Knee Alignment Shifts After Partial Knee Replacement Do Not Worsen Outcomes

Patients whose knee alignment phenotype changed after medial unicompartmental knee arthroplasty fared just as well as those whose original alignment was preserved, according to a retrospective cohort study from Chulalongkorn University in Bangkok published in the journal Arthroplasty. At a mean follow-up of 23.2 months, the 41 patients whose coronal plane alignment of the knee phenotype shifted after surgery and the 35 patients whose phenotype was restored reported nearly identical scores on the Forgotten Joint Score-12, the Oxford Knee Score, and the 12-Item Knee Injury and Osteoarthritis Outcome Score, along with similar satisfaction ratings and knee flexion. The finding challenges the assumption that preserving a patient’s native knee alignment category is essential for a good result after partial knee replacement.

Unlike total knee arthroplasty, which resurfaces the entire joint and often requires extensive ligament release, unicompartmental knee arthroplasty replaces only the damaged medial compartment and preserves both cruciate ligaments, the native meniscus in the unaffected compartments, and much of the knee’s proprioceptive and kinematic function. Because of this, postoperative coronal alignment is expected to remain close to the patient’s native alignment, making the procedure a natural setting in which to test whether alignment phenotype preservation matters.

To classify each knee, the researchers used the coronal plane alignment of the knee system, a phenotype-based classification introduced to describe native knee alignment using two radiographic parameters: the arithmetic hip-knee-ankle angle, which represents constitutional limb alignment, and joint line obliquity. Combining three categories of the arithmetic hip-knee-ankle angle with three categories of joint line obliquity yields nine distinct knee phenotypes, providing a standardized framework for evaluating coronal alignment in both native and post-arthroplasty knees.

Of the 145 enrolled patients, 65 were excluded from the radiographic phenotype analysis because their pre- and postoperative radiographs were unsuitable, showing intra- and inter-observer intraclass correlation coefficients below 0.5 for all measured angles and poor Kappa agreement for phenotype classification. Among the remaining 80 patients with suitable radiographs, four were excluded from the final analytic cohort: three because of gait-aid use that could affect weight-bearing radiographs and one because of incomplete questionnaires. That left 76 patients in the final phenotype analysis, with strong measurement reliability, including intra-observer intraclass correlation coefficients of 0.93 for the medial proximal tibial angle and 0.87 for the lateral distal femoral angle, and inter-observer values of 0.90 and 0.84 respectively. Agreement for phenotype classification was substantial to almost perfect, with Kappa values of 0.82 intra-observer and 0.78 inter-observer.

The mean age of the studied group was 69.5 years and the mean body mass index was 27.19 kilograms per square meter, with a female predominance. Baseline characteristics, including age, sex, body mass index, operated side, follow-up duration, and osteoarthritis grading according to the Kellgren-Lawrence and Ahlbäck systems, showed no significant differences between the restored and altered phenotype groups. Preoperatively, the two most common phenotypes were type I, accounting for 52.6 percent of knees, and type II, accounting for 32.9 percent. After surgery, the most common phenotypes were type I at 28.9 percent and type VII at 22.4 percent. Thirty-five patients, or 46 percent, retained their original phenotype postoperatively, while 41, or 54 percent, shifted to a different one. Among the altered group, type I most commonly changed to types IV or VII, and type II most commonly changed to types VII or VIII.

Postoperatively, the arithmetic hip-knee-ankle angle did not differ significantly between the two groups, at minus 3.11 plus or minus 2.21 degrees in the restored group versus minus 2.27 plus or minus 1.88 degrees in the altered group, and neither group was overcorrected toward valgus alignment, defined as an arithmetic hip-knee-ankle angle greater than 2 degrees. However, the altered group showed a significantly higher postoperative medial proximal tibial angle, at 88.80 versus 87.63 degrees, and higher joint line obliquity, at 179.88 versus 178.37 degrees. The mean change in the medial proximal tibial angle was also significantly larger in the altered group, at 3.98 versus 2.86 degrees, while the lateral distal femoral angle did not differ significantly between groups.

Despite these radiographic differences, functional outcomes at the mean follow-up of 23.2 months were favorable in both groups. The Forgotten Joint Score-12 averaged 85.34 in the restored group versus 86.93 in the altered group, the Oxford Knee Score 41.06 versus 40.73, and the KOOS-12 86.24 versus 85.66, with none of the differences reaching statistical significance. Patient satisfaction, measured on a 0-to-10-point scale, was 8.94 versus 8.93, and maximum knee flexion was 124.29 versus 124.88 degrees. Using the validated Patient Acceptable Symptom State threshold of 72.9 points for the Forgotten Joint Score-12, most patients in both groups achieved a satisfactory outcome, with no significant difference between the groups. When the 65 patients with unsuitable radiographs were added for a separate clinical comparison, forming a cohort of 141 patients, outcomes remained similar across the restored, altered, and unsuitable-radiograph groups on all measured measures.

The high rate of phenotype change, 54 percent, deserves careful interpretation, and the authors themselves urge caution. A recent systematic review reported high variability in medial proximal tibial angle measurements after unicompartmental knee arthroplasty, with different landmarks used to draw the proximal tibial reference line leading to differences of up to 3 degrees, enough to shift a knee across a phenotype boundary. The present study used the line from the lateral native tibial plateau to the lowest point of the femoral component as the reference line for the medial proximal tibial angle, and the line from the lowest point of the lateral femoral condyle to the lowest point of the femoral component for the lateral distal femoral angle, a choice the authors believed corresponded to the preoperative measurement method. This landmark-dependence means some proportion of the observed change likely reflects measurement-method sensitivity rather than true postoperative phenotype migration, a caveat the researchers state explicitly in their discussion.

The study also has other limitations that shape how the results should be read. Its retrospective observational design carries an inherent risk of selection and information bias. The follow-up duration was limited to approximately two years because full-body standing radiographs only became available at the institution in mid-2022, which precludes any prediction of long-term clinical outcomes or further radiographic changes in limb alignment and component fixation. The investigation was single-center and conducted in a Thai population, limiting generalizability to other ethnic groups, and no biomechanical or gait analysis was performed to assess joint loading and kinematics. Longer follow-up is also needed to evaluate implant survivorship and progression of lateral compartment osteoarthritis, and the authors note that coronal spinal imbalance detected on full-body radiographs in some patients might affect postoperative lower-limb alignment or clinical outcomes, warranting further study.

Within those constraints, the message for clinicians and patients is that a shift in coronal alignment phenotype after medial unicompartmental knee arthroplasty, at least in the short term, does not translate into worse patient-reported outcomes. The authors conclude that medial unicompartmental knee arthroplasty can alter the preoperative phenotype distribution while short-term clinical outcomes between those with restored and altered phenotypes remain similar, and they call for studies with longer follow-up to determine the impact of these alignment changes on implant survivorship and disease progression. The results align with a recent study of more than 300 patients in which over 60 percent of medial unicompartmental knee arthroplasties showed phenotype changes yet all clinical outcome parameters were similar between the altered and restored groups, adding weight to the view that the relationship between coronal alignment phenotype and clinical outcomes remains controversial and that preserving the native phenotype may not be the decisive factor it is sometimes assumed to be.

Subject of Research: Changes in coronal plane alignment phenotype after medial unicompartmental knee arthroplasty and their effect on patient-reported clinical outcomes

Article Title: Changes in knee phenotype in coronal plane alignment after medial unicompartmental knee arthroplasty did not affect clinical outcomes

Article References: Lapanikorn, S., Tanavalee, A., Ngarmukos, S., Jaruthien, N., Tanavalee, C., & Amarase, C. (2026). Changes in knee phenotype in coronal plane alignment after medial unicompartmental knee arthroplasty did not affect clinical outcomes. Arthroplasty, 8(1), Article 84. https://doi.org/10.1186/s42836-026-00447-x

Image Credits: AI Generated

DOI: 10.1186/s42836-026-00447-x

Keywords: unicompartmental knee arthroplasty, coronal plane alignment, CPAK phenotype, knee osteoarthritis, patient-reported outcomes, Forgotten Joint Score-12, Oxford Knee Score, KOOS-12, radiographic alignment, medial proximal tibial angle, joint line obliquity, retrospective cohort study