1 Introduction

The practice of robot-assisted total knee arthroplasty (raTKA) has gained significant popularity in the last few years due to its potential for higher alignment accuracy and precision. ^1–4 There is an increasing number of robotic systems available, including both image-based and imageless technology.⁵ In image-based robotic systems, a preoperative supine computed tomography (CT) scan is used to generate a three-dimensional (3D) model of the patients’ bony anatomy. This enables the surgeon to make a preoperative patient-specific plan by planning the component size, position of the implant and alignment of the lower leg. Functional alignment is applied to restore the plane and obliquity of the knee joint and balancing the soft tissue by manipulating bone resection and implants position in three planes, eliminating the need for soft tissue release.^6,7 The advantage of raTKA is that alignment evaluation can be assessed intra-operatively through soft tissue balancing and quantitative feedback provided by the system.^3,8 A previous study has shown that platforms such as MAKO achieve a high accurate knee alignment and implant position on postoperative CT imaging, highlighting the technological advantages of robotic assistance.⁹

However, this study relies on non-weight-bearing CT scans to evaluate postoperative alignment. While CT provides highly detailed information about bone and implant orientation, it does not recreate the physiologic loading conditions of daily activities. Alignment parameters measured in non-weight-bearing versus weight-bearing conditions may therefore have differences which carry important clinical implications.^10,11 Differences between supine and upright radiographs in native knees have been previously studied, with variations up to 2.0°, which can possibly be explained by the gravity that created a moment in the knee joint.^11,12 However post-arthroplasty knees have stiff polyethylene inserts, yet differences can still arise from soft-tissue behaviour like ligament laxity or muscle force.¹¹ The effect of weight-bearing should therefore be understood by the surgeon, as achieving the intended functional alignment it is a key determinant of both implant survival and patient satisfaction.

The objective of the present retrospective study was to¹; determine the accuracy of an image based system by correlating the intra-operative coronal knee alignment with the postoperative alignment from standing, weight-bearing radiographs,² if the difference in postoperative coronal knee alignment could possibly be explained by asymmetry in ligament laxity, and³ if any differences are seen in the implant positions in postoperative weight-bearing radiographs.

2 Methods

This retrospective observational study screened the database of all patients who underwent robot-assisted total knee arthroplasty at a single hospital performed by a single surgeon, between 01/2023 and 04/2024. Included patients had surgery using the MAKO robot with implantation of a cemented Triathlon CS® prosthesis (Stryker, Kalamazoo, MI, USA). The functional alignment philosophy and patella-in-place balancer was used in each case.¹³

Patients were excluded if a different type of prosthesis was used, in cases of revision surgery or post-traumatic procedures, if insufficient data was captured during surgery, or if the postoperative radiographs did not meet the criteria for optimal alignment evaluation. No other exclusion criteria based on intra-operative alignment, comorbidities, age or body mass index were implemented. A total of 69 patients were included. This study was approved by the AZ Alma Ethics Committee (reference number CvE/EH/20241113-SVO).

Full-leg weight bearing anteroposterior radiographs and lateral knee radiographs were obtained preoperative and postoperative as standard of care at the hospital. A supine CT of the entire lower limb is taken preoperatively according to the MAKO Knee CT Scanning protocol.¹⁴ The mechanical hip-knee-ankle angle (mHKA), medial proximal tibia angle (MPTA), lateral distal femur angle (LDFA) and tibial slope angle are measured on the preoperative and postoperative radiographs. The femoral flexion angle is measured exclusively on the postoperative radiographs. All measurements were performed by a single observer. The ligament laxity in extension and all angles are captured intra-operatively by the MAKO system during final implant testing.¹⁵

The mHKA angle on pre- and postoperative radiographs was defined as the smallest angle formed by the intersecting lines of the mechanical axes of the femur (centre of the femoral head to the most proximal point of the intercondylar notch) and the mechanical axis of the tibia (tibial intercondylar eminence or centre of the tibial prosthetic plateau to the centre of the proximal talus).

Fig. 1 shows a magnified view of the landmarks used to measure the LDFA, MPTA, femoral flexion and tibial slope. The MPTA is the medial angle formed between the joint line of the proximal tibia and the mechanical axis of the tibia. Preoperatively, this joint line is defined in this study as the line connecting the deepest points of the medial and lateral tibial plateau. On the prosthesis, the underside of the tibial baseplate is used as the joint line. The LDFA is defined as the lateral angle formed between the mechanical axis of the femur and the joint line of the distal femur. The joint line connecting the most distal points of the medial and lateral condyle.

Close

Fig. 1 Radiographic measurements. Magnified view of the (A) preoperative and (B) postoperative landmarks of LDFA, MPTA and (C) postoperative landmarks of femoral flexion and tibial slope.

Export to PPT

Femoral flexion is calculated on lateral radiographs by subtracting the anterior angle between the mechanical axis of the femur and the distal plane of the femoral prosthesis from 90°. The tibial slope is the posterior angle between the mechanical axis of the tibial and the underside of the plateau of the tibia prosthesis subtracted from 90°. The mechanical axis in lateral view is determined by the intersecting line between the anterior and posterior cortex.

3 Results

Sixty-nine cases, 36 women (52%) and 33 men (48%), were included in the study, with 31 left knees (44.9%) and 38 right knees (55.1%). Mean age at surgery of 67.6 years ±10.9. Three patients had bilateral surgery. Preoperative radiographs were taken 95.1 days ±93.1 before surgery and postoperative radiographs were taken around first follow-up consultation on average 41.2 days +9.9 after surgery.

The mean preoperative mHKA angle was 2.4° ± 6.2° of varus, corrected intra-operatively to 2.2° ± 2.7° varus. Postoperative measurements show a mean mHKA of 1.4° ± 3.2° varus, not significantly different from the intra-operative values. Table 1 shows the mean perioperative values based on the five preoperative alignment groups. No significant intra-to postoperative mHKA differences were observed in any subgroup.

The surgical plan followed the functional alignment ideology. We aimed for a stable knee targeting 0 mm medial and 0 mm lateral laxity in extension. And 0 mm posteromedial laxity and up to 1 mm posterolateral laxity in flexion to accommodate for physiological lateral posterior translation.¹⁶ The maximal intra-operative captured laxity in extension was no more than 2 mm. Fig. 2 illustrates the relationship between the asymmetry in soft tissue laxity and postoperative mHKA change. Pearson correlation revealed no significant association (r = 0.038, p = 0.754). Asymmetry in laxity accounted for <0.1% of the variance in postoperative mHKA.

Close

Fig. 2 Scatterplot of postoperative mHKA difference on laxity asymmetry. Subtraction from intra-operative ligament laxity in extension (medial laxity subtracted from lateral laxity; positive value shows excess lateral laxity, negative value shows excess medial laxity) plotted on postoperative difference in mHKA. Linear fit line (R2 = < 0.001).

Export to PPT

The mean perioperative implant position for all cases and a detailed breakdown by preoperative mHKA subgroups is summarised in Table 2. The LDFA increased significantly from preoperative measurement (87.8° ± 3.2°) to intra-operative values (90.1° ± 2.2°), a mean increase of 2.3° towards varus alignment of the femoral prosthetic component. LDFA remained stable on postoperative radiographs with no changes within any subgroup. The original MPTA angle was maintained during the operation with a mean angle of 87.7° ± 1.7°. But the MPTA angle exhibits significantly (p < 0.005) less varus on postoperative measurements, with an increase of 0.9° to a mean angle of 88.6° ± 1.9. Postoperative measurements showed a significant (p < 0.001) decrease in the femoral flexion angle in contrast to the final implant position showing a mean decrease of 1.9°. The tibial prosthetic component was placed during surgery with an average of 1.9° less slope compared to the preoperative value. No significant difference was found between the intra-operative plan and postoperative radiographs, with a slope of 6.4° ± 2.1°.

4 Discussion

The main finding of this study was the high accuracy with which the planned knee alignment was reproduced on postoperative weight-bearing radiographs when using an image-based robotic system combined with a functional alignment philosophy and a patella-in-place balancer. The surgical goal of achieving stable knees with minimal extension laxity was met, and even with a maximal intra-operative laxity of 2 mm, the JLCA and mHKA remained consistent between the plan and postoperative measurements. A strong correlation between planned and measured LDFA confirmed stable coronal femoral alignment under load, while slight variations in MPTA were likely attributable to rotational or flexion differences during radiographic acquisition.

We found that LDFA increased from preoperative values toward a more neutral joint-line orientation intra-operatively, and that this alignment was fully maintained on postoperative weight-bearing radiographs. This confirms that coronal femoral component alignment is reproducible under physiological loading conditions. Comparable LDFA stability has been reported in the robotic validation studies of Glowalla et al.,¹⁷ who demonstrated a similar non-significant deviation of 0.1° on postoperative radiographs.

The mean preoperative LDFA of 87.8° ± 3.2° in this study reflects the slight valgus alignment as described in native knees, typically ranging from 0° to 4°.⁶ The aim of functional alignment is to restore the physiologic plane and joint line obliquity by manipulating bone resection and implants position in three planes without the need for soft tissue releases.^6,18,19 This may suggest that the femoral component is used to manipulate the gap balancing and tissue laxity as a mean intra-operative LDFA of 90.1° is used.⁷

These findings show that image-based robotic systems provide reliable femoral coronal alignment data. This is clinically important because suggest the distal femoral joint-line is a major determinant in overall limb alignment and contributes to coronal gap symmetry.

Our results show that preoperative MPTA matched the planned values, reflecting the intent to preserve the constitutional alignment, resulting in a mean MPTA of 87.7°. The subgroup analysis data supports the functional alignment strategy based on equal medial and lateral resection heights while compensating for the medial or lateral tibial bone defects as result of varus or valgus osteoarthritis.²⁰ Patients with (severe) valgus showed a tendency towards decreased planned MPTA, whereas those with (severe) varus show an increase. Oussedik et al.⁶ supporting this concept and further reported that balancing of a tight medial extension gap in functional alignment can be made by tibial resection of up to 3° of varus. Winnock de Grave et al.²¹ described the inverse kinematic alignment technique in which the functional alignment is based on tibia-first resection with equal medial and lateral resection compensating for cartilage and bone loss, with possible better clinical outcome and long-term survival over existing alignment strategies.

Postoperative radiographs showed a small but statistically significant shift of approximately 0.9° toward less varus. Similar postoperative increase in MPTA is reported in the studies of Glowalla et al.¹⁷ and Sires et al..⁹ Glowalla et al.¹⁷ reporting a reduction of 1.2° varus on postoperative radiographs, while Sires et al.⁹ found a 0.5° reduction in varus when comparing intra-operative plan with a postoperative supine CT. As these consistent reductions were observed both weight- and non-weight bearing, it is plausible that the MPTA is largely independent of loading conditions. The clinical relevance of this shift is relative, as functional alignment aims for stability rather than strict alignment parameters. Malrotation in postoperative radiographs could increase the varus alignment, as further discussed in the limitations section.

We found that postoperative mHKA closely matched planned values across all preoperative alignment categories, proving that image-based robotic are accurate even in weight-bearing situations if functional alignment strategies with stable knees are applied.

Previous studies of Glowalla et al.¹⁷ and Yang et al.²² have reported significant variations in postoperative mHKA. The retrospective study of Yang et al.²² including 100 consecutive varus deformity patients showing 0.8° more valgus deformity on postoperative radiographs (p = 0.009). The study Glowalla et al.¹⁷ prospectively analysed 36 patients with varus deformities undergoing MAKO raTKA and reported a 1.8° higher valgus deformity on postoperative radiographs (p < 0.001). Both studies reporting that patients with higher preoperative mHKA or increased tibial slope exhibit higher postoperative discrepancies. And suggest the surgeon to prevent excessive overcorrection in varus knees, as higher level of ligamentous imbalance is involved.²³

One of the questions that can be asked is why there was a significant difference between the intra-operative alignment and postoperative alignment, but not in our study. One of the explanations could be the balancing technique during our surgery. As a patella-in-place balancing technique is used, flexion first with patella-in-place and a maximal laxity of 2 mm that was allowed intra-operatively.¹³ Asymmetry in medial-lateral laxity could not explain a possible deviation from planned mHKA, as only 0.1% of mHKA difference could be explained.

These findings suggest the clinical implication that a maximal laxity of 2 mm measured with a patella-in-place balancing technique could assure an accurate mHKA measured by the image-based robotic knee system. These finding indicate that the knee alignment is being executed to the intra-operative plan. And that postoperative weight-bearing radiographs can still function as the golden standard to evaluate a coronal alignment after raTKA.

Femoral flexion showed a significant decrease of 1.9° between planned and postoperative measurements. This finding mirrors the results from the CT-based validation study of Sires et al.,²⁴ with a similar decreased flexion of 1.48°. Suggesting that weight-bearing does not fully account for the difference and other factors tend to play a role. It should be noted that in our study the lateral radiographs did not always capture a full view of the femur, and the intramedullary method was used to determine the mechanical axis. But the effect of bowing of the femur could potentially alter the measurements.²⁵ As CT imaging allows to reconstruct the axes of the MAKO robot, this consistency suggest that the observed reduction may reflect a genuine difference between intra-operative values and postoperative measurement, although definitive conclusions can not be made from this present study design. Although this does not impact coronal stability, it emphasizes the need for more precise 3D measurement tools.

The tibial slope showed no significant changes on postoperative measurements. However, it is important to consider that the MAKO robot references a transmalleolar sagittal axis through 3D reconstruction, whereas lateral radiographs use an intramedullary sagittal axis and more susceptible to landmark positions.¹⁵ Chalmers et al.²⁴ reported that the tibial slope using the transmalleolar method decreases 1.9° ± 1.3° compared to the intramedullary method. It is therefore important for surgeons to fully understand what landmarks and axes are used by the robotic system.²⁴

Taken together, these results demonstrate that image-based robotic systems provide accurate reproduction of planned alignment across the coronal alignment parameters. The combination of functional alignment, patella-in-place balancing and maximal laxity of 2 mm appears to produce a stable knee under weight-bearing conditions.

This study has several limitations. Alignment was measured on two dimensional (2D) radiographs who are sensitive for limb rotation and incomplete knee extension, in cases of flexion contractures. ^25–28 As demonstrated by Brunner et al.²⁸ who reported MPTA increase on standing radiographs with external rotation of the leg in full extension. And significant increase of MPTA when combining internal rotation and knee flexion. The systematic review of Ahrend et al.²⁹ reports that most studies show less varus of the knee with internal rotation, although no clear consensus is found. In cases of combined knee flexion (5° - 15°), changes in mHKA exceeding 2° are possible. Although LDFA remained unchanged in our study, this is consistent with the 3D analysis of Jamali et al.,³⁰ who found LDFA to be minimally affected by rotational change. As recommended by Lonner et al.,²⁵ standardised AP radiographs with clear visualisation of hip, ankle and patella remains essential.

Second, all components were implanted using cement. Uneven cement distribution could theoretically result in a minor deviation in final placement. But according to Sires et al. this variation would need to be substantial to produce a postoperative measurement difference of ≥1°.^9,31

Other limitations include that all procedures were performed by a single surgeon at one institution, and all measurements were performed by a single observer. These findings also apply specifically to the MAKO robotic platform, which may differ from other image-based systems.

Further studies could incorporate the use of weight-bearing CT imaging to generate postoperative 3D reconstructions, enabling reproducing of the axes used by the image-based robotic system. This approach would minimize the impact of limb rotation and knee flexion on the measurements, while retaining the weight-bearing advantages. Additionally, it would be possible to evaluate the accuracy of axial rotation of the components.

5 Conclusion

This study shows that a functional alignment philosophy in combination with a patella-in-place balancing device produces stable and reproducible limb alignment and implant positioning. Intra-operative limb alignment closely matched postoperative radiographic measurements, with no significant alignment changes and no meaningful influence of soft-tissue laxity asymmetry. The planned increases in LDFA and preservation of native MPTA were reliably achieved, with only small postoperative variations likely attributable to radiographic measurement differences rather than true mechanical changes. Overall, the accuracy and consistency observed support the combination of functional alignment philosophy with a patella-in-place balancers as a reliable approach for achieving targeted alignment in TKA, warranting further investigation into its long-term clinical impact.

Patient consent statement

Written informed consent was obtained from all patients included in the study. No minors or legally incapacitated individuals were involved.

CRediT author statement

Dries Verbeke: Conceptualization, Data Curation, Visualisation, Formal Analysis, Writing – Original Draft.

Stefaan Van Onsem: Conceptualization, Methodology, Resources, Supervision.

Hans Van den Wyngaert: Resources, Writing - Review & Editing.

Alex Demurie: Resources, Writing – Review & Editing.

Ignance Ghijselings: Resources, Writing – Review & Editing.

All authors reviewed and approved the manuscript.

Data availability statement

Data available on request due to privacy/ethical restrictions.

Permission to reproduce material from other sources

N/A.

Level of evidence

Level IV.

Ethical approval and patient consent

This study was approved by the ‘XXBLINDEDXX’ (reference number CvE/EH/20241113-SVO). All procedures were conducted in accordance with the Declaration of Helsinki and international ethical standards. Written informed consent was obtained from all patients included in the study. No minors or legally incapacitated individuals were involved.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.