Changes in lower extremity joint moments one-year following osseointegration in individuals with Transfemoral lower-limb amputation: A case series

Abstract

Background:

Dissatisfaction with socket prostheses has led to the development of bone-anchored prostheses through osseointegration for people with transfemoral amputation, eliminating the need for a prosthetic socket. Gait deviations of transfemoral prosthesis users may be linked to increased risk of osteoarthritis, and it remains unknown if gait biomechanics change following osseointegration. The purpose of this case series was to evaluate the longitudinal changes in joint kinetics one year post-osseointegration in patients with transfemoral amputation during walking.

Methods:

Knee, hip, and trunk internal moments were evaluated in the prosthetic and intact limbs during walking at a self-selected speed in four participants pre- and one-year post-osseointegration. Longitudinal changes were quantified using the percent change (%Δ) in peak joint moments between the two time points and Cohen’s d (d) effect size was used to determine the magnitude of effect on joint moments during walking one year following osseointegration.

Findings:

Participants demonstrated increased peak knee extension moment (224 ± 308%Δ, d = ₋ 1.31) in the prosthetic limb, while demonstrating reduced peak knee extension moment (− 43 ± 34%Δ, d = 1.82) in the intact limb post-osseointegration. Participants demonstrated bilateral reduction of peak hip extension moment (prosthetic: − 22 ± 37%Δ, d = 0.86; intact: − 29 ± 10%Δ, d = 1.27) and bilateral increase of peak hip abduction moment (prosthetic: 45 ± 40%Δ, d = 1.20; intact: 23 ± 44%Δ, d = 0.74) post-osseointegration. Participants demonstrated reduced peak trunk moments on both the prosthetic (extension: − 31 ± 16%Δ, d = 1.51; lateral flexion: − 21 ± 20%Δ, d = 0.63) and intact side (extension: − 7 ± 22%Δ, d = 0.38; lateral flexion: − 22 ± 18%Δ d = 1.12) post-osseointegration.

Interpretation:

This case series suggests improved gait symmetry in individuals with transfemoral amputation one year following osseointegration, justifying future investigation.

1. Introduction

Individuals with transfemoral amputation (TFA) are routinely prescribed a socket prosthesis, with the prosthetic limb attaching to the body through a custom designed socket (Thomson et al., 2019). However, most individuals who use a socket prosthesis demonstrate gait deviations (i.e. changes in joint moments) during ambulation, and approximately 25–30% of socket prosthesis users report dissatisfaction with their prosthesis fit, including pain and wounds from the socket (Hagberg and Brånemark, 2001; Pezzin et al., 2004). Dissatisfaction with socket prostheses and poor physical function outcomes has led to the development of transfemoral osseointegrated implants (OI) for individuals with TFA, a surgical technique that creates a connection between the bone and the prosthetic limb, eliminating the need for a socket interface (Leijendekkers et al., 2019; Li and Brånemark, 2017). This direct fixation between the femur and prosthesis improves the load transmission between the ground and residual limb, likely improving femoral control of the prosthesis. It remains unknown if OI alters joint loading in individuals with TFA.

Aberrant joint loading during the weight acceptance phase of gait persists long-term in individuals with TFA (Harandi et al., 2020; Nealy and Gard, 2008; Segal et al., 2006). Aberrant and asymmetrical joint loading is hypothesized to be an underlying factor related to high levels of secondary complications (i.e. osteoarthritis, low back pain, increased energy demand) in individuals with TFA (Burke et al., 1978; Kulkarni et al., 1998; Mahon et al., 2019; Nolan et al., 2003; Struyf et al., 2009). Compared to healthy individuals, people with TFA demonstrate reduced peak internal knee extension moment (Segal et al., 2006) and greater peak internal hip extension moment in the prosthetic limb during weight acceptance (Nealy and Gard, 2008). People with TFA also demonstrate reduced peak internal hip abduction moment (Heitzmann et al., 2020) and greater peak internal lateral trunk flexion moment during weight acceptance on the prosthetic side (Heitzmann et al., 2020). It is hypothesized that the direct prosthesis to femur connection created with OI, along with the improved alignment between anatomical hip and prosthetic knee, may reduce aberrant loading patterns associated with socket prostheses users. Although prior work has evaluated lower extremity joint moments in patients with transfemoral OI (Dumas et al., 2017), to our knowledge, no studies have evaluated how hip and knee moments during gait change following OI. Accounting for patient-specific compensations that have been habituated within their socket-suspended prostheses are important to fully investigate the effect of OI.

Our primary aim in this case series was to evaluate peak internal joint moments during weight acceptance phase of gait one-year post-OI compared to pre-OI. Our secondary aim was to evaluate interlimb symmetry for the previously mentioned kinetics during weight acceptance phase of gait one-year post-OI compared to pre-OI. We hypothesized decreases in knee extension moment on the intact limb, increases in knee extension moment in the prosthetic limb, bilateral reductions in hip extension moment, trunk lateral flexion moment and trunk extension moment, and bilateral increases in hip abduction moment post-OI compared to pre-OI. We also hypothesized all kinetic variables would demonstrate improved symmetry.

2. Methods

2.1. Participants

Four participants undergoing unilateral transfemoral OI at the University of Colorado were recruited for this prospective cohort case series at the time of their pre-surgery appointment in August 2019. Participants were included if they were deemed eligible for OI implant surgery by the Limb Restoration Clinical Team at the University of Colorado Health. Participants were excluded if they were unable to complete walking biomechanics testing without the use of an assistive device. Surgery selection criteria are listed in Supplementary Table 1. The Colorado Multiple Institutional Review Board approved all methods, and all participants provided written informed consent prior to participation.

2.2. Surgery and rehabilitation

The titanium press-fit implant (OTN Implant BV, The Netherlands) was implanted by the same orthopaedic surgeon (JSW) in a two-step series of surgeries performed six to eight weeks apart (48 ± 1 days), as described elsewhere (Aschoff et al., 2009; Leijendekkers et al., 2019; Van De Meent et al., 2013) (Fig. 1). All participants underwent daily rehabilitation sessions with one of two physical therapists for three weeks, starting two days following the second surgery. Alignment of the prosthesis was performed by the same prosthetist for all participants.

Fig. 1.

Anteroposterior view of osseointegrated prosthesis for an individual with transfemoral lower-limb amputation (note: images are taken from different individuals).

2.3. Walking biomechanics

Gait biomechanics were collected 2 ± 2 days prior to the first OI surgery and one-year (367 ± 4 days) following the second OI surgery. Prosthesis componentry are in Table 1, and minimal to no changes were made in prosthetic fitting or componentry following OI surgery. Twenty-six retroreflective markers were placed on the upper extremities, lower extremities, and trunk (Davis et al., 1991). Marker positions were collected using eight infrared cameras (Vicon, Centennial, CO, USA; 120 Hz), synchronized with six force plates (Bertec, Columbus, OH, USA; 2160 Hz) embedded in a 10-m walkway while participants walked at a self-selected speed (Christensen et al., 2020; Gaffney et al., 2018). Each participant practiced walking (approximately 5 min) and then completed five valid trials where each limb cleanly contacted the force plates. Five steps were averaged for data analysis for each limb. No participant used an assistive device during trials.

Table 1.

Prosthetic components pre and post osseointegration.

Component	Pre osseointegration	Post osseointegration
Socket	Ischial Containment, n = 3 Unknown, n = 1	none
Suspension	Suction, n = 2 Pin, n = 1 Unknown, n = 1	none
Foot	Dynamic Carbon Fiber, n = 4	Dynamic Carbon Fiber, n = 4
Positional Rotator	No, n = 1 Yes, n = 3	Yes, n = 4
Knee	Microprocessor Knee, n = 4	Microprocessor Knee, n = 4
Shock/Torque	No, n = 3 Yes, n = 1	No, n = 1 Yes, n = 3

Kinetic data were low-pass filtered (4th-order Butterworth, 20 Hz cutoff). A subject-specific model was created in Visual 3D (C-Motion Inc., Germantown, MD, USA) from a static calibration trial. Kinetics were analyzed using standard inverse dynamic calculations (Christensen et al., 2020; Gaffney et al., 2018; Murray et al., 2017) and normalized to body mass and height (kg*m), with prosthesis mass included in body mass. Peak kinetics were identified during loading phase of gait (first 50% of stance phase) via custom scripts (MATLAB R2020a, MathWorks, Natick, MA, USA). Peak internal moments (i.e., the net sum of all internal moments delivered by all internal structures around the joint) included knee extension, hip extension, hip abduction, trunk extension, and trunk lateral flexion. Symmetry between the non-amputated limb (NA) and amputated limb (A) for each walking trial (t) for each variable (X) was calculated using the normalized symmetry index (NSI) formula (Queen et al., 2020). A positive NSI indicates the intact limb demonstrated the larger moment compared to the prosthetic limb, and an NSI of zero indicates absolute symmetry.

$$NSI=\frac{X_{NA,t}-X_{A,t}}{ma{x}_{t=1:n}\left(max\left(0,X_{NA,t},X_{A,t}\right)\right)-mi{n}_{t=1:n}\left(min\left(0,X_{NA,t},X_{A,t}\right)\right)}\ast 100$$ (1)

2.4. Statistical analysis

Means, standard deviations, and percent change scores (%Δ) (Eq. (2)) were calculated from pre-OI to post-OI. Cohen’s d effect sizes (d) (Eq. (3)) were reported for change between pre-OI and post-OI (small = 0.20, medium = 0.50, large = 0.80) (Lakens, 2013). A priori sample size calculations and inferential statistical testing were not performed for this case series.

$$PercentChange=\frac{(Post-Pre)}{Pre}\ast 100$$ (2)

$$Cohen\text{'}sd=\frac{Mea{n}_1-Mea{n}_2}{PooledStandardDeviation}$$ (3)

3. Results

Each participant decreased their self-selected walking speed (d = 0.81) post-OI (Table 2). Individual results and average results are presented for knee moments (Table 3), hip moments (Table 4), and trunk moments (Table 5) in tabular form.

Table 2.

Characteristics of participants.

Participant	Age (years)	Sex	Height (m)	Pre-OI Mass (kg)	Post-OI Mass (kg)	Percent change (%Δ) in mass	Pre-OI speed (m/s)	Post-OI speed (m/s)	Percent change (%Δ) in mass	Years since amputation	Residual Limb Length prior to Osseointegration (cm)	Reason for Osseointegration
A	55	F	1.7	68.2	68.0	− 0.29	1.08	0.96	− 11.1	3	28.8	Pain related to socket in the groin and bottom of amputated limb
B	52	M	1.8	99.8	96.2	− 3.61	1.11	0.92	− 17.2	9	25.3	Pain and tenderness related to socket at bottom of amputated limb and poor socket fit due to sweating
C	43	M	1.8	97.5	95.3	− 2.25	1.39	1.23	− 11.5	16	35.1	Poor fitting socket, wound development on amputated limb and groin
D	36	F	1.6	73.5	73.5	0	0.99	0.95	− 4.0	6	23.6	Poor fitting socket, pain and wound development on amputated limb, back and ischium pain
Mean (SD)	(9)		1.8 (0.1)	88.5 (16.2)	86.5 (14.6)	− 2.05 (1.70)	1.19 (0.17)	1.03 (0.14)	− 13.3 (5.4)	(6)	28.2(5.1)

Table 3.

Peak internal knee extension moment (Nm/kg*m).

	Amputated Limb			Intact Limb
Participant	Pre	Post	Percent change (%Δ)	Pre	Post	Percent change (%Δ)
A	− 0.018	0.105	683.3	0.195	0.193	−1.0
B	0.049	0.065	32.6	0.385	0.075	− 80.5
C	0.112	0.136	21.4	0.468	0.196	− 58.1
D	0.073	0.193	164.4	0.377	0.254	− 32.6
Group Mean (SD)	0.054 (0.054)	0.125 (0.054)	225.4 (312.1)	0.356 (0.115)	0.179 (0.075)	− 43.1 (34.2)

Table 4.

Peak internal hip moments (Nm/kg*m).

	Amputated peak internal hip extension moment			Intact peak internal hip extension moment			Amputated peak internal hip abduction moment			Intact peak hip abduction moment
Participant	Pre	Post	Percent change (%Δ)	Pre	Post	Percent change (%Δ)	Pre	Post	Percent change (%Δ)	Pre	Post	Percent change (%Δ)
A	0.307	0.232	− 24.4	0.410	0.317	− 22.7	− 0.135	− 0.195	44.4	− 0.305	− 0.300	− 1.6
B	0.196	0.253	29.1	0.647	0.381	− 41.1	− 0.247	− 0.270	9.3	− 0.335	− 0.512	52.8
C	0.170	0.108	− 36.5	0.810	0.538	− 33.6	− 0.251	− 0.316	25.9	− 0.369	− 0.272	− 26.3
D	0.283	0.119	− 57.9	0.727	0.589	− 19.0	− 0.133	− 0.267	100.7	− 0.303	− 0.503	66.0
Group mean (SD)	0.224 (0.066)	0.198 (0.075)	− 10.6 (37.0)	0.622 (0.172)	0.412 (0.128)	− 32.4 (10.1)	− 0.211 (0.066)	− 0.260 (0.050)	26.5 (39.8)	− 0.336 (0.031)	− 0.361 (0.128)	8.3 (43.9)

Table 5.

Peak internal trunk moments (Nm/kg*m).

	Amputated peak internal trunk extension moment			Intact peak internal trunk extension moment			Amputated peak internal trunk lateral flexion moment			Intact peak internal trunk lateral flexion moment
Participant	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)
A	0.505	0.396	− 21.6	0.640	0.554	− 13.4	0.400	0.248	− 38	0.414	0.269	− 35.0
B	0.742	0.638	− 14.0	0.400	0.468	17	0.148	0.157	6.1	0.353	0.214	− 39.4
C	0.777	0.454	− 41.6	1.048	0.681	− 35.0	0.124	0.102	− 17.7	0.287	0.281	− 2.1
D	1.032	0.547	− 47.0	0.932	0.937	0.5	0.253	0.170	− 32.8	0.459	0.402	− 12.4
Group mean (SD)	0.675 (0.216)	0.496 (0.106)	− 25.7 (15.7)	0.696 (0.292)	0.567 (0.204)	− 10.5 (22.0)	0.224 (0.126)	0.169 (0.060)	− 16.5 (19.8)	0.351 (0.074)	0.255 (0.079)	− 25.5 (17.9)

3.1. Kinetics

Following OI, participants demonstrated increased peak knee extension moment in the prosthetic limb (d = ₋ 1.31) and reduced peak knee extension moment (d = 1.82) in the intact limb (Table 3). Participants demonstrated bilateral reduction of peak hip extension moment (prosthetic: d = 0.86; intact: d = 1.27) and bilateral increase of peak hip abduction moment (prosthetic: d = 1.20; intact: d = 0.74; Table 4) post-OI. Participants demonstrated bilateral reduction of peak trunk extension moment (prosthetic: d = 1.51; intact: d = 0.38) and peak trunk lateral flexion moment (prosthetic: d = 0.63; intact: d = 1.12, Table 5) post-OI.

3.2. Interlimb symmetry

Peak knee extension moment (d = 5.95), peak hip extension moment (d = 0.24), peak hip abduction moment (d = − 0.34), and peak trunk lateral flexion moment (d = 0.41) became more symmetrical post-OI. Symmetry became worse (i.e., greater between-limb difference) in peak trunk extension moment (d = − 0.69) post-OI. NSI values are provided in Table 6 and shown in Fig. 2.

Table 6.

Lower-extremity joint moment normal symmetry index values.

	Peak internal hip extension moment			Peak internal hip abduction moment			Peak internal knee extension moment			Peak internal trunk extension moment			Peak internal trunk lateral flexion moment
Participant	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)	Pre	Post	Percent Change (%Δ)
A	22.68	21.78	−4.0	−46.98	−27.78	−40.9	74.76	29.34	−60.7	13.77	16.21	17.7	5.62	0.76	−86.5
B	64.34	33.46	−48.0	−21.35	−42.24	97.8	76.84	7.58	−90.1	−31.95	−23.89	−25.2	40.16	10.73	−73.3
C	71.46	64.03	−10.4	−22.23	11.61	−152.2	60.78	21.58	−64.5	16.08	29.45	83.1	43.74	36.09	−17.5
D	55.92	73.4	31.3	−39.22	−43.48	10.9	65.42	20.67	−68.4	−12.72	30.73	−341.6	25.44	37.71	48.2
Group mean (SD)	53.60 (21.60)	48.17 (24.50)	−7.77 (32.47)	−32.44 (12.71)	−25.47 (25.73)	−21.10 (104.49)	69.45 (7.62)	19.79 (9.02)	−70.94 (13.17)	−3.70 (22.92)	13.12 (25.53)	−66.49 (188.74)	28.74 (17.33)	21.32 (18.45)	−32.25 (61.43)

Fig. 2.

Changes in lower-extremity joint moment symmetry following osseointegration (OI). A smaller absolute normal symmetry index value represents more symmetrical joint moments. Post-OI, participants demonstrated improved interlimb symmetry of hip extension moment, hip abduction moment, knee extension moment, and trunk lateral flexion moment compared to pre-OI.

* Indicates a large (d ≥ 0.80) effect size.

4. Discussion

Our results agreed with our primary hypothesis that kinetics would change (decreases in knee extension moment on the intact limb, increases in knee extension moment in the prosthetic limb, bilateral reductions in hip extension moment and trunk lateral flexion moment, and bilateral increases in hip abduction moment) one-year post-OI compared to pre-OI. Large effect sizes were measured for increases in peak knee extension moment in the prosthetic limb post-OI compared to pre-OI. We also hypothesized all kinetic variables would demonstrate improved symmetry. Our results agreed with our secondary hypothesis that symmetry of knee extension, hip extension and abduction, and trunk lateral flexion moment would improve. These preliminary results have generated a hypothesis that OI may improve gait symmetry and reduce mechanical overloading of joints at high risk of osteoarthritis development one-year post-OI. Contrary to our initial hypothesis, trunk flexion moment symmetry did not improve post-OI.

To our knowledge, this case series is the first to evaluate changes in bilateral internal joint moments, particularly knee and hip kinetics, one-year post-OI in individuals with TFA. Individuals with TFA fitted with a socket prosthesis in a previous study demonstrated similar peak knee extension moment values of 0.067 Nm/kg⁷ to our participants pre-OI. In healthy gait, the knee extension moment from eccentric quadriceps muscle activity controls knee flexion motion during weight acceptance of gait (Shelburne et al., 2006). However, participants with TFA have poor control of their prosthetic knee joint, leading to a “stiffened” knee gait, where the prosthetic knee undergoes limited flexion excursion (Segal et al., 2006). OI results in improved load transmission from the prosthesis to the femur, which we hypothesize increases prosthesis control and weight transfer over the prosthetic limb from the hip musculature, leading potentially to improved knee flexion control during stance shown in the current study. OI may also result in improved perception of the prosthetic limb. For example, participants have described feeling that the OI prosthesis became an integrated part of their lower-limb and they were more able to identify sensory information transmitted through the prosthesis (Lundberg et al., 2011). We hypothesize that improved hip muscle control and perception of the prosthetic limb post-OI may have allowed for increased flexion motion of the prosthetic knee during stance.

Participants demonstrated bilateral reduction of peak hip extension moment and peak knee extension moment on the intact limb post-OI, potentially due to increased weight acceptance from the prosthetic limb. Additionally, participants demonstrated increased hip abduction moment and decreased trunk extension moment and trunk lateral flexion moment post-OI. Internal hip extension moments in our pre-OI participants are similar to values found in individuals with TFA (0.46 Nm/kg) when accounting for difference in units (Heitzmann et al., 2020). Individuals with TFA often avoid hip adduction positions and ambulate with a reduced hip abduction moment due, in part, to weak hip abductors (along with other factors such as poor socket alignment) (Heitzmann et al., 2020). Avoidance of hip adduction during gait leads to greater sagittal and frontal trunk motion (Tazawa, 1997). We hypothesize the direct connection between the femur and the prosthesis may result in a more physiological lower limb segment (i.e., no interposing soft tissue surrounding a traditional prosthesis socket), resulting in more efficient transmission of energy between the ground and skeleton. More efficient load transfer may reduce the need for movement compensation and contribute to the increased hip abduction moment demonstrated following OI, allowing for bilateral reductions in peak trunk extension moment and trunk lateral flexion moment. Our results suggest bilateral reductions in mechanical loading of both hips, the intact knee, and trunk, all joints at risk of osteoarthritis for people with lower limb amputation (Burke et al., 1978; Kulkarni et al., 1998; Nolan et al., 2003; Struyf et al., 2009), in individuals with TFA following OI.

There were certain limitations from this case series that can be improved upon in future studies. Pain during walking was not systematically evaluated pre- and post-OI. Walking speed during testing was not controlled. These preliminary results will inform the design of an adequately powered study evaluating joint loading following OI while controlling for changes in walking speed. Our hypotheses of improved energy transfer and control of the osseointegrated limb during gait should be evaluated in appropriately powered efficacy studies. Future research may also evaluate joint loading using wearable sensors during daily living activities or instrumented implants to measure in-vivo loading (Dumas et al., 2009; Lee et al., 2008) as well as detailed information regarding prosthetic fitting. The small sample size limits our ability to generalize findings.

5. Conclusions

Participants with TFA one-year post-OI demonstrated more symmetrical peak knee extension, hip extension, hip abduction, and trunk lateral flexion moments. These changes in gait post-OI provide preliminary evidence that will be used in subsequent studies to improve clinical outcomes following OI surgery.