Patellofemoral Joint Loading During Stair Ambulation in People With Patellofemoral Osteoarthritis

Authors


Abstract

Objective

To determine whether people with patellofemoral (PF) joint osteoarthritis (OA) ascend and descend stairs with different PF joint loading, knee joint moments, lower limb kinematics, and muscle forces compared to healthy people.

Methods

We recruited 17 participants with isolated PF joint OA, 13 participants with concurrent PF joint OA and tibiofemoral (TF) joint OA, and 21 age-matched controls. Joint kinematics and ground reaction forces were measured while participants ascended and descended stairs at a self-selected speed. Musculoskeletal computer modeling was used to determine lower limb muscle forces and the PF joint reaction force, and these parameters were compared between groups by analysis of variance.

Results

Compared to their healthy counterparts, participants with isolated PF joint OA and participants with concurrent PF and TF joint OA ascended and descended stairs with lower knee extension moments, lower quadriceps muscle forces, lower PF joint reaction forces, and increased anterior pelvic tilt. Participants with OA also ascended stairs with increased hip flexion angles and descended stairs with smaller knee flexion angles and smaller hip abductor muscle forces. No differences were evident between the two groups with OA.

Conclusion

Compared to their healthy counterparts, people with PF joint OA (with or without concurrent TF joint OA) exhibit lower PF joint reaction forces during stair ascent and descent, in conjunction with lower knee extension moments and lower quadriceps muscle forces.

Osteoarthritis (OA) frequently affects the patellofemoral (PF) joint ([1, 2]), and PF joint OA is more strongly associated with knee pain and stiffness than is tibiofemoral (TF) joint OA ([3, 4]). While much is known about the biomechanics of TF joint OA, especially the knee adduction moment and its effect on medial compartment loading ([5, 6]), the same cannot be said for PF joint OA ([7]). The PF and TF joints have different biomechanical characteristics, and individuals with PF joint OA may adopt unique gait modifications. Stair ambulation exposes the PF joint to high loads ([8]), while pain and difficulty with negotiating steps are hallmarks of PF joint disorders ([9]). Therefore, people with PF joint OA may demonstrate joint loading impairments during stair ambulation.

The biomechanics of stair ambulation has been studied in individuals with TF joint OA ([10, 11]). Individuals with more severe TF joint OA (Kellgren/Lawrence [K/L] grade 3 or 4 [12]) exhibited greater trunk flexion and a lower internal knee extension moment than did controls during stair ascent ([10]), indicating an endeavor to reduce knee joint loading. However, those with less severe TF joint OA (K/L grade 1 or 2) did not differ from controls. In contrast, Kaufman and colleagues observed lower knee extension moments in individuals with mild TF joint OA (K/L grade 2) than in controls during stair ascent and descent ([11]). Considering that stair ambulation also loads the PF joint, it is likely that individuals with PF joint OA, with or without mild TF joint OA, exhibit biomechanical alterations during stair ambulation.

Quantification of PF joint loading can be determined from the PF joint reaction force, which is the compressive force acting between the patella and femoral trochlea. The PF joint reaction force magnitude is primarily influenced by the knee flexion angle and the magnitude of the quadriceps muscle force ([13]). Attempts to minimize the PF joint reaction force involve alterations to one or both of these quantities. While lower quadriceps strength and smaller vastus muscle volume are found in people with PF joint OA ([14, 15]), we recently reported no difference in peak vastus muscle forces between individuals with and those without PF joint OA during level walking ([16]). Instead, peak forces for the gluteus medius and minimus muscles were lower, suggesting that hip abductor muscle dysfunction may be a characteristic feature of PF joint OA. This finding supports reports of hip abductor muscle dysfunction in younger individuals with PF joint pain ([17]). Given that ascending and descending stairs is more physically challenging than level walking, it is possible that differences in muscle forces, and therefore in PF joint loading, become more evident during stair ambulation in individuals with PF joint OA.

The primary aim of this study was to compare lower limb joint kinematics, knee joint moments, lower limb muscle forces, and PF joint reaction forces between individuals with isolated PF joint OA, individuals with concurrent PF joint OA and TF joint OA, and individuals without OA. Secondary aims were to identify potential contributors to the PF joint reaction force and to explore the relationship between the PF joint reaction force and patient-reported outcomes (e.g., pain, stiffness, function). We hypothesized that during stair ambulation, individuals with PF joint OA would exhibit smaller knee flexion angles, lower quadriceps and hip muscle forces, and lower PF joint reaction forces when compared to their healthy counterparts.

PATIENTS AND METHODS

Participants

Thirty participants with PF joint OA (17 with isolated PF joint OA and 13 with concurrent PF joint OA and TF joint OA [hereinafter referred to as combined OA]) and 21 age-matched controls with no lower limb joint pain were recruited from the community and provided written consent (Table 1). Participants with PF joint OA represented a sample from a randomized controlled trial ([18]). Inclusion criteria for patients with PF joint OA were age ≥40 years, anterior patellar or retropatellar pain severity ≥4 on an 11-point numerical pain scale during at least 2 activities that load the PF joint (e.g., stair ambulation, squatting), pain during these activities on most days of the previous month, and K/L grade ≥1 at the PF joint ([9]). Exclusion criteria were body mass index ≥35 kg/m2; concomitant pain in other knee structures, hips, or lumbar spine; previous lower limb arthroplasty or osteotomy; recent knee injections (within 3 months of study commencement); moderate-to-severe TF joint OA (K/L grade >2 on a posteroanterior radiograph [9]); and inability to understand written or spoken English.

Table 1. Characteristics of the participants in the control group, the group with isolated PF joint OA, and the group with combined PF joint OA and TF joint OA*
 Control (n = 21)Isolated PF joint OA (n = 17)Combined PF joint OA and TF joint OA (n = 13)P
  1. Except where indicated otherwise, values are the mean (95% confidence interval). PF = patellofemoral; OA = osteoarthritis; TF = tibiofemoral; KOOS = Knee Injury and Osteoarthritis Outcome Score (100 = no symptoms; 0 = maximum symptoms) (see ref.[20]); ADL = activities of daily living; QOL = quality of life; IQR = interquartile range.
  2. aBy analysis of variance.
  3. bBy chi-square test.
  4. cKellgren/Lawrence (K/L) scale adapted for PF joint (0 = no OA; 4 = severe OA) (see ref.[9]).
  5. dK/L scale (0 = no OA; 4 = severe OA) (see ref.[12]).
Age, years56 (52, 61)56 (51, 61)60 (54, 66)0.504a
Height, meters1.68 (1.65, 1.72)1.66 (1.61, 1.70)1.66 (1.61, 1.71)0.549a
Weight, kg71 (65, 77)73 (66, 80)76 (69, 84)0.527a
Female, no. (%)13 (62)13 (76)10 (77)0.523b
Stair ascent speed, meters/second0.51 (0.47, 0.57)0.46 (0.41, 0.51)0.45 (0.39, 0.51)0.105a
Stair descent speed, meters/second0.76 (0.67, 0.84)0.67 (0.60, 0.75)0.68 (0.57, 0.80)0.308a
KOOS subscale scores, 0–100
Pain99 (97, 100)65 (57, 72)70 (60, 80)
Symptoms96 (93, 99)65 (56, 73)67 (54, 79)
ADL100 (99, 100)73 (66, 81)70 (59, 81)
Sport and recreation97 (93, 100)47 (38, 57)47 (31, 62)
Knee-related QOL99 (97, 100)42 (34, 49)49 (35, 62)
PF joint K/L grade, median (IQR)c0 (0, 0.5)2 (1, 2)4 (2, 4)<0.001b
K/L grade, median (IQR)d0 (0, 1)1 (0, 1)2 (2, 2)<0.001b

All participants with OA fulfilled the combined clinical and radiographic American College of Rheumatology criteria for the diagnosis of knee OA (knee pain, stiffness ≤30 minutes, and osteophytes seen on radiography) ([19]). No participants in the control group fulfilled these criteria. Ethical approval was obtained from The University of Melbourne Human Research Ethics Committee prior to study commencement, and all human testing procedures undertaken conformed to the principles of the Declaration of Helsinki.

Clinical assessment

Using the K/L grading system ([12]), 2 investigators (1 of whom was KMC) assessed radiographic TF joint OA severity from a semiflexed, posteroanterior, weight-bearing short film view. Radiographic PF joint OA severity was assessed from weight-bearing skyline radiographs using the K/L grading system adapted to the PF joint ([9]). Participants in the group with isolated PF joint OA had a K/L grade of ≥1 for the PF joint and a K/L grade of ≤1 for the TF joint in the same knee, whereas participants in the group with combined OA had a K/L grade of ≥1 for the PF joint and a K/L grade of 2 for the TF joint in the same knee. The interrater reliability for grading radiographic OA was acceptable (κ = 0.745–0.843) ([16]).

Knee OA symptoms

The Knee Injury and Osteoarthritis Outcome Score (KOOS) ([20, 21]) is a reliable patient-reported outcome measure that comprises 5 subscales: pain, symptoms, function in activities of daily living (ADL), function in sport and recreation, and knee-related quality of life (QOL). A normalized score between 0 (maximum symptoms) and 100 (no symptoms) was calculated for each subscale.

Gait experiments

Quantitative 3-dimensional (3-D) gait analysis during stair ambulation took place in the Biomotion Laboratory at The University of Melbourne. Details of the experimental protocol have been described previously ([16]). Reflective markers were placed at specific anatomic landmarks on the trunk, pelvis, and upper and lower limbs. Marker trajectories were captured from 9 cameras at 120 Hz (Vicon; Oxford Metrics), while ground reaction force data were recorded simultaneously at 1,080 Hz from a ground-embedded force plate and 2 portable AccuGait force plates (AMTI) mounted in a custom-built staircase. Marker and ground reaction force data were low-pass filtered at 4 Hz and 60 Hz, respectively, using a fourth-order Butterworth filter. Muscle electromyographic (EMG) data were collected at 1,080 Hz with pairs of Ag/AgCl surface electrodes (MediMax Global) mounted over the gluteus maximus, gluteus medius, vastus lateralis, and soleus muscles. An initial static trial was performed with the participant standing in a neutral pose. Participants subsequently ascended and descended a flight of 3 steps (step height 16.5 cm) wearing standardized footwear at a self-selected speed (calculated from the average horizontal velocity of the posterior pelvis marker). All participants ascended and descended stairs without aids (e.g., handrails or canes).

Lower limb muscle forces and PF joint loading

A generic, 3-D musculoskeletal model was implemented in OpenSim ([22]), an open-source musculoskeletal modeling software program, to calculate lower limb joint kinematics, joint moments, and muscle forces. The skeleton was represented as a 12-segment, 31df mechanical linkage actuated by 92 muscle-tendon units. Each muscle-tendon unit was modeled as a 3-element Hill-type muscle, which takes into account the physiologic force-length-velocity properties of muscle, in series with an elastic tendon ([23, 24]). The full body model has been reported previously ([25]). Participant-specific musculoskeletal models were generated by scaling the generic model based on body segment dimensions recorded from the static trial ([22]).

For each participant, a representative trial for stair ascent and descent was chosen arbitrarily from a group of successful trials. A successful trial was one with valid foot contacts on all force plates and no missing force plate or marker data. Joint angles were computed using an inverse kinematics analysis that minimized the sum of the squared differences between the positions of virtual markers on the model and the experimental markers ([26]). Internal joint moments were calculated using an inverse dynamics approach, and muscle forces were computed using a static optimization algorithm ([27]). The static optimization solution was constrained to the force-length-velocity properties of each muscle ([24, 27]). Lower limb muscle forces of interest were those of the gluteus maximus, hip abductors (gluteus medius and gluteus minimus), quadriceps (vastus medialis, vastus lateralis, vastus intermedius, and rectus femoris), and ankle plantar flexors (soleus and gastrocnemius). The temporal validity of the predicted muscle forces was evaluated against EMG onset and offset times.

The PF joint reaction force was determined using a separate empirically based model of the PF joint ([13]) and was calculated with the following equation:

display math

where PFJRF is the PF joint reaction force, FQ is the quadriceps force, FP is the patellar tendon force, and β is the angle between the quadriceps muscle and the patellar tendon (patellar mechanism angle). The patellar tendon force and patellar mechanism angle were calculated using data from an in vitro study ([13]). Based on these data, empirical equations were established for the patellar mechanism angle as a function of knee flexion angle and for the patellar tendon force–to–quadriceps force ratio as a function of knee flexion angle ([13]). Quadriceps force and knee flexion angle calculated from the musculoskeletal model were applied to these empirical equations to determine the patellar tendon force and patellar mechanism angle, which were then used in the above equation to calculate the PF joint reaction force.

Data management and statistical analysis

Data were analyzed using the more affected limb in the groups with OA and a randomly chosen limb in the control group. Data were time-normalized to the stance phase and averaged across all participants for each group. The computed PF joint reaction force, lower limb muscle forces, and ground reaction force were normalized to body weight, while joint moments were normalized to body weight multiplied by height. Variables of interest were measured at the time of contralateral toe-off for statistical analysis, since the peak knee extension moment coincides closely with this time point during stair ambulation ([28-30]), and because biomechanical differences between healthy individuals and individuals with pathology have been observed around the time of contralateral toe-off for stair ambulation ([30]).

Mean values and 95% confidence intervals (95% CIs) are reported. Participant characteristics were compared using one-way analysis of variance (ANOVA) or chi-square test, as appropriate. An ANOVA was used to determine group differences in dependent variables (PF joint reaction force, lower limb muscle forces, sagittal plane joint kinematics, and knee joint moments) during stair ascent and descent. Age was the only participant characteristic that correlated significantly with PF joint reaction force during stair descent (r = –0.414, P = 0.023); therefore, between-group comparisons for PF joint reaction force were conducted with and without age as a covariate (analysis of covariance [ANCOVA]). Post hoc comparisons involving the least significant difference test were used to determine between-group differences, where appropriate. P values less than 0.05 were considered significant. The contributions of the ground reaction force magnitude and moment arm (defined as the perpendicular sagittal plane distance from the ground reaction force line of action to the knee joint center) to the PF joint reaction force were evaluated using forward stepwise multiple regression analyses. The relationships between the PF joint reaction force and the subscales of the KOOS in participants with OA were evaluated using Pearson's correlation coefficient (r). Relationships were defined a priori: >0.75 = good to excellent; >0.50–0.75 = moderate to good; 0.25–0.50 = fair ([31]).

RESULTS

Participant characteristics

There were no significant differences in age, sex, height, weight, or stair ambulation speed between groups (Table 1). Participants in the groups with OA exhibited greater radiographic disease severity than those in the control group.

Stair ascent

Group differences were observed for pelvic tilt and hip flexion angles (Table 2 and Figure 1). When compared to controls, participants with isolated PF joint OA ascended stairs with greater anterior pelvic tilt (mean difference 4.1° [95% CI 1.6, 6.7]; P = 0.002) and hip flexion (mean difference 4.6° [95% CI 0.7, 8.5]; P = 0.023). Similarly, when compared to controls, those with combined OA ascended stairs with greater anterior pelvic tilt (6.9° [95% CI 4.1, 9.7]; P < 0.001) and hip flexion (7.1° [95% CI 2.8, 11.3]; P = 0.002). No significant differences were observed between the group with isolated PF joint OA and the group with combined OA for pelvic and hip kinematics. There were no group differences for trunk flexion, knee flexion, or ankle dorsiflexion angles.

Table 2. Between-group comparisons of sagittal plane joint kinematics, knee extension moments, and lower limb muscle forces*
 ControlIsolated PF joint OACombined PF joint OA and TF joint OAPa
  1. Values are the mean (95% confidence interval). See Table 1 for definitions.
  2. aBy analysis of variance.
  3. bGluteus medius and gluteus minimus combined.
  4. cVastus medialis, vastus lateralis, vastus intermedius, and rectus femoris combined.
  5. dSoleus and gastrocnemius combined.
Stair ascent    
Kinematics, degrees
Trunk flexion17.9 (15.4, 20.5)17.7 (14.9, 20.5)20.2 (17.0, 23.5)0.436
Pelvic tilt8.1 (6.4, 9.8)12.2 (10.3, 14.1)15.0 (12.8, 17.2)<0.001
Hip flexion48.0 (45.4, 50.6)52.6 (49.6, 55.5)55.0 (51.7, 58.4)0.004
Knee flexion58.5 (56.5, 60.4)55.8 (53.6, 57.9)54.9 (52.4, 57.4)0.053
Ankle dorsiflexion19.8 (15.3, 24.3)17.3 (12.3, 22.3)15.17 (9.5, 20.9)0.433
Joint moments, % body weight × height
Knee extension5.38 (4.83, 5.93)3.82 (3.21, 4.44)3.00 (2.30, 3.70)<0.001
Forces, body weight
Gluteus maximus0.87 (0.76, 0.98)0.96 (0.84, 1.09)0.95 (0.81, 1.10)0.493
Hip abductorsb1.04 (0.89, 1.19)1.00 (0.83, 1.16)0.86 (0.67, 1.05)0.333
Quadricepsc2.48 (2.25, 2.70)1.82 (1.57, 2.08)1.62 (1.33, 1.90)<0.001
Ankle plantar flexorsd1.54 (1.33, 1.76)1.43 (1.18, 1.67)1.42 (1.14, 1.67)0.696
Stair descent    
Kinematics, degrees
Trunk flexion6.5 (4.0, 9.0)5.4 (2.6, 8.1)8.6 (5.5, 11.8)0.298
Pelvic tilt0.1 (–1.9, 1.8)3.7 (1.6, 5.7)5.8 (3.5, 8.2)0.001
Hip flexion16.7 (13.9, 19.6)16.1 (13.0, 19.3)20.2 (16.6, 23.8)0.198
Knee flexion36.5 (33.7, 39.3)28.8 (25.7, 31.9)32.3 (28.7, 35.8)0.003
Ankle dorsiflexion18.2 (13.5, 23.0)13.8 (8.5, 19.1)14.4 (8.4, 20.5)0.402
Joint moments, % body weight × height
Knee extension5.43 (4.42, 6.44)2.65 (1.52, 3.77)3.39 (2.10, 4.67)0.002
Forces, body weight
Gluteus maximus0.31 (0.19, 0.42)0.37 (0.24, 0.49)0.49 (0.34, 0.64)0.160
Hip abductorsb1.81 (1.59, 2.03)1.33 (1.09, 1.58)1.18 (0.89, 1.46)0.001
Quadricepsc2.24 (1.81, 2.62)1.29 (0.87, 1.72)1.46 (0.98, 1.95)0.004
Ankle plantar flexorsd2.96 (2.59, 3.34)2.67 (2.25, 3.09)3.02 (2.54, 3.50)0.456
Figure 1.

Mean sagittal plane joint angles for stair ascent and descent. Results are shown for participants with isolated patellofemoral (PF) joint osteoarthritis (OA) (dotted lines), participants with combined PF joint OA and tibiofemoral joint OA (dashed lines), and control participants (solid lines). FS = foot strike; CTO = contralateral toe-off; CFS = contralateral foot strike; TO = toe-off; flex = flexion; ext = extension; ant = anterior; post = posterior; dorsi = dorsiflexion; plant = plantar flexion.

Group differences were observed for the knee extension moment (Table 2 and Figure 2). The group with isolated PF joint OA and the group with combined OA ascended stairs with a knee extension moment that was 29% lower (–1.56% body weight multiplied by height [95% CI –2.38, –0.74]; P < 0.001) and 44% lower (–2.38% body weight multiplied by height [95% CI –3.27, –1.49]; P < 0.001), respectively, than that of the control group. No significant differences in the knee extension moment were found between groups with OA.

Figure 2.

Mean sagittal plane knee extension moments, ground reaction forces (GRFs), moment arms of the GRF calculated about the center of the knee joint, and PF joint reaction forces (PFJRFs) calculated for stair ascent and descent. Results are shown for participants with isolated PF joint OA (dotted lines), participants with combined PF joint OA and tibiofemoral joint OA (dashed lines), and control participants (solid lines). %BW-Ht = percent body weight multiplied by height; vert = vertical; m = meters (see Figure 1 for other definitions).

Group differences were also observed for quadriceps muscle forces (Table 2 and Figure 3). Compared to the control group, the group with isolated PF joint OA displayed a 26% lower quadriceps force (–0.65 body weight [95% CI –0.99, –0.31]; P < 0.001), whereas the group with combined OA displayed a 35% lower quadriceps force (–0.86 body weight [95% CI –1.16, –0.53]; P < 0.001). There were no significant differences in quadriceps forces between groups with OA. No between-group differences were observed for gluteus maximus, hip abductor, or ankle plantar flexor muscle forces.

Figure 3.

Mean stance leg muscle forces calculated for stair ascent and descent. Horizontal bars shown below each plot indicate the periods of electromyographic (EMG) activity recorded for each muscle group. Results are shown for participants with isolated PF joint OA (dotted lines), participants with combined PF joint OA and tibiofemoral joint OA (dashed lines), and control participants (solid lines). Glut Max = gluteus maximus; BW = body weight; Hip Abduct = gluteus medius and gluteus minimus combined; Quad = vastus medialis, vastus lateralis, vastus intermedius, and rectus femoris combined; Ankle Plant = soleus and gastrocnemius combined (see Figure 1 for other definitions).

Plots displaying the PF joint reaction force profile across the stance phase of stair ascent for each group are contained in Figure 2. During stair ascent, the mean PF joint reaction force in the group with isolated PF joint OA (1.61 body weight [95% CI 1.41, 1.81]) and the group with combined OA (1.50 body weight [95% CI 1.28, 1.73]) was significantly lower than that in the control group (2.15 body weight [95% CI 1.97, 2.33]) (overall P < 0.001). Compared to the control group, the PF joint reaction force was 25% lower in the group with isolated PF joint OA (–0.54 body weight [95% CI –0.81, –0.28]; P < 0.001) and 30% lower in the group with combined OA (–0.65 body weight [95% CI –0.94, –0.37]; P < 0.001). No significant differences in the PF joint reaction force were evident between groups with OA.

Stair descent

Group differences were observed during stair descent for pelvic tilt and knee flexion angles (Table 2 and Figure 1). Anterior pelvic tilt was significantly higher during stair descent in participants with isolated PF joint OA (3.7° [95% CI 1.0, 6.5]; P = 0.009) and participants with combined OA (5.9° [95% CI 2.9, 8.9]; P < 0.001) compared to controls. Participants with isolated PF joint OA displayed smaller knee flexion compared to controls (7.7° [95% CI 3.5, 11.9]; P = 0.001). Such differences were not found for participants with combined OA (P = 0.068). No differences in pelvic or knee kinematics were observed between groups with OA. Furthermore, trunk flexion, hip flexion, and ankle dorsiflexion angles were not significantly different between groups.

Group differences were observed for the knee extension moment during stair descent (Table 2 and Figure 2). Compared to controls, the group with isolated PF joint OA and the group with combined OA descended stairs with a knee extension moment that was lower by 51% (–2.78% body weight multiplied by height [95% CI –4.30, –1.27]; P < 0.001) and 38% (–2.04% body weight multiplied by height [95% CI –3.68, –0.41]; P = 0.016), respectively, with no significant difference evident between groups with OA.

Participants with PF joint OA descended stairs with significantly lower quadriceps and hip abductor muscle forces than did controls (Table 2 and Figure 3). Specifically, the group with isolated PF joint OA demonstrated a 42% lower quadriceps force (–0.94 body weight [95% CI –1.52, –0.37]; P = 0.002) and a 27% lower hip abductor force (–0.47 body weight [95% CI –0.81, –0.14]; P = 0.006) compared to the control group. Similarly, the group with combined OA showed a 35% smaller quadriceps force (–0.77 body weight [95% CI –1.40, –0.16]; P = 0.015) and a 35% smaller hip abductor force (–0.63 body weight [95% CI –0.99, –0.28]; P = 0.001) compared to the control group, with no significant differences evident between groups with OA. No between-group differences were observed for gluteus maximus or ankle plantar flexor muscle forces.

Plots displaying the PF joint reaction force profile across the stance phase of stair descent for each group are contained in Figure 2. The PF joint reaction force was significantly different (overall P = 0.001) between the control group (1.72 body weight [95% CI 1.43, 2.01]), the group with isolated PF joint OA (0.92 body weight [95% CI 0.59, 1.24]), and the group with combined OA (1.12 body weight [95% CI 0.74, 1.49]). Compared to control participants, participants with isolated PF joint OA descended stairs with a 47% lower PF joint reaction force (–0.80 body weight [95% CI –1.24, –0.37]; P = 0.001), whereas those with combined OA descended stairs with a 35% lower PF joint reaction force (–0.60 body weight [95% CI –1.08, –0.13]; P = 0.013). No significant differences in PF joint reaction force were observed between groups with OA. The results from the ANCOVA were no different when age was included as a covariate (P = 0.001).

Predictors of PF joint reaction force

For stair ascent, linear regression modeling revealed that ground reaction force magnitude (B = 1.845, P < 0.001) and ground reaction force moment arm (B = –6.784, P = 0.004) were strongly associated with the PF joint reaction force (R2 = 0.697, P < 0.001). Similarly, for stair descent, the ground reaction force moment arm (B = –14.686, P < 0.001) and ground reaction force magnitude (B = 0.520, P = 0.024) were strongly associated with the PF joint reaction force (R2 = 0.822, P < 0.001).

Relationship between PF joint reaction force and knee OA symptoms

Among the 30 participants with OA, fair correlations were found between higher PF joint reaction force during stair descent and higher scores on the KOOS sport and recreation subscale (r = 0.483, P = 0.007) and KOOS ADL subscale (r = 0.368, P = 0.045). There were no significant associations between PF joint reaction force and scores on the KOOS pain, symptoms, or knee-related QOL subscales. For stair ascent, PF joint reaction force was not significantly correlated with any of the KOOS subscales.

DISCUSSION

This study investigated lower limb biomechanics during stair ambulation in individuals with isolated PF joint OA, concurrent PF joint OA and TF joint OA, and no OA. Compared to controls, individuals with PF joint OA exhibited lower knee extension moments, quadriceps forces, and PF joint reaction force at contralateral toe-off during stair ascent and descent, and lower hip abductor forces during stair descent. Both groups with OA used increased anterior pelvic tilt during stair ascent and descent and increased hip flexion during stair ascent, while those with isolated PF joint OA descended stairs with smaller knee flexion (i.e., a more extended knee). The findings from this study demonstrate that people with PF joint OA have characteristic biomechanical impairments during stair ambulation.

The lower PF joint reaction force observed in both groups with OA can be partially explained by the changes in the magnitude of the ground reaction force and its line of action with respect to the knee joint center, with the latter determining the moment arm of the ground reaction force relative to the knee. During stair ascent, participants in both groups with OA pushed on the ground less forcefully and used movement patterns, such as an increased anterior pelvic tilt angle and greater hip flexion (Figure 1), to shift the body's center of mass anteriorly, reducing the moment arm of the ground reaction force at the knee (Figure 4). In contrast to stair ascent, the ground reaction force magnitude contributed less to the PF joint reaction force during stair descent. Primarily, participants with PF joint OA reduced the ground reaction force moment arm during stair descent by ambulating with less knee flexion (Figures 1 and 4).

Figure 4.

Schematic figures showing the body configuration for each group at the time of contralateral toe-off during stair ascent (left) and stair descent (right). Grey, red, and blue bodies represent controls, participants with isolated PF joint OA, and participants with combined PF joint OA and tibiofemoral (TF) joint OA, respectively. Participants with OA are overlaid on the control participant to aid comparison between the groups. Green, red, and blue arrows represent the ground reaction force for controls, participants with isolated PF joint OA, and participants with combined PF joint OA and TF joint OA, respectively. Arms are not shown, for purposes of clarity. Participants with isolated PF joint OA and those with combined PF joint OA and TF joint OA ascended stairs by exerting lower ground reaction forces (left), and they descended stairs by reducing knee flexion (right). See Figure 1 for other definitions.

This study expands on present knowledge of stair ambulation in TF joint OA and provides unique insights into the pathomechanics of PF joint OA. Our finding of a lower knee extension moment confirms earlier observations in individuals with more severe TF joint OA (K/L grade 3 or 4) ([10]) and mild TF joint OA (K/L grade 2) ([11]). We observed similar findings in both groups with OA, including those with isolated, mild radiographic PF joint OA. In our study, the severity of TF joint OA or PF joint OA contributed little to the results, suggesting that differences in gait biomechanics could be ascribed mostly to the presence or absence of PF joint disease. Given that knee OA frequently involves the PF joint ([1, 2]), it is possible that the findings from previous knee OA studies are partly attributable to PF joint OA. However, the results from this study provide only indirect evidence for this premise, and future research should include a group of individuals with isolated TF joint OA to further understand compartment-specific gait changes.

Our finding of lower quadriceps force and knee extension moment in both groups with OA is consistent with the lower quadriceps volume we have reported in this cohort ([15]) and extends earlier findings of weak quadriceps muscles in those with PF joint OA ([14, 32]). Due to the cross-sectional nature of our study, we cannot determine the temporal relationship between PF joint OA and lower quadriceps force. However, since stair ambulation results in high PF joint reaction force and is a common source of PF joint OA pain ([9]), it is likely that those with PF joint OA modulate their stair ambulation mechanics to reduce pain. This possibility is plausible when one considers that lower quadriceps force was not observed during a less challenging task (i.e., walking) in a larger group of people with PF joint OA ([16]). Furthermore, we did not see differences in the knee extension moment, quadriceps force, or PF joint reaction force at contralateral foot strike during stair descent, an instant when adaptations such as modulating knee flexion would be difficult. Experimental pain studies indicate that quadriceps strength ([33]), activation ([34]), and the knee extension moment ([35]) are all reduced in the presence of knee pain. However, pain resulted in modest (<10%) reductions in the knee extension moment during level walking ([35]), indicating that knee pain itself cannot fully explain the differences (29–51%) observed in the current study.

Importantly, while reducing the knee extension moment, quadriceps force, and the PF joint reaction force may be instinctive adaptations to reduce PF joint–related pain, the persistence of such adaptations may result in maladaptive effects, such as lower quadriceps strength and size. Compromised quadriceps function may increase susceptibility to further joint damage over time, through altered sensorimotor function ([36]), reduced shock absorption ([37]), and/or joint stiffening, especially if such impairments eventually become evident in more frequent daily tasks, such as walking. Increased joint stiffness may be associated with the reduced knee flexion observed during stair descent. Such effects can subsequently result in frequent loading of a smaller area of articulation and/or the loading of surfaces that were previously unaccustomed to repetitive loading, with further deterioration of the articular cartilage. As the properties of cartilage and bone may be mechanically conditioned to the type of loading these tissues experience ([38, 39]), muscle atrophy–related changes in cartilage and bone loading could render these structures more vulnerable to injury. Furthermore, limited prospective evidence from a cohort study ([32]) and an animal model ([40]) supports the notion of a causal relationship between lower quadriceps strength and incidence or progression of PF joint OA.

Lower hip abductor force was observed in both groups with OA during stair descent. This reduction was evident throughout the stance phase of stair descent and may represent a frontal plane adaptation to offload the PF joint and/or the TF joint. Lower hip abductor force is consistent with our findings during level walking in this population ([16]) and is consistent with deficits seen in hip abductor strength in cohorts of patients with PF joint OA ([41]) and PF joint pain syndrome ([42, 43]). The finding of lower hip abductor force, combined with altered sagittal plane pelvic kinematics, suggests that the dynamics of the hip joint and pelvis should be considered in the evaluation of people with PF joint OA. Further studies may elucidate the temporal relationship between such findings and the development or progression of PF joint OA.

Individuals with PF joint OA (isolated and combined with TF joint OA) exhibited considerable OA symptoms, as evidenced by the mean scores on the KOOS, mostly in the subscales of sport and recreation and function in ADL. Notably, we observed a significant correlation between greater PF joint reaction force at contralateral toe-off during stair descent and less difficulty in ADL (including stair descent) and in sports and recreation. Although the strength of this relationship was fair, it appears as though those who had greater difficulty with functional activities exhibited lower PF joint reaction force during stair descent, possibly due to greater adaptations to offload the PF joint during this task.

Musculoskeletal modeling provides a noninvasive method for determining muscle and joint reaction forces in vivo. The lower limb muscle and PF joint reaction forces calculated in this study are extensions of the net joint moment calculations, since an infinite combination of muscle forces may contribute to a measured net joint moment. However, as with all modeling approaches, there are several limitations and assumptions that must be acknowledged.

First, the muscle-tendon properties in all musculoskeletal models were scaled according to each participant's anthropometry ([24]) rather than measured directly, which may affect the predictions of muscle forces. Second, the static optimization approach used to calculate muscle forces may have limited capacity to predict cocontraction of antagonistic muscles. Hence, the quadriceps muscle forces and PF joint reaction forces may have been underestimated if there was substantial cocontraction between the quadriceps and its antagonists (hamstrings and gastrocnemius) during these activities. Nevertheless, the muscle force predictions in this study were in good temporal concordance with the patterns of EMG activity (Figure 3), and the same scaling approach has been applied successfully in a number of other studies ([16, 44-46]). Third, the assumption of a planar representation of the PF joint in the PF joint model is a simplification of the true 3-D nature of PF joint biomechanics. However, the largest variation in PF kinematics occurs in the sagittal plane ([47]), implying that the PF joint reaction force is primarily a sagittal plane quantity. Furthermore, the PF joint reaction force model incorporated parameters (e.g., the patellar mechanism angle and patellar tendon force–to–quadriceps force ratio) that were based on cadaveric measurements ([13]).

The extent to which these cadaveric measurements accurately represent the biomechanical behavior of the PF joint in vivo is unknown. Nevertheless, our estimations of PF joint reaction force fall within the range of magnitudes calculated in stair ambulation studies ([8, 28, 29, 48-50]). Furthermore, variations in the patellar mechanism angle and patellar tendon force–to–quadriceps force ratio of 5% and 10%, respectively, resulted in a change in the PF joint reaction force of no greater than 7.5% for stair ascent and 15% for stair descent at contralateral toe-off. Since such changes are less than the differences found between the control group and the groups with OA, it is unlikely that the main findings were significantly influenced by the use of a model based on cadaveric measurements. It should be noted that a small number of the control participants (n = 5) exhibited mild radiographic OA, which may have affected their gait. However, it is challenging to recruit an older cohort of people with no radiographic evidence of OA, and none of the control participants reported pain in the knee or other lower limb joints, which was the primary eligibility criterion. All other participant characteristics were similar between groups. Finally, the sample size was modest, and future studies should be undertaken to replicate these findings in larger cohorts.

People with isolated PF joint OA or concurrent PF joint OA and TF joint OA ascend and descend stairs with a lower PF joint reaction force when compared to a group of age-matched healthy controls. These reductions were associated with gait modifications, which were activity dependent. Quadriceps and hip abductor forces were also lower in individuals with OA during stair ambulation, suggesting that hip and knee muscle dysfunction may be characteristics of the population with PF joint OA. While these modifications may be intuitive adaptations to lower the PF joint reaction force in those with PF joint pain and OA, such changes may not be benign and may potentially have long-term deleterious consequences. Further research is needed to understand the temporal relationship between gait biomechanical adaptations and the PF joint OA disease process.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Pandy 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.

Study conception and design. Fok, Schache, Crossley, Pandy.

Acquisition of data. Schache.

Analysis and interpretation of data. Fok, Schache, Crossley, Lin, Pandy.

Acknowledgments

The authors thank Jonathan Lentzos and Hannah Ozturk for their substantial contributions to acquisition of data by assisting in patient recruitment, imaging procurement, and collection of gait data. The authors also thank Rana Hinman for her substantial contributions to study conception and design and for grading all radiographs.

Ancillary