Three-Dimensional gait biomechanics in patients with mild knee osteoarthritis

Abstract

Knee osteoarthritis (KOA) is a common degenerative joint disease in older adults that causes pain and functional impairment. Gait biomechanics in early-stage KOA (Kellgren–Lawrence grades I–II) are understudied. This study aimed to examine differences in three-dimensional gait biomechanics and muscle activation in mild KOA to inform early detection and intervention. Twenty-four patients (aged 55–70 years) with unilateral mild KOA and twelve age- and sex-matched healthy older adults were recruited for evaluation. Participants were instructed to walk at a self-selected, comfortable speed along a 6-meter walkway, and at least five valid gait trials were collected for each individual. Three-dimensional gait analysis was conducted using a motion capture system synchronized with force plates to measure spatiotemporal parameters (gait cycle, step width, walking speed), joint kinematics (ROM, peak angular velocity), and joint kinetics (peak joint moments). Muscle activation levels (normalized to %MVIC) and muscle onset times were recorded using a 16-channel wireless surface electromyography system. Between-group differences were assessed using independent-samples t-tests (p < 0.05), with effect sizes calculated using Cohen’s d. Mild KOA patients had a significantly longer gait cycle (1.12 ± 0.13 vs. 1.10 ± 0.07 s, p = 0.04, d = 0.18) and a wider step width (0.09 ± 0.03 vs. 0.07 ± 0.04 m, p < 0.01, d = 0.60) than healthy controls, while walking speed remained similar between groups. They exhibited reduced knee and ankle range of motion and lower peak angular velocities at the hip and knee joints compared to controls. For example, sagittal-plane knee flexion-extension ROM was 61.5°±5.1 vs. 65.1°±2.8 (KOA vs. control, p < 0.01, d = 0.85), and ankle plantarflexion-dorsiflexion ROM was 53.4°±8.1 vs. 59.1°±6.5 (p < 0.01, d = 0.77). Peak knee angular velocity was also lower in KOA patients (372 ± 58 vs. 399 ± 72 °/s, p = 0.01, d = 0.41). KOA patients also generated lower peak flexion and extension moments at the hip and knee joints compared to controls, along with a higher external rotation moment at the affected knee. Muscle activation patterns differed between the groups. The KOA group showed reduced activation of the gluteus medius and medial gastrocnemius (GMed: 27% ± 23 vs. 38% ± 19; MG: 41% ± 24 vs. 62% ± 16; both p < 0.01) but higher activation of the biceps femoris and tibialis anterior. Furthermore, the gluteus maximus on the affected side activated later in the gait cycle, whereas the semimembranosus and the contralateral gluteus maximus activated earlier than in controls. Even at a mild stage, KOA is associated with distinct gait and neuromuscular alterations. These findings underscore the importance of early gait assessment and targeted interventions to improve dynamic stability and potentially slow the progression of osteoarthritis.

Introduction

Knee osteoarthritis (KOA) is a chronic degenerative joint disease common in middle-aged and elderly individuals, characterized by articular cartilage degeneration, inflammation, pain, and functional impairment¹. KOA often causes joint pain, stiffness, swelling, and reduced range of motion (ROM), which severely affects daily activities such as walking, stair climbing, and working. Over time, these symptoms can lead to a loss of the ability to live independently in older adults​². The incidence of KOA has been shown to increase with age³. In recent years, with the aging global population, the prevalence of KOA has steadily risen, making it one of the major conditions affecting quality of life in the elderly². In 2020, approximately 595 million people worldwide were affected by osteoarthritis, accounting for 7.6% of the global population⁴.

Current treatment approaches for KOA can be broadly divided into surgical or pharmacological interventions and non-pharmacological strategies​⁵. Unfortunately, although medications and surgical treatments can alleviate pain to some extent, they cannot cure the disease and often carry significant side effects​⁶. For example, non-steroidal anti-inflammatory drugs (NSAIDs) are a first-line treatment for KOA pain and inflammation​⁵, but long-term NSAID use may lead to adverse effects such as gastrointestinal ulcers and bleeding​⁶. Therefore, more effective treatments and improved diagnostic tools are needed to better manage KOA and slow its progression​.

Walking is the most fundamental form of physical activity for adults, and gait analysis has become a valuable tool for evaluating functional limitations and treatment outcomes in KOA patients⁷. Studies have shown that KOA patients exhibit altered lower-limb biomechanics during walking^8,9. For instance, compared to healthy individuals, KOA patients demonstrate reduced knee ROM during the swing phase, lower heel-strike and toe-off angles, increased knee ROM in the coronal plane, and reduced hip ROM​^10,11. Muscle imbalances may also contribute to KOA progression; evidence indicates that during gait, KOA patients have reduced quadriceps activation and increased hamstring activation^8,12.

However, the aforementioned studies mainly focused on moderate to severe KOA and primarily on sagittal-plane abnormalities. Research on the biomechanical and functional characteristics of the lower limbs in mild KOA—particularly potential changes in the coronal and transverse planes—is limited​​^9,11. Moreover, few studies have simultaneously analyzed both lower-limb kinematics and muscle activation patterns in mild KOA​. This knowledge gap hinders our understanding of KOA across different stages and may impede early rehabilitation efforts​. A recent review highlighted the importance of tailoring physical activity programs according to knee OA severity (defined by Kellgren–Lawrence grade) to achieve better outcomes​¹³. In this context, biomechanical analysis of early-stage KOA could play a key role in identifying functional impairments and optimizing physical activity prescriptions based on early disease biomechanics.

Therefore, this study aimed to compare the three-dimensional gait biomechanics of individuals with mild KOA and healthy controls, in order to fill the knowledge gap regarding the relationship between physical activity type and intensity and KOA progression as defined by the Kellgren–Lawrence grading system​. This research provides a theoretical basis for health management and clinical rehabilitation strategies in early-stage KOA​. We hypothesized that patients with mild KOA would exhibit a longer gait cycle, greater step width, and slower gait speed, potentially accompanied by compensatory changes in sagittal- and coronal-plane parameters and abnormal muscle activation patterns in certain lower-limb muscles​^8–12.

Methods

Participant selection and study protocol

This study received ethical approval from the Guangzhou Sport University Ethics Review Committee for Human Experiments (No. 2022LCLL-32). Based on prior research​¹⁴, we recruited a higher number of KOA participants (using a 2:1 ratio of patients to controls) to account for potential variability in gait patterns among KOA patients.

Inclusion criteria

Participants in the KOA group were 55–70 years old with a body mass index (BMI) ≤ 28, met the 1995 American College of Rheumatology diagnostic criteria for KOA, and had unilateral knee OA confirmed by X-ray (Kellgren–Lawrence grade I or II)​​¹⁵. Additionally, the affected knee had to be the dominant side to eliminate the influence of limb dominance. Furthermore, KOA participants had no injuries to the hip or ankle, and no other conditions that could cause knee pain. Healthy control participants were 55–70 years old with BMI ≤ 28 and had no clinical symptoms or radiographic evidence of KOA. Controls also had no history of significant lower-limb or lower-back pain, and no other painful or neurological conditions that could affect gait biomechanics.

Exclusion criteria

Participants (in either group) were excluded if they were unable to walk independently for at least 20 m, had undergone knee surgery within the past 6 months or received a corticosteroid injection in the knee within the past 3 months, or were currently taking pain-relieving medication​. Individuals with any primary or secondary neuromuscular disorder (e.g., Parkinson’s disease or severe muscle spasms) were also excluded, as were those with severe psychiatric disorders, such as major depression, obsessive-compulsive disorder, or schizophrenia.

Out of 40 initial patients with KOA who were assessed for eligibility, 24 met all criteria and were ultimately included as the KOA group (8 males, 16 females). The remaining 16 candidates were excluded for not meeting the inclusion criteria or other reasons (e.g., presence of bilateral KOA, radiographic grade higher than II, recent knee treatments, or inability to complete the gait protocol). Recruitment and screening were conducted by specialists in the musculoskeletal system, rheumatic diseases, and geriatrics in accordance with the study objectives to minimize selection bias during enrollment. Twelve healthy older adults (5 males, 7 females) from the same region, matched by age and sex, were recruited as the control group.

Gait mechanics were assessed using two force platforms (AMTI OR6-7, 60 × 40 cm, Watertown, MA, USA) embedded in the floor, recording at 1000 Hz​. Ground reaction force data from these platforms were used to calculate joint moments, which were normalized to body weight. Kinematic data, including joint angles and angular velocities, were collected with a 10-camera Vicon Nexus motion capture system (Vicon, Oxford, UK) operating at 100 Hz​. The Vicon system utilized 36 retroreflective markers (14-mm diameter) placed on specific anatomical landmarks to model seven lower-limb segments; the marker placement is illustrated in Fig. 1A and detailed in Table 1​.

Fig. 1.

(A) The positions of retroreflective markers. IAS, Anterior superior iliac spine. PS, Posterior superior iliac spine. TH1-5 Cluster, Cluster of four markers placed on the lateral surface of the thigh. FLE, Lateral epicondyle. FME, Medial epicondyle. SK1-4 Cluster, Cluster of four markers placed on the lateral surface of the shank. FAL, Lateral prominence of the lateral malleolus. TAM, Medial prominence of the medial malleolus. FCC, Aspect of the Achilles tendon insertion on the calcaneus. FM1, Dorsal margin of the first metatarsal head. FM5, Dorsal margin of the fifth metatarsal head. (B) The placement of the sensors. GMed, Gluteus medius. GM, Gluteus maximus. BF, Biceps femoris. ST, Semimembranosus. LG, Lateral gastrocnemius. MG, Medial gastrocnemius. VL, Vastus Lateralis. RF, Rectus femoris. VM, Vastus Medialis. TA, Tibialis anterior. Made by Visual 3D v6 Professional.

Table 1.

Retroreflective markers’ name and body segment (Left).

Muscle activity was measured using a 16-channel Delsys Trigno Avanti wireless surface EMG (sEMG) system (Delsys, Boston, MA, USA) at a sampling rate of 2000 Hz. Sensor placement followed the guidelines from the Surface Electromyography for the Non-Invasive Assessment of Muscles (SENIAM) project​ (Fig. 1B)¹⁶. Prior to sensor attachment, hair at each electrode site was removed and the skin was gently abraded and cleansed to reduce impedance, then allowed to dry. The sensors were attached to the skin and secured with adhesive tape. For KOA patients, EMG data were collected on the affected side; for control participants, data were collected on the dominant side (as determined by tasks such as kicking a ball or stepping onto an object)​¹⁷.

In the preparation phase, participants changed into black tight-fitting shorts and performed warm-up exercises in the laboratory to become familiar with the environment. During this time, the researchers explained the experimental procedures and requirements to each participant. After the warm-up, the reflective markers and sEMG sensors were applied as described, and each participant’s height and weight were recorded.

Before the walking trials, a static calibration trial was conducted with the participant standing in an anatomical position (as shown in Fig. 2) for 5 s; these static data were used for subsequent data normalization​. Next, participants stood at a designated starting position and, upon the “start” command, walked along a 6-meter walkway at a comfortable, self-selected speed. A trial was considered valid if the first force platform captured the contact of one foot and the second force platform captured the contact of the opposite foot. A total of five valid trials were collected for each participant, and participants were allowed to practice beforehand to ensure natural walking​.

Fig. 2.

Static data collection.

Gait analysis procedure

After data collection, the Vicon Nexus software was used to preprocess the motion capture data, including noise filtering and gap-filling of marker trajectories​. The processed kinematic and force data were then imported into Visual3D software (C-Motion, Germantown, MD, USA) for analysis​. Heel-strike and toe-off events were identified from the kinematic and kinetic data to define each gait cycle​. Using Visual3D’s built-in gait analysis functions, we calculated spatiotemporal gait parameters, including total gait cycle time (s), step length (m), step width (m), walking speed (m/s), single-limb support phase (%), swing phase (%), and double support phase (%)​. Joint angles and angular velocities of the hip, knee, and ankle throughout the gait cycle were obtained by computing the relative angles between adjacent segments. Figure 3 illustrates the definitions of joint movement directions​. Using the 3D kinematic and force plate data, we calculated the internal joint moments at the hip, knee, and ankle during walking via inverse dynamics. All joint moments were normalized to body weight (reported in N/kg). Details of these calculations are provided in the Visual3D Inverse Dynamics documentation - Inverse Dynamics [HAS-Motion Software Documentation].

Fig. 3.

Definition of joint movement directions. Flexion: Decreasing the angle between two body segments at a joint; Adduction: Movement toward the midline of the body; Internal Rotation: Rotation of a limb toward the body’s midline, causing inward turning. Made by Visual 3D v6 Professional.

The raw EMG signals were processed in Visual3D with the following steps: high-pass filtering at 400 Hz, full-wave rectification, low-pass filtering at 20 Hz, calculation of the root mean square (RMS), normalization, and determination of muscle onset time. The RMS value reflects the level of motor unit activation during walking¹⁸​. All muscle activation levels were expressed as a percentage of each muscle’s maximal voluntary isometric contraction (MVIC)¹⁹​. For MVIC testing, each target muscle was gradually engaged to a maximal effort while a researcher provided resistance to the corresponding limb; the participant maintained a contraction for 3–5 s at maximum effort before relaxing. Each MVIC trial was performed three times for each muscle, with 30–60 s of rest between trials, to ensure accuracy and repeatability. All MVIC tests were conducted manually by the same researcher.

Muscle onset time was defined as the point in the gait cycle when a muscle’s EMG signal first continuously exceeded a baseline threshold (the resting mean + 2 standard deviations), indicating the beginning of muscle activation²⁰​. Activation times were normalized to the gait cycle, with the heel strike of the affected (or dominant) limb defined as 0% and the subsequent heel strike of the same limb defined as 100% of the gait cycle​.

Data analysis

Statistical analyses were performed using IBM SPSS Statistics (version 26.0; IBM Corp., Armonk, NY, USA)​. The Shapiro–Wilk test was used to assess the normality of the data. For normally distributed data, independent-samples t-tests were conducted to compare group differences in demographic variables, spatiotemporal gait parameters, kinematic data, and kinetic data​. For data not following a normal distribution (such as certain EMG parameters), the values were log-transformed before applying t-tests​. A two-tailed p < 0.05 was considered statistically significant. Cohen’s d was calculated to estimate effect sizes, where d < 0.2 indicates a small effect, 0.2 ≤ d < 0.5 a medium effect, and d ≥ 0.5 a large effect.

Results

Demographic characteristics

There were 24 participants in the KOA group and 12 in the control group. The two groups showed no significant differences in age, height, body mass, or BMI, indicating they were well-matched demographically​. This comparability suggests that any observed differences in gait or muscle function can be attributed to the presence of KOA rather than to differences in baseline characteristics (Table 2)​.

Table 2.

Descriptive participant demographics (± SD). BMI body mass index.

Parameters	KOA Group	Control Group	p-value	Cohen’s d
	(n = 24)	(n =12)
Age (years)	63.00 ± 4.50	63.50 ± 4.00	0.7	0.11
Height (cm)	162.00 ± 6.50	161.50 ± 6.00	0.8	0.08
Body Mass (kg)	62.00 ± 6.00	61.50 ± 5.50	0.75	0.09
BMI (kg/m²)	24.00 ± 1.50	23.50 ± 1.50	0.6	0.33
K-L grade I/II	May-19	-	-	-

Walking parameters

As shown in Table 3, patients with mild KOA exhibited a significantly longer gait cycle and a greater step width compared to healthy controls​. Walking speed tended to be lower in the KOA group on average, but this difference was not statistically significant.

Table 3.

Walking parameters between KOA and control group (± SD).

Parameters	KOA Group	Control Group	p-value	Cohen’s d
	(n = 24)	(n = 12)
Gait cycle time (s)	1.12±0.13	1.10±0.07	0.04	0.18
Stride/Height	0.74±0.08	0.73±0.06	0.25	0.13
Speed (km/h)	1.07±0.15	1.08±0.10	0.06	0.07
Step width (m)	0.09±0.03	0.07±0.04	<0.01	0.6
Double stance phase (%)	21.87 ± 3.56	21.33 ± 2.08	0.9	0.17
Affected/Dominant side
Stance phase (%)	61.59± 6.78	60.31 ± 5.44	0.3	0.2
Swing phase (%)	40.37 ± 4.38	40.19 ± 3.53	0.49	0.04
Non-affected/Non-dominant
Stance phase (%)	61.31 ± 7.14	61.41 ± 4.61	0.76	0.02
Swing phase (%)	40.53 ± 4.02	39.63 ± 3.05	0.09	0.24

Kinematic parameters

According to the results in Table 4, individuals with KOA demonstrated smaller knee and ankle ROM in both the sagittal and coronal planes compared to controls. Additionally, in the transverse plane, hip joint ROM was decreased while knee joint ROM was increased in the KOA group. The KOA patients also showed lower peak angular velocities at the hip and knee joints than the control group​.

Table 4.

Kinematic parameters between KOA and control group (± SD).

Paraments	KOA Group	Control Group	p-value	Cohen’s d
	(n = 24)	(n = 12)
ROM (°) in Sagittal Plane
Hip	42.06 ± 4.13	44.10 ± 7.66	0.09	0.33
Knee	61.54 ± 5.12	65.06 ± 2.84	<0.01	0.85
Ankle	53.44 ± 8.09	59.08 ± 6.53	<0.01	0.77
ROM (°) in Coronal Plane
Hip	13.30 ± 3.53	13.65 ± 3.17	0.52	0.1
Knee	7.12 ± 2.59	8.12 ± 2.77	0.02	0.37
Ankle	20.64 ± 7.13	25.23 ± 4.29	<0.01	0.78
ROM (°) in Horizontal Plane
Hip	11.80 ± 3.28	14.44 ± 4.65	<0.01	0.66
Knee	11.95 ± 3.51	10.95 ± 3.33	0.09	0.29
Ankle	7.68 ± 3.28	7.70 ± 2.71	0.96	0.01
Maximum Angular Velocity (deg/s) in Sagittal Plane
Hip	188.21 ± 26.28	198.46 ± 27.72	0.03	0.38
Knee	372.34 ± 58.13	399.28 ± 71.81	0.01	0.41
Ankle	160.73 ± 48.34	160.04 ± 47.79	0.93	0.01

Kinetic parameters

Figure 4 illustrates that the peak extension and flexion moments at the hip and knee joints were lower in the KOA group than in the control group. The affected knee of KOA patients exhibited a higher peak external rotation moment, whereas the corresponding joints on the unaffected side had lower peak moments across multiple movement directions.

Fig. 4.

Joint peak moment between two groups (N/kg). * means p < 0.05. PEM: peak extension moment; PFM: peak flexion moment; PABM: peak abduction moment; PADM: peak adduction moment; PERM: peak external rotation moment; PIRM: peak internal rotation moment.

Surface EMG parameters

Table 5 shows that the KOA group had lower normalized RMS values (%MVIC) in several muscles compared to controls – specifically, the gluteus medius (bilaterally) and the gluteus maximus, medial gastrocnemius, and rectus femoris on the affected side. In contrast, the biceps femoris and tibialis anterior exhibited higher RMS values in the mild KOA group​. Additionally, the timing of muscle activation differed between groups: the gluteus maximus on the affected side had a delayed muscle onset time, whereas the semimembranosus and gluteus maximus on the unaffected side showed earlier onset times (Table 6)​.

Table 5.

Muscle activation level (%MVIC) in two groups of participants (± SD). GMed, Gluteus medius. GM, Gluteus maximus. BF, Biceps femoris. ST, Semimembranosus. LG, Lateral gastrocnemius. MG, Medial gastrocnemius. VL, Vastus Lateralis. RF, Rectus femoris. VM, Vastus Medialis. TA, Tibialis anterior. † means that the original data has already been logarithmically transformed.

Paraments	KOA Group	Control Group	p-value	Cohen’s d
	(n = 24)	(n = 12)
Affected/Dominant side
BF (%) †	25.31 ± 9.27	21.90 ± 15.45	<0.01	0.27
ST (%) †	19.16 ± 6.10	19.58 ± 5.39	0.5	0.07
LG (%) †	44.31 ± 13.10	47.26 ± 13.76	0.3	0.22
MG (%) †	41.32 ± 23.76	62.35 ± 16.13	<0.01	1.04
VL (%) †	25.24 ± 17.50	26.78 ± 10.48	0.12	0.11
RF (%) †	22.21 ± 5.75	24.71 ± 5.28	0.01	0.45
VM (%) †	22.41 ± 6.17	21.40 ± 7.40	0.22	0.15
TA (%) †	30.68 ± 9.54	27.67 ± 10.02	0.04	0.31
GMed (%) †	27.37 ± 23.05	38.27 ± 18.59	<0.01	0.52
GM (%) †	20.88 ± 9.52	24.24 ± 8.58	0.02	0.37
Non-affected/Non-Dominant side
GMed (%) †	25.19 ± 6.58	34.38 ± 23.86	0.02	0.53
GM (%) †	23.80 ± 18.40	24.15 ± 9.10	0.2	0.2

Table 6.

Muscle onset time in two groups of participants (± SD). † means that the original data has already been logarithmically transformed.

Paraments	KOA Group	Control Group	p-value	Cohen’s d
	(n = 24)	(n = 12)
Affected/Dominant side
BF (%) †	51.84 ± 27.02	53.25 ± 29.30	0.88	0.34
ST (%) †	53.92 ± 30.90	77.63 ± 19.13	<0.01	0.05
LG (%) †	39.32 ± 35.05	40.87 ± 35.77	0.91	0.92
MG (%) †	31.60 ± 26.33	23.26 ± 17.98	0.38	0.04
VL (%) †	53.14 ± 30.34	53.06 ± 35.97	0.26	0.37
RF (%) †	42.63 ± 28.11	44.75 ± 26.10	0.9	0.01
VM (%) †	48.44 ± 26.61	48.04 ± 25.32	0.81	0.08
TA (%) †	28.47 ± 27.36	35.10 ± 28.25	0.22	0.02
GMed (%) †	42.99 ± 31.83	50.03 ± 34.79	0.66	0.21
GM (%) †	38.40 ± 32.27	23.61 ± 26.76	0.01	0.04
Non-affected/Non-Dominant side
GMed (%) †	32.49 ± 23.08	31.68 ± 20.39	0.92	0.21
GM (%) †	36.84 ± 22.10	43.44 ± 16.89	0.03	0.5

Discussion

This study investigated the three-dimensional gait biomechanics of individuals with mild KOA. The results indicate that even at an early stage of the disease, KOA alters gait dynamics—evident in changes to step timing and spatial parameters—and affects neuromuscular function by prompting compensatory muscle activation​. These insights provide guidance for developing rehabilitation strategies aimed at early-stage KOA.

Broadly speaking, the gait alterations identified in mild KOA can be categorized into primary changes directly resulting from knee joint degeneration and secondary compensatory adjustments by the neuromuscular system. For example, knee pain and stiffness (primary symptoms of KOA) may directly limit joint motion¹, whereas increases in step width or altered muscle activation patterns are secondary adaptations aimed at maintaining stability and reducing joint loads^7,21.

In terms of spatiotemporal gait parameters, we found that patients with mild KOA had a significantly prolonged gait cycle and increased step width compared to controls. Fundamentally, these changes reflect an adaptive adjustment by the neuromuscular system in response to reduced knee joint stability resulting from early osteoarthritic changes²²​. Diminished activation of the hip stabilizing muscles (gluteal muscles) and delayed activation of the gluteus maximus can lead to weakened proximal control and reduced balance​²³, forcing patients to compensate by lengthening the stance phase and widening their base of support to enhance dynamic stability and reduce the risk of falling²⁴​. However, this secondary compensatory strategy may carry long-term consequences: an increased step width can elevate the knee adduction moment (KADM), concentrating load on the medial compartment of the knee²⁵​ and potentially accelerating cartilage degeneration. In our study, the peak adduction moments (PADM) at the hip and knee in the KOA group were not significantly higher than those in controls, but they showed an upward trend​, consistent with the concern that a wider gait could increase medial knee loading.

Notably, although KOA patients in our study tended to walk more slowly than controls, the difference in walking speed was not statistically significant​. This contrasts with findings in patients with moderate or severe KOA, who typically exhibit significantly reduced gait speeds²⁶​. A previous stratified analysis found that the magnitude of gait speed reduction correlates positively with the Kellgren–Lawrence radiographic grade of KOA²⁷​. The maintenance of nearly normal walking speed in the mild KOA group suggests that when joint structural damage is still relatively limited, patients can preserve their walking pace​. It also implies that gait alterations observed in KOA are more likely a consequence of the disease (i.e., compensatory changes) rather than a precipitating cause of knee osteoarthritis​.

From a kinematic perspective, in the sagittal and coronal planes we observed a general reduction in knee and ankle ROM in the KOA group, indicative of a classic “stiff-knee gait” pattern²⁸​. Landry et al. reported that faster walking speeds demand greater joint moments, particularly noting that knee flexion-extension moments increase substantially as speed increases​²⁹. Thus, a stiff-knee gait can be interpreted as a pain-avoidance strategy: patients may deliberately walk slower and limit their knee motion to reduce patellofemoral joint stress and thereby alleviate pain²⁸​. Consistent with this notion, our results showed that mild KOA patients generated significantly lower flexion-extension moments at the hip and knee than healthy individuals​, reinforcing the presence of a stiff-knee, pain-avoidance gait pattern in early-stage KOA.

In the long term, however, this protective gait strategy (a secondary adaptation) may become counterproductive. A reduced sagittal-plane ROM can shift joint contact pressure from a well-distributed pattern to a more localized concentration, increasing the accumulation of micro-damage in the cartilage over time³⁰​. Furthermore, a decrease in joint angular velocity can increase the damping effect within the joint, leading to higher energy expenditure and an earlier onset of fatigue, which may ultimately facilitate disease progression​³¹.

Interestingly, the transverse-plane kinematics exhibited a paradoxical pattern in the mild KOA patients​. Specifically, these patients had decreased rotational ROM at the hip but increased rotational ROM at the knee, suggesting a pathological coupling between hip and knee motion. On one hand, insufficient activation of the hip muscles can result in reduced hip rotation, which the body may compensate for by allowing greater rotation at the knee³²​. On the other hand, earlier activation of the semimembranosus (ST) muscle can cause the hamstrings–gastrocnemius muscle couple to engage too early; this premature ST activation, combined with increased biceps femoris (BF) activation, tends to increase external rotation of the knee​²².

This compensatory mechanism in the transverse plane appears to be an adaptive strategy employed by the central nervous system in the context of quadriceps weakness and insufficient sagittal-plane propulsion³³​. However, the increased activation of BF and the concomitant reduction in medial gastrocnemius (MG) activation disrupt the normal balance between lateral and medial musculature of the leg​. This imbalance reduces control over external rotation of the knee, leading to further increases in the peak external rotation moment (PERM) at the knee and greater transverse-plane motion. Additionally, the suppressed activation of the MG (reflected by a lower RMS value) may contribute to an imbalance between medial and lateral support around the knee, causing the patella to track more laterally. This lateral tracking increases shear stress on the patellofemoral joint and exacerbates wear of the patellar cartilage³⁴​. This biomechanical pathway helps explain the clinical observation that a reduced knee extensor/flexor strength ratio is associated with worsening anterior knee pain in KOA patients​.

Aside from alterations in rotational moments, the knee adduction moment is known to play a key role in KOA progression. Kutzner et al. found that during early stance phase, patients with mild KOA experience significantly higher medial tibiofemoral contact forces and lower lateral contact forces compared to healthy individuals; importantly, the degree of this medial–lateral imbalance correlated with the magnitude of the knee varus moment²⁵​. Moreover, peak medial contact force has been shown to correlate with the loss of medial tibial cartilage volume over time​³⁵. In our study, the KOA group exhibited a similar tendency toward higher peak KADM than controls, although the difference was not statistically significant​. This finding suggests that changes in KADM may be a secondary effect of joint degeneration rather than an initial cause; however, such changes can nonetheless accelerate disease progression, creating a vicious cycle. In the long run, an elevated KADM is considered an independent risk factor for accelerated knee cartilage degeneration^25,35​.

An increased activation of the tibialis anterior (TA) muscle in KOA patients might indirectly affect knee loading by altering ankle mechanics. Computational models have shown that greater TA activation can shift the center of pressure at the knee forward, causing the ground reaction force vector to move laterally relative to the knee and thereby reducing the medial knee load (lowering PADM)³⁶​. Considering the altered activation patterns of the quadriceps and hamstrings observed in this study, we speculate that early-stage KOA patients may be increasing their knee external rotation moment as a strategy to counteract excessive varus (adduction) stress in an attempt to slow disease progression. However, as KOA advances and neuromuscular function deteriorates, such compensatory mechanisms likely become ineffective, failing to break the vicious cycle between an elevated KADM and ongoing joint degeneration^25,35​.

Moreover, while the compensation strategies discussed might offload the affected knee in the short term, they could have detrimental effects on the contralateral limb and other joints in the long term​. Studies have indicated that although increasing external rotation or altering gait can transiently reduce medial knee loading, over time these changes may transfer abnormal loads through the lower-limb kinetic chain, potentially increasing the risk of degenerative changes in the contralateral knee²¹​. Likewise, earlier activation of the semimembranosus and gluteus maximus in the contralateral limb—which helps maintain gait symmetry in the short term—has been shown to increase contact stress in the hip joint of that limb, adding to cartilage load​³⁷. This provides a biomechanical explanation for the clinical observation that KOA often progresses to involve both knees over time​.

Despite the insights gained from this study, there are several limitations that should be acknowledged. First, the small sample size may limit generalizability to other populations or disease stages. Second, the cross-sectional design precludes causal inferences. Third, gait analysis was performed in a controlled laboratory setting, which may not reflect real-world walking conditions. Future studies should include larger, more diverse samples and longitudinal designs to track biomechanical changes across KOA progression. Interventions targeting early gait and muscle activation abnormalities—combined with advanced tools such as wearable sensors or musculoskeletal modeling—may help validate and enhance early rehabilitation strategies.

Conclusion

In summary, individuals with mild KOA already exhibit notable gait and neuromuscular alterations, including a longer gait cycle, wider step width, reduced knee and ankle ROM, altered joint moments, and abnormal muscle onset times. These findings emphasize the importance of early identification and intervention. Targeted rehabilitation focusing on gait retraining, improving joint mobility, and strengthening key muscles may help correct these deviations and delay disease progression. This study provides important biomechanical evidence to support early-stage KOA management and lays a foundation for further clinical intervention research.

Abbreviations

KOA

Knee osteoarthritis

ROM

range of motion

sEMG

Surface Electromyography

RMS

Root Mean Square

PEM

peak extension moment

PFM

peak flexion moment

PABM

peak abduction moment

PADM

peak adduction moment

PERM

peak external rotation moment

PIRM

peak internal rotation moment

GMed

Gluteus medius

GM

Gluteus maximus

BF

Biceps femoris

ST

Semimembranosus

LG

Lateral gastrocnemius

MG

Medial gastrocnemius

VL

Vastus Lateralis

RF

Rectus femoris

VM

Vastus Medialis

TA

Tibialis anterior

Author contributions

J.P. was responsible for experimental design, manuscript drafting, and revisions. Z.X., H.S., and J.L. contributed to data collection, analysis, and interpretation. X.Z. and B.L. coordinated the research, supervised the project, and critically reviewed the manuscript. All authors approved the final manuscript and agreed to be accountable for their respective contributions.

Funding

This work was supported by the Guangzhou Sports Science and Technology Collaborative Innovation Center (2023B04J0466), Open Fund of the Guangdong Provincial Key Laboratory of Physical Activity and Health Promotion (2021B1212040014), Higher Education Teaching Reform Project of Guangdong Provincial Department of Education (Strategy and Practice Exploration of Innovative Training Mode for Sports Rehabilitation Professionals Under the Guidance Of Supply- Side Reform), The Basic and Applied Basic Research Project of Guangzhou City (2023A04J0554), and The Key Scientific Research Project of Guangdong Province Education Department (2024ZDZX2064). These funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The sponsor approved the design, methods, subject recruitment, data collections and analysis in the grant proposal.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to privacy issues. However, data will be available upon formal request to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval and consent to participate

The project was approved by the Human Subject Committee of Guangzhou Sport University (NO: 2022LCLL-32). All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Written informed consent was obtained from all participants for the publication of any potentially identifiable data included in this article.