Participants
Eighteen patients with medial KOA (4 men and 14 women; mean age: 60.17±6.96 years) were recruited from the outpatient department of orthopedics at Kashiba Asahigaoka Hospital; 18 healthy adults (4 men and 14 women; mean age: 62.78±8.58 years) were recruited from participants in a health promotion project conducted by Koryo Town, Nara, Japan, and from the staff members of Kashiba Asahigaoka Hospital. Participant characteristics are shown in Table 1.
Table 1
Participant characteristics
| Controls (n=18) | KOA (n=18) | p-value |
Age (years) | 62.78 (8.58) | 60.17 (6.96) | 0.323 |
Height (cm) | 157.83 (8.89) | 161.44 (7.38) | 0.194 |
Weight (kg) | 54.33 (10.60) | 65.97 (11.07) | 0.003* |
Female, n (%) | 14 (78%) | 14 (78%) | 1.000 |
Walking Speed (m/s) | 1.30 (0.15) | 1.18 (0.13) | 0.016* |
Tibiofemoral alignment (deg) | 177.35 (1.95) | 180.28 (1.87) | |
K/L grade, 2/3/4 | N/A | 11/ 7/ 0 | |
K/L grade, Kellgren and Lawrence grade. Values are presented as mean (SD). P-values were obtained from the independent t-test. *P<0.05 |
The inclusion criteria for the participants with medial KOA were as follows: Age >45 years and radiographic changes with Kellgren-Lawrence (KL) grade of 2 or higher in the medial tibiofemoral compartment. The exclusion criteria for the participants with medial KOA were as follows: A history of any other orthopedic injury in the lower extremities, neurological injury, rheumatoid arthritis, joint surgery in lower extremities, lateral KOA, or use of an assistive device. For control participants, the inclusion criteria were as follows: age >45 years, walking independently without any ambulatory assistive device, and no pain in the knee joint, and we excluded those with any previous surgical treatment for the lower limbs or trunk or known neuromuscular or musculoskeletal pathologies.
Gait analysis
Six VICON MX-F20 cameras (Vicon Metrics, Oxford, UK) and four floor-embedded force platforms (2 × 9281B, Kistler Instrument Corporation, Switzerland; 2 × AMTI BP400600, Advanced Mechanical Technology Inc., USA) were used to measure 3D kinematics and GRF with a sampling rate of 120 Hz, respectively. The plug-in-gait full body 39 markers set was used. These markers were attached to each participant’s skin at the following anatomical landmarks: forehead, back of the head, shoulder, 7th cervical vertebra (C7), 10th thoracic vertebra, upper arm, elbow, forearm, wrist, finger, right back, anterior superior iliac spines (ASIS), posterior superior iliac spines (PSIS), thigh, lateral knee, tibia, lateral ankles, heels, and second metatarsal heads.
Participants were asked to walk at a self-selected speed, barefoot, along a flat 8-m walkway. Three successful walking trials were recorded. Stance phases during gait were determined using vertical GRF data. GRFs were low-pass filtered using 10-Hz cut-off frequencies for walking. The threshold value for the stance phase was a GRF of 25N or more in the vertical direction. The walking speed was calculated from the distance of the right and left second metatarsal head marker positions during 6 steps and the time taken.
Musculoskeletal model
A musculoskeletal model with 18 body segments and 92 muscle-tendon actuators in OpenSim was used to compute the CF [12, 14]. In this knee mechanism, the medial and lateral TFJs share all forces transmitted between the femur and tibia and resolve them as the medial and lateral CFs required to balance the net reaction forces and frontal-plane moments across the TFJ. Correspondingly, the knee remains a single degree of freedom (DOF) joint with motion only in the sagittal plane. Therefore, the model includes a spline function on the knee axis, which takes into account the anterior-posterior translations [12, 14]. The model has only a single DOF for dorsi-/plantarflexion and a single axis for the ankle joint. For each participant, subject-specific models were created based on the approach reported by Lerner et al. [12]. In the KOA group, standing anatomical motion capture markers were supplemented with anteroposterior weight-bearing radiograph of the lower-extremity [12, 15]. The dimensions of each body segment in the model were scaled based on relative distances between pairs of markers obtained from a motion-capture system and the corresponding virtual marker locations in the model. Next, we modified each participant’s scaled model and created a subject-specific model by specifying their lower extremity alignment from the radiographs in KOA patients. An anteroposterior radiograph of the participant's lower-extremity was used to determine the subject-specific alignment for the musculoskeletal model. The tibiofemoral alignment was found by drawing lines connecting the hip, knee, and ankle joint centers, which were defined as the center of the femoral head, center of the femoral condyles, and midpoint of the medial and lateral margins of the ankle, respectively. The tibiofemoral alignment was adjusted only for the healthy adults using the marker information [16]. The model was then used to calculate joint angles and moments during walking trials using OpenSim inverse kinematics and inverse dynamics, respectively. MF were estimated using static optimization. Mini-mizing of activation squared was used as the static optimization criterion for the cal-culation of MF. These results were used to calculate the joint reaction force (JRF) in OpenSim.
Foot assessment
The FPI was calculated in a relaxed standing position to assess foot posture based on a six-item standard protocol [17]. The FPI included the following: 1) talar head palpation, 2) supra and infra lateral malleolar curvature, 3) calcaneal frontal plane position, 4) prominence in the region of the talonavicular joint, 5) congruence of the medial longitudinal arch, and 6) abduction/adduction of the forefoot on the rear foot. Each item was scored on a scale of –2, –1, 0, +1, and +2 (0 for neutral, –2 for clear signs of supination, and +2 for clear signs of pronation) [17]. The total score ranged from –12 to +12, with a larger positive value indicating a more pronated foot. For foot type identification, normative values with scores of ≥6 represent a pronated foot type; those with scores of 0–5 represent a neutral foot; and those with scores of ≤−1 represent a supinated foot [17]. The inter-rater and intra-rater reliability for measuring FPI was excellent (ICC: 0.98 and 0.91, respectively).
A force platform (WinFDM; Zebris, Isny im Allgäu, Germany) was used to measure static and dynamic footprint data. Static footprint data were recorded for 10 seconds with a participant standing on the platform with double limb support. Dynamic footprint data were collected during walking at a self-selected speed on the platform. The Staheli Arch Index (SAI), the ratio of the smallest width of the midfoot to the greatest width of the rearfoot [18], was calculated from footprint data. The value of the SAI increases with increasingly planus foot morphology and takes on a value of zero with cavus foot morphology. The widths were measured by ImageJ software [19]. A previous validity study indicated a moderately significant correlation between SAI measurements of foot morphology and radiological measurements of total foot morphology [18]. The average value of two measurements was used in the analysis. The reliability of SAI measurements using ImageJ was calculated by 2 testers in 20 participants. The inter-rater and intra-rater reliability for measuring the SAI was excellent (ICC: 0.96 and 0.91, respectively).
The hallux valgus angle (HVA) and calcaneus inverted angle relative to the floor (CIA) were measured using a goniometer. HVA was the angle formed by the first metatarsal bone and the proximal phalanx of the hallux [20]. Janssen et al. [20] investigated the reliability and concurrent validity of universal goniometer measurements of the HVA compared to those of radiographs and found a good intraclass correlation coefficient. CIA was defined as the angle formed by the bottom of the calcaneal tuberosity and the enthesis of the Achilles tendon [21]. The reliability of CIA measurements was calculated by 2 testers for 20 participants. The inter-rater and intra-rater reliability for measuring the CIA was good (ICC: 0.92 and 0.86, respectively).
Navicular height (NH)/foot length was measured as the length from the floor to the top of the navicular tuberosity [22]. The measure of NH in a standing or sitting position demonstrated good intra- and inter-rater reliability [22].
Navicular drop test (ND-t) was recorded as the difference in NH between relaxed standing and sitting position [21]. ND-t had moderate intra- and inter-rater reliability [22].
TGS was measured using a toe grip dynamometer (T.K.K.3362; Takei Scientific Instruments Co., Ltd., Niigata, Japan) with participants sitting upright [23]. Participants gripped the grip bar with maximal effort for about 3 seconds. The average value of two measurements was used in the analysis. In this measurement protocol, substantial to almost perfect inter- and intra-rater reliability was found in people aged 60–79 years [23].
Statistical analysis
Gait trial data were normalized by time to 0%–100% of the stance phase in the gait cycle. Each CF, normalized by body weight (BW), was represented as a time series throughout the stance phase (SP). The first and second peaks of the TCF, MCF, and LCF were identified. The maximum CF during 0%–50% of SP was defined as the first peak, and the maximum CF during 51%–100% of SP was defined as the second peak.
Differences in demographic data, CF, FPI, Standing SAI, Walking SAI, HVA, CIA, NH, ND-t, TGS, and gait speed between the groups were evaluated by a t-test. Sex differences were examined by chi-square test. For the relationship between foot posture or function and CF, correlation coefficients of CF with foot parameters that were significantly different between the groups were calculated. If a significant correlation between CF and a foot parameter was found for the first peak, we then additionally performed a correlation between the MF at the peak and the foot parameter. The correlation between walking speed and CF or foot parameters was further calculated. The Shapiro-Wilk test was used to determine the normal distributions. In normally distributed data, Pearson’s correlation tests were used. The power analysis was performed by selecting a desired power of 0.8.
All statistical analyses were performed using SPSS Statistics for Windows, version 22.0 (IBM Corp, Tokyo, Japan). A p-value of <0.05 was considered significant.