Quantitative Whole-Leg Radiographic Assessment of Coronal Knee-Ankle Alignment: Associations with Talar Inclination...

Background/Objectives: Standing whole-leg radiographs are central to coronal deformity assessment, but mechanical axis deviation (MAD) may not fully characterize the orientation of individual joint surfaces. We evaluated ankle joint orientation across coronal alignment phenotypes and, secondarily, explored descriptive changes after knee realignment osteotomy. Methods: This retrospective single-center study included 382 whole-leg radiographs from 374 patients and an exploratory longitudinal subgroup of 80 limbs from 75 patients before and after high tibial osteotomy (HTO) or distal femoral osteotomy (DFO). Limbs were classified as varus, valgus, or near-neutral by MAD. Talar inclination was the primary imaging outcome; knee-ankle joint line angle (KAJA), talar tilt, and segmental alignment parameters were assessed by three independent readers. Results: Talar inclination was greater in valgus than varus or near-neutral limbs (10.5 ± 5.9° vs. 4.7 ± 3.3° and 4.7 ± 3.5°; p < 0.001). Inter-reader reliability was excellent (ICC 0.91-0.96). MAD correlated with talar inclination (ρ = -0.426, p < 0.001), while KAJA showed a larger univariable correlation (ρ = 0.611, p < 0.001). Adding KAJA to age, sex, and MAD increased model R² from 0.237 to 0.551; because these measures are geometrically interrelated, this increment was interpreted as descriptive model fit rather than an independent biological contribution. In the exploratory postoperative subgroup, talar inclination decreased in valgus limbs (-3.2 ± 5.2°, p = 0.012), whereas changes were not significantly associated with the magnitude of MAD correction. Conclusions: Whole-leg radiography demonstrates distinct knee-ankle alignment patterns not captured by MAD alone. KAJA provides complementary descriptive morphologic information regarding talar orientation, but should not be interpreted as an independent biological predictor. Postoperative subgroup findings are hypothesis-generating because of small, unbalanced groups and procedure-related confounding; no clinical outcome or three-dimensional hindfoot inference can be drawn from these data.