The Knee 17 (2010) 287–290

Contents lists available at ScienceDirect

The Knee

Hamstring antagonist torque generated in vivo following ACL rupture andACL reconstruction

Adam L. Bryant a,⁎, Mark W. Creaby a, Robert U. Newton b, Julie R. Steele c

a Centre for Health, Exercise and Sports Medicine, School of Physiotherapy, The University of Melbourne, Australiab School of Biomedical and Sports Science, Edith Cowan University, Australiac Biomechanics Research Laboratory, School of Health Sciences, University of Wollongong, Australia

⁎ Corresponding author. Centre for Health, Exercise aPhysiotherapy, Faculty of Medicine, Dentistry and HeaMelbourne, Victoria, Australia 3010. Tel.:+61 3 8344 41

E-mail address: albryant@unimelb.edu.au (A.L. Brya

0968-0160/$ – see front matter. Crown Copyright © 20doi:10.1016/j.knee.2010.02.004

a b s t r a c t

a r t i c l e i n f o

Article history:Received 22 September 2009Received in revised form 22 January 2010Accepted 8 February 2010

Keywords:Anterior cruciate ligamentReconstructionHamstring antagonist torqueSensory feedback

Hamstring motor behaviour and resultant antagonist torque during knee extension has been quantified inuninjured individuals however, the effect of ACL rupture and ACL reconstruction (ACLR) on the morphologyof hamstring antagonist torque generated in vivo is unknown. The purpose of this cross-sectional study wasto quantify the hamstring antagonist torque generated in vivo during isokinetic knee extension in ACLD andACLR patients relative to uninjured control subjects. Ten male ACL deficient (ACLD) subjects (18–35 years),14 matched males who had undergone ACLR using the bone–patellar tendon–bone graft and 22 matchedmale control subjects participated. We used a mathematical model to estimate the opposing torquegenerated by the hamstrings during isokinetic knee extension in 10° intervals from 80° to 10° knee flexion.Control group hamstring antagonist torque was significantly lower at 80–70° knee flexion compared withthat of the ACLD (% Diff=40.2; p=0.019) and ACLR (% Diff=34.8; p=0.036) groups. For all subject groups,hamstring antagonist torque demonstrated a descending–ascending curve; decreasing significantly from 80–70° to 50–40° knee flexion (% Diff=40.8 to 63.3; p=b0.001 to 0.009) but then increasing significantly from50–40° to 20–10° knee flexion (% Diff=37.6 to 59.0; p=b0.001 to 0.012). ACL status and therefore, the ACL-hamstring reflex has little effect on the magnitude of hamstring antagonist torque generated duringquadriceps-induced knee joint loading. Capsular afferents are thought to dictate the hamstring torque profilewhich decreased then increased during knee extension to maintain dynamic joint stability.

Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved.

1. Introduction

In large weight-bearing joints, loads often exceed the capacity ofthe passive non-contractile elements and, in order to protect againstinjury, assistance from active muscles is required [1]. At the knee,loading of the anterior cruciate ligament (ACL) and knee capsularstructures – given the existence of excitatory reflexes (including theACL-hamstring reflex) – has been reported to elicit afferent,mechanoreceptor-mediated feedback innervating alpha and gammamotor neurons within the hamstring muscles [2,3]. As such, thehamstrings have been described as ‘load regulators' of the knee,capable of increasing their activity on demand to decrease anteriortibial translation (ATT) and, in turn, strain within the ACL [4,5].

Indeed, during maximal effort isokinetic knee extension in healthysubjects, hamstring electromyographic (EMG) activation levels rangefrom 15 to 58% of themaximumagonist value and vary as a function ofjoint angle, demonstrating a descending–ascending curve [5–7]. By

nd Sports Medicine, School oflth Sciences, The University of37; fax: +61 3 8344 4188.nt).

10 Published by Elsevier B.V. All rig

establishing the EMG–force relationship, two studies have alsoapplied mathematical models to predict the opposing torquegenerated by the hamstrings during isokinetic concentric kneeextension [5,7]. In both studies, the predicted hamstring torqueswere nearly constant throughout the range of motion (ROM),although between-study comparisons are difficult due to the disparityof methods. Nevertheless, the magnitudes of the reported antagonisttorques versus the agonist (i.e. quadriceps) torques confirm thenotion that the hamstrings function to maintain dynamic knee jointstability and, in doing so, protect integrity of the ACL and other passiveknee joint structures.

In the ACL ruptured knee, thehamstrings take an even greater role inmaintaining dynamic knee joint stability. Certainly, the effectiveness ofhamstring torque in reducing laxity in ACL deficient (ACLD) knees hasbeen demonstrated in studies incorporating cadaveric [8] and computer[9] models. Nonetheless, the likelihood of potential neurosensorydeficits caused by ACL rupture and capsular stretching, a sequela ofincreased anterior knee joint laxity, and the resultant effect onhamstring firing patterns and antagonist torque has not beenentertained in these previous studies. Certainly, the effect(s) of theseneurosensory deficits on hamstring behaviour post-ACL injury can onlybe accurately assessed in vivo.

hts reserved.

288 A.L. Bryant et al. / The Knee 17 (2010) 287–290

Interestingly, there is some evidence to suggest that suchneurosensory deficits are partly reversible via the process of ACLreconstruction (ACLR) given that a proliferation of ACL mechanor-eceptors has been identified during the assimilation of the ACL grafttissue [10]. Moreover, the decrease in quadriceps-induced ATT, andthus capsular stretching following ACLR [11], may conceivably restorethe sensitivity of afferent feedback to the gamma motor neurons.Hence, it is plausible that the morphology of the hamstring antagonisttorque generated in vivo following ACLR is more similar to those ofuninjured individuals compared with those who are ACLD.

Given these considerations, the aim of this studywas to quantify thehamstring antagonist torque generated in vivo during isokinetic kneeextension (80° to10° kneeflexion) inACLDandACLRpatients relative touninjured control subjects. Accordingly, our hypotheses were asfollows: (i) hamstring antagonist torque of ACLD subjects would behigher compared with ACLR subjects (H1), (ii) hamstring antagonisttorque of ACLD subjects would be higher compared with uninjuredcontrol subjects (H2), and (iii) there would be no difference inhamstring antagonist torque generated by ACLR subjects and uninjuredcontrol subjects (H3).

2. Materials and methods

2.1. Subjects

The ACLD (n=10), ACLR (n=14) and control (n=22) subjectsincluded in this study were the same as those described in a previousstudy [12]. These male subjects were matched for age, activity level,and anthropometric characteristics. Female subjects were excludedgiven the confounding effects of estrogen on tissue mechanicalproperties [13] and neuromuscular function [14–18].

On clinical examination, all ACLR subjects had regained full ROMand were stable in flexion and extension (i.e. negative Lachman andpivot shift tests). For the ACLD subjects, isolated complete rupture ofthe ACL was confirmed by previous arthroscopy and initial injuryoccurred at least 1 year before testing. All ACLD subjects demonstrat-ed minimum grade two Lachman (mode=2.5; range=2 to 3) andpivot shift (mode=2.5; range=2 to 3) tests. Subjects in the ACLDand ACLR groups also met the following inclusion criteria: aged 18 to35 years; no evidence of collateral ligament, meniscal or posteriorcruciate ligament damage or repair at the time of arthroscopy/surgery; no evidence of osteoarthritis on current plain radiographs;no previous ACL knee surgery or subsequent knee surgery on theinvolved leg; no history of surgery or traumatic injury to thecontralateral knee, and; participation in a physical therapy rehabil-itation program for the injured knee following knee injury/surgery.

The 22 control subjects demonstrated no history of trauma ordisease in either knee and no evidence of abnormality on clinicalexamination. Our protocol was approved by the University humanethics committee and all subjects read and signed an approvedconsent document prior to participating in the study.

2.2. Surgical procedure and rehabilitation

All ACL reconstructions were performed by the same orthopaedicsurgeon using an arthroscopically assisted two-incision technique aspreviously described [19]. Briefly, the central third of the patellatendon (10 mm wide) of the ipsilateral knee together with boneblocks (20×10 mm) from the patella and tibial tuberosity washarvested as part of a bone–patellar tendon–bone graft. Tibial andfemoral tunnels (drilled using a 4.5 mm gauge drill) were fashionedwith guide pins, and the graft was passed into the joint and the boneplug was fixed at the femoral site with an interference screw.Similarly, the tibial bone plug fixation was accomplished with aninterference screw at 10° of knee flexion. The graft was observed

under a final arthroscopic examination to determine its ability toresist anterior displacement during a Lachman test [20–22].

A comprehensive description of the physical therapy rehabilitationprograms for theACLD andACLR subjects has beenpublished elsewhere[19]. In brief, all ACLR subjects participated in the same standardaccelerated physical therapy rehabilitation program, as originallydescribed by Shelbourne and Nitz [23], for the injured knee followingknee surgery. On average, ACLR subjects underwent supervisedrehabilitation for 12 weeks post-surgery and then entered the ‘returnto sport’ phase where they gradually increased the complexity andintensity of activities. ACLD subjects were also subjected to anaccelerated protocol with immediate training of ROM and weightbearing. Supervised rehabilitation continued for 8–10 weeks followinginjury and primarily consisted of quadriceps/hamstring musclestrengthening (open and closed kinetic chain) together with agilityskill training and sport specific training in the latter part of the program.

2.3. Experimental protocol

Following standard preparation to reduce cutaneous impedancebelow 5 kΩ (Metex M-3650 impedance meter), silver–silver chloridepre-amplified surface electrodes (Quantec™, Brisbane, Australia; 5 mmdiameter) were placed over the bellies of the semitendinosus (ST) andbiceps femoris (BF) muscles of the involved limb of the ACLD and ACLRsubjects and the dominant and non-dominant limbs of the controlsubjects. A bipolar electrode configuration (inter-electrode distance=20 mm [24]) was used and a reference electrode was placed equidistantwith respect to the differential electrodes [25]. Saline gel was placedbetween each electrode and the skin to enhance conductivity.

The warm-up, isokinetic test protocol and data collection methodshave been previously described [19]. Briefly, subjects completed a10 minute warm-up consisting of low-resistance ergometer cyclingfollowing static stretching of the quadriceps and hamstring muscles.Following familiarization, each subject then performed 2 sets of 5maximal extension and flexion repetitions at 180°s−1. The test orderfor the non-dominant and dominant limbs of the control group wasalternated. Following amplification, the EMG, isokinetic torque andknee joint displacement data were analog-to-digital (A/D) converted(12-bit, 1000 Hz) by an AMLAB® signal acquisition system (Associa-tive Measurement, Sydney, Australia).

2.4. Data analysis

The pre-amplified EMG signals from the ST and BF muscles werefiltered using a fourth-order zero-phase-shift Butterworth filter (highpass fc=15 Hz; low-pass fc=250 Hz) [26]. To create a linear envelope(mV), the filtered EMG data were full wave rectified and then filteredagain using a fourth-order zero-phase-shift Butterworth low-passfilter (fc=30 Hz). To convert any negative values arising at theextreme ends, the linear envelope was again full wave rectified.

Hamstring antagonist torque was estimated as previously describedin the literature [19], across 10° intervals from 80° to 10° knee flexion[5,7]. Antagonist iEMG activity for the ST and BF muscles for each 10°interval was normalized relative to their activity during flexion [5,7,27],and converted to a torque based on the linear relationship betweeniEMG and hamstring torque during open kinetic chain knee extension[7,19]. Individual hamstring torque contribution was based on a stressdistribution approach [19], andmoment armswere assumed to be equal[5,28,29]. Thus, the contribution of each hamstring muscle towards thenet knee jointflexion–extension torquewas assumed tobe proportionalto the muscles' physiological cross-sectional area (PCSA). Using thisapproach, hamstring antagonist torque at each 10° knee flexion intervalwas obtained using the following equation:

HamTORQUE =BFTOR × BFPSCAð Þ + STTOR × STPSCAð Þ + STTOR × SMPSCAð Þ

∑PSCA:

Fig. 1. Hamstring antagonist torque (mean±standard deviation) generated by ACLD(n=10), ACLR (n=14) and control (n=22) subjects between 10 and 80° knee flexion.*Between-group contrasts revealed a significant (pb0.05) difference between thecontrol group and the groups with a deficient or reconstructed ACL; ‡Within-groupcontrasts revealed significant differences (for all groups) between 80–70° and 50–40°knee flexion, 50–40° and 20–10° knee flexion and 80–70° and 20–10° knee flexion.

289A.L. Bryant et al. / The Knee 17 (2010) 287–290

where: HamTORQUE=Hamstring antagonist torque in Nm; TOR=estimated torque (Nm) generated by the individual hamstrings, and;PCSA=physiological cross-sectional area of the hamstrings.

Hamstring torque at each flexion interval was normalized to thenet extension torque for that interval. Data from all 10 trials of eachsubject's test limb were averaged. Custom written software (VisualBasic 5.0, Microsoft Corp., WA, USA) was used to analyze the data.

2.5. Statistical analysis

Descriptive statistics for the age, height and body mass for eachsubject group were calculated and a one-way ANOVA design was thenused to compare these variables among the ACLD, ACLR and controlgroups.

Means and standard deviations were calculated for the hamstringantagonist torque generated in the seven intervals between 80° and10° knee flexion. Preliminary statistical analysis of the control groupdata revealed no significant differences between the dominant andnon-dominant limbs for hamstring antagonist torque and, as aconsequence; the data were pooled across test limbs, producing“standardised” data sets for comparisonwith the involved limbs of theACLD and ACLR groups. A repeated measures ANOVA design was thenused to compare the hamstring antagonist torque produced duringmaximal effort knee extension throughout the ROM for the involvedlimbs of the ACLD and ACLR groups with the control group. Therefore,each ANOVA design included one within factor (knee flexion interval:80–70°, 70–60°, 60–50°, 50–40°, 40–30°, 30–20° and 20–10°) and onebetween factor (subject group: ACLD, ACLR and control). Prior torunning each ANOVA, normality and equal variance were confirmedusing a Kolmogorov–Smirnov test (with Lilliefor's correction) and aLevene Median Test, respectively. In the event of a significant maineffect or interaction (pb0.05) following ANOVA contrasts, post hoccomparisons of the means were conducted using the least significantdifference (LSD) test to delineate differences among subject groups orknee flexion intervals. All analyses were conducted using statisticalpackage, SPSS Version 14.0 (SPSS for Windows, SPSS Inc., Chicago, IL).

3. Results

Descriptive data pertaining to the physical characteristics of theACLD, ACLR and control subjects, together with F-ratios and alphalevels derived for each source of variance in the one-way ANOVAanalyses, are presented in Table 1. Statistical analysis revealed nosignificant difference in the physiological characteristics betweensubject groups. For the ACLD group, initial ACL injury occurred 75.6(±72.5) months prior to testing whilst the ACLR group underwentACLR surgery 15.1 (±5.0) months prior to testing.

Hamstring antagonist torque (mean±standard deviation) gener-ated at 180°s−1 between 80–70°, 70–60°, 60–50°, 50–40°, 40–30°,30–20° and 20–10° knee flexion for the non-involved and involvedlimbs for the ACLD, ACLR and control groups is presented in Fig. 1. Asignificant knee flexion interval × subject group interaction(F=3.104; p=0.011) was evident for the hamstring antagonisttorque produced between 80 and 10° knee flexion.

Between-subject group post hoc analyses indicated that thehamstring antagonist torque generated by the control group at 80–70° knee flexionwas significantly lower comparedwith that produced

Table 1Mean (±standard deviation) of the age, height and mass for the ACLD (n=10), ACLR(n=14) and control (n=22) subjects.

Variable ACLD ACLR Control F-value p-value

Age (years) 30.7±8.6 30.9±7.3 29.0±8.2 0.302 0.741Height (cm) 176.2±5.0 180.2±4.7 178.5±6.2 1.498 0.235Mass (kg) 72.4±4.8 87.1±19.6 78.6±14.4 2.970 0.062

by the ACLD (95% CI=2.387 to 14.237; % Diff=40.2; p=0.019) andACLR (95% CI=−10.973 to−0.351; % Diff=34.8; p=0.036) groups.No other between-subject group contrasts reached significance.Within-subject group post hoc contrasts indicated a significantdecrement in hamstring antagonist torque from 80–70° to 50–40°knee flexion for the ACLD (95% CI=8.971 to 18.207; % Diff=63.3;pb0.001), ACLR (95% CI=4.878 to 12.684; % Diff=51.2; pb0.001)and control (95% CI=2.187 to 8.413; % Diff=40.8; p=0.009) groups.In contrast, a significant increase in hamstring antagonist torque wasidentified between 50–40° and 20–10° knee flexion for the ACLD (95%CI=−15.338 to −5.145; % Diff=59.0; pb0.001), ACLR (95% CI=−11.669 to −3.055; % Diff=37.6; p=0.012) and control (95% CI=−12.423 to −5.552; % Diff=53.9; pb0.001) groups. No significantdifferences were identified between torques generated at 70–80° and20–10° knee flexion for any subject group.

4. Discussion

Histological studies have confirmed the existence of sensoryreceptors within the ACL [30] and capsular structures [31], the role(s)of which have been the focus of several neurophysiological investiga-tions [2,32]. Whilst it has been suggested that capsulo-ligamentousreceptors contribute to a negative feedback system which modulatesreflexive activity of the hamstrings in order to protect the ACL fromexcessive loading and risk of injury [32]; our results demonstrate that,for the most part, ACL status and therefore, the ACL-hamstring reflexdoes not wield a particularly strong effect on the magnitude ofnormalized hamstring antagonist torque generated between 80 and10° knee flexion.

To expand, no significant ACL treatment-specific effects were iden-tified (in disagreement with H1) and the only part of the movementwhere ACLD and ACLR subjects generated significantly higher ham-string antagonist torque compared with uninjured subjects was duringthe acceleratoryphase (i.e. 80–70° kneeflexion; largely in disagreementand agreement with H2 and H3, respectively). Concordant resultsbetween70and10° kneeflexion suggest that afferent nervefibers in thenative ACL exert little effect on the excitability of the skeletomotorneurons in the hamstrings; at least during quadriceps-induced ACLloadingwithin normal physiological ranges. In support, several previousattempts to elicit the ACL-hamstring reflex by evoking excitatory EMGresponses from the hamstrings via direct mechanical/electrical stimu-lation of the human ACL have been unsuccessful [2,33]. Without thenative ACL, amplified hamstring antagonist torque at the start of kneeextension is thought to be attributed to excess stretching of the joint

290 A.L. Bryant et al. / The Knee 17 (2010) 287–290

capsule (and collateral ligaments) as the quadriceps contract forcefullyto overcome the inertia of the lower limb. Via reflexive action, thiswould amplify the afferent output from the sensory receptors in thejoint capsule eliciting excitatory potentials to the skeletomotor neuronsof the hamstrings, resulting in increased hamstring activity and antag-onist torque.

Normalized hamstring antagonist torque for ACLD, ACLR andcontrol subjects decreased and increased widely as a function of kneeflexion angle, demonstrating a descending–ascending curve. Thisleads to the question: physiologically, what is the major sourcedictating this torque profile? As mentioned with respect to severalprevious studies, hamstring skeletomuscular neurons appear to beinfluenced little by ACL loading within normal physiological ranges.Alternatively, varying amounts of capsular stretching throughoutknee extension appears to exert a more potent effect on hamstringexcitability and resultant antagonist torque. Specifically, quadriceps-induced capsular stretching at the start and end of the movement asthe lower limb is being accelerated and decelerated, respectively; isthought to trigger a powerful capsulo-muscular reflex to enhancedynamic knee joint stability [5]. In contrast, the mid-range of themovement was associated with significantly lower hamstring antag-onist torque because constant lower limb velocity decreases capsularstretching which, in turn, diminishes the magnitude of reflexive inputfrom the capsular afferents.

Although novel, our study has several limitations. Our results maynot generalize tomale ACLD and ACLR individuals with characteristics –namely, time since ACLD or ACLR and the type of ACL tendon graft –different to those of our subjects. Moreover, the applicability of ourfindings to femaleACLD, ACLR anduninjured individuals is questionablegiven that female's possess hamstring activation strategies different tothose of their male counterparts [34]. Our mathematical model did notaccount for potential individual differences in hamstring anatomy andfuture studies comparing ACLD, ACLR and uninjured individuals shouldinclude magnetic resonance imaging to determine hamstring cross-sectional area.

In conclusion, our findings indicate that, apart from the accel-eratory phase, ACL status and therefore the ACL-hamstring reflex haslittle effect on the magnitude of hamstring antagonist torquegenerated during quadriceps-induced knee joint loading. Stretch-induced, sensory feedback from the capsular structures of ACLD andACLR subjects is thought to be responsible for the augmentedhamstring antagonist torque during knee joint acceleration. More-over, capsular afferents are thought to dictate the hamstring torqueprofile, which decreased then increased during knee extension inorder to maintain dynamic joint stability.

5. Conflict of interest

There are no conflicts of interest associated with our paper.

References

[1] Granata KP, Padua DA, Wilson SE. Gender differences in active musculoskeletalstiffness. Part II. Quantification of leg stiffness during functional hopping tasks.J Electromyogr Kinesiol 2002;12:127–35.

[2] Dyhre-Poulsen P, Krogsgaard MR. Muscular reflexes elicited by electricalstimulation of the anterior cruciate ligament in humans. J Appl Physiol 2000;89:2191–5.

[3] Johansson H, Sjolander P, Sojka P. Receptors in the knee joint ligaments and theirrole in the biomechanics of the joint. Crit Rev Biomed Eng 1991;18:341–68.

[4] Solomonow M, Baratta R, Zhou BH, Shoji H, Bose W, Beck C, et al. The synergisticaction of the anterior cruciate ligament and thigh muscles in maintaining jointstability. Am J Sports Med 1987;15:207–13.

[5] Baratta R, Solomonow M, Zhou BH, Letson D, Chuinard R, D'Ambrosia R. Muscularcoactivation: the role of the antagonist in maintaining knee stability. Am J SportsMed 1988;16:113–22.

[6] Aagaard P, Simonsen EB, Andersen JL, Magnusson SP, Bojsen-Moller F, Dyhre-Poulsen P. Antagonist muscle coactivation during isokinetic knee extension. ScandJ Med Sci Sports 2000;10:58–67.

[7] Kellis E, Baltzopoulos V. The effects of antagonist moment on the resultant kneejoint moment during isokinetic testing of the knee extensors. Eur J Appl Physiol1997;76:253–9.

[8] More RC, Karras BT, Neiman R, Fritschy D, Woo S, Daniel DM. Hamstrings — ananterior cruciate ligament protagonist: an in vitro study. Am J Sports Med 1993;21:231–7.

[9] Pandy MG, Shelburne KB. Dependence of cruciate-ligament loading on muscleforces and external load. J Biomech 1997;30:1015–24.

[10] Barrack RL, Lund PJ, Munn BG, Wink C, Happel L. Evidence of reinnervation of freepatellar tendon autograft used for anterior cruciate ligament reconstruction. Am JSports Med 1997;25:196–202.

[11] Woo SL, Kanamori A, Zeminski J, Yagi M, Papageorgiou C, Fu FH. The effectivenessof reconstruction of the anterior cruciate ligament with hamstrings and patellartendon. A cadaveric study comparing anterior tibial and rotational loads. J Bone JtSurg Am 2002;84-A:907–14.

[12] Bryant AL, Newton RU, Steele J. Successful feed-forward strategies following ACLinjury and reconstruction. J Electromyogr Kinesiol 2009;19:988–97.

[13] Bryant AL, Clark RA, Bartold S, Murphy A, Bennell KL, Hohmann E, et al. Effects ofestrogen on the mechanical behavior of the human Achilles tendon in vivo. J ApplPhysiol 2008;105:1035–43.

[14] Hewett TE. Neuromuscular and hormonal factors associated with knee injuries infemale athletes: strategies for intervention. Sports Med 2000;29:313–27.

[15] Romani W, Curl LA, Lovering R. The effect of endogenous estradiol levels at threephases of the menstrual cycle on anterior cruciate ligament stiffness in activefemales. J Athl Train 2001;2:S62.

[16] Romani W, Patrie J, Curl LA, Flaws JA. The correlations between estradiol, estrone,estriol, progesterone and sex hormone-binding globulin and anterior cruciateligament stiffness in healthy, active females. J Women's Health 2003;12:287–98.

[17] Shultz SJ, Kirk SE, Johnson ML, Sander TC, Perrin DH. Relationship between sexhormones and anterior knee laxity across the menstrual cycle. Med Sci SportsExerc 2004;36:1165–74.

[18] Eiling E, Bryant AL, Petersen W, Murphy A, Hohmann E. Effects of menstrual-cyclehormone fluctuations on musculotendinous stiffness and knee joint laxity. KneeSurg Sports Traumatol Arthrosc 2007;15:126–32.

[19] Bryant AL, Creaby MW, Newton RU, Steele JR. Dynamic restraint capacity of thehamstring muscles has important functional implications after anterior cruciateligament injury and anterior cruciate ligament reconstruction. Arch Phys MedRehabil 2008;89:2324–31.

[20] Buss DD, Warren RF, Wickiewicz TL, Galinat BJ, Panariello R. Arthroscopicallyassisted reconstruction of the anterior cruciate ligament with use of autogenouspatellar-ligament grafts. Results after twenty-four to forty-two months. J Bone JtSurg Am Vol 1993;75:1346–55.

[21] Barber-Westin SD, Noyes FR, Andrews M. A rigorous comparison between thesexes of results and complications after anterior cruciate ligament reconstruction.Am J Sports Med 1997;25:514–26.

[22] Beynnon BD, Johnson RJ, Fleming BC, Kannus P, Kaplan M, Samani J, et al. Seecomment. J Bone Jt Surg Am Vol 2002;84-A:1503–13.

[23] Shelbourne KD, Nitz P. Accelerated rehabilitation after anterior cruciate ligamentreconstruction. Am J Sports Med 1990;18:292–9.

[24] Daanen HA, Mazure M, HolewijnM, Van der Velde EA. Reproducibility of themeanpower frequency of the surface electromyogram. Eur J Appl Physiol Occup Physiol1990;61:274–7.

[25] De Luca CJ. Control properties of motor units. J Exp Biol 1985;115:125–36.[26] Winter DA. Biomechanics and Motor Control of Human Movement. New York:

Wiley-Interscience; 1990. Edited.[27] Kellis E, Baltzopoulos V. The effects of normalization method on antagonistic

activity patterns during eccentric and concentric isokinetic knee extension andflexion. J Electromyogr Kinesiol 1996;6:235–45.

[28] Kellis E, Baltzopoulos V. In vivo determination of the patella tendon andhamstrings moment arms in adult males using videofluoroscopy duringsubmaximal knee extension and flexion. Clin Biomech 1999;14:118–24.

[29] Smidt GL. Biomechanical analysis of knee flexion and extension. J Biomech1973;6:79–92.

[30] Kennedy JC, Alexander IJ, Hayes KC. Nerve supply of the human knee and itsfunctional importance. Am J Sports Med 1982;10:329–35.

[31] Halata Z, Haus J. The ultrastructure of sensory nerve endings in human anteriorcruciate ligament. Anat Embryol (Berl) 1989;179:415–21.

[32] Tsuda E, Okamura Y, Otsuka H, Komatsu T, Tokuya S. Direct evidence of theanterior cruciate ligament-hamstring reflex arc in humans. Am J Sports Med2001;29:83–7.

[33] Biedert RM, Zwick EB. Ligament-muscle reflex arc after anterior cruciate ligamentreconstruction: electromyographic evaluation. Arch Orthop Trauma Surg 1998;118:81–4.

[34] Cowling EJ, Steele JR. Is lower limb muscle synchrony during landing affected bygender? Implications for variations in ACL injury rates. J Electromyogr Kinesiol2001;11:263–8.