,9] The extra-articular surface of the RC and intrasubstance ligamentous lesions are better evaluated with MRA. It can document abnormalities of the cuff muscles and surrounding structures, show the extent and configuration of RC abnormalities, and indicate mechanical imbalances in the cuff.[10,11] A meta-analysis showed that the sensitivity (Sn) and specificity of 3-Tesla (3T)-MRA for diagnosing shoulder injuries vary from 88% to 100% and 90% to 100%, respectively.[12] However, 3T holds its own disadvantages as compared to 1.5T in terms of time taken and the magnetic field.

Thus, we conducted this study to know whether 1.5T0 MRA carries the feasibility of application as a diagnostic imaging modality for shoulder injuries in the Indian population. This study aimed at determining the Sn, positive predictive value (PPV), and diagnostic accuracy (DA) of MRA for diagnosing different types of shoulder injuries, considering arthroscopy as the gold standard. The novelty of this study is that it shall provide data for the accuracy of 1.5T-MRA in use as an imaging technique against arthroscopy (invasive method), 3T-MRA, and MRI for evaluating shoulder injuries. Moreover, research published till now is skewed in favor of positive findings and the use of 3T-MRA or MRI, which may not be the best method, considering the advantages of MRA in screening and evaluating the intricate and complex anatomy of ligament tears in the shoulder joint.

MATERIALS AND METHODS

This cross-sectional study was conducted at the Department of Radiodiagnosis at a tertiary health care setup in North India for a 24-month duration beginning from June 30, 2017, to May 31, 2019. Patients coming to the department with RC injuries (shoulder injuries) with signs and symptoms of pain or restriction in the movement or trauma were screened for eligibility.

Inclusion criteria

• Adult patients (>18 years).

• Both males and females.

• Patients in whom RC pathologies, including acute and chronic, were clinically suspected.

• Patients with an indication of MRA for shoulder injury. The indications of MRA were cases of suspected shoulder injuries—to get confirmation of the tears in suspected or equivocal cases and to get additional information in already diagnosed cases of RC tears, such as specific tears in other tendons or muscles, condition of ligaments differentiating between fractures and contusion in bones, who had a history of shoulder injury with positive findings on MRI, or cases where there was a discrepancy between clinical and MRI findings.

Exclusion criteria

• Patients who had any type of implant, such as cochlear, metallic, or cardiac pacemaker

• Posttreatment or post-surgery patients

• Claustrophobic patients and those who declined to participate in the study

• Pregnant patients.

Sample size

The sample size was based on the values of El-Liethy et al.,[13] the study where it was found that MRA demonstrated Sn of 78% and specificity of 100% to diagnose shoulder injuries. With the above numbers as a guide, the minimal sample size needed to achieve the desired accuracy of 20%, research power of 90%, and level of significance of 5% was 38 patients. Fifty samples were taken in total to lower the margin of error.

Methodology

Before beginning the study, certification and clearance from the institute’s Ethical Committee were acquired. The study comprised all eligible participants who were referred to the Department of Radiodiagnosis and met the established inclusion and exclusion criteria. Informed consent was acquired from patients before enrollment. The study pro forma contained information on demographic, clinical, and imaging parameters.

Imaging and clinical history were recorded. The features of various RC disorders were explained. For examination, the patient was placed in a supine posture and had their arm at their side in either a neutral or slightly external rotation. The GE-Wipro Signa HDxt 1.5T MRI Scanner was used to do the MRI. Images of the axial, coronal, and sagittal T1, T2, proton density (PD), and STIR were obtained. The patients were subjected to MRA and diagnostic arthroscopy as per the standard procedure.

MRI was interpreted by one radiologist with 4 years of clinical experience, and MRA was interpreted by another radiologist with 7 years of clinical experience. A third radiologist was taking records of positive cases of MRI and allotting these cases for MRA. The first and second radiologists were unaware of the finding of each other. The third radiologist was only aware of the findings of the MRI.

For performing MRA, approximately 0.2 milliliters of gadolinium-based contrast was diluted in 20 milliliters of sterile saline and combined with approximately 5 milliliters of iodine-based contrast. Once the joint capsule was adequately distended, about 10 ml of solution was administered slowly. The injection site was covered with dressing when the needle was taken out.

Arthroscopy was performed in all those cases who underwent MRA by an orthopedician (Orthopedics Department) independently without knowing the clinical findings of MRA. Outcome measures were Sn, PPV, and DA of MRA for diagnosing shoulder injuries against the gold standard shoulder arthroscopy.

Statistical analysis

Number and percentage (%) were used to represent categorical variables, whereas mean ± standard deviation (SD) and median were used to represent continuous variables. The diagnostic test was used to determine the MRA’s Sn, PPV, and DA. A P value of less than 0.05 was deemed significant. The Statistical Package for Social Sciences (SPSS) version 21.0, IBM, Chicago, USA, was used to conduct the analysis after the data were imported into a Microsoft (MS) Excel spreadsheet.

RESULTS

The study considered 65 patients who had shoulder injury to be screened for this study, among whom eight did not need MRA and seven did not give consent for the study. Fifty patients who underwent MRA and shoulder arthroscopy were enrolled. The patients’ mean ± SD age was 37.32 ± 12.78 years, with most (30%) patients belonging to the age group of 31 to 40 years and 28% patients from 21 to 30 years. Most of the patients (43 (86%)) were males.

Recurrent and primary shoulder injuries were present in 66% and 34% of patients, respectively. Superior labral anterior–posterior (SLAP) tear was the most common pathology in 24 (48%) patients, with other being Bankart lesion (19 (38%)), anterior labral periosteal sleeve avulsion (ALSPA) (4 (8%)), partial articular supraspinatus tendon avulsion (PASTA) lesion (2 (4%)), and Buford lesion (1 (2%)). It is shown in Figure 1.

Figure 1.

Distribution of pathology of study subjects

MRA was positive in 49 (98%) patients, where it missed a single case of PASTA lesion. To diagnose shoulder injury, MRA demonstrated Sn of 98%, PPV of 100%, and DA of 98%. In particular, for individual lesions, MRA had DA of 100% for the Bankart lesion, SLAP tear, ALSPA lesion, and Buford lesion and 50% for the PASTA lesion [Table 1].

Table 1.

Diagnostic power of MRA in patients with shoulder injury

Injuries	Sensitivity	PPV	DA
Shoulder injury	98%	100%	98%
SLAP	100%	100%	100%
Bankart	100%	100%	100%
ALSPA	100%	100%	100%
PASTA	50%	50%	50%
Buford	100%	100%	100%

A single case of PASTA lesion was false negative on MRA

DISCUSSION

The preferred approach for diagnosing glenohumeral instability is MRA, which offers superior delineation of associated intra-articular lesions, differentiation of complex anatomic components of the joint, and presentation of subtle abnormalities. As contrast solution causes distension of joint capsule, highlights intra-articular structures, and penetrates abnormal regions, MRA enhances the potential of conventional MRI. Any joint that undergoes conventional arthroscopy can undergo MRA.[14]

MRA was found to be positive among 49 (98%) patients in the present study. To diagnose the shoulder injury, MRA showed Sn of 98%, PPV of 100%, and DA of 98%.

As for the different types of injuries, SLAP tear was the most common injury reported in our study. It is an injury to the superior labrum, which runs from anterior to posterior within the superior labrum. Based on the type of lesion present, they are divided into four different groups, emphasizing that this lesion can hinder the long head of the bicep brachia’s origin.[14,15] In our study, to diagnose SLAP tear, Sn, PPV, and DA were 100%.

Apart from SLAP tears, Bankart lesions were the second most common in our study, which refer to injuries, particularly at the anteroinferior aspect of the glenoid labral complex. These lesions are a frequent complication of anterior shoulder dislocation.[15] For diagnosing BANKART lesions, in our study, the Sn, PPV, and DA were 100%. In our study, MRA missed a single case of PASTA lesion. In comparison, Magee[9] reported that for the diagnosis of shoulder injuries, the Sn of MRA in comparison with arthroscopy for anterior labral tear was 98%, for posterior labral tear was 95%, for SLAP tear was 98%, for supraspinatus tendon tear was 100%, for partial-thickness articular surface tear was 97%, and for partial-thickness bursal surface tear was 84%. The specificity of MRA for all these tears was 100%, except for the SLAP tear, which was 99%. The study reported that 3T-MRA demonstrated a statistically significant improvement in Sn for the identification of anterior labral, partial-thickness supraspinatus, and SLAP tears (P < 0.05).

El-Liethy et al.[13] mentioned that MRA had Sn, specificity, PPV, and negative predictive value of 78%, 100%, 100%, and 21%, respectively. In a study by Stendel et al.,[16] Sn, specificity, and DA for detecting shoulder injuries involving the articular surface of the tendon were 72%, 97%, and 86% on MRA. In a study by Stendel et al.,[16] Sn, specificity, and DA for detecting tears involving the bursal surface of the tendon were 81%, 97%, and 89% on MRA.

de Jesus et al.[3] found that Sn and specificity of partial RC tear were 91.7% and 96.5% on MRA, respectively. McGarvey et al.[12] reported that concerning partial-thickness supraspinatus tears, Sn and specificity of 3T 2D MRA were 86.6% and 95.2%, respectively. The findings of these studies and our study show the usefulness in terms of the accuracy of MRA for diagnosing shoulder injuries. However, the DA varies in the range of 86–98% and Sn in the range of 72–100%. Notwithstanding, the specificity remains high in the range of 95–100%, allowing for better ruling out of any specific tendon or ligament injury in the suspected cases of RC pathology, thus allowing for a better diagnosis and management.

In the current medical practice, the increasing evidence of the good accuracy of MRA has made it more popular and widely applicable as a diagnostic modality, specifically for shoulder injuries, as it provides excellent contrast-based delineation of the structures and proper differentiation of the injury.[11] However, its use demands certain restrictions and holds disadvantages, such as need for fluoroscopy (depending on patient tolerability) and a longer examination time.[17,18] Still, its applicability rests on the individual cases and the setup and the findings of the MRI; its efficacy must be taken into account.[18]

Strengths of the study

The study holds strength as the use of MRA for the evaluation of shoulder injuries is relatively less invasive than arthroscopy, more informative, and can be used at 1.5Tmachine, and this reduces the dependency on the 3T machine, which has more acquisition time. Moreover, it has easy availability and cost-effectiveness, allowing for wider use in developing countries.

Limitations

This study was conducted in a setting that caters to patients mainly from lower or middle socioeconomic status, and the results clearly illustrate the particulars of this study population. The reliability of diagnostic procedures for RC and labral injuries is not reported in this study.

As this study was a single-center, hospital-based investigation, its conclusions cannot be used to investigate the diagnostic assessment of shoulder injuries in all types of populations.

As ultrasonography is an evolving type of musculoskeletal imaging technique that has recently demonstrated potential accuracy in the diagnosis of RC injuries, the results of MRI and MRA were not compared with ultrasonography.

There was no control case without any shoulder injury and thus no true negative, because of which specificity and negative predictive value could not be assessed.

Confounding factors, such as history of trauma, body mass index, and the 1.5T vs 3T-MRA, were not taken into account.

CONCLUSION

MRA was found to be accurate in the diagnosis of shoulder injuries, against arthroscopy as the gold standard. MRA showed a negative finding in only one patient with a PASTA lesion. MRA showed excellent Sn, that is, 98%, in detecting shoulder injuries. MRA was 100% sensitive in the diagnosis of Bankart lesions, SLAP tears, ALSPA, and Buford lesions. Based on the results of the study, it is recommended that MRA be advised and can be used as an additional novel imaging method for the evaluation of cases of shoulder injury for confirming the suspected cases, for better management.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.