Abstract
Objectives:
It is estimated that up to 2% of the general population will experience a glenohumeral joint dislocation in their lifetime, with active adolescents and young adults making up a large proportion of those events. When encountered by an orthopaedic surgeon, patients with shoulder instability typically undergo MRa or MRI for diagnosis and to aid in surgical planning. Recently, MRa has been the advanced imaging of choice as it is considered superior to MRI in detecting labral pathology, although both are still used in clinical practice, and with limitations to accurately detect labral pathology. Thus, shoulder arthroscopy remains the gold standard for diagnosing labral pathology. In regards specifically to the pediatric population, relatively few studies exist in determining the sensitivity and specificity of MRa/MRI compared to arthroscopy in detecting anterior and SLAP lesions in cases of shoulder instability. To our knowledge, there are no studies that assess the ability of MRa/MRI in detecting posterior labral pathology in this specific cohort. The purpose of this study was to evaluate the effectiveness of MRa and MRI as diagnostic tools prior to shoulder stabilization surgery in the pediatric population.
Methods:
This is a retrospective review of data obtained between January 1, 2012 to June 30, 2022 from the IRB-approved Shoulder Stabilization Study. Query CPT codes to identify patients treated with shoulder stabilization surgery included both arthroscopy CPT codes (29806, 29807, and 29999) and open procedure CPT codes (23450, 23455, 23460, 23462, 23065, 23066, 23929). Inclusion criteria were any patient treated with shoulder stabilization surgery with an MRa or MRI obtained at a single institution. Exclusion criteria were greater than six months between imaging and surgery, MRa or MRI reports that were inconclusive due to motion, prior ipsilateral shoulder stabilization surgery, or charts with incomplete or insufficient data. Data collected and reviewed included description of the labral pathology identified from pertinent imaging reports from non-MSK focused radiologists and operative notes from one of three board certified orthopaedic surgeons at a single institution. Demographics and office notes were also analyzed. Lesion type was classified based on location of tear using conventional clock face nomenclature with lesions defined as follows regardless of laterality: Superior 10-2 o’clock, Anterior 3-6 o’clock, Posterior 6-10 o’clock. For patients that had two or more lesions, each lesion was counted separately. The use of the term fraying was classified as negative whereas the use of the terms “maybe” or “possible” was considered positive. A tear that partially extended into another region as defined by the clock face classification was considered positive only for the primary region. Outcome measures included sensitivity and specificity. For each imaging method, the sample size for sensitivity was the number of positives according to the arthroscopic findings which was considered the “gold standard” (TP + FN); for specificity, it was the number of negatives according to the “gold standard” (TN + FP).
Results:
340 cases met inclusion criteria (297 MRa, 43 MRI). Average age at surgery was 17 ±1.8 (10-25 years old). 350 lesions were identified on arthroscopy: 200 (57%) anterior; 90 (26%) posterior; 60 (17%) SLAP. See the attached table for a comparison of sensitivity and specificity of MRa versus MRI in identifying anterior, posterior, and SLAP lesions relative to the arthroscopic findings as the “gold standard” (Table I). When comparing between MRa versus MRI, the only statistically significant differences in sensitivity or specificity were for SLAP lesions. Sensitivity was greater for MRa compared to MRI (74% vs. 29%, respectively; p=0.028). However, the specificity for a SLAP lesion was greater for MRI compared to MRa (97% vs. 84%, respectively; p=0.039). When comparing all imaging reports to the “gold standard” arthroscopic findings, the imaging was correct in fully diagnosing the integrity of the glenoid labrum in all three regions only 60% of the time.
Conclusions:
The only statistically significant differences in sensitivity/specificity of MRa and MRI were for SLAP lesions- with MRa being more sensitive and MRI being more specific. In general, this study includes the largest pediatric cohort to date and is the first to assess the ability of MRa/MRI to detect posterior labral pathology in the pediatric patient. The results consistently demonstrated lower sensitivities and specificities for both MRa and MRI in identifying various labral pathologies compared to other publications which almost ubiquitously have used MSK focused radiologists as part of its methodology. Our findings may more accurately represent clinical practice as access to MSK focused radiologists is limited. Lastly, the imaging report was accurate in fully diagnosing the existing labral pathology only 60% of the time. This is critically important for patients, families, and providers to be aware of, as the imaging reports available in the electronic medical record may misdiagnose one or more regions of the glenoid labrum prior to definitive diagnostic arthroscopy.
