Changes in kinematic behavior and collateral ligament strain after medially stabilized TKA using a novel intraoperative navigation platform: a cadaveric study Authors: Arne Van de Vyver, Orçun Taylan, Geert Peersman, Margot Demeulenaere, Maciej Bialy, Thomas Lauwagie, Lennart Scheys Published: 2026 ABSTRACT AbstractObjectivesThis study aimed to analyze to what extent a novel intraoperative navigation platform (Next-AR, Medacta) for total knee arthroplasty (TKA) allows to restore the native knee joint kinematics and strains in the medial collateral ligament (MCL) and lateral collateral ligament (LCL) throughout a squatting motion.Materials and methodsComputed tomography (CT) scans of 6 native cadaver legs were used to design patient-specific guides. Bony landmarks and virtual single-line collateral ligaments were identified to acquire real-time intraoperative feedback on bone resection, implant alignment, tibiofemoral kinematics, and collateral ligament elongations using the Next-AR system. The specimens were subjected to squatting (35°-100°) motion using a physiological ex vivo knee simulator while maintaining a constant vertical ankle load of 110 N through active quadriceps and bilateral hamstring controls. Subsequently, each knee underwent a medially-stabilized TKA (Medacta) with mechanical alignment technique and was retested under the same conditions as in their native state. The tibiofemoral and patellofemoral kinematics, along with collateral ligament strains, were computed from 3D marker trajectories using a six-camera optical system (Vicon). MCL and LCL insertions—ant, mid, and post bundles—were identified in relation to bone-pin markers using a wand.ResultsBoth native and post-operative conditions exhibited similar tibial valgus orientation (Root Mean Square Error (RMSE = 1.7°), patellar flexion (RMSE = 1.2°), abduction (RMSE = 0.5°), and rotation (RMSE = 0.4°) during squatting (p > 0.13). However, a significant difference was observed in tibial internal rotation between 35° and 62° (p < 0.048, RMSE = 3.2°). MCL strains in anterior (RMSE = 1.5%), middle (RMSE = 0.7%), and posterior (RMSE = 0.8%) were closely matching in both conditions, showing no statistical differences (p > 0.05). Contrary, LCL strain in all bundles (RMSE < 4.6%) differed significantly from mid to deep flexion in both conditions (p < 0.048).ConclusionThe novel intraoperative Next-AR system not only targets planned knee alignment but also aids in restoring native knee kinematics and elongation of the collateral ligaments through real-time feedback.Keywords : Collateral ligament kinematics, Knee arthroplasty, Navigation, Computer-assisted surgery Introduction Total knee arthroplasty (TKA) is a highly successful treatment for end-stage knee osteoarthritis and aims to restore pre-operative biomechanical- and kinematic properties mimicking the native knee joint [1]. Despite the overall success of TKA, a notable proportion of patients may remain dissatisfied post-surgery. Possible factors contributing to joint pain, reduced motion or instability are implant loosening, poor soft tissue balance, or implant choice [2]. Advancements have been made in various aspects of TKA technologies to address dissatisfaction and alignment outliers, advancements have been made in various aspects of TKA technology. These include improvements in implant design and materials, image-based cutting guides, patient-specific implants, and computer-navigated surgery [3]. Navigation systems, whether imageless or image-based (CT or MRI) assist in achieving desired alignment goal, thus recreating optimal ligament tension and knee kinematics [4]. Current literature has yet to provide definitive evidence comparing image-based and imageless navigation systems in terms of accuracy, patient-reported outcomes (PROMs), range of motion (ROM), or postoperative satisfaction. However, differences have been noted between planned and achieved bone cuts, with femoral and tibial coronal cuts more likely to result in unintended varus using imageless systems. This is likely because imageless systems rely on intra-articular anatomical landmarks, making their accurate identification critical for precise implantation [5]. There is ongoing debate about which type of implant design is superior. Currently, the most popular designs are cruciate retaining (CR) and posterior stabilized (PS) implants. Additionally, medially stabilized implants have been introduced to mimic natural knee kinematics, such as femoral rollback and ligament behavior, more closely [6, 7]. Previous studies have shown that none of the ligament bundles of MCL and LCL showed isometric behavior after TKA, regardless of implant design. Achieving symmetrical medial and lateral joint spaces in both extension and flexion intraoperatively does not necessarily equate to optimal soft tissue balance [8, 9] Preservation of ligament tension during knee motion is challenging in TKA and new technologies are emerging to cope with these challenges. Medacta International has recently introduced the Next-Ar Navigation Platform (Medacta International, Strada Regina, Switzerland), a wireless single use tracking system mounted directly onto bone. This system provides real-time intraoperative feedback by using preoperative CT to identify bony landmarks and ligamentous insertions. Consequently, the surgeon can monitor bone resection, knee alignment, tibiofemoral kinematics and collateral ligament elongation. Precise adjustments can be made during the procedure. This study aims to assess the efficacy of a novel Next-AR system for medially stabilized TKA in restoring native joint kinematics and collateral ligament strains in a cadaveric set-up. In this study, cadaveric knees were used for testing as they complement in vivo research and represent the most physiologically representative option for evaluatin knee joint function and collateral ligament behavior under reproducible and standardized conditions. Materials and methods Preoperative specimen imaging and surgical planning Seven fresh-frozen full cadaveric legs (age = 80.8 ± 9.59, three unilateral female donors, one unilateral and two bilateral male donors, height = 165 cm ± 0.5, and weight = 66 kg ± 10.5) were obtained following ethical approval from the local Ethics Committee (NH019-2021-04-01). The specimens were screened for potential lower limb disorder and prior surgical intervention. No significant deformities were present, coronal alignment was < 5° for all specimens. Bicortical bone pins were fixed into the femur, tibia, and patella to rigidly attach dedicated rigid marker frames, each containing four reflective spheres. Each specimen’s computer tomography (CT) scans (Siemens Somatom Force, Siemens Healthcare, Erlangen, Germany) were then acquired in full extension with a slice thickness of 0.6 mm. Segmentation software (Mimics 25.0, Materialise, Leuven, Belgium) was used based on the well-established algorithm to identify the location of spherical markers and the required anatomical landmarks from CT images [10–12]. These identified landmarks were used to define bone-specific anatomical joint coordinate systems (ACS). For the femur and tibia, the joint coordinate systems were defined according to the Grood and Suntay convention for the femur, tibia, and patella [13]. For the patella, we used a Grood and Suntay–based extension as described by Bull et al. The patellar anatomical coordinate system was constructed by defining the z-axis (proximal–distal) along the central cartilage ridge, defining the y-axis (mediolateral) perpendicular to this ridge using the line between the most lateral patellar edge and the central medial patellofemoral ligament insertion, and obtaining the x-axis (anterior–posterior) as the cross-product of the y- and z-axes. Patellofemoral kinematics were then expressed relative to the femoral coordinate system using the Grood–Suntay floating-axis convention. More specifically, flexion/extension corresponds to rotation about the femoral mediolateral axis, medial–lateral tilt corresponds to rotation about the patellar long (z) axis, and abduction-adduction rotation corresponds to rotation about the floating axis, defined as the cross product of the patellar long axis (z-axis) and the femoral anterior–posterior (x) axis. Translations of the patella were calculated along the femoral axes as anterior–posterior, medial–lateral, and superior–inferior translations [10]. Additional CT scans of the hip, knee, and ankle of each subject were acquired conforming to the MyKnee protocol (Medacta International SA, Castel San Pietro, Switzerland) to design patient-specific surgical guides [14], including the proprietary holders for the Next-AR motion trackers (a camera and a receiver unit), for mechanical alignment technique and to generate input parameters, such as 3D models of the bones, bony landmarks, and insertions of the collateral ligaments [15–17]. Subsequently, these input parameters were exported into the Next-AR Navigation Platform. Specimen preparation Each specimen was thawed 24 h before testing, with a single freezing-thawing cycle. The femur and the tibia were resected 32 cm proximally and 28 cm distally to the knee joint line, respectively. Care was taken to preserve the joint capsule, ligaments and tendons while removing the skin and subcutaneous tissue around the knee joint. The exposed quadriceps tendon was clamped within a custom metal clamp, and suture loops (2 × 2 non-absorbable polyester braided suture wire; Cardioxyl, Peters Surgical, Bobigny Cedex, France) were passed through the medial and lateral hamstring tendons. The tibia and femur were embedded in metal containers using an acrylic resin (Struers, Ballerup, Denmark) while positioning the femur in 6° of valgus. Preoperative testing The native specimens were mounted into an Oxford-based physiological ex-vivo knee simulator and subjected to passive flexion (10°–120°) and squatting (35°–100°), respectively. During passive flexion, the femur pot was rigidly attached to the knee simulator, with the tibia and the tendons left unconstrained. Subsequently, the tibia was manually cycled from maximum extension to maximum flexion angle. For squatting motion, the tibial pot was mounted to the ankle assembly. The suture loops fixed to the medial and collateral hamstring tendons were connected to the medial and lateral electromechanical hamstring actuators through a steel pulley and cable system. The quadriceps clamp was affixed to the quadriceps electromechanical actuators directly through a clevis and pin assembly. During squatting motion, a computer-controlled electromechanical actuator was used to apply a dynamic load to the quadriceps tendon. while computer-controlled electromechanical actuators individually applied a constant 50 N load to the medial and lateral hamstrings. In conjunction with simultaneously controlling the vertical translation of the ankle assembly to facilitate flexion and extension, the controller ensured a constant vertical ankle load of 110 N throughout the squatting motion. Each motion was performed in triplicate, and the trajectories of the retro-reflective spheres attached to the specimens were recorded using six infrared cameras (Vero v1.3, Vicon Motion Systems, Oxford, UK) operating at 100 Hz [18]. After performing functional motion tests, the superficial insertions of the medial collateral ligament (MCL) and lateral collateral ligament (LCL) were carefully exposed. Subsequently, the coordinates of each ligament bundle (anterior, middle, and posterior) were recorded using a dedicated optical motion capture system’s wand, equipped with five spherical retro-reflective markers. Moreover, for precise and physiologically accurate computation of Vicon-based ligament elongations, while using a shortest path calculation, the tibial plateau edge was defined as a 3D spline using Mimics (Materialise, Leuven, Belgium). It is noteworthy that such measures were not required for the LCL, given its absence of anatomical wrapping around the tibial plateau. Surgical technique and postoperative testing Two knees were excluded from pre- and postoperative squatting motion analysis due to early quadriceps tendon failure after surgery. All eight knees were included in pre- and postoperative passive flexion analysis. After testing the native knees, each knee underwent a cemented cruciate-sacrificing total knee arthroplasty (medially stabilized, GMK Sphere, Medacta International) using a mechanical alignment technique with neutral tibial rotation (according to Akagi’s line) and 5° tibial slope. First, the patient-specific surgical guides were rigidly fixed to the femur and tibia with bone pins. Following, the Next-AR motion tracker unit—a camera and a receiver—were rigidly attached to detachable device housings mounted on the guides (Fig. 1A). To minimize potential discrepancies between the positions reported by the motion tracker unit and the actual bone positions, fine-tuning correction was performed throughout the registration process [17, 19]. Subsequently, the surgical guides were removed while maintaining the camera and receiver units fixed. Before resecting the bones, the absolute change in the collateral ligament lengths to an initial value (extended knee (0°−10°)) was determined and tibiofemoral kinematics were documented during passive flexion-extension. The surgical procedure and ex-vivo testing. A Patient-specific surgical guide and holders for the camera and the receiver, B registration of the bone surface to the model, C resection of the bone using dedicated cutting guides, D implantation of a cemented total knee arthroplasty (GMK Sphere, Medacta International) and E quasistatic squatting in an Oxford-based physiological ex-vivo knee simulator The cutting jigs were affixed to the femur first, followed by the tibia, and then sequential bone resections were carried out. Subsequently, a cemented TKA was implanted. Postoperative collateral ligament elongations and tibiofemoral kinematics were digitally recorded during passive flexion-extension cycles. These recordings were then compared to the preoperatively measured reference curves. No additional cuts or ligament releases were performed during this experiment. Data processing The trajectories of reflective markers captured with the Vicon optical motion capture system were processed using motion tracking software (Nexus 2.9, Vicon, Oxford, UK). Based hereon, the tibio- and patellofemoral kinematics and the elongation of each collateral ligament bundle (anterior, middle and posterior) were computed. For the latter, the method described by Marai et al. was used [20]. The kinematic variables, along with the collateral ligament elongations, were downsampled and interpolated at 1° intervals of flexion, all within the common knee flexion range shared by all specimens: 10° to 100° for passive flexion and 35° to 10° for squatting. Furthermore, the absolute difference between the output measured at the initial reference knee flexion angle (10° for passive flexion and 35° for squatting) and the value at each subsequent angle was used to compute the engineering strain of collateral ligament bundles throughout the motion. The initial knee flexion angle was considered 0% and all subsequent measurements were expressed to this initial value using a custom code in Matlab (R2022a, Mathworks Inc, Natick, MA, USA). Statistical analysis A linear mixed model was used to compare measured outputs derived from native and postoperative conditions, such as tibio- and patellofemoral kinematics and collateral ligament strains. Subsequently, Tukey’s HSD post hoc test was performed for pairwise comparisons at each knee flexion angle. The residuals were assessed for normality, multicollinearity, and heteroscedasticity. No transformations were required as assumptions were met. The Cohen’s d effect size (ES, large ≥ 0.8 and small ≤ 0.2) was computed for the statistically different groups. Additionally, the mean Root-Mean-Square Error (RMSE) for the entire motion was calculated to quantify the difference between the mean curves of the conditions. All statistical analyses were performed in R (R-Studio Version 2023.03.1, Boston, MA), and the significance level was determined at p < 0.05. Results Tibiofemoral Kinematics During passive flexion, both native and postoperative conditions exhibited a similar trend in varus-valgus behavior (p > 0.508, mean Root-Mean-Square Error (RMSE) = 0.763°): an initial increase in varus until mid-flexion, followed by a subsequent decrease throughout the remaining range of motion (Fig. 2A). For tibial internal-external rotation, the native and postoperative conditions showed a tendency towards increased internal rotation (Fig. 2C). Nonetheless, a significant difference between the native and postoperative conditions was observed after 81° of knee flexion angle (0.028 < p < 0.048, RMSE = 2.862°, 0.765 < Effect size (ES) > 0.713). In terms of AP translation of the femur knee center with respect to the tibia knee center, significant differences were observed between 52° and 120° (0.001 < p < 0.045, RMSE = 5.639 mm, 1.014 < Effect size (ES) > 3.854). Tibiofemoral kinematics of the knee in the native condition (black) and TKA (red) conditions during passive flexion (left column) and squatting (right column). A, B tibial varus-valgus orientation (varus < 0, valgus > 0), C, D tibial internal-external rotation (external rotation < 0, internal rotation > 0), and E, F femur AP translation (Posterior < 0, Anterior > 0). The AP translation was computed as the position of the femoral knee centre relative to the tibial knee centre. Data are represented as mean (solid) ± SD (shaded) across six specimens. The asterisk indicates a significant difference < 0.05, while the dashed black line illustrates the range of knee flexion where this significant difference is found During squatting, both native and postoperative conditions consistently maintained varus orientation across the entire range of motion (Fig. 2B). Nonetheless, the native condition tended towards reduced varus compared to TKA. However, no significant difference was detected (p > 0.157, RMSE = 1.734°). In terms of internal-external rotation, both conditions displayed internal rotation, with the native condition exhibiting decreased internal rotation after mid-flexion. However, significant differences between these two conditions were evident only between 35° and 61° of knee flexion (p < 0.045, EMSE = 3.284, ES > 0.973). On the other hand, significant differences in AP translation between both conditions were found within the knee flexion range of 35° and 100° (0.001 < p < 0.044, RMSE = 6.426 mm, 1.920 < Effect size (ES) > 3.344), with the native condition displaying lower magnitude translations compared to those of the postoperative condition. Patellofemoral kinematics In terms of patellar kinematics during squatting, both native and postoperative conditions displayed a closely matched linear increase in patellar flexion (Fig. 3A), with no significant difference observed (p > 0.140, RMSE = 1.217°). For patella abduction (Fig. 3B), both conditions maintained a plateau, preserving a consistent offset throughout the entire squatting motion (p > 0.710, RMSE = 0.465°). On the other hand, for patellar tilting (Fig. 3C), both conditions showed a similar tendency towards the external rotation of the patella to the femur throughout squatting (p > 0.421, RMSE = 0.507), while remaining within the internal rotation region. Patellofemoral kinematics in the native (black) and TKA (red) conditions during squatting motion. A Patella flexion, B abduction (medial (+)/lateral (−)), C tilting (medial (+)/lateral (−)), D anterior (+)/posterior (−) translation, E medial (+)/lateral (−) translation, and F inferior (−)/superior (+) translation. Data are represented as mean (solid) ± SD (shaded) across six specimens. The black arrows on the 3D model of the patella exhibit the motion direction relative to the femur, while blue solid lines display the axis of motion. A-P Anterior-posterior, M-L Medial-Lateral, I-S Inferior-Superior In addition, although there was no statistical difference observed between both conditions for patellar translations (Fig. 3D-F) in the anterior-posterior (p > 0.162, RMSE = 1.677 mm), medial-lateral (p > 0.155, RMSE = 2.541 mm), and inferior-superior directions (p > 0.099, RMSE = 1.330 mm), the largest offset between the two conditions was found in the medial-lateral translation, with the native condition showing an increased medial translation compare to the postoperative condition. Collateral ligament strains During passive flexion, the strain in the anterior bundle increased (lengthening or tensed) in the native and TKA conditions, with the former showing a greater lengthening. Furthermore, a statistical difference was observed between 55° and 68° (0.045 < p < 0.049, RMSE = 3.792%, 0.397 < ES < 0.358). For the middle bundle, although the native condition showed a slight lengthening until late mid-flexion, both conditions demonstrated decreased strain (shortening or relaxation) during the remaining range of motion, and no statistical difference was observed (p > 0.131, RMSE = 2.69%). In contrast, both conditions exhibited a similar trend for the posterior bundle, decreasing strain (shortening) throughout the entire motion, with no statistical difference (p > 0.127, RMSE = 2.04%). On the other hand, each bundle measured similar strain for LCL strain for both conditions until early mid-flexion. Moreover, the native condition consistently displayed decreasing strain compared to the postoperative condition after mid-flexion. Nevertheless, none of the measured outcomes reached statistical difference (p > 0.589, RMSE < 3.684%) (see Fig. 4). Absolute change in medial collateral ligament strain (%, top row) and lateral collateral ligament strain (%, bottom row) in anterior (blue), middle (red) and posterior (pink) bundle (e.g. antMCL) across six specimens during A, C passive flexion and B, D squatting for the native condition (solid), and following total knee arthroplasty using medially-stabilized GMK sphere (dashed). The data is represented as mean (solid and dashed) ± SD (shaded) across six specimens. The asterisk indicates a significant difference < 0.05, while the dashed black line illustrates the range of knee flexion where this significant difference is found. Moreover, the significant differences across groups are represented based on their associated colors. MCL medial collateral ligament, LCL lateral collateral ligament, ant anterior, mid middle, post posterior During squatting, the anterior (p > 0.078, RMSE = 1.542%), middle (p > 0.204, RMSE = 0.762%) and posterior (p > 0.351, RMSE = 0.827%) bundles of MCL resulted in increasingly similar strain behavior (shortening) from anterior to posterior in both native condition and postoperative conditions, with no statistical difference found. In contrast, the anterior, middle and posterior bundles of LCL exhibited a greater offset between the native and postoperative conditions after early mid-flexion, albeit the pattern was comparable. Moreover, significant differences were detected between the native and the postoperative conditions in terms of the strains in the anterior (69°−100°, 0.012 < p < 0.046, RMSE = 3.013%, 0.640 > ES > 0.505), middle (73°−93°, 0.037 < p < 0.048, RMSE = 3.022%, 0.797 > ES > 0.606) and posterior (61°−100°, 0.002 < p < 0.046, RMSE = 4.628%%, 1.138 > ES > 0.505) bundles of the LCL. Discussion The primary finding of this study was that overall tibiofemoral kinematics and collateral ligament strains demonstrated similar patterns throughout full motion range in passive conditions, as targeted using the Next-AR navigation system. Comparable tibiofemoral kinematic behaviors were also observed during squatting motion, albeit with a higher offset, especially in varus-valgus orientation and tibial internal-external rotation. In terms of collateral ligament strains during squatting, the native and TKA conditions exhibited decreasing and similar MCL strain in all bundles. LCL strain was increased after surgery both in passive flexion and squatting in all bundles particularly from early mid-flexion to deep flexion range. During passive motion, the native condition exhibited a marginally higher strain pattern on the MCL bundles (3.792% ≥ RMSE ≤ 2.04%), especially in the anterior and middle bundles. This non-significant difference between the native and TKA conditions could potentially be explained by a reduced varus orientation (RMSE = 0.763°) and increased tibial internal rotation (RMSE = 2.862°) in the native knee compared to TKA. Additionally, the increased anterior translation of the femur up to late mid-flexion observed in TKA may have further contributed to the disparity between the two conditions. As anticipated based on this kinematics behavior in the native and TKA conditions, the strain pattern on the LCL exhibited an opposite trend, with the TKA condition showing increased strain across all LCL bundles (RMSE < 3.684%) throughout knee flexion. This suggests that the kinematic variations in TKA influence the LCL in a manner analogous to the MCL for passive flexion. During squatting motion, the slightly reduced strain in all MCL bundles in the TKA condition, compared to the native condition, may be due to the increased varus orientation following TKA, which likely minimized tension on the medial side. Additionally, the greater tibial internal rotation in the TKA condition might lead to less direct engagement of the MCL bundles. Other studies using different bearing designs showed more pronounced elongation of the anterior MCL during flexion motion. They confirmed higher MCL strains at > 20° flexion, however, only in mechanically aligned TKA and not in kinematic TKA [19, 20]. In all studies, the primary determinant of ligament elongation was increased knee flexion. Another explanation for this difference in strain curves might be component design. Elongation of the anterior MCL from extension to 90° flexion was seen in active squatting motion of a CR design [8, 21, 22]. Posterior stabilized designs have been shown to double MCL strain during squatting in a cadaveric setup with peak strains at 60° flexion [23]. On the contrary, an in vivo gait study of 30 subjects compared PS, medial pivot and ultracongruent designs in terms of MCL and LCL changes. They found that component design did not demonstrate statistically significant effects on elongation patterns of collaterals. For the LCL bundles, the consistently higher strain (lengthening) observed in the TKA condition seemed to be related to the combined effects of increased varus orientation and tibial internal rotation. Another explanation for higher LCL strain curves might be the external rotation of the femoral component as performed in mechanical alignment. Together, these kinematic behaviors likely pull the lateral side further outward relative to the femur, placing additional stretch on the LCL. Although the native knee exhibited comparable kinematic behavior, the reduced varus orientation and increased posterior translation of the femur relative to the tibia may explain the decreased LCL strain (shortening) observed in the native condition. The effect of patellar kinematics and contact pressure on clinical function after TKA is an ongoing debate. It is hypothesized that higher contact pressures related to prosthesis design, femoral roll forward and internal rotation of the tibia can lead to postoperative anterior knee pain. Other in vitro studies showed higher reduction in mediolateral laxity after TKA and attributed this to overstuffing of the patellofemoral joint after patellar resurfacing [24, 25]. Our results showed a reduction in mediolateral translation of only 2 mm after surgery. Similar pre- and postoperative patellar kinematics concerning patellar tilt, -abduction, or -flexion were seen in our series indicating a correct restoration of anatomy and contact pressures. Our study shows some limitations. First, we had only a small number of specimens. Second, the Oxford-based physiological ex vivo knee simulator used in this study restricted analysis to 35°−100° of knee flexion. As a result, the squatting motion did not fully capture the knee joint kinematics and ligament behavior occurring in the early flexion range. Cadaveric preparation might influence joint- and ligament behavior. The use of the physiological knee simulator mimics loaded squatting motion, however the loads used in the muscles might not accurately reflect the natural loads experienced in vivo. We know that structural damage is occurring in ligaments from 5.14% strain levels and mechanoreceptors in ligaments have been documented [22], however the clinical value of ligament strain is unknown. On the other hand, a comparison between native and postoperative knees is valuable in terms of recreating native joint conditions. Conclusion We believe that Next-AR provides a useful tool for the surgeon to balance TKA and achieve targeted biomechanical properties. Appendix See Table 1. Table 1The knee alignment parameters for seven fresh-frozen full cadaveric legs, comparing the surgical planning (mechanical alignment technique) with the postoperative CT-based achieved alignmentTarget (mean ± std)Postoperative (mean ± std)p-valueRMSEHKA (°)180 ± 0178.5 ± 1.870.122.29Femoral varus (°)0 ± 00.5 ± 0.960.371.01Femoral flexion (°)1.93 ± 0.452 ± 1.040.990.98Femoral rotation (°)3 ± 02.86 ± 1.280.991.19Tibial varus (°)0 ± 00.79 ± 1.150.181.32Tibial slope (°)2 ± 01.64 ± 1.070.391.05HKA Hip-knee-ankle angle, mean average across specimens, std standard deviation, RMSE Root-Mean-Square ErrorStatistical analyses were performed using a non-parametric two-sided Wilcoxon Signed Rank testThe p-value < 0.05 indicates a significant difference. The knee alignment parameters for seven fresh-frozen full cadaveric legs, comparing the surgical planning (mechanical alignment technique) with the postoperative CT-based achieved alignment HKA Hip-knee-ankle angle, mean average across specimens, std standard deviation, RMSE Root-Mean-Square Error Statistical analyses were performed using a non-parametric two-sided Wilcoxon Signed Rank test The p-value < 0.05 indicates a significant difference. Author contributions A.V.D.V.: Investigation, Data Curation, Writing-Original Draft, Resources, O.T.: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing-Original Draft, Visualization, Project administration. G.P.: Conceptualization, Methodology, Investigation, Resources, Data Curation, Writing-Review & Editing, Supervision, Project administration, Funding acquisition. M.B.: Investigation, Writing-Review & Editing, Supervision, Project administration.: M.D.: Investigation, Writing - Review & Editing.: -T.L.: Software, Investigation, Writing - Review & Editing, Project administration. L.S.: Conceptualization, Methodology, Investigation, Resources, Data Curation, Writing-Review & Editing, Supervision, Project administration. Data availability No datasets were generated or analysed during the current study. Declarations Conflict of interest Geert Peersman receives funding as a consultant and speaker for Medacta. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.