2.1 Study design and ethics

This single-center retrospective cohort study was approved by the institutional review board (Approval ID: 2025-19), and the requirement for individual informed consent was waived due to the retrospective design. The study adhered to the Declaration of Helsinki. The methodological framework (eligibility, measurements, endpoints, and analysis flow) was developed with reference to previous CR-TKA examining combined pain and ROM outcomes.²

2.2 Participants

We retrospectively identified 93 knees with advanced-to-end-stage KOA (Kellgren–Lawrence grade 3–4) that underwent primary MP-TKA at our affiliated institution between October 2018 and October 2024. Indications included persistent pain and functional decline despite conservative treatment. Exclusion criteria were prior TKA, high tibial or other knee osteotomy, ligament reconstruction, or active infection. For the 1-year outcome analysis, knees were additionally excluded if 1-year follow-up was unavailable or if 1-year knee flexion ROM and/or pain (NRS) data were missing (Fig. 1). Baseline characteristics (age, sex, body height, body weight, and BMI) and limb alignment (hip–knee–ankle, HKA; varus recorded as positive) were evaluated.

Close

Fig. 1 Patient flow chart of this study.

Export to PPT

Knees were categorized into two groups according to outcomes at 1 year postoperatively: a Good outcome group (knee flexion ROM ≥120° and NRS ≤1.0) and a Non-Good group (knee flexion ROM <120° or NRS >1.0). This definition mirrors prior CR-TKA studies and reflects functional thresholds relevant to Asian populations.²

2.3 Surgical technique (MP-TKA with patient-specific instrumentation)

All procedures used a fixed-bearing Evolution® medial-pivot implant (MicroPort Orthopedics, Arlington, TN, USA) with cement through a medial parapatellar approach. The Prophecy® patient-specific instrumentation (PSI) system was employed for preoperative planning and fabrication of patient-specific cutting guides. Both anterior and posterior cruciate ligaments were resected, and patellar resurfacing was not performed. The infrapatellar fat pad was preserved whenever possible and excised only when required for adequate visualization, in accordance with the manufacturer's instructions and prior MP-TKA protocols.⁷

The PSI-assisted workflow followed a measured-resection concept using patient-specific cutting guides referenced to bony landmarks, with extension-gap assessment performed first. Distal femoral resection was followed by proximal tibial resection according to the preoperative plan. If the extension gap remained tight, additional proximal tibial resection was performed in 2-mm increments without routine soft-tissue release, and component sizing was adjusted as needed to optimize gap balance and implant fit. After achieving a satisfactory extension gap, posterior femoral resection was performed at 90° of knee flexion; if the flexion gap was too narrow, additional posterior femoral resection was similarly performed in 2-mm increments. Following the final resections, extension and flexion gaps were reassessed.

The preoperative PSI plan incorporated a saw-blade thickness allowance (up to 1.28 mm), which was considered when comparing PSI-planned and intraoperatively measured osteotomy thicknesses. Osteophytes were preserved as reference landmarks for guide placement whenever feasible.

Preoperative flexion contracture >20° was considered carefully, as it has been associated with component-size mismatch in PSI-guided MP-TKA.⁷

2.4 Rehabilitation protocol

Postoperative rehabilitation followed our standardized program: supervised physical therapy was initiated approximately 3–4 weeks postoperatively, emphasizing seated pedaling, knee extension stretching, and heel-to-toe gait drills performed at least three times daily as tolerated. Outpatient sessions were individualized by the supervising therapist.

2.5 Clinical evaluations

Preoperative and postoperative knee ROM (extension, flexion) were recorded using a double-armed goniometer by trained staff, consistent with previous protocols.^2,3,5 Knee extension deficit was defined as the degrees short of full extension (0°) and recorded as a positive value. Anteroposterior (AP) translation was measured at 30° and 90° of knee flexion using a Rolimeter, with three repeated measurements averaged for analysis. Osteotomy thickness (distal/dorsal femur; proximal tibia, medial/lateral) was recorded intraoperatively with a Vernier caliper. Comparisons between the PSI plan and intraoperative findings (component sizes and osteotomy thicknesses) were obtained from Prophecy planning sheets and operative records.^3,5 Operative time was also recorded.

2.6 Statistical analysis

Continuous variables are summarized as mean (SD) and categorical variables as counts (%). Between-group comparisons between the Good and Non-Good groups were performed using Student's t-test or the Mann–Whitney U test for continuous variables and the χ² test or Fisher's exact test for categorical variables.

Univariate logistic regression analyses were conducted to explore the association of each variable with the Good outcome. Multivariable logistic regression analysis was performed to identify independent predictors (reporting odds ratios [ORs] with 95% confidence intervals [CIs]). Variables included in the multivariable model were selected based on clinical relevance and prior literature, with univariate results considered. Multicollinearity was assessed using variance inflation factors (VIFs). Model calibration and discrimination were evaluated using the Hosmer–Lemeshow test and the area under the receiver operating characteristic curve (AUROC), respectively. For selected continuous variables, exploratory ROC analyses were performed to evaluate discriminatory performance, and optimal cut-off values were determined using the Youden index.

Missing data were handled using multiple imputation (m = 100) when the proportion of missing data exceeded 5% for variables included in regression analyses, and regression estimates were pooled across imputations using Rubin's rules. P values were pooled using Rubin's rules with t-based degrees of freedom, whereas 95% confidence intervals were computed from pooled standard errors using normal approximation. P < 0.05 was considered statistically significant. All analyses were performed using EZR.⁸

Because of the retrospective design, the sample size was determined by the number of knees available during the study period. Given the limited number of outcome events, the number of predictors in the multivariable models was kept small to minimize overfitting, and effect estimates are presented with 95% confidence intervals.

3.1 Baseline characteristics

Among 93 knees that underwent PSI-assisted MP-TKA during the study period, 64 knees (21 in the Good group [G], 43 in the Non-Good group [NG]) were included in the analysis after excluding knees without 1-year follow-up and those with missing 1-year knee flexion ROM and/or pain (NRS) data (Table 1). Baseline characteristics were similar between groups, except for height (G: 156.4 ± 6.6 cm vs NG: 151.3 ± 7.1 cm, p = 0.020) and extension deficit (G: 12.6 ± 8.4° vs NG: 8.3 ± 8.1°, p = 0.049). Age, sex, BMI, preoperative flexion, and HKA angle showed no significant differences (all p > 0.05; Table 1).

3.2 Component matching and intraoperative findings

Operative time did not differ significantly between groups. Concordance between the PSI-planned and implanted component sizes differed between groups. The Good group showed a higher frequency of overall intraoperative component size adjustment (G: 17/21 [81.0%] vs NG: 19/43 [44.2%], p = 0.007) and tibial size adjustment (G: 16/21 [76.2%] vs NG: 19/43 [44.2%], p = 0.018) (Table 2). Among knees with overall intraoperative component size adjustment, the majority of modifications were tibial downsizing (Table 3). Osteotomy thickness deviation from the PSI plan (Actual − Plan) did not differ between groups at any femoral or tibial site (all p > 0.1; Table 2).

3.3 Postoperative outcomes at 1 year

At one year postoperatively, the Good group demonstrated significantly greater flexion ROM than the NG group (G: 123.5 ± 4.5° vs NG: 114.6 ± 10.6°, p < 0.001) and significantly lower NRS pain scores (G: 0.2 ± 0.4 vs NG: 2.2 ± 1.9, p < 0.001). AP translation at 90° flexion was greater in the Good group one year postoperatively (G: 2.77 ± 0.71 mm vs NG: 2.11 ± 0.91 mm, p = 0.017), whereas AP translation at 30° did not differ significantly between groups (Table 4).

3.4 Logistic regression analysis

In univariate analysis, overall component size adjustment and body height were significantly associated with achieving a favorable postoperative outcome. Preoperative knee extension deficit and AP translation at 90° of flexion at 1 year showed a trend toward an association and were considered as candidate variables for multivariable modeling based on clinical relevance (Table 5).

In multivariable logistic regression analysis including three pre- and intraoperative variables, intraoperative component size adjustment was independently associated with the achievement of pain-free and good flexion recovery (OR = 4.32, 95% CI 1.07–17.45, p = 0.040). Body height (OR = 1.04, 95% CI 0.95–1.14, p = 0.37) and preoperative extension deficit (OR = 1.02, 95% CI 0.95–1.10, p = 0.58) were not significantly associated with the favorable outcome (Table 6). The model demonstrated good calibration (Hosmer–Lemeshow test, p = 0.471) and moderate discrimination (AUC = 0.734, 95% CI 0.600–0.869).

When postoperative anteroposterior translation at 90° of flexion was additionally included in the model, intraoperative component size adjustment showed a trend toward an association with the favorable outcome (OR = 4.23, 95% CI 1.01–17.75, p = 0.054), whereas AP translation itself was not significantly associated (OR = 1.57, 95% CI 0.76–3.25, p = 0.233) (Table 7). This expanded model demonstrated comparable calibration (Hosmer–Lemeshow test, p = 0.664) and discrimination (AUC = 0.746, 95% CI 0.624–0.868).

ROC analysis demonstrated that overall component size adjustment predicted the favorable outcome with an AUC of 0.684 (95% CI 0.570–0.798). When overall component size adjustment was treated as a binary predictor (adjustment present vs absent), the sensitivity and specificity were 55.8% and 81.0%, respectively. For reference, the AUCs for body height ≥158 cm and preoperative extension deficit ≥10° were 0.642 (95% CI, 0.500–0.783) and 0.650 (95% CI, 0.510–0.789), respectively.

4.1 Interpretation of intraoperative component size adjustment

PSI is designed to improve surgical accuracy and reproducibility.^9,10 However, it cannot fully account for intraoperative findings such as soft-tissue tension, flexion–extension gap balance, or subtle variations in bone morphology, and a mismatch between the planned and implanted component size can still occur, particularly in the presence of flexion contracture.⁷ Therefore, intraoperative modification of the PSI-planned component size based on intraoperative reassessment—rather than rigidly adhering to the preoperative PSI-plan—may be clinically meaningful.

In the present study, intraoperative component size adjustment, particularly tibial downsizing, was more frequently observed in patients who achieved pain-free recovery and good knee flexion. This suggests that preoperative PSI plans may not always reflect the intraoperative optimal component fit in some cases, and that intraoperative sizing modifications help optimize tibial fit and soft-tissue balance. However, component sizing should be individualized rather than uniformly downsized.

Liu et al. reported that medial tibial overhang was associated with worse PROMs, whereas medial and lateral underhang increased the risk of tibial bone resorption at 2 years after TKA, supporting the need to individualize tibial sizing to avoid both soft-tissue irritation from overhang and excessive undercoverage.¹¹

Accordingly, intraoperative reassessment of tibial sizing may play a crucial role in optimizing tibial coverage and joint balance, thereby contributing to a more favorable flexion-gap environment and postoperative function.

4.2 Role of AP laxity in deep flexion

In the present study, patients with favorable outcomes demonstrated greater AP translation at 90° flexion at 1 year postoperatively. However, AP translation was not independently associated with the favorable outcome in multivariable analysis, indicating that AP laxity in deep flexion should be interpreted as a postoperative kinematic consequence rather than a primary determinant of achieving pain-free recovery and good flexion.

Implant design may influence sagittal stability at different flexion angles. In a propensity-matched comparison, MP-TKA showed smaller AP laxity at 30° flexion but larger AP laxity at 90° flexion than CR-TKA, indicating design-related differences in sagittal stability across flexion angles.⁵

Nevertheless, even within the same implant design, excessive tightness may still be unfavorable. In CR-TKA, reduced AP laxity at 30° flexion has been associated with a postoperative decline in knee flexion ROM, suggesting that insufficient sagittal laxity in mid-flexion may reflect a constrained flexion-gap environment.³ Excessive tightness at 90° flexion may restrict femoral rollback and contribute to residual pain or limited flexion. These observations should not be interpreted as suggesting that excessive laxity is desirable. Instead, they suggest that an overly tight flexion gap should be avoided to achieve an appropriate balance between stability and motion.

Taken together, appropriate intraoperative balancing and sizing may contribute to a more favorable flexion-gap environment, resulting in sufficient AP laxity in deep flexion and improved postoperative comfort and range of motion.

4.3 Patient-related factors as background modifiers

Greater body height and greater preoperative extension deficit were associated with favorable outcomes in univariate analyses but did not remain significant in multivariable models. This suggests that these factors may increase the likelihood that intraoperative adjustment is required, rather than directly determining postoperative outcomes. In other words, patient morphology and preoperative stiffness may act as background modifiers that necessitate careful intraoperative judgment.

Consistent with this, Hiyama et al. reported that flexion contracture (>20°) increases the risk of PSI component size mismatch in MP-TKA and recommended anticipating an undersized tibial component during planning.⁷

4.4 Clinical implications

Our findings emphasize that successful MP-TKA depends not only on implant design or preoperative planning but also on the surgeon's ability to interpret intraoperative conditions and adjust component sizing accordingly. Surgeons using PSI should be prepared to modify the planned component size when intraoperative assessment indicates suboptimal balance, particularly in deep flexion. This is especially relevant in patients with flexion contracture, since PSI planning may be less reliable and tibial downsizing may be more likely.⁷ When modifying tibial size, coverage should be individualized to avoid overhang while maintaining adequate tibial coverage.¹¹

4.5 Limitations

This study has limitations inherent to its retrospective single-center design and modest sample size. Only short-term outcomes were assessed, and patient-reported outcome measures beyond pain were not included. In addition, component size adjustment was analyzed as an overall binary variable, so the relative contributions of femoral versus tibial changes could not be determined. Finally, our findings were obtained from one PSI system and implant design, which may not be generalizable to other systems. Despite these limitations, the consistent surgical technique and detailed intraoperative data strengthen the internal consistency of our findings.