Medial opening wedge high tibial osteotomy: A prospective cohort study of gait, radiographic, and patient-reported outcomes

Authors

  • Trevor B. Birmingham,

    Corresponding author
    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    • Elborn College, University of Western Ontario, London, Ontario, Canada N6G 1H1
    Search for more papers by this author
  • J. Robert Giffin,

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Bert M. Chesworth,

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Dianne M. Bryant,

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Robert B. Litchfield,

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Kevin Willits,

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Thomas R. Jenkyn,

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Peter J. Fowler

    1. The Wolf Orthopaedic Biomechanics Laboratory and Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author

Abstract

Objective

To evaluate the effect of medial opening wedge high tibial osteotomy on gait, radiographic, and patient-reported outcomes over a 2-year postoperative period in patients with varus alignment and medial compartment knee osteoarthritis, and to identify significant predictors of outcome.

Methods

We used an observational cohort study design and prospectively administered 3-dimensional quantitative gait analysis, hip to ankle weight-bearing radiographs, and patient-reported outcomes preoperatively and 6, 12, 18, and 24 months postoperatively. Observed changes with 95% confidence intervals (95% CIs) were calculated. Multivariate linear regression and cluster analysis were used to evaluate associations between patient characteristics and 2-year outcomes in dynamic knee joint load (external knee adduction moment during gait) and Knee Injury and Osteoarthritis Outcome Scores (KOOS).

Results

A total of 126 patients (mean age 47.48 years) were included in the study. Mean changes suggested clinically important improvements in malalignment (change in mechanical axis angle 8.04° [95% CI 7.16°, 8.93°]), medial compartment load during gait (change in knee adduction moment −1.38 [95% CI −1.53, −1.22] percentage body weight × height), and all KOOS domain scores (change in pain 23.19 [95% CI 19.49, 26.89] KOOS points). A small (13%) increase in knee adduction moment was observed from 6 to 24 months postoperatively. Few preoperative clinical and/or gait characteristics assessed at baseline were significantly associated with 2-year outcomes.

Conclusion

A medial opening wedge high tibial osteotomy with correction to approximately neutral alignment produces substantial and clinically important changes in dynamic knee joint load and patient-reported measures of pain, function, and quality of life 2 years postoperatively. Changes in knee adduction moment observed in the first 2 years postoperatively should be explored as potential predictors of longer-term success and subgroups of patients with poor outcomes.

INTRODUCTION

Knee osteoarthritis (OA) ranks among the most common disabling and costly health conditions in adults, and its prevalence is projected to increase sharply over the next 2 decades (1–3). There are currently no interventions proven to alter the course of knee OA, leaving total joint replacement for end-stage disease the most widely accepted surgical treatment option (4). The substantial burden of knee OA provides strong support for implementing treatment strategies earlier in the disease process. High tibial osteotomy (HTO) is a biomechanically focused surgical intervention intended to improve pain and function and delay joint degeneration in patients with OA located primarily in a single knee compartment (5). Studies suggest that HTO procedures performed using a variety of techniques can result in large improvements in self-administered, knee-specific questionnaires (5–7). For example, when converted to overall scores out of 100, patient-reported status can increase from a mean of approximately 30 points preoperatively to a mean of approximately 70 points postoperatively (6, 7). However, concerns regarding variability in outcomes among patients and the duration of these improvements exist, and administrative databases suggest reluctance in North America to perform these procedures (8).

HTO alters knee alignment in an attempt to redistribute the weight-bearing load on the tibiofemoral joint. Classically, a valgus-producing osteotomy for varus gonathrosis was performed by removing a lateral wedge of bone from the proximal tibia to correct coronal malalignment (9). During the past decade, interest in a medial opening wedge valgus-producing osteotomy technique has increased, owing to claims that it offers improvements over the lateral closing technique (10). Potential detrimental effects of HTO include delayed union or nonunion and inadvertent changes in the posterior tibial slope and patellar height (11). Optimal indications for surgery, particularly age and extent of disease, and the amount of correction in the alignment necessary to adequately unload the medial compartment are controversial. Despite the proposed benefits of medial opening wedge HTO, a very limited number of prospective cohort studies (12) evaluating this newer technique exist (6).

Recent studies evaluating mechanical risk factors for progression of knee OA confirm earlier hypotheses regarding the important role of malalignment and distribution of tibiofemoral loads (13–17). In particular, the magnitude of the external knee adduction moment during walking, measured during quantitative gait analysis, has emerged as a valid and reliable proxy for the dynamic load on the medial tibiofemoral joint and a potential risk factor for disease progression (18, 19). Although decreasing the load on the medial knee compartment is the primary rationale for valgus HTO, we are aware of few studies with relatively small sample sizes evaluating changes in the knee adduction moment postoperatively. Published studies include 5 cohorts ranging in size from 7 to 32 patients, a variety of HTO techniques, and one postoperative gait analysis, often performed at variable lengths of followup (20–25). Together, these studies provide convincing evidence that valgus HTO procedures produce statistically significant decreases in the knee adduction moment, although the magnitude and stability of these reductions and potential changes in other gait characteristics known to affect the knee adduction moment remain unclear (20–25). Additionally, previous studies evaluating whether preoperative quantitative gait analysis provides useful predictors of clinical outcome have produced contrary results (23, 24, 26, 27).

Our objectives were to 1) evaluate the effect of medial opening wedge HTO on gait, radiographic, and patient-reported outcomes over a 2-year postoperative period in patients with varus alignment and medial compartment knee OA, and 2) identify significant predictors of outcome. We hypothesized that medial opening wedge HTO would result in substantial, clinically important improvements in dynamic knee joint load and patient-reported pain. We also hypothesized that preoperative patient characteristics would be associated with these improvements.

PATIENTS AND METHODS

Study design.

We used an observational cohort study design (12) to evaluate patients undergoing medial opening wedge HTO. Four surgeons (JRG, RBL, KW, PJF) at a tertiary care center specializing in adult reconstruction and orthopedic sports medicine surgery participated. We prospectively assessed gait, radiographic, and patient-reported outcomes preoperatively and at 6, 12, 18, and 24 months postoperatively. Data were collected between November 2002 and December 2007. Patients provided informed consent to participate in the study. The study was approved by the Research Ethics Board for Health Sciences Research Involving Human Subjects of the University of Western Ontario.

Patients.

We recruited patients presenting to this center for treatment of unresolved knee pain located primarily in the medial compartment. Patients were referred from family physicians, rheumatologists, and primary care sports medicine specialists for consultation with an orthopedic surgeon regarding treatment options. Inclusion criteria consisted of mechanical varus alignment and knee OA according to the American College of Rheumatology classification criteria (28), with the greatest severity in the medial compartment of the tibiofemoral joint. Patients with concomitant disease in the lateral compartment were considered eligible as long as pain and radiographic disease were more severe in the medial compartment. Patients with concomitant chronic anterior cruciate ligament (ACL) deficiency undergoing simultaneous ACL reconstruction were included. We excluded patients age ≥60 years with grade 4 degenerative changes in ≥2 knee compartments because they were considered better candidates for arthroplasty. Other exclusion criteria included inflammatory or infectious arthritis of the knee, end-stage disease in the patellofemoral joint, prior HTO on the contralateral extremity, multiligamentous instability, major neurologic deficit that would affect gait, major medical illness with a life expectancy <2 years or with an unacceptably high operative risk, pregnancy, inability to speak or read English, and psychiatric illness that limited informed consent.

Intervention.

Using an opening wedge osteotomy system and a nonlocking plate (Arthrex Opening Wedge Osteotomy System; Arthrex, Naples, FL), the surgeons used a similar operative technique to that previously described by Fowler et al (29). The desired angle of correction was calculated preoperatively according to the method described by Dugdale et al (30), with the goal of creating very slight valgus by moving the weight-bearing line laterally up to a maximum position of 62.5% of the medial to lateral width of the tibia, depending on the magnitude of malalignment and status of the articular cartilage in the lateral tibiofemoral compartment. The osteotomy was performed using both flexible and rigid osteotomes below a fluoroscopically placed guide pin, then opened slowly to the predetermined width for placement of the plate that was fixed proximally and distally with cancellous and cortical screws, respectively. Cancellous allograft bone was used in osteotomies >7.5 mm.

Postoperative management included use of a hinged postoperative knee brace and feather-touch weight bearing with crutch use for at least 6 weeks. With clinical and radiographic evidence of healing, partial weight bearing was permitted at 6 weeks, and full weight bearing was permitted at 12 weeks postoperatively. At ∼3 weeks postoperatively, the patients started exercises in the brace until healing of the osteotomy site had occurred. Goals of these exercises during this time period of limited weight bearing were to limit swelling, joint contracture, and disuse muscle atrophy. Range of motion exercises consisted of active and passive knee and hip movements. Strengthening exercises initially consisted of isometric knee extension/flexion and hip extension. At ∼8 weeks postoperatively, non–weight-bearing concentric exercises using weights or Thera-Band (Hygienic Corporation, Akron, OH) were added and progressed until weight bearing was permitted. At ∼12 weeks postoperatively, weight-bearing exercises emphasizing proprioceptive control and balance were then implemented and progressed until patients demonstrated a normal walking gait pattern.

Gait analysis.

Gait was evaluated using an 8-camera motion capture system (Eagle EvaRT; Motion Analysis Corporation, Santa Rosa, CA) synchronized with a floor-mounted force platform (Advanced Mechanical Technology, Watertown, MA). We used a modified Helen Hayes 22 passive-reflective marker set (31). Additional markers were placed over the medial knee joint line and medial malleolus bilaterally while the patient stood on the force platform to determine body mass, marker orientation, and positions of joint centers of rotation for the knee and ankle. These 4 markers were removed prior to gait testing.

Patients walked barefoot across the laboratory while 3-dimensional kinetic (sampled at 1,200 Hz) and kinematic (sampled at 60 Hz) data were recorded during the middle of several strides for at least 5 trials from each extremity. We calculated external knee moments from the kinematic and kinetic data using commercial software (Orthotrak 6.0, Motion Analysis Corporation), custom postprocessing, and previously described data reduction techniques (32, 33).

Based on the external knee adduction moment waveform, we identified the peak magnitudes in the first and second halves of stance and the area under the curve (impulse), and normalized these values to body weight and height. We also calculated gait speed, toe-out angle, and lateral trunk lean because of their influence on the knee adduction moment in patients with knee OA (33–35). Walking speed was calculated as the average walking speed between successive foot contacts of the tested extremity. Toe-out angle (positive angle) was calculated as the angle between a line drawn between the center of the ankle and the head of the second metatarsal joint and the forward progression of the body. Lateral trunk lean over the stance extremity (positive angle) was calculated as the angle of a line drawn from the midpoint of the anterior superior iliac spines to the midpoint of the anterior tips of the acromion processes with respect to vertical. All gait variables were calculated by averaging across 5 trials for each patient. We have previously reported excellent test–retest reliability of the first peak knee adduction moment (intraclass correlation coefficient [ICC2,1] = 0.86) with a minimum detectable change (95% confidence level) of 1% body weight (BW) × height (Ht) (36). We have also previously reported appropriate reliability of gait speed (ICC2,1 = 0.92), toe-out angle (ICC2,1 = 0.69), and trunk lean angle (ICC2,1 = 0.91) measurements (34).

Radiographic assessment.

Long cassette bipedal standing anteroposterior and lateral digital radiographs were assessed using custom computerized software (37). For anteroposterior radiographs, the patients stood with the patellae centered over the femoral condyles and feet straight ahead to attain a true anteroposterior image and to control for effects of foot rotation on measures of lower extremity alignment (15). The x-ray beam was centered on the knee at a distance of ∼2.5 meters and beam exposure was determined based on each patient's leg mass.

The mechanical axis angle of the lower extremity was used to quantify alignment in the frontal plane and was defined as the angle formed between a line drawn from the center of the hip to the center of the knee and a line drawn from the center of the ankle to the center of the knee (38, 39). Negative values indicated varus alignment. The center of the hip was identified as the geometric center of the femoral head using a circular template, the center of the knee was identified as the midpoint of the tibial spines extrapolated inferiorly to the surface of the intercondylar eminence, and the center of the ankle was defined as the mid-width of the tibia and fibula at the level of the tibial plafond. We have previously reported excellent reliability of mechanical axis angle measurements (ICC2,1 = 0.97) (37).

Other radiographic measures of knee geometry relevant to HTO included the femoral anatomic axis angle (the angle between the anatomic axes of the femur and tibia), the proximal tibial articular angle (formed by the tibial anatomic axis and a line tangent to the tibial plateau on the medial side), the posterior tibial slope (the angle formed by a line perpendicular to the lateral longitudinal axis of the tibia and a line parallel to the medial tibial plateau joint surface), and patellar height (Blackburn-Peel and Insall-Salvati ratios) (40). Measures of joint degeneration included the Kellgren/Lawrence (K/L) scale grade (41) in the medial and lateral compartments and medial and lateral joint space in millimeters. Although we have not evaluated the reliability of these additional radiographic measures, the same experienced and independent evaluator measured all radiographic variables at all followup visits.

Patient-reported outcomes.

We used the Knee Injury and Osteoarthritis Outcome Score (KOOS) to assess patient-reported outcomes (42, 43). The KOOS is a 42-item knee-specific questionnaire with 5 separately reported domains of pain (9 items), other symptoms (7 items), function in daily living (17 items), function in sports/recreation (5 items), and knee-related quality of life (4 items). Each item has 5 response options. Domain scores represent the average of all items in the domain standardized to a score of 0–100 (where 0 = worst and 100 = best). This instrument has face validity and has demonstrated construct validity, excellent test–retest reliability for each domain (range 0.75–0.93), and has been shown to be responsive to change (7, 42, 43). Other patient-reported outcomes included the Lower Extremity Functional Scale (44), a region-specific measure of lower extremity function that provides a single value for function that varies from 0 to 80 (where 0 = worst and 80 = best), and the generic physical function and mental health domains of the Short Form 36 health survey (45), standardized to scores from 0 to 100 (where 0 = worst and 100 = best).

Statistical analysis.

We first calculated the mean ± SD for all variables assessed preoperatively and 2 years postoperatively. We then calculated means and 95% confidence intervals (95% CIs) for changes after 2 years and evaluated statistical significance using the dependent t-test for continuous variables. We also plotted the mean and 95% CI for the mechanical axis angle, the first peak knee adduction moment on both extremities, and each of the KOOS domain scores preoperatively and 6, 12, 18, and 24 months postoperatively.

We defined 2 primary outcome measures (dependent variables) observed at the final followup: the first peak knee adduction moment and the KOOS pain score. Linear regression was used to quantify the magnitude of the association between predictors and outcomes. Potential predictors (independent variables) were chosen based on previous reports and consisted of factors related to disease severity and knee joint loading (6, 23, 24, 26, 27), including gait characteristics demonstrated to affect the knee adduction moment (33–35). Independent variables included preoperative scores for knee adduction moment, KOOS pain domain, age, sex, body mass index, mechanical axis angle, medial compartment K/L scale grade and lateral compartment K/L scale grade (both entered as binary variables; grade 1 or 2 versus grade 3 or 4), gait speed, toe-out angle, and trunk lean angle. To specifically investigate the effect of the postoperative alignment achieved with surgery, we also included the 6-month postoperative mechanical axis angle as an independent variable. We also repeated the analyses, replacing the first peak knee adduction moment with the area under the curve and replacing the change in the KOOS pain domain with the change in the KOOS function in daily living domain. We then repeated analyses, excluding patients who underwent simultaneous ACL reconstruction.

We used exploratory cluster analysis to identify groups of patients who were similar with respect to the 2-year change in the first peak knee adduction moment and the 2-year change in the KOOS pain score. We used a 2-step hierarchical clustering model with the unstandardized Euclidean distance algorithm as the similarity index, Akaike's information criterion as the measure of goodness of fit, and the auto-clustering feature of the software (TwoStep method, Statistical Package for the Social Sciences, version 16; SPSS, Chicago, IL). We repeated the initial analysis twice after randomizing the order of cases. Between-cluster comparisons were then made with the independent t-test, Wilcoxon's rank sum test, and the chi-square test as required. The cluster analysis method used screened-out patients who could not be assigned to a cluster because they were not similar with respect to the variables of interest. Because of this dissimilarity, these outliers were not evaluated further.

RESULTS

A total of 258 potential patients were screened for eligibility. Fifty-two patients were ineligible and 22 patients elected to not undergo HTO. Of the remaining 184 patients entered in the study, 36 were lost to followup, 3 died before completing the study, and 2 withdrew from the study. Fifteen patients elected to undergo HTO on their opposite extremity within the 2-year followup, and 2 patients required revision surgery for nonunion and loss of correction. These 17 patients were excluded from the present analysis, although we continue to follow them. A total of 126 patients with 2-year postoperative data remained. Six of these patients had some missing data and were excluded from the regression analyses (no data imputation methods were used). All 126 patients were included in the cluster analysis.

Three patients had delayed union, which resulted in missing the 6-month gait test only. Four patients had infections within 2 weeks of surgery that were treated with irrigation and debridement and a short course of intravenous antibiotics. Three patients had delayed infections at their tibial incision ∼1 year postoperatively (likely following a bacteremia) that required surgical removal of hardware and intravenous antibiotics. One patient had a hematoma 2 weeks postoperatively that required surgical washout. These complications did not require patients to miss followup tests.

Table 1 shows the preoperative demographics and clinical characteristics of this sample. The patients were predominantly male and, on average, relatively young and overweight. They had substantial varus alignment and tibiofemoral degeneration primarily of the medial compartment. Although these characteristics are consistent with other studies evaluating HTO, it should be noted that ∼10% of the present sample also had substantial degeneration in the lateral compartment (K/L scale grade ≥3).

Table 1. Baseline demographics and clinical characteristics (n = 128)*
CharacteristicValue
  • *

    Values are the mean ± SD unless otherwise indicated. BMI = body mass index.

  • Kellgren/Lawrence (K/L) scale grade of osteoarthritis severity.

Sex, no. (%) 
 Male102 (79.7)
 Female26 (20.3)
Age, years47.48 ± 9.53
Height, meters1.75 ± 0.09
Weight, kg90.54 ± 16.60
BMI, kg/m229.50 ± 4.82
Mechanical axis angle, degrees−7.50 ± 4.10
Femoral anatomic axis angle, degrees−2.20 ± 4.50
Medial compartment K/L scale grade, no. (%) 
 14 (3.13)
 232 (25.00)
 341 (32.03)
 451 (39.84)
Lateral compartment K/L scale grade, no. (%) 
 08 (6.25)
 144 (34.38)
 263 (49.22)
 311 (8.59)
 42 (1.56)

Table 2 shows the changes in all gait, radiographic, and patient-reported outcomes at the 2-year end point. With the exception of the Insall-Salvati ratio for patellar height, 95% CIs excluded the value 0 and were quite narrow. Changes in mechanical axis angle, first peak knee adduction moment, and KOOS scores at all test sessions are shown in Figure 1. The mechanical axis angle plot shows that surgery produced a large reduction in varus alignment that was maintained throughout the 24 months of followup. The knee adduction moment plot also shows a large reduction in dynamic knee joint load on the operative extremity 6 months postoperatively. Unlike the mechanical axis angle, however, the knee adduction moment slightly but consistently increased throughout followup. A post hoc comparison indicated that the increase from 6–24 months (mean increase 0.19 [95% CI 0.10, 0.28] %BW × Ht) was statistically significant (P < 0.001). The KOOS plot also shows large changes in all domains that continued to improve throughout followup.

Table 2. Gait, radiographic, and patient-reported outcome measures (n = 120)*
Outcome measureBaseline, mean ± SD24 months, mean ± SDChange, mean (95% CI)
  • *

    95% CI = 95% confidence interval; BW = body weight; Ht = height; KOOS = Knee Injury and Osteoarthritis Outcomes Score; SF-36 = Short Form 36; LEFS = Lower Extremity Functional Scale.

  • P < 0.01.

Gait   
 Knee adduction moment   
  First peak, %BW × Ht2.99 ± 0.921.62 ± 0.69−1.38 (−1.53, −1.22)
  Second peak, %BW × Ht2.37 ± 1.081.30 ± 0.78−1.07 (−1.26, −0.88)
  Area under the curve, %BW × Ht × seconds1.45 ± 0.490.77 ± 0.36−0.68 (−0.76, −0.60)
 Speed, meters/second1.10 ± 0.171.16 ± 0.160.06 (0.08, 0.04)
 Toe-out angle, degrees12.00 ± 5.7713.20 ± 5.401.23 (0.54, 1.92)
 Lateral trunk lean, degrees3.45 ± 2.971.96 ± 2.07−1.49 (−2.00, −0.99)
Radiographic   
 Mechanical axis angle, degrees−8.00 ± 4.09−0.05 ± 3.058.04 (7.16, 8.93)
 Femoral anatomic axis angle, degrees−2.15 ± 4.455.11 ± 3.447.26 (6.24, 8.28)
 Proximal tibial articular angle, degrees81.44 ± 4.5188.47 ± 3.767.02 (7.90, 0.44)
 Medial joint space width, mm2.86 ± 1.623.30 ± 1.480.44 (0.19, 0.68)
 Lateral joint space width, mm6.30 ± 1.825.91 ± 1.81−0.39 (−0.71, −0.07)
 Posterior tibial slope, degrees5.15 ± 4.666.37 ± 4.221.22 (0.24, 2.20)
 Blackburn-Peel ratio0.76 ± 0.140.71 ± 0.14−0.05 (−0.08, −0.01)
 Insall-Salvati ratio1.05 ± 0.181.05 ± 0.180.00 (−0.03, 0.03)
Patient reported   
 KOOS (range 0–100)   
  Pain51.71 ± 17.2574.90 ± 20.2523.19 (19.49, 26.89)
  Other symptoms52.62 ± 17.2570.76 ± 19.9218.14 (14.65, 21.63)
  Function in daily living60.60 ± 18.6482.26 ± 19.3221.66 (18.13, 25.20)
  Function in sports/recreation25.88 ± 20.4454.04 ± 29.8728.16 (22.84, 33.49)
  Quality of life25.59 ± 16.8653.70 ± 24.2128.12 (23.96, 32.27)
 SF-36 physical function (range 0–100)46.99 ± 24.1872.20 ± 24.1725.22 (20.25, 30.18)
 SF-36 mental health (range 0–100)75.83 ± 14.9482.99 ± 14.697.16 (3.72, 10.61)
 LEFS (range 0–80)42.62 ± 13.5059.64 ± 16.0117.02 (14.43, 19.61)
Figure 1.

Means and 95% confidence intervals for A, mechanical axis angle, B, first peak external knee adduction moment on both extremities, and C, Knee Injury and Osteoarthritis Outcomes Scores (KOOS) assessed at each test session. The figures show large, statistically significant (P < 0.001) changes in mechanical axis angle, knee adduction moment on the operative extremity, and all KOOS domains 6 months postoperatively. Small increases in the knee adduction moment from 6 to 24 months were statistically significant (P < 0.01). BW = body weight; Ht = height.

Table 3 shows that the predictor variables explained 43% of the variation in the 2-year postoperative first peak knee adduction moment. While adjusting for other variables, higher preoperative knee adduction moment, lower lateral trunk lean, higher age, and lower 6-month postoperative mechanical axis angle (i.e., more varus) were significantly associated with higher 2-year knee adduction moments. Analyses that excluded patients who underwent simultaneous ACL reconstruction revealed no additional insight into significant predictors, but slightly reduced the explained variance. Similarly, there were no meaningful differences when the value for the first peak was replaced by the area under the knee adduction moment curve. With the exception of their corresponding preoperative scores, all linear regression analyses yielded no other significant predictors of 2-year KOOS pain or KOOS function in daily living scores. Variance inflation factor and tolerance levels for all regression analyses suggested acceptable levels of multicolinearity.

Table 3. Predictors of 2-year first peak knee adduction moment (n = 120)*
Potential predictor variablesβ (95% CI)P
  • *

    95% CI = 95% confidence interval; BMI = body mass index; OA = osteoarthritis; BW = body weight; Ht = height; KOOS = Knee Injury and Osteoarthritis Outcomes Score.

  • All were assessed preoperatively unless stated otherwise.

  • Effect on 2-year peak knee adduction moment, adjusting for all other predictors in model. Adjusted r2 = 0.43.

  • §

    Less severe = Kellgren/Lawrence scale grade of OA severity 0, 1, 2; more severe = 3, 4.

Demographic  
 Age, years0.02 (0.00, 0.03)0.03
 Sex (0 = female, 1 = male)−0.10 (−0.39, 0.20)0.51
 BMI, kg/m2−0.01 (−0.04, 0.02)0.37
Radiographic  
 Mechanical axis angle, degrees  
  Preoperative measure0.01 (−0.03, 0.06)0.56
  6-month postoperative measure−0.14 (−0.18, −0.10)<0.01
 OA severity  
  Medial compartment−0.17 (−0.47, 0.12)0.25
  Lateral compartment (0 = less, 1 = more)§0.07 (−0.16, 0.29)0.57
Gait  
 First peak knee adduction moment, %BW × Ht0.24 (0.08, 0.39)0.01
 Speed, meters/second−0.09 (−0.79, 0.60)0.79
 Toe-out angle, degrees0.01 (−0.01, 0.03)0.45
 Lateral trunk lean, degrees−0.04 (−0.07, 0.00)0.05
Patient-reported outcome  
 KOOS pain domain0.00 (−0.01, 0.01)0.58

Figure 2 shows that the cluster analysis identified 2 clusters (Akaike's information criterion = 1,368.57) of patients who had similar values for their 2-year changes in first peak knee adduction moment and KOOS pain score. The larger patient cluster (n = 92) shows patients with substantial decreases in both knee joint load (mean ± SD change in first peak knee adduction moment 1.33 ± 0.61) and knee pain (mean ± SD change in KOOS 24.47 ± 14.45). The smaller cluster (n = 10) highlights a group of patients that leaned toward relatively smaller decreases in knee joint load (mean ± SD change in first peak knee adduction moment 0.66 ± 0.73) and increases in knee pain (mean ± SD change in KOOS −15.27 ± 14.00). Consistent results were achieved when the order of cases in the sample was randomized. Consistent with the clustering criteria, 2-year change in first peak knee adduction moment and 2-year change in KOOS pain scores were significantly different between these groups (P < 0.01). However, no other demographic and clinical characteristics were statistically different. Outliers (n = 24) were not evaluated.

Figure 2.

Scatter plot of 2-year changes in knee adduction moment and Knee Injury and Osteoarthritis Outcomes Score (KOOS) pain domain scores shows results of the cluster analysis. The larger patient cluster (large open circles, n = 92) shows patients with substantial decreases in both knee joint load and knee pain. The smaller patient cluster (solid circles, n = 10) shows a group of patients that leaned toward relatively smaller decreases in knee joint load and increases in knee pain. Outliers did not fit into either cluster (small open circles, n = 24). BW = body weight; Ht = height.

DISCUSSION

We hypothesized that medial opening wedge HTO would result in substantial, clinically important improvements in dynamic knee joint load and patient-reported pain. The present findings strongly support this hypothesis. Effect sizes and standardized response means for the first peak knee adduction moment (1.50 and 1.58, respectively) and each of the KOOS domains (0.89–1.38) are above the threshold value for a large effect (0.8) (46). Additionally, even the lower ends of the 95% CIs for the 2-year changes in the KOOS pain, function in daily living, function in sports/recreation, and knee-related quality of life domains (Table 2) indicate improvements beyond the minimum to suggest clinically important changes have occurred in the majority of patients (43). These confidence intervals are also consistent with previously reported changes in the KOOS after HTO procedures (6, 7). We believe that the large improvements in both biomechanical and patient-reported outcomes observed 2 years postoperatively underscore the potential benefit of surgically restoring neutral lower extremity alignment.

It is important to note that these changes were achieved with intentional minimal to no overcorrection of alignment. Figure 1 shows that, on average, alignment was altered to a position of only slight valgus (mechanical axis angle 1°, anatomic axis angle 5°), which was maintained throughout the 2-year followup. The absolute magnitude of change in the mechanical axis angle achieved with surgery is approximately equal to the amount of preoperative varus (Table 2). This is consistent with our aim of HTO, which generally attempts to establish neutral alignment in all patients despite the preoperative measure. Change in alignment was accompanied by a very small change in tibial slope and patellar height. However, even the upper ends of the 95% CIs for change in tibial slope (2.2°) and Blackburn-Peel ratio (<0.1) were extremely small and likely not clinically important (Table 2). Change in alignment was also accompanied by a very small increase in medial joint space and a decrease in lateral joint space. Overall, these radiographic findings suggest that the change in alignment produced a sustained lateral shift in weight-bearing load without producing detrimental changes in knee geometry. These findings are consistent with recent reports of favorable radiographic results with minimal complications after medial opening wedge HTO (11, 47).

We observed several changes in gait that are also consistent with the biomechanical rationale for this surgery (Table 2). Despite a slight increase in speed and a decrease in lateral trunk lean toward the stance extremity (both reflect a more normal gait pattern and would increase medial joint load), there was a large reduction in all measures of the external knee adduction moment postoperatively. Interestingly, there was also a small increase in toe-out angle that would serve to further reduce the knee adduction moment. It is unclear whether the change in toe-out angle was a structural change as a result of surgery or a gait adaptation adopted by patients. Because toe-out angle has been demonstrated to have a protective effect on future knee joint degeneration (48), the effect of medial opening wedge HTO on toe-out angle should be evaluated further.

Importantly, unlike alignment, we observed an unexpected increase in the knee adduction moment from 6 to 24 months postoperatively (Figure 1). The increase was not observed on the opposite extremity, and therefore is not likely an overall response to the small increase in gait speed observed over the same postoperative time period. The magnitude of the increase in postoperative knee adduction moments from 6 to 24 months was small (13%), and its value at 2 years postoperatively was still far less than those observed preoperatively or on the opposite extremity (55% and 65%, respectively). However, we believe that these relatively early postoperative increases in the knee adduction moment are of potential concern and may be a precursor to poorer longer-term outcomes. Changes in the knee adduction moment without concomitant changes in mechanical axis angle are consistent with previous reports of the low to moderate correlation between these measures, and further emphasize the difference between dynamic measures of knee joint load and static measures of alignment (48–50). Further research evaluating changes in knee joint load observed within 2 years postoperatively, including potential underlying mechanisms, is warranted. Most importantly, the potential for these changes to affect long-term outcomes needs to be evaluated.

We also hypothesized that preoperative patient characteristics would be associated with 2-year outcomes, which was only partly supported. Linear regression analyses yielded no significant predictors of 2-year KOOS pain or KOOS function in daily living scores other than their preoperative values. Only preoperative knee adduction moment, trunk lean, age, and mechanical axis angle at the 6-month followup (i.e., alignment achieved with surgery) explained substantial variance in 2-year knee adduction moments. It is possible that other factors not evaluated in the present study, including other clinical characteristics or more elaborate measures of neuromuscular function and motor control, contributed to changes. However, we believe that the relatively little variance in all outcomes 2 years postoperatively is the primary reason for our findings and we continue to evaluate this objective with longer-term followup.

Owing to our 2 primary goals of HTO, decreasing the load on the medial tibiofemoral compartment and improving patient-reported outcomes, we performed a cluster analysis to identify patients who experienced similar changes in these outcomes. A large cluster of 92 patients experienced substantial improvements in both knee adduction moment and KOOS pain domain scores, and reflect the general success of this procedure. However, a cluster of 10 patients who had poor results in both outcomes was also identified. Given the small sample size, we could not identify any significant differences in patient characteristics between these clusters. Similarly, no trends for increased prevalence of patient characteristics, including disease severity in the lateral compartment, age, sex, or correction angle, were evident. The exploratory nature of cluster analysis should be acknowledged. Given the goals of surgery, however, we believe that this small cluster represents a subgroup of patients with particularly poor results that require further investigation.

Inclusion and exclusion criteria for the present study are also somewhat different from previous reports and should be highlighted. We recruited patients of all ages with varus gonarthrosis and greatest radiographic severity and pain in the medial compartment, but did not exclude those with concomitant lateral compartment disease. Typically, these were young patients with substantial varus deformities who were not candidates for total knee arthroplasty. We do not advocate HTO as a widespread treatment option for patients with substantial bicompartmental tibiofemoral disease. However, 2 patients (ages 43 and 50 years) had lateral compartment OA that also met the K/L scale criteria for grade 4 severity (although medial compartment degeneration was greater) and underwent the procedure. Both patients experienced results similar to many in the sample, with reductions in the first peak knee adduction moment of approximately 1% BW × Ht and improvements in the KOOS pain score of approximately 30 points 2 years postoperatively. We continue to closely follow patients meeting these criteria.

Strengths of our study include its prospective design that adhered to published guidelines for conducting observational studies (12), its relatively large sample size, and the spectrum of validated outcome measures relevant to HTO. A limitation in this design is the inability to compare the effects of HTO with other surgical and nonsurgical treatment options. Similarly, although the meaning and expected outcome patients attribute to treatment is likely to influence outcomes (51), comparison of HTO with sham surgery is not realistic. However, the importance of comparing other less invasive surgical procedures for knee OA with medical and physical therapies (52) and even sham treatment (53) has been established. Clearly, future research comparing HTO with different treatment strategies is required. Other limitations in the present study include the relatively small number of women in our sample and the inability to generalize the results beyond 2 years postoperatively. Overall, the present findings suggest that improvements in malalignment achieved by medial opening wedge HTO produce substantial and clinically important changes in knee joint loading during gait and patient-reported measures of pain, function, and quality of life 2 years postoperatively. Changes in knee adduction moment observed within the first 2 years postoperatively should be explored as potential predictors of long-term success and subgroups of patients with poor outcomes.

AUTHOR CONTRIBUTIONS

All authors were involved in contributions to study conception and design, acquisition of data, or analysis and interpretation of data, and drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Birmingham had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Ancillary