Full-limb and knee radiography assessments of varus-valgus alignment and their relationship to osteoarthritis disease features by magnetic resonance imaging

Authors


Abstract

Objective

To examine the correlation between hip-knee-ankle and femur-tibia radiograph angles, calculate the offset of the femur-tibia angle with respect to the hip-knee-ankle angle, calculate the sensitivity and specificity and area under the receiver operating characteristic (ROC) curve of the femur-tibia angle, and examine the relationship of malalignment by each approach with osteoarthritis (OA) tissue pathology in the mechanically stressed compartment using magnetic resonance imaging (MRI).

Methods

Individuals with knee OA underwent full-limb and knee radiographs and knee MRI. Linear regression was used to determine if the 2 angles differed systematically and to identify the cutoff. Alignment means for MRI grades were compared using Dunnett's t-test.

Results

In the 146 participants (109 women, mean age 70 years, body mass index 30.6 kg/m2), femur-tibia and hip-knee-ankle angles correlated (r = 0.86; 95% confidence interval [95% CI] 0.81, 0.90). On average, the femur-tibia angle was 3.4° more valgus (3.0° in women and 4.7° in men); after correction, its sensitivity and specificity (to predict the hip-knee-ankle angle) were 0.84 and 0.84 for identifying varus and 0.98 and 0.73 for valgus, respectively. The area under the ROC curve (95% CI) was 0.91 (0.86, 0.96) for varus and 0.94 (0.89, 0.99) for valgus. Varus severity worsened comparably with each alignment measure as medial lesion score on MRI worsened. Laterally, as lesion score worsened, comparably worse valgus was seen with either assessment approach.

Conclusion

In knee OA, the knee radiograph femur-tibia and full-limb radiograph hip-knee-ankle angles were correlated. The femur-tibia angle, corrected for mean offset, was sensitive, specific, and had excellent discriminative ability for identifying varus and valgus alignment evidenced by area under the ROC curve. The relationship between alignment and specific OA MRI features was comparable with the 2 approaches. Use of the femur-tibia angle, corrected for offset, should be considered in research and clinical settings.

Ancillary