Internal femoral component rotation adversely influences load transfer in total knee arthroplasty: a cadaveric navigated study using the Verasense device

This work was performed under formal ethical approval and UK HTA licence within the surgical training facility xxxxxxxxxxx (Human Tissue Act 2004, section 16/2, licence number 12148). Whole lower limb cadaveric material without pre-existing radiographic evidence of malrotation/fracture/deformity or arthritic disease was used. Eight fresh frozen lower limbs disarticulated at the hip (three right, five left from Caucasian donors with median BMI 22, range 17–28, male/female ratio 7:1, and mean age 74, range 64–79 years) were prepared in a standardised manner as reported in previous published work [6, 9, 14]. Limbs were mounted onto a custom rig with the tibia hung vertically and secured using set screws to prevent specimen rotation. All muscle groups acting across the knee joint were loaded throughout the experimental procedure using previously validated methodology for the study of cadaveric knee kinetics. In particular, the iliotibial band (ITB), quadriceps muscle group, biceps tendon and medial hamstring apparatus were individually loaded in a physiological manner direction (Fig. 1). Navigation trackers (Stryker eNdtrac Knee Navigation System, Michigan USA) were fixed to the femur and tibia 25 cm from the joint line, respectively, and in such a manner as to avoid interfering with muscle pull [5]. Previous work had validated the use of eight limbs and confirmed that this provides sufficient power to identify significant differences using this technique with 95% confidence presuming 80% power [8, 22].

Insertion of a Stryker Triathlon (Michigan USA) single-radius cruciate-retaining TKA (CR-TKA) via a medial parapatellar approach was undertaken using a measured resection technique [27, 34]. The femoral component rotation was determined using the transepicondylar axis, validated using the mid-trochlear axis. A balanced knee was confirmed as that where after minimal soft tissue release, there was passive full extension, full flexion with normal ‘no touch’ patellar tracking, less than grade 1 laxity for varus valgus stressing at 0°, 30°, 60° and 90° of flexion, firm endpoint to Lachman testing and no laxity for anteroposterior stressing at 90° of flexion. To achieve such, 8 mm of distal femoral resection was performed at 5° valgus intramedullary alignment. Femoral sizing was undertaken using anterior referencing. The mid-point of the medial epicondyle may be difficult to localise but to reduce any peroperative error both senior authors agreed the mid-point of such so as to define the epicondylar axis. The rotation of the final 4 in 1 femoral cutting block was further validated using the mid-sulcus or Whiteside line. The tibia was then prepared using a 9-mm resection from the superiormost tibial condyle in a plane perpendicular to the anatomical axis with an associated 3° posterior slope achieved via an extra-medullary jig. The tibial cutting block was centred at the mid-third of the tibial tubercle and formal resection performed. The trial tibial baseplate was centred without restraint on the cut surface. Final orientation of the tibial component was determined after cycling the knee >20 times. The final seated position which achieved greatest conformity was marked on the anterior tibial crest using diathermy. For each experiment and each position of femoral component rotation, the tibial component was allowed to find its most conforming position. A standard polyethylene insert was selected and inserted with trial components to ensure the flexion and extension gaps were stable for varus valgus and anteroposterior stressing at the defines points of flexion stated earlier. Full extension was confirmed. The flexion gap allowed for full flexion but was balanced such as to ensure no opening in deep flexion. So as to optimise femoral component stability on the distal femoral resection, lughole screws were inserted and captured the femoral component whilst being buried and not interfering in articulation with the poly-component. This ensured absence of component rotation or glide or translation at any point of the experimental stressing in 6° of motion during the flexion/extension arc of the experiment (Fig. 2). A Verasense (Orthosensor, Dania FL) of appropriate thickness then replaced the polyethylene insert. With the knee held in 10° of flexion, the tibial rotation was adjusted and secured when the Verasense system registered a parallel contact point rotation [12]. Medial and lateral tibiofemoral contact forces were recorded as the loaded knee was taken through a range of passive flexion without stressing. No procedure required more than a simple peri-articular medial or lateral capsular release to satisfy our criteria for balancing as defined by confirming the load across the medial and lateral compartments were within 15lbs (pounds-force) of each other, respectively, through a full arc of motion (Fig. 2) [12, 45]. This state defined the datum for femoral component rotation for each knee. To allow for ±3° of femoral rotation whilst maintaining the original component size, femoral cuts were downsized by 2 mm. The externally rotated (ERF-TKA) and internally rotated (IRF-TKA) states were achieved with the insertion of custom wedges (Fig. 3). The femoral component was fast secured following each rotation using custom cancellous thread lughole screws engaging the stronger subchondral bone. Care was taken to perform surgical closure of the arthrotomy via interrupted mattress sutures respecting the anatomy of the parapatellar tissues prior to any testing [31]. The senior authors (xxx/xx) performed all surgical procedures and stress testing. For all experiments, there were two senior surgeons, a surgical assistant, a manual operator of the orthosensor system, a senior scientist recording the output from the navigation system.

Data were captured, as per previous validated work, using a standard computerised navigation system with orthosensor provided range of compatible tibial trials [14, 37]. Knees were manually stressed to mimic intraoperative laxity assessment. A datum was taken from the knee in extension from which maximal displacements of the tibia in relation to the fixed femur were tracked via computer navigation (Stryker eNdtrac Knee Navigation System, Michigan, USA) to an accuracy of ±0.5 mm in 6° of freedom [5, 10, 28]. For each TKA condition, maximal displacements (anteroposterior, varus, valgus, internal and external rotation) were each recorded at five angles of flexion (0°, 30°, 60°, 90° and 110°). To reduce hysteresis, repeated flexion cycles were undertaken between measurements and ensured compartment forces remained constant during passive flexion. After each set of measurements, the output instrumentation was reset at zero. The Verasense device recorded tibiofemoral contact force (lbs/force) and contact points continuously during testing (millimetre accuracy ±2 mm—C. Anderson, OrthoSensor, personal communication 03.08.2015) with additional data capture under maximal stress at each of the five angles of flexion. Three knee conditions were defined with the internal and external rotatory states compared with the neutral or datum. These were the internally rotated femoral component (IRF-TKA) and the externally rotated femoral component (ERF-TKA) (Fig. 3).

Mixed-effect modelling was used to quantify the effect of flexion angle, direction of movement and implantation of TKA upon laxity [30, 32]. Displacements were used as the response variable, with TKA and flexion as covariates. Student’s t test was used to compare differences in tibiofemoral force and contact point measurements. Significance was set at a level of p < 0.05.

Total tibial rotation (combined internal and external rotation for each point of knee flexion of replaced state) and maximal varus/valgus laxity were found to increase consistently from full extension to deep flexion for all knee specimens. Rotational and coronal laxity did also increase with knee flexion within the native knee. This increase in laxity/rotation comparing the native versus the replaced state was significant for maximal anteroposterior movement (AD) when comparing 110°–30° (Fig. 4a).

A decrease in laxity was found for all knees after implantation of a single-radius (CR-TKA) cruciate-retaining knee design when compared to the native state. This reduction in maximal laxity only reached significance for rotatory laxity in higher levels of knee flexion beyond 90° (Fig. 4a–c).

Assessment of the unstressed flexion arc showed that contact forces across the medial and lateral compartments showed no statistical difference at each angle of flexion in the CR-TKA (Fig. 5a). Tibiofemoral contact force demonstrated equilibrium of load in the primary CR-TKA at each angle of flexion. The total force across the tibiofemoral joint was seen to decrease with flexion. An inverse relationship for total contact load versus maximal laxity was, as expected, found consistently for each knee.

Maximal anteroposterior was significantly increased for the internally rotated femoral component state beyond 60° when compared to both the neutrally and externally rotated femoral component states. Varus valgus maximal movement and maximal rotatory motion assessment failed to identify any difference for the internally rotated component state when compared to neutral or external rotation (Fig. 4a–c).

Load across the medial compartment was significantly increased at and beyond 60° of flexion with the internal femoral component state (Fig. 5b). This was most marked when stressing the knee with maximal internal rotation beyond 60° of flexion. At 90° flexion, a mean of 83% of the total contact load was borne by the medial compartment in the ERF-TKA compared to 47 and 43% in the CR-TKA and ERF-TKA, respectively (p < 0.05).

The laxity pattern for the externally rotated femoral component knee arthroplasty (ERF-TKA) did not significantly alter when compared to the primary neutral position of the femoral component. Whilst the greatest difference between the ERF-TKA and CR-TKA was found at 110° with an increase in rotational laxity, this failed to reach statistical significance. No significant difference at any angle of flexion was found between the neutrally and externally rotated femoral component arthroplasty states for maximal anteroposterior movement or varus valgus stressing (Fig. 4a–c).

Similar load was found across the medial and lateral femoral compartments between 0° and 30° of flexion when comparing the ERF-TKA versus CR-TKA states. Between 60° and 110° load increased across the lateral compartment under rotational stress for the ER-TKA when compared to the CR-TKA. This change did not, however, reach significance (Fig. 5c). It was representative of a proportional increase in lateral compartment loading in the ER-TKA state compared to the CR-TKA in mid to deep flexion. Overall, the ERF-TKA load pattern very closely resembled the pattern of contact force recorded for the neutrally aligned femoral component (CR-TKA) state.

There was consistency of load transfer across the medial and lateral compartments for both neutral datum femoral position and external rotation of the femoral component under both valgus (Fig. 5d) and varus maximal stress testing (Fig. 5e). Interesting, whilst this did not reach significance, there was a consistent increase in load transfer on the medial compartment in higher degrees of knee flexion for both varus and valgus load (Fig. 5d, e).