Prosthetic Accuracy Depends on the Design of Patient-Specific Instrumentation: Results of a Retrospective Study Using Three-Dimensional Imaging

To determine accuracy of patient-specific instrumentation (PSI), the preoperative three-dimensional (3D) plan should be superimposed on the postoperative 3D image to compare prosthetic alignment. We aimed to compare prosthetic alignment on a preoperative 3D computed tomography (CT) plan and postoperative 3D-CT image, and evaluate the accuracy of PSI during total knee arthroplasty (TKA). Thirty consecutive knees (30 patients) who underwent TKA using PSI were retrospectively evaluated. The preoperative plan was prepared using 3D CT acquisitions of the hip, knee, and ankle joints. The postoperative 3D CT image obtained 1 week after surgery was superimposed onto the preoperative 3D plan using computer software. Differences in prosthetic alignment between the preoperative and postoperative images were measured using six parameters: coronal, sagittal, and axial alignments of femoral and tibial prostheses. Differences in prosthetic alignment greater than 3 degrees were considered outliers. Two observers performed all measurements. All parameters were repeatedly measured over a 4-week interval. This measurement method's intraobserver and interobserver reliabilities were more than 0.81 (very good). For the femoral and tibial prostheses, absolute differences between the preoperative and postoperative 3D CT images were significantly larger in the sagittal than in the coronal and axial planes (p < 0.001). The outlier rate for the sagittal alignment of femoral and tibial prostheses was significantly higher than that for the alignment of coronal and axial planes (p < 0.001). However, there were no significant differences in the range of motion (ROM) before and after TKA when comparing cases with and without outliers in the sagittal plane. Even though the present study did not reveal any issues with the ROM that depended on the presence of an outlier, accurate verification of prosthetic alignment for individual PSI models may be necessary because the designs, referenced images, and accuracy are different in each model.