Distribution of coronal plane alignment knee phenotypes in Japanese patients with knee osteoarthritis: a comparison of two-dimensional and three-dimensional measurements.

Abstract

This study aimed to determine whether the distribution of coronal plane alignment of the knee (CPAK) classification differs in Japanese patients with knee osteoarthritis (OA) when assessed using two-dimensional (2D) standing long-leg radiographs versus three-dimensional (3D) supine computed tomography-based measurements. This retrospective study included 100 knees of 86 patients who underwent primary total knee arthroplasty for OA (mean age, 72.9 years; 27 men, 73 women; body mass index, 27.3 kg/m²). The lateral distal femoral angle, medial proximal tibial angle, arithmetic hip-knee-ankle angle (aHKA), and joint line obliquity (JLO) were measured on 2D and 3D images. Corresponding parameters were compared, and correlations were evaluated. Using aHKA and JLO, differences in CPAK phenotypes between 2D- and 3D-based classifications were assessed. All parameters differed significantly between 2D and 3D measurements. Mean lateral distal femoral angle was 88.2° on 2D and 87.9° on 3D (p < 0.01), medial proximal tibial angle was 85.3° and 83.8° (p < 0.01), aHKA was -2.9° and -4.1° (p < 0.01), and JLO was 173.5° and 171.7° (p < 0.01), respectively. The distribution of CPAK phenotypes did not differ significantly between 2D- and 3D-based classifications, with good agreement (Cohen's kappa = 0.615). All parameters demonstrated strong correlations between 2D and 3D measurements (r > 0.80). In Japanese patients with knee OA, the distribution of CPAK phenotypes showed good agreement between 2D- and 3D-based assessments. Although 2D measurements may underestimate varus alignment and JLO, they appear sufficient for CPAK-based phenotypic assessment in clinical practice.