Poor reproducibility of the MRI measurement of distal femoral torsion

Abstract

Recent publications on patient-specific instrumentation for total knee arthroplasty have reported considerable variability in the axial positioning of the cutting guides for the femoral component. These personalized cutting guides are manufactured based on bone shape data, generated from magnetic resonance imaging (MRI) or computed tomography (CT). The goal of this study was to compare the reproducibility and accuracy of distal femoral torsion (DFT) values measured using these two imaging modalities. We hypothesized that MRI does not reproducibly and consistently measure DFT and is not as accurate as CT scan. MethodsAnonymized radiology records from 54 patients that included MRI and CT scans of the knee were read in random order by two observers, on two separate occasions. These records were from patients being considered for a meniscal or osteochondral graft and who had their knee explored, but who had not undergone femoral or tibial surgery and were free of osteoarthritis. The DFT was estimated using the posterior condylar angle (PCA), using both its anatomical and surgical definitions. The intra- and inter-observer reproducibility of the MRI and the differences relative to CT scan measurements were analysed. ResultsThe average intra-observer difference for the MRI evaluation of the anatomical PCA was 0.8±1.2°; it was 0.4±0.9° for the surgical PCA. More than 1° difference from the average was found in 8 cases (14%) using the anatomical PCA measurement and 4 cases (7.4%) when using the surgical PCA (P=0.4). The intraclass correlation coefficients (ICCs) were 0.67 (95% CI: 0.33–0.85) and 0.74 (95% CI: 0.47–0.89) for the anatomical and surgical PCA, respectively. The average inter-observer difference for the MRI evaluation of the anatomical PCA was 1.6±1.4°; it was 1.5±1.0° for the surgical PCA. More than 1° difference from the average was found in 27 cases (50%) using the anatomical PCA measurement and 22 cases (40%) when using the surgical PCA (P=0.4). The ICCs were 0.31 (95% CI: 0.14–0.65) and 0.48 (95% CI: 0.06–0.75) for the anatomical and surgical PCA, respectively. The average differences between the CT and MRI measurements were 1.4±1.1° (0.2–5°) and 1.1±0.8° (0–3.6°) for the anatomical and surgical PCA, respectively. Greater than 1° difference between CT and MRI was found in 29 records (54%) for the anatomical PCA and in 18 records (33%) for the surgical PCA (P=0.03). ConclusionDFT measurement on MRI is more reproducible and consistent when using the surgical PCA. MRI measurements differed by more than 1° relative to CT measurements in more than one-third of cases. Case control studyLevel III.