Quadriceps Strength Does Not Associate With Gait Adaptation Ability in Individuals With Anterior Cruciate Ligament Reconstruction Open Access

• Regardless of quadriceps strength, after anterior cruciate ligament reconstruction, individuals can meet real-time gait biofeedback cues to mitigate aberrant gait patterns associated with posttraumatic osteoarthritis.

• Individuals may benefit from early interventions of gait retraining after anterior cruciate ligament reconstruction.

• It may be unnecessary to wait for quadriceps strength to return before implementing a gait retraining regimen after anterior cruciate ligament reconstruction.

Quadriceps weakness is a common clinical impairment that presents acutely after anterior cruciate ligament (ACL) injury and reconstruction (ACLR).¹  As quadriceps strength is required for joint stability, quadriceps weakness is understood to contribute to poor knee function after ACLR.²  Specifically, quadriceps weakness is linked to gait pattern changes in individuals with and without ACLR. Authors of a previous crossover study on able-bodied individuals demonstrated that experimental knee effusion models that are known to acutely impair quadriceps force generation capacity result in immediate aberrant gait alterations (ie, lesser peak vertical ground reaction forces [vGRFs] and internal knee extension moments).³  These same aberrant gait features, as well as less peak knee flexion and excursion, have been reported in cross-sectional studies of ACLR patients with quadriceps weakness.^1,4  As such, it is hypothesized that quadriceps weakness caused by ACL injury and ACLR is a catalyst for the development of aberrant knee biomechanics that, if not corrected, will contribute to the development of knee osteoarthritis.⁵  Lesser vGRF, peak knee flexion and excursion, and knee extension moment are associated with early biochemical and compositional changes linked to cartilage breakdown in the first 12 months after ACLR,^6,7  as well as the development of symptomatic and radiographic knee osteoarthritis.^8,9  As such, traditional ACL injury rehabilitation regimens primarily focus on restoration of quadriceps strength, with the hypothesis that improved quadriceps strength will transfer to normalizing gait biomechanics and mitigate development of knee osteoarthritis.¹⁰ 

Despite undergoing traditional therapeutic exercise after ACLR, 50% to 90% of all ACL injuries progress to knee osteoarthritis, suggesting that current intervention strategies are not sufficient to prevent osteoarthritis.¹¹  Moreover, evidence has shown traditional rehabilitation regimens do not restore normal gait biomechanics.¹²  Specifically, despite having completed traditional rehabilitation regimens, ACLR limb gait biomechanics remain altered for years after ACLR compared with the contralateral uninjured limb and matched controls.¹²  Therefore, interventions that directly target aberrant gait biomechanics after ACLR may be needed to normalize gait and mitigate the risk of osteoarthritis development.

In previous literature, on average, real-time gait biofeedback (RTGBF) has been shown to be an effective modality for acutely modifying gait, and as such, is a promising modality for normalizing gait biomechanics in patients with ACLR.^13,14  Specifically, RTGBF that cues an increase in the first peak of the vGRF (pvGRF) waveform increases knee flexion excursion (KFE) and peak knee extension moment (pKEM) in individuals with ACLR.^13,14  Additionally, cueing an increase in pvGRF during a single 20-minute session resulted in acute biological benefits including decreased serum cartilage oligomeric matrix protein, a biomarker of cartilage breakdown, in individuals with ACLR.^14,15  As such, using RTGBF to normalize gait biomechanics after ACLR has important implications for joint tissue changes relevant to osteoarthritis development. However, using RTGBF to increase pvGRF in ACLR individuals exhibits greater root mean square error (RMSE) of pvGRF (37.98 ± 22.42) than other RTGBF targets such as cueing symmetrical pvGRF (26.68 ± 10.49).¹⁶  Further, when cued to increase pvGRF by 5% body weight (BW), the ability to meet the cue remains variable among those with ACLR, as the standard deviations are half as large (2.49% BW) as the desired change in pvGRF.¹⁶  Therefore, it is critical to understand the factors that influence the ability to meet a pvGRF target. It is unknown if quadriceps strength influences the ability of individuals with ACLR to meet a pvGRF target, thus influencing the efficacy of RTGBF. If quadriceps strength does influence the efficacy of RTGBF, it may be important to regain full strength of the quadriceps before gait retraining. Alternatively, should quadriceps strength not influence the efficacy of RGTBF, gait retraining with RTGBF may be beneficial independent of an individual’s quadriceps strength.

Therefore, the primary purpose of the current study was to determine the association between quadriceps strength and the ability to modify gait biomechanics with RTGBF in individuals with ACLR. Specifically, we evaluated the associations between isometric knee extension strength (Nm/kg) and changes in pvGRF (ΔpvGRF; BW) and target error (pvGRF RMSE) when using RTGBF to cue an increase of pvGRF by 5%, 10%, and 15% BW. The secondary aim was to evaluate the association between quadriceps strength (Nm/kg) and the acute changes in biomechanical variables, which characterize the aberrant gait patterns common in ACLR individuals (ie, peak knee flexion angle [pKFA], °); KFE (°); knee extension excursion (KEE, °), and pKEM (BW × height) in response to the same RTGBF protocol. We hypothesized that lesser quadriceps strength would be associated with smaller changes in biomechanical variables and larger target errors for each of the 3 RTGBF walking conditions.

Participants completed 1 laboratory session during which quadriceps strength and walking gait biomechanics were collected. Cohort demographics (eg, height, weight, age) were collected upon the participants’ arrival. Next, we measured strength of the ACLR and contralateral limbs and determined self-selected walking speed over ground, which was subsequently used during 4 treadmill walking testing trials. The university’s institutional review board approved all methods, and all participants provided written informed consent before participation.

Individuals were included if they had undergone unilateral ACLR 6 months to 5 years before data collection, had physician approval to return to unrestricted participation in physical activity, were between the ages of 18 to 35 years, and had a body mass index ≤35 kg/m². Individuals were excluded if they were currently pregnant or planning to become pregnant while enrolled in the study or if they had a history of (1) musculoskeletal injury to either leg (eg, ankle sprain, muscle strain) within 6 months before participation in the study, (2) lower extremity surgery other than the primary ACLR, (3) a previous diagnosis of or current self-reported symptoms related to knee osteoarthritis (pain, swelling, or stiffness), or (4) cardiovascular restrictions that limited the ability to participate in physical activity. Participants were recruited from lists of previous participants, word of mouth, flyers, and classroom announcements around the university population. Additionally, participants were identified and recruited from 3 local surgeons affiliated with the study. Demographic data are provided in Table 1. No previous authors have evaluated the association between quadriceps strength and the ability to modify gait patterns after ACLR; however, moderate associations were reported between quadriceps strength and similar gait biomechanical variables (ie, peak knee flexion moment and angle) after ACLR.¹  We determined that 35 individuals with ACLR would be sufficient to determine statistical significance using a 2-tailed linear regression for a moderate association (r = 0.6, R² = 0.36) detected between quadriceps strength and the ability to modify gait patterns after ACLR with an α level set at .05 and 80% power (G*Power, v3.1.9.2).

Self-selected walking speed was determined by instructing participants to walk a 6 m distance between 2 infrared timing gates (TF100, Trac Tronix) at the pace they would “comfortably walk over a sidewalk.”¹³  After participants felt comfortable walking in the laboratory, the speed of 5 walking trials was averaged and used to determine the speed of the treadmill during the subsequent treadmill walking protocol. After a 5-minute acclimatization period on a force instrumented treadmill, habitual walking biomechanics were collected as individuals walked without feedback for 250 steps to serve as a control trial.¹⁴  Finally, 3 separate 250-step RTGBF trials (ie, 5%, 10%, and 15% BW increase) were completed in a randomized order. The order of RTGBF trials was randomized in blocks of 6 using a computerized random number generator.

A custom MATLAB (version R2016b, Math-works Inc) script was used to calculate average baseline pvGRF of the ACLR and contralateral limbs from each participant by extracting and averaging pvGRF from the 250-step control trial. Using a second custom MATLAB program, targets for the RTGBF trials were calculated as 5%, 10%, and 15% BW above the respective baseline pvGRF and used throughout the respective trial. The 5%, 10%, and 15% increases in pvGRF targets were chosen to represent the range of changes that might be used in a gait retraining regimen, which seeks to match vGRF to that of asymptomatic ACLR individuals and uninjured controls.⁹  A single red horizontal target line representing the appropriate target pvGRF for each limb was shown on a 55-inch screen directly in front of the treadmill.¹³  To account for baseline differences in pvGRF between limbs, and therefore target differences, the MATLAB code concurrently calculated and displayed scaled right and left bar graphs representing the continuously changing average of the previous 2 pvGRFs for each limb. Therefore, the RTGBF walking trials cued the same percent change in pvGRF magnitude for each limb rather than cueing to prescribe interlimb symmetry.

Muscle strength of both the ACLR and contralateral limbs, expressed as torque normalized to body mass (Nm/kg), was measured with a maximal voluntary isometric contraction (MVIC) of the quadriceps using a reliable visual feedback methodology (ICC [2,K] = 0.994; ICC [2,1] = 0.838) and a HUMAC isokinetic dynamometer.^19,20  As participants sat with the hips and knees flexed to 85° and 90°, respectively, a padded Velcro strap was secured around the ankle at the height of the gastrocnemius-soleus muscle-belly junction to secure the lower limb to the arm of the dynamometer. Participants were instructed to extend the knee maximally “as hard and as fast as possible” against the dynamometer. Participants performed a graded warm-up (25%, 50%, and 75% of perceived MVIC).⁴  After the warm-up, practice MVICs were performed until the provided methodology was successfully implemented and the torque outputs between 2 consecutive practice MVICs did not change more than 5% between contractions. Authors of a prior study found an average of 4 trials were needed to establish the practice MVIC. The average maximal torque from the last 2 practice MVICs was calculated and subsequently used to create a visual MVIC red target line that represented torque output 10% above what the participants achieved in practice. Participants were able to visualize the torque magnitude during the trials and were instructed to try to produce enough force to get above the red target line. Participants were unaware that reaching the target line may have been unattainable, and subsequently, the target encouraged maximal effort from participants.¹⁹  Participants were also given verbal encouragement to increase motivation to achieve maximal effort. Two valid MVICs were performed. To be considered valid, the MVIC was required to be, at minimum, equal to the average torque output previously calculated from the practice trials. To minimize the risk of fatigue for the final recorded MVIC, participants were given as much time as needed to recover from previous trials, with a minimum resting period of 60 seconds between each contraction and a minimum resting period of 5 minutes between completion of the practice MVIC trials and testing. These methods for testing the MVIC demonstrate strong reliability between sessions and raters.¹⁹  Each individual’s ACLR limb strength was assessed first, followed by the contralateral limb. The highest peak torque from the 2 MVIC trials for each limb was used as a measurement of strength in subsequent data analysis.

All statistical analyses were performed using the SPSS (version 21; IBM Corp.). For our primary analyses, we conducted separate univariate linear regression analyses between our predictor variable of strength (Nm/kg) and each of our primary criterion variables (ie, ΔpvGRF and pvGRF RMSE for each of the 5%, 10%, and 15% RTGBF trials). For our secondary analyses, we conducted separate univariate linear regression analyses between strength (Nm/kg) and each of our secondary criterion variables (ie, ΔpKFA, ΔKFE, ΔKEE, and ΔpKEM for each of the 5%, 10%, and 15% RTGBF trials). We reported the R², the unstandardized β and its standard error, and the associated P value for each linear regression analysis. We conducted separate stepwise linear regression models as sensitivity analyses to determine the unique variance predicted by strength for each primary and secondary criterion variable (ie, ΔpvGRF, pvGRF RMSE, ΔpKFA, ΔKFE, ΔKEE, and ΔpKEM) by accounting for the variance explained by potential covariates (ie, the time post-ACLR, sex, graft type, and walking speed of individuals). Owing to the preliminary nature of this study and due to our limited sample size, we conducted separate linear regression models with each covariate rather than including all covariates within a single multiple regression model. Future research with an increased sample size would be required to include all covariates within a single multiple regression model, and as such, the results of the current sensitivity analyses are only hypothesis generating to inform the use of covariates in future studies. For each sensitivity analysis, we first entered the relevant covariate (ie, the time post-ACLR, sex, graft type, or walking speed of individuals) and next entered the predictor variable of strength (Nm/kg). We then determined the unique association between the criterion variable and strength after accounting for the variance attributed to the model by each covariate. To do so, we calculated the change in R² (ΔR²) as the difference in R² between the primary model (association between strength and a change of biomechanics) and the model with the addition of the respective covariate (either time post-ACLR, sex, graft, or walking speed). The level of significance for all analyses was determined a priori at P < .05.

An additional post hoc supplemental analysis was conducted to determine if the consistency of the actual pvGRF, irrespective of the distance from the target pvGRF, was associated with strength in any of the RTGBF conditions. Thereby, we evaluated the association between the standard deviation of pvGRF and strength (Nm/kg) for each of the 5%, 10%, and 15% RTGBF trials. Finally, paired-samples t-tests compared pvGRF between the baseline walking trial and each of the 3 RTGBF trials to demonstrate that our cohort exhibited biomechanical changes in response to RTGBF.

Thirty-five individuals completed all aspects of the study. Descriptive, predictor, and criterion variables are shown for primary and secondary aims in Table 1.

No significant associations were identified between strength (Nm/kg) and the biomechanical variables of interest (ie, ΔpvGRF and pvGRF RMSE) for any of the 3 RTGBF trials (range R², 0.003–0.025, all P’s ≥ .37; Table 2).

No significant associations were identified between strength (Nm/kg) and changes in sagittal plane biomechanical variables (ie, ΔpKFA, ΔKFE, ΔKEE, ΔpKEM) for any of the 3 RTGBF trials (range R², <0.001–0.046, all P’s ≥ .21; Table 3).

There continued to be no significant associations between strength and any of the biomechanical variables of interest (ie, ΔpvGRF, pvGRF RMSE, ΔpKFA, ΔKFE, ΔKEE, ΔpKEM) after adjusting for the covariates (ie, time post-ACLR, sex, graft type, and walking speed) for each of the 3 RTGBF trials (range ΔR², <0.001–0.160, P = .060–1.00; Table 4).

No significant associations were found between pvGRF standard deviation and strength for any RTGBF condition (range R², <0.001–0.075, all P’s ≥ .11; see Supplemental Table). Statistically significant increases for pvGRF were found between the baseline walking trial and each of the 3 RTGBF conditions (all P’s < .001; Table 1).

Contrary to our hypothesis, we determined that the ability (ΔpvGRF, pvGRF RMSE, and pvGRF standard deviation) of individuals to adjust pvGRF during gait using RTGBF is not associated with quadriceps strength. Similarly, strength did not predict changes in sagittal plane kinematics and kinetics during any of the RTGBF conditions. Further, after accounting for time post-ACLR, sex, and walking speed of the individual, still no associations were found between strength and any of the biomechanical variables of interest. Our results indicate that quadriceps strength is not associated with the ability to change gait biomechanics and meet the given RTGBF target in individuals with an ACLR. As such, the results from the current study suggest that RTGBF could be prescribed to normalize aberrant gait patterns common after ACLR irrespective of quadriceps strength.

While the mechanistic pathway linking quadriceps weakness to osteoarthritis development remains uncertain, the association between decreased quadriceps function and changes to gait biomechanics related to osteoarthritis development has been established.³  Specifically, quadriceps inhibition acutely caused by hypertonic saline solution causes 5% to 6% lesser pvGRF and 14% to 15% lesser pKEM during walking, characteristics of the same aberrant gait patterns common post-ACLR and linked to posttraumatic osteoarthritis (PTOA).^{3,6–9,11,13,14}  This mechanistic link, as well as the myriad of associations between quadriceps function and aberrant biomechanics, has contributed to a hypothesis that improving quadriceps strength after ACLR would concurrently improve gait biomechanics.^1,4,20  However, this hypothesis likely overlooks the complexity of neuromuscular control and the multifaceted nature of gait. Aberrant biomechanics persist even after individuals undergo therapeutic exercise to address postoperative quadriceps weakness.²¹  Despite measured improvements of quadriceps strength in individuals 6 months post-ACLR, asymmetric pKFA, KFE, and pKEM during gait remain.²²  Further, quadriceps strength symmetry is not correlated with the asymmetries in the pKFA, KFE, or pKEM identified in individuals 6 months post-ACLR.²¹  This suggests that restoring quadriceps strength alone does not fully restore gait biomechanics but rather a presence of other neuromuscular deficits contributing to aberrant gait biomechanics after ACLR remains.

The reason improving quadriceps function does not always normalize aberrant gait biomechanics after ACLR remains unknown, yet we hypothesize that the aberrant motor patterns are established early after ACLR during a period of substantial quadriceps weakness. We further hypothesize that increases in quadriceps strength are not the catalyst capable of reversing these aberrant gait biomechanics. Chan et al demonstrated that, despite having the ability to employ limb loading symmetry during various submaximal tasks, individuals 3 months post-ACLR employed aberrant movement patterns, shifting loading away from the surgical limb.²³  This nonuse behavior was also demonstrated in the ACLR limb in our cohort. Specifically, participants demonstrated lower pvGRF during baseline walking than uninjured controls despite having relatively high average quadriceps strength symmetry (0.94 Limb Symmetry Index) and the ability to reach higher pvGRF targets in the RTGBF conditions.²⁴  Therefore, an alternative approach to ACLR rehabilitation has been developed, which proposes that targeted gait rehabilitation is needed after ACLR.²¹  Both fundamentals of motor learning as well as the results of the current study support this alternate approach. Primarily, the specificity-of-practice hypothesis suggests that motor learning improves as the specificity of practice to the desired motor skill increases.^25,26  Therefore, instead of relying on the motor patterns established during quadriceps strengthening to transfer and subsequently improve gait patterns, practicing the specific gait patterns may be a more effective way of establishing improved gait motor patterns. Moreover, the lack of significant associations between quadriceps strength and any changes in biomechanics across 3 RTGBF trials in this study indicate that the ability to improve gait may be independent of the magnitude of quadriceps strength. This suggests that gait retraining could be conducted without first achieving full quadriceps strength. As such, early gait interventions after ACLR may be needed to prevent the development of habitual aberrant gait patterns and may contribute to decreased risks of PTOA after ACLR.

Despite quadriceps strength not contributing to the ability to modify gait, improving strength after ACLR remains important. Specifically, an isometric MVIC ≥ 3.1 Nm/kg² has been identified as a clinical cutoff, which significantly improves the odds of higher patient-reported outcomes.²⁷  Similarly, a cutoff of ≥3.0 Nm/kg² discriminates uninjured controls without knee-related symptoms from ACLR individuals with knee-related symptoms.²⁸  Further, while individuals from our cohort demonstrated relatively high strength symmetry (0.94 ± 0.17 Limb Symmetry Index), the average isometric MVIC of the ACLR limb normalized to body mass (2.56 ± 0.75 Nm/kg) did not meet strength sufficiency cutoff scores of 3 Nm/kg that are linked to high patient-reported function.^27,28  Therefore, while individuals in our study may still benefit from interventions targeting improvements in quadriceps strength, on average, the degree of quadriceps strength in our cohort was not linked to their ability to modify gait.

With this study, we are the first to demonstrate that quadriceps strength is not associated with the ability to modify gait patterns, yet limitations should be addressed in future research. In this study, we required all participants to be between 6 months and 5 years post-ACLR and to have finished formal physical therapy, and in doing so, we may have excluded individuals with the worst quadriceps strength and dysfunction and contributed to the relatively high average quadriceps strength symmetry demonstrated by our cohort. As such, it remains unknown if a minimal value of strength is required to modify gait biomechanics that was outside of the range of strength values included in our cohort and if the relationship between quadriceps strength and the ability to modify gait biomechanics may be different in individuals with greater quadriceps weakness at earlier time frames post-ACLR. To address what constitutes adequate quadriceps strength for starting gait training post-ACLR, authors of future studies should evaluate the relationship between quadriceps strength and the ability to modify gait biomechanics in individuals earlier post-ACLR. Additionally, in this study, we focused on maximal isometric quadriceps strength measured at 90° of knee flexion, and we did not measure the submaximal neuromuscular control of individuals’ quadriceps, which may be better established with electromyography. Authors of future work may seek to determine if an association exists between varied measurements of quadriceps strength and function and the ability to modify gait post-ACLR. Meniscal status after ACL injury has been associated with aberrant gait patterns after ACLR.²⁹  As such, the role of meniscal status on the ability to modify gait post-ACLR should be established. Findings from this study are limited to treadmill walking, and the relationship between quadriceps strength and ability to modify gait biomechanics may be different during over-ground walking or inclined and declined walking conditions. Finally, in this study, we only evaluated individuals’ ability to increase pvGRF. Therefore, the association between quadriceps strength and an individual’s ability to decrease pvGRF is unknown and may be evaluated in future studies.

No statistically significant associations were found between quadriceps strength and the gait biomechanics when cued to increase pvGRF during walking in individuals 6 months to 5 years post-ACLR. These data may suggest that regardless of quadriceps strength, individuals after ACLR can meet RTGBF cues to mitigate the PTOA-associated aberrant gait patterns common after ACLR (ie, the stiffened knee gait and underloading strategies). We speculate that individuals may benefit from early interventions of gait retraining post-ACLR in addition to interventions aimed at improving quadriceps function.

Dr. Evans-Pickett reports grants from UNC CCCR Grant during the conduct of the study (NIH/NIAMS P30AR072580). The funding source had no role in the study design; collection, analysis, and interpretation of data; writing of the report; nor the decision to submit the article for publication.

Dr. Franz reports grants from the NIH during the conduct of the study, others from VETTA Solutions outside the submitted work. In addition, Dr Franz has a patent to Serial No. 63/400,163 pending and is Cofounder and CTO of VETTA Solutions, which has developed a wearable sensor platform for gait retraining.

Dr. Kiefer reports a patent to U.S. Patent No. 11,350,854, augmented neuromuscular training system and method, issued, licensed, and with royalties paid. Dr Schwartz reports grants from the NIH during the conduct of the study.

Supplemental Table. Associations Between Strength (Nm/kg) and pvGRF Standard Deviation for 5%, 10%, and 15% RTGBF Trials.