1 Introduction

Total knee arthroplasty (TKA) is a highly effective intervention for end-stage knee osteoarthritis, offering substantial pain relief and functional improvements. However, a substantial subset of patients continues to experience suboptimal outcomes, including persistent pain, instability, and dissatisfaction.^1–3 These inconsistencies are often attributed to variations in surgical technique, intraoperative decision making, and soft tissue management.⁴ Unlike other joint arthroplasties, TKA involves a broad range of surgical workflows, with difference in bone cut sequencing, alignment strategies, and balancing techniques that significantly influence component positioning and joint function.^5,6

Two of the most commonly adopted surgical workflows in total knee arthroplasty (TKA) are measured resection (MR) and gap balancing (GB), each representing a distinct approach to achieving component alignment and soft tissue balance.⁷ The MR technique relies primarily on anatomical bony landmarks to determine femoral resections, with ligament releases performed as needed to restore balance.⁸ In contrast, GB prioritizes the native soft tissue envelope, using tensioning devices either mechanical or force-controlled to guide femoral component positioning after the tibia is resected.⁷ In the traditional femur-first MR workflow, distal and posterior femoral cuts are made before tibial planning, which can lead to flexion imbalance if ligament tension is suboptimal. Conversely, the tibia-first GB approach begins with the tibial resection and leverages soft tissue-guided planning to inform subsequent femoral alignment. Due to these differences in terms of the sequence of the bone cuts and the definition of the references, both the position and orientation of the final implants are expected to vary based on these workflow options. Although existing literature often emphasizes changes in femoral external rotation angle between these two techniques,^9–11 limited information is available about the impact on the other bone cut parameters and joint gaps.

Modern computer assisted orthopaedic surgery (CAOS) platforms provide surgeons with quantitative, real-time intraoperative feedback on implant positioning, alignment, and soft tissue balance, helping reduce variability introduced by manual techniques.^12,13 The recent integration of force-controlled distractors into these systems has further enhanced reproducibility and enabled a more data-driven approach to gap balancing, potentially leading to improved consistency and long-term outcomes.¹⁴ However, transitioning from one surgical workflow introduces a learning curve that can influence surgical execution and patient outcomes. These transitions may affect implant positioning parameters such as sagittal and coronal alignment, femoral component flexion, and joint line preservation, all of which are known to influence joint kinematics,¹⁵ load distribution,¹⁶ and implant survivorship.¹⁷

This study investigates the impact of transitioning from a femur-first MR workflow to a tibia-first GB workflow using a consistent CAOS platform augmented with a force-controlled distractor. We evaluate differences in femoral and tibial resection parameters, joint gaps, and patient-reported KOOS Jr. (Knee injury and Osteoarthritis Outcome Score for Joint Replacement) outcomes across three surgical cohorts: MR, GB early (first 75 cases), and GB late (subsequent 75 cases). By comparing cases from the early and later phases of GB adoption, we aim to characterize how surgical planning evolves with experience. In addition to technical factors, we assess the clinical impact of these workflow transitions on early patient outcomes, thereby linking intraoperative decisions with functional recovery.

2 Methods

2.2

2.2 Surgical workflows

The femur-first MR workflow is presented in Fig. 1a. After registration of landmarks using imageless CAOS system, the femoral cut parameters were planned based on predefined anatomical references following a mechanical alignment philosophy.¹⁸ The goal was to achieve neutral limb alignment without adjusting for soft tissue tension or joint line obliquity. Femoral planning and distal femoral cuts were performed first, followed by final femoral preparation. Subsequently, the tibial component was planned and resected based on varus/valgus, tibia slope, and thickness considerations, without soft tissue balancing prior to bony resections. Although the mechanical alignment techniques was used in MR cohort, small deviations from 0° in planned resection parameters reflect intraoperative anatomical variation, referencing limitations, and surgeon discretion while aiming to restore neutral mechanical alignment.

Close

Fig. 1 Overview of a) femur-first MR workflow with CAOS and b) tibia-first GB workflow with CAOS and force-controlled distractor.

Export to PPT

Fig. 1b outlines the tibia-first GB workflow. After attaching active tracking arrays to the femur and tibia, anatomical landmarks were registered using an imageless CAOS system. The proximal tibial resection was performed based on the surgeon's preferences for thickness, coronal alignment, and posterior slope. Following the control of the tibial cut, a force-controlled distractor (Newton, Exactech, Gainesville, FL, USA) was placed between the resected tibia and the native femur. This device applied a constant distraction force of 90 N per compartment (180 N total), independent of the gap size. The knee was moved through the full range of motion while the CAOS system recorded medial and lateral joint gaps (planned gaps). Based on these measurements, femoral resection parameters were planned to optimize soft-tissue balance, alignment, and sizing. After completing the femoral cuts, a trial component was placed, and the distractor was reinserted. The knee was again taken through the full range of motion, allowing the system to capture the relationship between the trial femoral component and the tibial cut surface. This yielded the second set of gap measurements (checked gaps), which were compared to the planned gaps. Final implant parameters were then recorded.

3 Results

3.1

3.1 Clinical outcomes

Demographic comparisons showed a statistically significant difference in patient age across the cohorts (ANOVA, p=0.012). Tukey HSD post-hoc analysis revealed that patients in the GB late group were significantly younger than those in the GB early group (mean difference = 4.1 years, p=0.0096). No significant age differences were observed between the MR and either GB group. Body mass index (BMI) did not differ significantly between cohorts (ANOVA, p=0.36), and all pairwise comparisons were non-significant (p<0.05).

KOOS Jr. outcomes across the three surgical cohorts are summarized in Fig. 2. Preoperative KOOS Jr. scores were comparable among the cohorts: MR (48.2±10.4), GB early (48.2±11.5), and GB late (45.5±13.1), indicating no significant difference (p=0.26). All cohorts demonstrated postoperative improvement in KOOS Jr. scores at one-year follow-up, with the GB late achieving the highest mean score (79.6±16.2), compared to GB early (75.9±13.9) and MR (75.5±15). The bottom panel of Fig. 2 shows KOOS Jr. improvement, with GB late showing the highest improvement (34.1±20.9), followed by GB early (27.7±16.1), and the MR (27.3±15.8). Statistical comparisons revealed that the GB late cohort improved by 6.4 points more than the GB-early cohort (p=0.036) and 6.8 points more than the MR cohort (p=0.025). There was no significant difference between GB early and MR cohorts with negligible mean difference of 0.4 points (p=0.89), suggesting that early adoption phase of the new workflow yielded comparable outcomes to those from the prior workflow.

Close

Fig. 2 Preoperative (top), postoperative (middle), and improvement (bottom) in KOOS Jr. scores by surgical group. ∗ indicates statistically significant pairwise differences based on Welch's t-test (p<0.05).

Export to PPT

3.2

3.2 Bone resection parameters

Fig. 3 summarizes the planned bone resection parameters across the three surgical cohorts. Planned femoral external rotation was similar in mean values across groups, with MR (2.9±1°), GB early (2.6±2.5°), and GB late (3.0±2.1°) cohorts. However, both GB cohorts showed significant greater standard deviation compared to MR (FK: p<0.05), indicating increased dispersion in surgical planning associated with GB technique. A similar pattern was observed in planned femoral varus/valgus alignment, although the mean values shifted closer to neutral from MR (−1.1±0.7°) to GB late (−0.3±1.7°), both GB early and GB late exhibited significantly higher standard deviations compared to MR (FK: p<0.05), suggesting broader intraoperative adjustments enabled by the GB workflow. Planned femoral flexion showed both increased mean values and variability in GB groups, with GB late (2.8±0.5°) significantly higher than MR (2±0.3°) and GB early (2.2±0.8°), indicating a trend toward greater femoral flexion over time as experience with the GB workflow increased (Welch, MU, FK: p<0.05). Planned tibial parameters also reflected this evolution: tibial slope increased from MR (3±0.2°) to GB late (3.8±0.3°), with significant difference in both mean and variance (Welch, MU, FK: p<0.05), while tibial varus/valgus alignment showed significant mean differences between GB cohorts and MR (Welch, MU: p<0.05) but relatively stable variability across groups. In contrast, planned tibial insert thickness did not differ significantly in mean or variance across cohorts, indicating consistent insert selection despite evolving bone cut strategies. Collectively, these results suggest that adoption of the GB workflow introduced greater flexibility and variability in planning, particularly in femoral parameters, reflecting intraoperative responsiveness to soft tissue tension and surgical learning over time.

Close

Fig. 3 Comparison of planned bone resection parameters across surgical cohorts. Panels show: (a) femoral external rotation, (b) femoral varus/valgus alignment, (c) femoral flexion, (d) tibial varus/valgus alignment, (e) tibial slope, and (f) tibial insert thickness. ∗ indicates significant differences in mean or distribution (Welch's t-test and Mann–Whitney U test, p < 0.05). † indicates significant differences in variance (Fligner–Killeen test, p < 0.05).

Export to PPT

3.3

3.3 Planned and checked gaps

Figs. 4 and 5 compare the planned and final (checked) joint gap profiles between the GB early and GB late cohorts, highlighting statistically significant differences in both medial and lateral compartments across the full arc of flexion. For planned gaps (Fig. 4), the GB late group exhibited consistently smaller medial and lateral gap values compared to the GB early group at nearly all flexion angles. This trend is further supported by the ANOVA results, which demonstrated a significant main effect of surgical group on both medial and lateral gaps (p < 0.001). Post hoc Tukey HSD tests revealed that GB late achieved significantly reduced planned medial gaps by an average of 0.59 mm (p < 0.001) and lateral gaps by 0.46 mm (p < 0.001), relative to GB early. These reductions suggest a shift in planning behavior as the surgeon gained experience with the GB workflow, favoring tighter balancing targets in both compartments. While there was no fixed balancing target applied across all cases, the gap balancing strategy evolved over time. The planning generally aimed for approximately 1–2 mm greater gap in flexion compared to extension. A slight increase in lateral laxity in flexion was also accepted in some cases, provided the component rotation remained within the acceptable range. These balancing goals were adjusted based on real-time gap measurements and feedback from CAOS system, reflecting an individualized approach to soft tissue.

Close

Fig. 4 Comparison of planned medial and lateral joint gaps across flexion angles between GB early and GB late cohorts (N = 70 per group). Left and middle panels show boxplots of gap values at each flexion angle; right panels show corresponding median profiles.

Export to PPT

Close

Fig. 5 Comparison of checked medial and lateral joint gaps across flexion angles between GB early and GB late cohorts (N = 70 per group). Left and middle panels show boxplots of gap values at each flexion angle; right panels show corresponding median profiles.

Export to PPT

A similar but slightly attenuated pattern was observed in the final checked gaps (Fig. 5). ANOVA again showed significant group effects for both medial and lateral compartments (p = 0.017), indicating that the differences seen in planning were largely preserved at final implant trialing. Tukey HSD confirmed that GB late knees had significantly smaller final medial gap values by 0.53 mm (p < 0.001) and lateral laxity by 0.29 mm (p = 0.047), compared to GB early. Median laxity curves (rightmost panels in Figs. 4 and 5) further illustrate these findings. In both planned and checked states, GB late curves shift leftward compared to GB early, denoting uniformly tighter gaps across the flexion arc. Overall, these gap results combined with the consistent mean insert thickness between the two GB groups suggest that as the surgeon progressed from early to late GB adoption, there was a deliberate reduction in both planned and final laxity targets, likely reflecting increased confidence in achieving joint stability without over-releasing soft tissues.

4 Discussion

All surgical cohorts showed significant one-year improvements in KOOS Jr. scores, confirming the overall effectiveness of TKA across workflows. Notably, the GB late cohort demonstrated the greatest improvement, with a statistically and clinically meaningful gain of 6.4–6.8 points compared to the GB early and MR groups (Fig. 2). The GB early cohort achieved outcomes comparable to the MR group, suggesting that the initial adoption of the GB workflow did not negatively impact clinical results. The progressive improvement from GB early to GB late likely reflects increased surgical proficiency and more deliberate application of gap-driven planning principles.

Transitioning to a GB workflow introduced greater variability and flexibility in femoral resection parameters. Both GB cohorts showed significantly increased standard deviations in planned femoral external rotation and femoral varus/valgus compared to MR (Fig. 3). This reflects a fundamental difference in philosophy: while MR adheres to fixed anatomical landmarks, GB allows femoral component positioning to be customized based on soft tissue tension. The observed increase in femoral external rotation variability aligns with the findings from prior studies.^14,21 In femoral flexion, GB late cases demonstrated both increased mean values and reduced variability, suggesting more confident and targeted use of flexion to optimize the flexion gap. Prior studies have shown that increasing femoral flexion improves posterior rollback and flexion stability while reducing the risk of mid-flexion instability.^22,23

Tibial resection parameters also evolved with the adoption of the GB workflow. Tibial slope increased progressively from the MR group to GB late, with both mean and variance significantly elevated during the early GB phase (Fig. 3). This trend likely reflects the surgeon's transition from using a fixed slope value in the MR workflow to a more personalized approach in the GB cohorts. As the surgeon gained experience and trust in gap balancing workflow, posterior tibial slope was adjusted intraoperatively to optimize flexion gap and accommodate soft tissue tension. Posterior tibial slope plays a critical role in sagittal plane kinematics, particularly influencing flexion mechanics. Multiple studies have demonstrated that increasing posterior slope facilitates greater posterior femoral rollback and reduces anterior tibial translation, contributing to improved knee flexion and stability.^24–26 Despite these adjustments in slope and coronal alignment, tibial insert thickness remained consistent across all cohorts, suggesting that joint line height was preserved even as bone cut strategies evolved.

Analysis of joint gaps revealed a progressive reduction in both planned and checked gaps in the GB late group. Across the flexion arc, GB late exhibited significantly tighter and more symmetric medial and lateral gaps compared to GB early. These refinements suggest improved confidence in soft tissue tensioning and greater control over final balance. Prior work has shown that tighter, more symmetric joint gaps are associated with better proprioception, coronal stability, and improved patient satisfaction.^27–29

The technical refinements observed in the GB late group, including increased femoral flexion, tighter joint gaps, and steeper tibial slope, may collectively explain the improved KOOS Jr. outcomes. Our findings are consistent with our prior study demonstrating that the use of a dynamic, force-controlled balancer significantly improved postoperative Knee Society Scores (KSS) compared to traditional static tensioning techniques.¹⁴ That study emphasized the importance of dynamic gap data in guiding femoral rotation throughout the arc of motion, ultimately resulting in more stable knees and better functional outcomes. Similarly, in the present study, progressive refinement of the GB workflow characterized by tighter joint gaps and more controlled femoral flexion and tibial slope was associated with superior KOOS Jr. outcomes. These findings align with existing literature suggesting that improved mid-flexion stability^30–32 enhances proprioception and patient-reported function, likely due to more consistent ligament tension and joint mechanics during everyday activities.

The progressive improvement in KOOS Jr. scores observed in the GB late cohort likely reflects more than just technical adjustments. It also signifies a deeper shift in surgical philosophy and increased trust in the new planning system. The MR workflow was grounded in traditional mechanical alignment principles, mirroring intramedullary and extramedullary instrument-based approaches. The early GB cohort represented a transitional phase, where the surgeon began exploring gap-driven planning but still relied partly on conventional alignment instincts. By the GB late phase, the workflow had fully embraced soft tissue–guided planning, with the surgeon more confidently adapting resection parameters based on dynamic gap measurements. This trust in the intraoperative data, and willingness to deviate from familiar angles, was essential in achieving more balanced and personalized outcomes. While the learning curve for operating the system may be short, developing confidence in the system's feedback and trusting its recommendations may take longer and is a key component of clinical success.

This study has several limitations. First, as a retrospective analysis, it did not include a priori power calculations to determine optimal sample sizes. Although efforts were made to standardize the cohorts, not all intraoperative parameters were available for every case due to incomplete surgical logs, and analyses were therefore limited to cases with complete data. To ensure balanced comparisons, a final sample size of 70 cases per group was used for most intraoperative parameters, while insert thickness analyses were restricted to 62 cases. In contrast, KOOS Jr. outcomes were available for all 75 cases per group. Relevant statistical tests were applied for each comparison to assess significance within these available sample sizes. Second, all surgeries were performed by a single high-volume surgeon, which improves consistency but limits the generalizability of the findings across institutions and surgeon experience levels. Third, joint gap measurements were only available for the GB cohorts. Standardized intraoperative gap data were not collected during the MR workflows, which limits the direct comparison across all groups. Future studies may benefit from capturing comparable gap data during MR procedures to enable more comprehensive analysis.

Although KOOS Jr. is a validated and widely used metric, longer-term outcomes and additional functional scores could further elucidate the clinical implications of these technical refinements. Future multi-surgeon, multi-center prospective studies with standardized data collection and broader functional assessments will be essential to validate these findings and confirm the long-term advantages of force-controlled gap balancing workflows.

5 Conclusion

This study demonstrates that transitioning from a femur-first measured resection (MR) workflow to a tibia-first gap balancing (GB) workflow using a computer-assisted platform with a force-controlled distractor enables more individualized and precise intraoperative decision-making. Over time, adoption of the GB technique was associated with refinements in femoral and tibial resection parameters and joint gaps profiles, resulting in improved early clinical outcomes. These findings support the clinical value of dynamic, soft tissue-guided planning in total knee arthroplasty and emphasize the importance of surgical learning and intraoperative feedback in optimizing outcomes. While the results are promising, further prospective, multi-surgeon studies are warranted to confirm the generalizability and long-term benefits of force-controlled GB workflows.

Patient consent statement

Not applicable.

Data availability statement

Data is available on reasonable request by contacting the corresponding author.

Permission to reproduce material from other sources

Not applicable.

Author's contributions according to CRediT taxonomy-

PTC: conceptualization, data curation, formal analysis, writing-original draft, writing-review and editing; LDA: conceptualization, writing-review and editing, visualization, formal analysis; AJ: data curation, writing-review and editing; CAJ: writing-review and editing, formal analysis, methodology.

Ethical statement

Not applicable.

Guardian or Patients consent

Not applicable.

Funding statement

No funding was provided for this work.