Comparison of weight-bearing full-length radiography and three-dimensional computed tomography scan-based models to assess knee joint coronal alignment following total knee arthroplasty

Background: The validity of standing long-leg radiography (LLR) - the standard method for evaluating coronal alignment after total knee arthroplasty (TKA) - remains controversial. We evaluated the measurement reliability and validity of LLR following TKA by comparing postoperative radiographs with three-dimensional computed tomography (3DCT) scans of the lower extremities.

Methods: We retrospectively collected the data of 55 knees from 44 patients who underwent postoperative 3DCT and 5-day and 6-month LLR for coronal alignment evaluation following TKA. The coronal femoral component angle (CFA) and coronal tibial component angle (CTA) were measured using 3DCT and LLR images. Correlations between the hip-knee-ankle angle (HKAA), CFA and CTA were analysed using Pearson's correlation coefficient (PCC); Bland-Altman plots were constructed to assess agreement between 5-day and 6-month radiographic and 3DCT scan measurements.

Results: The mean difference in HKAA between 3DCT, and 5-day and 6-month LLR was 1.3 ± 1° and 1.1 ± 0.7°, respectively. Differences of >1° in HKAA between 3DCT, and 5-day and 6-month LLR were observed in 31 (56.4%) and 28 (50.9%) knees, respectively; differences of >2° in these parameters were observed in 15 (27.3%) and nine (16.4%) knees, respectively. The 3DCT scan and radiograph measurements of HKAA, CFA and CTA were strongly correlated (PCC, 0.81-0.92; P < 0.001). Bland-Altman plot validity was within acceptable limits.

Conclusion: LLR shows good reliability and validity for measuring coronal alignment. When comparing 3DCT scans and radiographs, a high number of HKAA values differed by more than 1° and 2°; this indicates that rigorous alignment evaluation requires 3DCT measurements.