OBJECTIVE Abnormal femoral antetorsion is associated with the development of femoroacetabular impingement (FAI). Anatomically correct antetorsion measurements are performed on transverse MR images over the proximal and distal femur, but some authors use alternative measurements on oblique images parallel to the femoral neck axis. We set out to assess the relationship between these two measurement methods and to obtain reference values for oblique measurements in patients with FAI and healthy control subjects. We also evaluated whether the oblique measurements could be used to predict the standard transverse measurements.
SUBJECTS AND METHODS MRI data of 126 individuals, 63 asymptomatic volunteers and 63 patients with FAI (age 20-50 years), were included in this prospective study. Two readers independently assessed antetorsion with oblique measurements and standard transverse measurements. Differences between subgroups were compared with the unpaired t test. Trigonometric calculations were used to predict standard antetorsion measurements on the basis of oblique measurements. Interobserver agreement and Bland-Altman plots were calculated.
RESULTS Reference values for assessing femoral antetorsion with the oblique method were established, with mean (± SD) values of 9.4° ± 7.9° (reader 1) and 9.8° ± 8.4° (reader 2) for patients and 9.2° ± 8.4° (reader 1) and 9.6° ± 9.1° (reader 2) for asymptomatic volunteers. The oblique method generated smaller antetorsion values than the standard transverse method (p < 0.001), with an average difference of 3.5° ± 3.2° for reader 1 and 3.6° ± 3.5° for reader 2. Differences between predicted antetorsion values based on oblique measurements and standard measurements were minimal: 0.1° ± 2.9° (p = 0.62) for reader 1 and 0.3° ± 3.3° (p = 0.29) for reader 2. Interobserver agreement was high for all antetorsion measurements (intra-class correlation coefficient, 0.945-0.977).
CONCLUSION Oblique measurements of femoral antetorsion were smaller than standard transverse measurements, but they can be used to accurately predict standard measurements.