UC Davis

BackgroundTotal knee arthroplasty (TKA) treats advanced osteoarthritis by replacing degenerated articular surfaces of the tibia and femur with implants. However, of approximately 1 million TKAs performed annually in the US, 20-25% of TKA patients are dissatisfied due to persistent pain, poor function, or implant failure which is conventionally performed using mechanical alignment (MA). MA TKA aligns the implants to establish a ‘straight’ limb (i.e. line connecting hip and ankle passes through center of knee); however, a straight limb is unnatural and requires ligament release in the majority of patients.

Unrestricted kinematic alignment (KA) TKA is an alternative alignment method which restores the native limb alignment and joint lines without ligament release. KA TKA is a viable alternative to MA TKA because multiple studies have reported that patients treated with KA TKA have significantly better pain relief, function, and a more normal feeling knee than patients treated with MA TKA. However, clinicians have not widely embraced KA TKA because of concern that KA TKA may compromise implant survivorship due to tibial baseplate loosening. This concern arises because baseplate loosening has been linked to varus alignment of the baseplate in MA and because most baseplates are aligned in varus in KA.

Furthermore, although promising clinical results have been achieved for KA TKA using conventional (i.e., low conforming) implants, such implants fail to restore native (i.e., healthy) knee kinematics. Medially-conforming (MC) implants better restore native knee kinematics but features of their design exacerbate the concern regarding tibial baseplate loosening.Assessing long-term risk of baseplate loosening can be evaluated in the short-term using model-based radiostereometric analysis (RSA) which measures in vivo tibial baseplate movement relative to the underlying tibia bone (termed migration). Baseplate migration is computed by acquiring biplanar radiographs postoperatively on the day of surgery and at follow-up time points over 2 years,followed by registering a 3D baseplate model onto the images to determine how its position and orientation relative to reference tibia markers change over time. To assess risk of loosening, the magnitude of the migration is compared to stability limits from the RSA literature. To assess this risk for a group of patients, mean maximum total point motion (MTPM) (i.e., largest displacement of any point on the baseplate) is compared to a limit of 0.5 mm at 1 year. To assess this risk in individual patients, there are two stability limits: a change in MTPM between 1 and 2 years (ΔMTPM) of 0.2 mm and anterior tilt at 2 years of 0.8°.

To address the concern that KA TKA using MC implants which retain the posterior cruciate ligament (KA+MC+PCL) increases risk of tibial baseplate loosening, aims were to: (1) set-up a clinical RSA laboratory and develop laboratory protocols which reduce error and (2) determine whether tibial baseplate migration for KA+MC+PCL is below the stability limits.

Aim 1: Error Reduction MethodsTo determine methods for error reduction, a protocol for quantifying bias and precision in clinical settings was developed and sources of registration error for model-based RSA were identified and evaluated. The first source of registration error was the orientation of the baseplate relative to the imaging planes. The primary objective was to determine the baseplate orientation which minimized mean registration error (i.e., bias error). A tibia phantom was rotated to achieve 24 different orientations with three independent pairs of radiographs acquired at each orientation. Bias error, quantified as mean artifactual (aMTPM), was plotted as a function of the rotation angles to determine the orientation with the lowest error. For the baseplate under study, the orientation with the lowest error was 10° rotation in flexion and 5° rotation in varus which decreased bias error by 85% relative to 0° rotation in flexion and varus. A consequence of baseplate reorientation during imaging is deviation between the globally-aligned baseplate coordinate system (BCS) (defined by calibration box) and cardinal body planes; therefore, computing migration using a local BCS (defined by baseplate) may be preferrable. Objectives were to (1) summarize the migration equations for the two BCSs and (2) provide a clinical example to demonstrate their differences. Differences in migration for the two BCSs ranged from about ±0.5 mm in translations and -0.4° to 0.7° in rotations, were largest for deviations in internal-external rotation, and were smallest for deviations in varus-valgus rotation. Migrations computed using the local BCS were independent of reorientation and therefore enhanced consistency.

The second source of registration error was the type of 3D baseplate model. The primary objective was to determine whether RE models reduced bias error relative to CAD models. Bias error, quantified as mean aMTPM, was computed between two pairs of independent biplanar radiographs from the same day for thirty-five patients. The biplanar radiographs were analyzed using each model type for all patients. Mean aMTPM was reduced by 24% when using RE models instead of CAD models.

Aim 2: Clinical Tibial Baseplate MigrationThirty-five patients enrolled in this prospective 2-year longitudinal study. Patients underwent cemented, unrestricted KA+MC+PCL TKA. Biplanar radiographs acquired on the day of surgery and at 1.5, 3, 6, 12, and 24 months were processed with model-based RSA software to determine migration in six degrees of freedom (6DOF) and in MTPM relative to the day of surgery.

For group assessment, objectives were to determine whether mean MTPM (1) was significantly < 0.5 mm at 1 year and (2) increased over time. At 1 year, mean MTPM of 0.35 mm was significantly < 0.5 mm (p = 0.0002). Mean MTPM did not increase from 6 weeks to 1 year (p = 0.3047).

For individual patient assessment of KA+MC+PCL, primary objectives were to determine the number of patients who exhibited ΔMTPM > 0.2 mm and the number of patients who exhibited anterior tilt > 0.8°. Because current stability limits linking baseplate migration to long-term risk of baseplate loosening only exist for ΔMTPM and anterior tilt, and because the concern of baseplate loosening for unrestricted KA TKA is pervasive particularly due to varus baseplate alignment, a final objective was to determine whether patient-specific alignment and loading variables were associated with migration in either 6DOF or MTPM. Hip knee ankle angle (HKAA), proximal medial tibial angle (PMTA), posterior slope angle (PSA), sex, and body mass index (BMI) were analyzed for an association with migration in 6DOF and MTPM. Thirty-two of the 35 patients were available for analysis at 2 years. One of 32 patients exhibited ΔMTPM > 0.2 mm and also exhibited anterior tilt > 0.8°. Varus rotation (p = 0.048, r ≤ 0.34) and medial translation (p = 0.0273, r ≤ 0.29) increased with increased varus baseplate alignment; however, the magnitudes of the migrations were minimal and did not manifest as increases in ΔMTPM and anterior tilt.

ConclusionsResults from Aim 1 inform RSA researchers on methods for reporting clinical bias in MTPM, reducing registration error (by reorienting the tibial baseplate during imaging and using RE models), and computing migration using local BCSs. Results from Aim 2 inform orthopedic surgeons that unrestricted KA+MC+PCL TKA does not increase risk of baseplate loosening on a group or individual patient level.