Deviation from native medial posterior tibial slope and dose-response relationships with forgotten joint score-12 in Robot-assisted versus conventional manual total Knee arthroplasty: A retrospective cohort study Authors: Boyi Chen, Lin Jing, Hongmei Zhang, Qi Yan, Yuanyuan Li, Siye Liu, Ximing Yu Published: 2026 ABSTRACT AbstractBackground Robot-assisted total knee arthroplasty (TKA) enables high-precision osteotomy, but the clinical choice between pursuing a population-wide optimal posterior tibial slope (PTS) and individualizing the restoration of the native anatomical PTS remains debated. This study quantified the dose-response relationship between PTS deviation from native anatomy and the Forgotten Joint Score-12 (FJS-12), comparing outcomes between robot-assisted and conventional manual TKA.MethodsA retrospective cohort study analyzed 240 patients who underwent primary unilateral posterior-stabilized TKA, divided equally into conventional manual (n = 120) and robot-assisted (n = 120) groups. Postoperative PTS and absolute osteotomy errors were measured via CT scans. At the 1-year follow-up, the primary outcomes assessed were maximum active flexion and FJS-12 scores. Multivariate linear and quadratic regressions were utilized to map the association between postoperative PTS, absolute native deviation, and functional outcomes. Subgroup analysis was conducted using a 2° PTS deviation (Δ) threshold.Results The robot-assisted group demonstrated a significantly lower absolute osteotomy error than the conventional group (1.2° ± 0.9° vs. 2.8° ± 1.6°, P < 0.001). Although the robotic cohort showed statistically greater maximum flexion and FJS-12 scores, objective functional metrics (KSS, OKS, WOMAC) did not significantly differ between groups. Regression analysis revealed a linear correlation between PTS and flexion (β = 1.9) and a quadratic (inverted U-shaped) relationship between PTS and FJS-12, peaking at approximately 5.3°. Crucially, greater deviation from the native PTS independently impaired FJS-12 scores (β = − 2.4). Subgroup analysis confirmed that patients with a precise reconstruction (Δ ≤ 2°) achieved distinctly higher FJS-12 scores compared to high-deviation groups (Δ > 2°), regardless of the surgical modality employed.ConclusionsThe postoperative PTS exhibits a nonlinear relationship with subjective joint awareness, with a population-level optimum of approximately 5.3°. However, anatomical deviation from the native PTS is an independent predictor of worsening FJS-12 scores. While robotic assistance is highly effective at minimizing extreme biomechanical outliers, surgical planning should prioritize the precise restoration of individualized native anatomy over universally fixed targets to maximize patient satisfaction.Keywords: Total knee arthroplasty, Posterior tibial slope, Robot-assisted surgery, Forgotten joint score, Individualized reconstruction, Regression analysis Introduction In modern artificial joint surgery, the ultimate clinical goal of total knee arthroplasty (TKA) has evolved from simply pursuing prosthesis survival to achieving excellent functional recovery and patient satisfaction. Although prosthesis design and surgical techniques have matured, approximately 15–20% of patients remain dissatisfied with their prognosis. This “subjective dissatisfaction” is typically attributed to loss of proprioception, limited flexion, or heightened self-awareness during joint motion (i.e., “foreign body sensation”) [1]. Among the parameters regulating postoperative kinematics, the posterior tibial slope (PTS) has emerged as a core research focus because of its decisive role in flexion-gap capacity, the femoral rollback trajectory, and extension-flexion balance [2–4]. PTS is defined as the posterior tilt angle of the tibial plateau relative to the tibial mechanical axis in the sagittal plane. Previous biomechanical studies have shown that a moderate PTS increase effectively enlarges the flexion gap and improves the knee range of motion (ROM); however, excessive posterior tilt may reduce mid-flexion stability and increase the risk of prosthesis wear and loosening, whereas insufficient PTS reconstruction often leads to a tight flexion gap and restricted motion [3, 5]. Therefore, precise intraoperative PTS reconstruction is critical for optimizing postoperative function. In conventional TKA, preparing the PTS relies heavily on manual instrumentation, such as intramedullary guides. Because these guides conform to the inherent anterior bowing of the tibial shaft, they often inadvertently introduce an intrinsic posterior slope. While robot-assisted total knee arthroplasty (RATKA) has been introduced with the goal of improving surgical precision, the outcomes are highly dependent on the specific technology employed. For instance, imageless systems have sometimes demonstrated less precision compared to experienced hand instrumentation. Conversely, CT-based robotic systems (e.g., MAKO) utilize a true mechanical axis rather than an intramedullary reference; consequently, they tend to produce a decreased posterior slope relative to manual techniques unless the surgeon explicitly accounts for this biomechanical discrepancy and targets the patient’s native anatomy during preoperative planning. Furthermore, while certain RATKA systems may offer enhanced radiographic accuracy, it is currently widely acknowledged that this radiographic precision does not necessarily translate into long-term clinical benefits or improved implant survivorship. Because accurately reconstructing this native anatomy requires a specific and reliable baseline, in this study, all radiological measurements and subsequent surgical planning targets were specifically defined by the native medial posterior tibial slope. The medial slope was chosen as the primary focus because it dictates medial pivot kinematics, serves as the primary anatomical reference in our 3D robotic planning algorithms, and ensures measurement consistency with historical 2D radiographic studies where the medial plateau contour predominantly forms the visible joint line. However, whether—and to what exact degree—specific deviations from this native medial PTS independently impact sensitive, patient-reported functional outcomes, such as the Forgotten Joint Score-12 (FJS-12), has not been systematically investigated. With the increasing importance of patient-reported outcome measures (PROMs) in joint surgery evaluation, the Forgotten Joint Score-12 (FJS-12) has become a key indicator of TKA postoperative function. By quantifying how often patients “forget” the artificial joint in daily life, the FJS-12 is considered more sensitive than traditional scores for detecting proprioceptive recovery and psychological integration. Therefore, on the basis of a retrospective cohort, this study aims to elucidate the dose‒response relationship between PTS reconstruction accuracy and FJS-12 scores and to explore whether the precision boundary provided by robotic technology can substantially improve patients’ postoperative subjective “forgotten” experience by precisely restoring the individualized native PTS and crossing the minimal clinically important difference (MCID) threshold. Methods Study design This investigation adhered to a single-center retrospective cohort design and was approved by the hospital’s ethics committee. Patient data from patients who underwent primary unilateral TKA from August 2021 to June 2024 were retrospectively analyzed. Inclusion criteria (1) Primary knee osteoarthritis with Kellgren–Lawrence grade III–IV [6]; (2) primary unilateral TKA; (3) age 50–80 years; (4) preoperative flexion contracture < 15°; (5) complete preoperative and postoperative imaging follow-up data. Exclusion criteria (1) History of ipsilateral knee surgery; (2) inflammatory arthropathy (e.g., rheumatoid arthritis); (3) severe osteoporosis or neuromuscular disease; (4) posterior cruciate ligament (PCL) insufficiency; (5) missing follow-up data. A total of 240 patients met the inclusion criteria and were included. Surgical technique The enrolled patients were divided into a conventional group (n = 120) and a robot-assisted group (n = 120). Grouping decisions were based on patients’ personal preferences and the availability of intraoperative resources; patients in the robot-assisted group actively chose to undergo the robot-assisted surgery system (RATKA), while the remaining patients entered the conventional surgery control group. This study was a nonrandomized retrospective study. All surgeries were performed by the same senior joint surgery team. A midline knee incision and the medial parapatellar approach were used uniformly [7]. In this study, the surgical philosophy for both the conventional and robot-assisted groups was strictly based on the restoration of standard mechanical alignment (MA), aiming for a neutral overall lower limb mechanical axis. Regarding component rotational alignment, the femoral component rotation was targeted parallel to the surgical transepicondylar axis (sTEA). In the conventional group, this was achieved using intramedullary guidance and a preset external rotation angle of 3° [8]. The tibial component rotation was aligned using the medial third of the tibial tubercle as the primary anatomical landmark. For the tibial osteotomy in the conventional group, standard extramedullary positioning instruments were used. All robot-assisted procedures were performed using the Joint Replacement Surgical Navigation and Positioning System (Hangzhou Jianjia Medical Technology Co., Ltd., Hangzhou, China). For these patients, preoperative 3D modeling was generated based on high-resolution computed tomography scans (slice thickness: 1.0 mm) obtained using a Siemens CT scanner (Siemens Healthineers, Erlangen, Germany) at our outpatient radiology department. Utilizing this personalized planning and real-time intraoperative optical navigation feedback, the haptic limiting technology of the robotic arm was used to precisely execute the osteotomy plan [9]. All patients in both groups received the Double Medical Posterior Stabilized (PS) Total Knee Prosthesis (Double Medical Technology Inc., Xiamen, China), which was fixed with bone cement. Postoperative rehabilitation Both groups followed a standardized perioperative pathway, including multimodal analgesia and routine thromboprophylaxis [10]. Quadriceps isometric exercises began on postoperative day 1; partial weight-bearing with a walker started on day 2. Progressive ROM training was supervised by rehabilitation therapists thereafter. Outcome measures Primary outcome measures Maximum active knee flexion angle and FJS-12 score at 1 year post-operatively. Flexion angle measurement: The measurement was performed independently by two senior physicians who did not participate in the surgery via a standard goniometer (range 0–180°, precision 1°). During measurement, the subject was placed in the supine position, and the affected knee was actively flexed to the limit angle, with the readings recorded. For each patient, each of the two physicians repeated the measurement 3 times; the mean of each physician’s measurements was taken, and then the average between the two physicians was calculated as the final outcome data. Reliability testing: To verify the consistency of the measurements, 30 cases were randomly selected (approximately 12.5% of the total number of follow-up cases). The above two physicians performed independent measurements after a 1-week interval. The intraclass correlation coefficient (ICC) was used to assess test‒retest reliability. According to the evaluation criteria, an ICC value > 0.75 indicates good reliability, whereas 0.50–0.75 indicates moderate reliability. Forgotten Joint Score: The Chinese version of the FJS-12 scale was used to evaluate postoperative proprioceptive recovery. The scale contains 12 items, with a total score ranging from 0 to 100 points. A higher score indicates lower postoperative joint awareness, that is, a better “forgotten” state. The FJS-12 has been validated in multiple cultural backgrounds, demonstrating excellent reliability and validity with a lower ceiling effect; its minimal clinically important difference (MCID) is defined as 13.7–16.6 points. Secondary outcome measures KSS, OKS, WOMAC index, anterior knee pain VAS, tibial anterior translation (stability), and perioperative/follow-up complication rates. Assessment of osteotomy accuracy The actual PTS was measured on postoperative CT. Osteotomy error (absolute error) = |actual PTS – target PTS|, with the target PTS determined from preoperative digital planning. Statistical analyses IBM SPSS 26.0 was used. Continuous variables are presented as the means ± SDs and were compared with independent-samples t tests; categorical variables are presented as frequencies (percentages) and were compared with χ² tests or Fisher’s exact tests. P < 0.05 was considered statistically significant. Propensity score matching (PSM): To balance baseline covariates between groups and reduce selection bias, PSM was performed at a 1:1 ratio to match the conventional and robot-assisted groups [11]. The propensity score model included age, sex, BMI, preoperative KSS score, preoperative flexion angle, degree of preoperative deformity, and preoperative native posterior tibial slope. Nearest-neighbor matching was used, with the caliper value set at 0.02. Balance after matching was assessed via the standardized mean difference (SMD), with SMD < 0.10 considered indicative of good balance between groups. Minimal clinically important difference (MCID): This study introduced the MCID as the threshold for clinical efficacy evaluation to distinguish statistical significance from clinical relevance. In accordance with previous studies, the MCID for FJS-12 was set at 12 points [12], and the MCID for the maximum active knee flexion angle was set at 5°. Core variable construction and regression analysis: To quantitatively evaluate the effect of deviation between actual osteotomy values and anatomical morphology on postoperative function, the core index deviation (Deviation) was calculated as Deviation = |Actual PTS − Native PTS|. Multivariate linear regression was used to analyze the influencing factors of functional outcomes, with the FJS-12 score as the dependent variable. The independent variables of the model included (1) actual postoperative PTS (containing both linear and quadratic terms to capture potential nonlinear effects); (2) deviation (continuous variable); and (3) covariates (age, sex, BMI, preoperative flexion angle, and surgical method). The partial regression coefficients (β) and their significance levels for each variable were used to explore the independent contributions of “population-wide target value” and “individualized native restoration” to patients’ subjective “forgotten” joint experience [13]. Sensitivity analysis: To further assess the potential independent interference of coronal alignment on clinical function and to verify the robustness of the model conclusions, postoperative lower-limb alignment deviation (the absolute value of the HKA angle deviation from the neutral position) was added as an additional control variable into the regression equation for sensitivity analysis [14]. Reliability testing: The intraclass correlation coefficient (ICC) was used to evaluate the test-retest reliability of knee flexion angle measurements, with an ICC > 0.75 considered indicative of good reliability [15]. Results Patient demographics and propensity score matching Initially, a total of 312 patients who underwent primary unilateral TKA between August 2021 and June 2024 were reviewed. After applying the strict inclusion and exclusion criteria, 286 eligible patients remained in the initial cohort. To mitigate selection bias and balance baseline confounders between the surgical modalities, a 1:1 propensity score matching (PSM) algorithm was executed. This rigorous matching process yielded a final cohort of 240 patients, comprising exactly 120 patients in the conventional group and 120 in the robot-assisted group (Fig. 1). Following PSM, baseline characteristics—including age, sex, BMI, preoperative flexion, KSS score, OKS score, WOMAC score, deformity, and native PTS—were excellently balanced with no significant intergroup differences (all P > 0.05) (Table 1). Flowchart of patient selection and propensity score matching (PSM). TKA: total knee arthroplasty Table 1Patient baseline characteristics (mean ± SD)IndicatorConventional(n = 120)Robot-assisted(n = 120)t/χ² P Age(years)67.5 ± 7.967.9 ± 8.10.380.704Sex(male/female)48/7252/680.270.603BMI(kg/m²)27.4 ± 3.727.7 ± 3.80.620.536Preoperative flexion(°)108.3 ± 10.5109.1 ± 10.80.580.562Preoperative FJS-1222.4 ± 5.122.8 ± 4.90.620.535Preoperative KSS48.3 ± 12.149.0 ± 12.50.440.660Preoperative OKS24.2 ± 5.524.6 ± 5.80.550.583Preoperative WOMAC52.5 ± 11.353.1 ± 11.80.400.689Preoperative deformity(°)8.6 ± 3.78.9 ± 3.90.610.542Native PTS(°)6.1 ± 2.36.3 ± 2.40.660.510 Patient baseline characteristics (mean ± SD) Osteotomy accuracy The preoperative target PTS did not differ significantly (5.8° ± 1.2° vs. 6.1° ± 1.3°, P = 0.065). The actual postoperative PTS was 6.8° ± 2.5° (conventional) vs. 6.5° ± 1.8° (robot-assisted) (P = 0.290). The absolute error was 2.8° ± 1.6° vs. 1.2° ± 0.9° (t = 9.46, P < 0.001) (Table 2). Table 2Osteotomy accuracy comparison (mean ± SD)IndicatorConventional(n = 120)Robot-assisted(n = 120)t P Target PTS(°)5.8 ± 1.26.1 ± 1.31.850.065Actual PTS(°)6.8 ± 2.56.5 ± 1.81.060.290Osteotomy error(°)2.8 ± 1.61.2 ± 0.99.46< 0.001 Osteotomy accuracy comparison (mean ± SD) Reliability of the flexion angle measurement For 30 patients, the interrater ICC was 0.89 (95% CI: 0.82–0.94); the intrarater ICCs were 0.92 (Physician A) and 0.91 (Physician B) (Table 3). Table 3Flexion angle measurement reliabilityReliability typeICC95% CIRatingInterrater0.890.82–0.94GoodIntrarater (Physician A)0.920.86–0.96GoodIntrarater (Physician B)0.910.85–0.95Good Flexion angle measurement reliability Clinical functional outcomes at 1 year post-surgery The robot-assisted group had greater maximum flexion (121.5° ± 9.2° vs. 118.3° ± 10.1°, P = 0.011) and FJS-12 (68.7 ± 15.8 vs. 64.2 ± 16.3, P = 0.030) values. No significant differences were found in the KSS, OKS, WOMAC, tibial translation score, or anterior knee pain VAS score (all P > 0.05) (Table 4). Table 41-year clinical outcomes (mean ± SD)IndicatorConventional(n = 120)Robot-assisted(n = 120)t P Maximum flexion (°)118.3 ± 10.1121.5 ± 9.22.550.011FJS-1264.2 ± 16.368.7 ± 15.82.180.030KSS85.3 ± 11.586.9 ± 11.81.060.290OKS40.1 ± 5.841.3 ± 5.91.580.115WOMAC19.8 ± 8.718.5 ± 8.41.180.239Tibial anterior translation (mm)5.6 ± 2.05.4 ± 1.90.800.424Anterior knee pain VAS1.4 ± 1.71.3 ± 1.60.470.639 1-year clinical outcomes (mean ± SD) Multivariate linear regression After adjusting for age, sex, BMI, and preoperative flexion, (1) PTS and flexion were positively linearly correlated (β = 1.9, 95% CI: 1.2–2.6, P = 0.008). (2) PTS and FJS-12: quadratic relationship (quadratic term β = − 0.95, 95% CI: − 1.58 to − 0.32, P = 0.012); peak at PTS = 5.3°(Fig. 2). (3) Deviation and FJS-12 score: negative correlation (β = − 2.4, 95% CI: − 3.8 to − 1.0, P = 0.006). (4) Sensitivity analysis for lower-limb alignment deviation (HKA): Postoperative HKA deviation was included in the model as an additional covariate. The results revealed that a quadratic relationship between actual PTS and the FJS-12 score (β = −0.89, P = 0.018) and a negative correlation between deviation and the FJS-12 score (β = −2.1, P = 0.012) still existed. In addition, HKA deviation was negatively correlated with the FJS-12 score (β = −1.8, 95% CI: −3.1 to −0.5, P = 0.009). Since the intraoperative soft-tissue pressure distribution was not quantified, this factor was not included in the model. (5) Group differences did not exceed MCID thresholds (flexion difference 3.2° < 5° MCID; FJS-12 difference 4.5 points < 12 points MCID). Scatter plot demonstrating the nonlinear dose-response relationship between actual postoperative posterior tibial slope (PTS) and the Forgotten Joint Score-12 (FJS-12). The quadratic trendline reveals an inverted U-shaped distribution, indicating that subjective joint awareness peaks when the postoperative PTS is restored to approximately 5.3° Subgroup analysis by surgical modality and PTS deviation To further isolate the influence of reconstruction accuracy from the surgical modality, clinical outcomes were evaluated across four subgroups based on the surgical method and the 2° deviation (Δ) threshold. The analysis revealed that patients with a precise PTS reconstruction (Δ ≤ 2°) achieved distinctly higher FJS-12 scores compared to the high-deviation groups (Δ > 2°), regardless of whether conventional or robotic techniques were utilized. In contrast, objective functional metrics, including maximum flexion, KSS, and OKS, did not exhibit significant variations across the four sub-cohorts (Table 5). Table 5Clinical outcomes stratified by surgical method and PTS deviation amplitudeOutcome MetricConventional Δ ≤ 2°(n = 13)Conventional Δ > 2°(n = 107)Robotic Δ ≤ 2°(n = 89)Robotic Δ > 2°(n = 31)FJS-12 Score70.5 ± 14.863.4 ± 16.371.3 ± 15.765.2 ± 15.8Maximum Flexion (°)119.8 ± 9.6118.1 ± 10.2120.9 ± 9.5123.2 ± 8.8KSS Score86.1 ± 10.585.2 ± 11.686.6 ± 11.485.2 ± 12.5Δ indicates absolute deviation from native PTS. Clinical outcomes stratified by surgical method and PTS deviation amplitude Δ indicates absolute deviation from native PTS. Adverse events There were no significant differences in infection, deep vein thrombosis, or stiffness rates (all P > 0.05) (Table 6). Table 6Adverse events [n (%)]EventConventional(n = 120)Robot-assisted(n = 120)χ² P Infection2 (1.7)1 (0.8)0.340.560Deep vein thrombosis2 (1.7)2 (1.7)0.001.000Stiffness4 (3.3)3 (2.5)0.150.699Total8 (6.7)6 (5.0)0.310.578 Adverse events [n (%)] Discussion This retrospective cohort study compared conventional and robot-assisted TKA with respect to PTS reconstruction accuracy and examined the relationship between postoperative PTS and clinical function. Robot-assisted surgery significantly reduced the osteotomy error (1.2° ± 0.9° vs. 2.8° ± 1.6°, P < 0.001), which is consistent with prior reports [16, 17]. The advantage extends beyond sagittal PTS to coronal alignment, femoral rotation, and soft-tissue balance. Furthermore, precise component-to-component rotational alignment is increasingly recognized as a critical determinant of postoperative kinematics and patellofemoral tracking. As recently highlighted by Strahovnik et al. [18], three-dimensional evaluation of rotational alignment in robotic TKA provides superior accuracy and correlates with distinct biomechanical advantages, underscoring the necessity of strict rotational targets in surgical planning. Conventional techniques rely on intramedullary or extramedullary references and are strongly affected by anatomical variation, whereas the robotic system achieves higher prosthesis implantation accuracy through preoperative three-dimensional digital modeling and intraoperative real-time optical navigation feedback [16]. Importantly, we transparently acknowledge that while the overall FJS-12 scores were statistically higher in the RATKA group (68.7 ± 15.8 vs. 64.2 ± 16.3, P = 0.030), the absolute mean difference of 4.5 points failed to reach the established minimal clinically important difference (MCID) threshold of 12 points. Although the MCID is conventionally utilized to assess within-patient longitudinal improvements, applying it as a strict benchmark for inter-group comparisons in this context suggests that this statistical advantage may not reliably translate into a distinctly perceptible functional benefit for the patient. Consistent with the lack of significant differences in traditional objective scores (KSS, OKS, and WOMAC), this echoes the growing consensus in the literature that enhanced robotic radiographic precision does not automatically guarantee universally superior clinical scores across an unstratified population. Furthermore, the absolute FJS-12 scores observed in our cohort (averaging in the mid-60s) warrant careful contextualization. While these scores may appear modest compared to certain highly selected or longer-term follow-up cohorts reported in the literature, they accurately reflect the real-world outcomes of an unselected, consecutive patient series. This scoring profile can largely be attributed to the relatively short 1-year follow-up duration of our study; current evidence indicates that joint awareness and proprioceptive integration often require two to three years to fully plateau following TKA. Additionally, baseline patient expectations, severity of preoperative functional deficits, and inherent cultural variations in interpreting and responding to subjective “forgotten” states likely contribute to this conservative scoring pattern. Regarding the linear relationshipbetween the postoperative PTS and knee flexion range of motion regression analysis revealed that for every 1° increase in the postoperative PTS, the maximum flexion angle increased by approximately 1.9° (β = 1.9). This conclusion validates the regulatory role of the PTS in the knee flexion dimension: increasing the posterior slope can effectively enlarge the flexion gap and reduce posterior bony impingement between the femur and tibia, thereby improving flexion activity. Biomechanical and finite element analyses also support this view, considering that a moderate increase in PTS is a key means to increase the postoperative flexion range (ROM) of PS-type prostheses [19]. However, despite the theoretical benefits of PTS optimization and the higher precision of robotic systems, the actual clinical gain in ROM observed between the surgical modalities was limited. Regarding postoperative range of motion, although the maximum flexion was statistically higher in the RATKA group compared to the conventional group (121.5 ± 9.2° vs. 118.3 ± 10.1°, P = 0.011), the absolute mean difference was only 3.2°. We calculated the effect size to be small (Cohen’s d = 0.33). Given that the minimal clinically important difference (MCID) for knee flexion is generally considered to be between 5° and 10°, this 3.2° difference, while statistically significant due to the sample size, is unlikely to translate into a perceptible functional advantage for the patient. By de-emphasizing the general superiority of RATKA, this study sharply focuses on a more critical, novel finding: the dose-response relationship between the deviation from the native medial PTS and subjective joint awareness. Regression analysis revealed a quadratic (inverted U) relationship between PTS and FJS-12 (peak at 5.3°), aligning closely with the cohort’s mean native PTS (≈ 6°). More critically, deviation analysis demonstrated that greater PTS deviation from native anatomy independently lowered FJS-12 scores (β = − 2.4). Therefore, we hypothesize that the true clinical value of robot-assisted systems lies not in uniformly elevating the mean postoperative FJS-12 score across all patients, but rather in their ability to reliably hit the target slope and prevent severe biomechanical outliers (e.g., > 3° or > 5° deviation). Preventing these extreme deviations is paramount, as they directly correlate with significantly worsened subjective outcome measures.Hierarchical influence of the PTS deviation amplitude (Δ) on subjective perception. The results of the subgroup stratified analysis revealed that when the postoperative PTS deviation was controlled within 2° (Δ ≤ 2°), patients’ FJS-12 scores exhibited a significant hierarchical advantage. This finding defines the “precision threshold window” for subjective functional benefit and confirms that the anatomical reduction accuracy of PTS is the key factor determining the postoperative “natural feel.” The extremely high achievement rate of the robot-assisted group within this window (87.3%) provided empirical support for its significant advantage in FJS-12 scores. Although the 2° threshold was set on the basis of clinical experience, its mutual corroboration with the continuous-variable regression results further reveals the value of “anatomical reconstruction” in subjective functional recovery [13]. Anatomical and biomechanical value of restoring the native posterior tibial slope This study revealed that the FJS-12 score in the Δ ≤ 2° group was significantly superior, indicating that precise anatomical alignment can more effectively place subjects into the “subjective functional benefit window.” From a biomechanical perspective, restoring the PTS close to the physiological state helps preserve joint proprioception and enhances patients’ bodily cognition of the artificial prosthesis. Given the significant anatomical heterogeneity of the native PTS in the population, compared with pursuing a population-wide target value, precise reconstruction based on individualized anatomical features has far-reaching clinical significance for maintaining the native kinematic characteristics of the knee joint and improving mid- to long-term postoperative function [13]. Analysis of inconsistency between objective domains and subjective sensitivity In this study, secondary outcome indicators, such as the KSS score, Oxford Knee score (OKS), and WOMAC, showed no statistically significant intergroup differences, which is in sharp contrast with the statistically significant differences in the primary outcomes (FJS-12 score and flexion angle). The potential mechanisms may include the following: (1) FJS-12 has greater detection sensitivity for subtle differences in joint perception, whereas traditional functional scales are limited by the “ceiling effect,” making it difficult to distinguish microdifferences among high-functioning patients [20]; (2) the clinical benefits of robot-assisted technology have certain functional specificity, mainly manifested in the patients’ subjective perception dimension rather than comprehensive objective functional domain improvement. In addition, this finding also suggests that within the retrospective study framework, differences in FJS-12 scores may be influenced by potential confounders such as selection bias, and causal inference should be interpreted with caution. Study limitations This study has the following limitations: (1) The primary limitation of this study is the exclusive use of posterior-stabilized (PS) implant designs. In PS TKA, the cam-post mechanism requires a standardized, decreased posterior slope (typically 0° to 3°) to prevent post impingement and flexion instability. Consequently, true individual restoration of the native slope is biomechanically contraindicated in patients with steep native anatomy. However, this design constraint highlights a critical clinical dilemma: patients with a high native PTS are forced to undergo a significant anatomical deviation when receiving a PS implant. Our findings demonstrate that when this forced deviation becomes too large, the FJS-12 significantly deteriorates. This suggests that the anatomical compromise inherent to PS designs may negatively impact joint awareness in this specific patient subset. Future studies comparing PS and cruciate-retaining (CR) implants—which may more safely accommodate the restoration of a steeper native slope—are warranted to further elucidate this dynamic. (2)Retrospective design and selection bias: The retrospective nature of this investigation inherently introduces both selection and observational biases. Although we attempted to mitigate these effects by enrolling a consecutive series of patients who met strict inclusion criteria, the lack of true randomization and blinding means that unrecognized confounding variables, patient selection, or surgeon preferences could have influenced the clinical outcomes. Specifically, the nonrandomized grouping driven by patients’ socioeconomic status and high functional expectations for robotic technology may introduce a “placebo effect,” weakening the causal inference power of the conclusions. Therefore, future prospective, randomized controlled trials (RCTs) are necessary to definitively validate these findings and eliminate these inherent biases. (3) Follow-up period limitations: the 1-year follow-up reflects only short-term efficacy and cannot yet assess long-term prosthesis survival, periprosthetic bone density changes, or long-term polyethylene liner wear.(4) Multidimensional confounding interference: Although this study adjusted for HKA deviation, it did not quantify dynamic mechanical indicators such as soft-tissue balance. Given that robot-assisted surgery has systemic technical advantages (such as femoral rotation positioning optimization), it is difficult to completely isolate the independent contribution of PTS to FJS-12 scores.(5) Lack of coronal morphotype stratification: Finally, our study did not stratify patients based on their native Coronal Plane Alignment of the Knee (CPAK) morphotypes. Because our surgical philosophy was strictly based on standard mechanical alignment (MA), the preoperative planning target uniformly aimed for a neutral axis, regardless of the patient’s individual CPAK classification. However, forcing various native coronal morphotypes into a standardized mechanical axis may introduce varying degrees of anatomical compromise that could independently confound functional outcomes. As recently demonstrated by Strahovnik et al. [21], CPAK classification plays a crucial role in determining individualized alignment strategies and directly influences postoperative kinematics and clinical success in robotic TKA. Future studies should integrate baseline CPAK classifications with sagittal slope analyses to provide a more holistic understanding of how multidimensional anatomical deviations impact patient-reported outcomes. Clinical implications and future directions Surgeons should measure native PTS preoperatively via 3D imaging and leverage robotic precision to achieve Δ ≤ 2° reconstruction, aiming to cross MCID thresholds. Future prospective RCTs stratified by native PTS, incorporating coronal alignment, sagittal PTS, and multidimensional soft-tissue data, are warranted. Artificial intelligence-assisted planning systems may further optimize individualized strategies. Conclusions In summary, under precise robotic osteotomy, the postoperative PTS is significantly nonlinearly related to the FJS-12 score, with an optimal population-level value of approximately 5.3°. PTS deviation from native anatomy is an independent predictor of FJS-12 score, underscoring the value of individualized native PTS restoration—particularly in patients with marked anatomical variation. However, the observed FJS-12 score improvement did not exceed the MCID, and most secondary outcomes were negative.Therefore, simply using robotic assistance is insufficient to guarantee clinically meaningful superiority; surgical planning must strictly target the individualized native anatomy to maximize patient satisfaction. Acknowledgements Not applicable. Abbreviations Body mass index Forgotten Joint Score-12 Hip-knee-ankle Intraclass correlation coefficient Knee society score Minimal clinically important difference Oxford Knee Score Posterior cruciate ligament Posterior stabilized Propensity score matching Posterior tibial slope Robot-assisted total knee arthroplasty Range of motion Standardized mean difference Total knee arthroplasty Visual analog scale Western Ontario and McMaster Universities Osteoarthritis Index Author contributions Authors’ contributions Boyi Chen: Conceptualization, Methodology, Formal analysis, Data curation, Writing—original draft, Writing—review & editing. Qi Yan, Yuanyuan Li, Siye Liu, Ximing Yu: Investigation, Data curation, Validation.Hongmei Zhang, Lin Jing: Performed the surgical procedures, Supervision, Project administration, Writing—review & editing.All authors have read and agreed to the published version of the manuscript. Data availability The data supporting the findings of this study are not publicly available due to patient privacy protection. These data were obtained from Wangjing Hospital, China Academy of Chinese Medical Sciences, and are used with the hospital’s authorization for this study only. Access to the data is restricted. However, data may be available from the corresponding authors upon reasonable request and with the permission of Wangjing Hospital, China Academy of Chinese Medical Sciences. Declarations Conflict of interest The authors declare that they have no competing interests. Ethics approval and consent to participate The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Wangjing Hospital, China Academy of Chinese Medical Sciences (approval numbers: WJEC-KT-2021-035-P003, WJEC-KT-2021-035-P004, WJEC-KT-2021-035-P005; dates of approval: July 15, 2021; July 28, 2022; July 24, 2023). Informed consent was obtained from all participants involved in the study. Consent for publication Written informed consent for publication was obtained from each patient. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Lin Jing, Email: 13718822649@163.com. Hongmei Zhang, Email: 1767814835@qq.com. References Associated Data Data Availability Statement The data supporting the findings of this study are not publicly available due to patient privacy protection. These data were obtained from Wangjing Hospital, China Academy of Chinese Medical Sciences, and are used with the hospital’s authorization for this study only. Access to the data is restricted. However, data may be available from the corresponding authors upon reasonable request and with the permission of Wangjing Hospital, China Academy of Chinese Medical Sciences.