Purpose or Case Report: Developmental dysplasia of the hip (DDH) is the leading cause of hip disease in infants. AVN is a potential complication of treatment for DDH, which is traditionally diagnosed on follow-up pelvic radiographs. But classic radiographic findings of AVN may be radiographically-occult for months to years after surgery. To investigate the performance of gadolinium-enhanced magnetic resonance imaging (MRI) for predicting avascular necrosis (AVN) of the femoral head following surgical hip reduction for developmental dysplasia of the hip (DDH) using qualitative and quantitative methods.
Methods & Materials: This IRB-approved, HIPAA compliant retrospective study included 47 children (7 boys and 40 girls; mean age 9.3 ± 4.6 months), who underwent contrast-enhanced MRI examinations immediately following unilateral surgical hip reduction between April 1, 2009 and June 30, 2018. AVN was determined on follow-up pelvic radiographs (mean 48.1 ± 23.0 months) using both Salter’s and Kalmachi and MacEwen Criteria. Two radiologists and 1 orthopaedic surgeon, blinded to clinical outcomes, independently classified the postoperative MRI enhancement pattern into normal, globally-decreased, focally-decreased, or near-absent. Regions of interests (ROIs) were used to quantify changes in T1-weighted signal intensities between treated and untreated sides and percent enhancement between AVN and non-AVN hips. Non-parametric tests and Fisher exact test were used to compare values between hips that did and did not develop AVN. Bonferroni correction were used for multiple comparisons.

Results: Ten (21%) femoral heads developed AVN. No significant differences between AVN and non-AVN groups were detected in sex (p=61), laterality (p=0.46), surgical technique (p=0.08), trial of harness treatment (p=0.72), or abduction angle (p=0.44). No significant differences were found between AVN and non-AVN hips in any of the enhancement grades (Grades 1-4) and at either early (p = 0.76) or late (p = 0.66) enhancement times. Signal intensities between treated and untreated sides were not significantly different (non-AVN group, p=0.09-1.0, AVN group, p=0.54-1.0) and percent enhancement between AVN and non-AVN hips were also not significantly different (early enhancement, p=0.41-0.88; late enhancement, p=0.53-0.74).

Conclusions: Qualitative enhancement patterns and quantitative enhancement values on post-operative MRI examinations were not significantly different between hips that did and did not subsequently develop AVN.