Knee Extensor Rate of Torque Development After Anterior Cruciate Ligament Reconstruction With Hamstring Tendon Autografts in Young Female Athletes

Background: For the measurement of functional decits after anterior cruciate ligament reconstruction (ACLR) with bone-patellar-tendon bone (BTB) grafts, the knee extensor rate of torque development (RTD) is known as a relevant outcome. However, it remains unclear if the RTD is also a relevant outcome after ACLR with semitendinosus and gracilis tendon (STG) autografts. The purpose of the present study was to compare the limb symmetry index (LSI) of the RTD of the quadriceps with that of the peak torque after ACLR using STG autografts and to investigate the relationship between the self-reported functional outcome and the LSI of the torque parameters. Methods: Eighteen young female athletes after ACLR with STG grafts (age: 16.8 ± 1.3 years; time after surgery: 8.6 ± 0.8 months). The participants had undergone three maximal voluntary isometric quadriceps tests using an isokinetic dynamometer. Using the torque-time curves, the peak torque, RTD 100 (0 to 100 ms) and RTD 200 (100 to 200 ms) were determined. The International Knee Documentation Committee subjective knee evaluation form (IKDC-SKF) was used to assess self-reported knee function. Comparisons of the LSI of the peak torque and RTD were performed by analysis of variance (ANOVA) with Bonferroni’s correction as a post-hoc test. The Pearson correlation coecient was used to examine the associations of the IKDC-SKF score with the LSI of the RTD and peak torque. Results: The LSI of the peak torque and RTD 200 was signicantly lower than that of the RTD 100 (peak torque vs RTD 100 : P = 0.017; RTD 200 vs RTD 100 : P = 0.015). The LSI of the peak torque was positively correlated with the IKDC-SKF score (R = 0.621; P = 0.006). Conclusions: The peak torque and RTD 200 were more sensitive for detecting inter-limb differences in quadriceps function than RTD 100 . The IKDC-SKF score was correlated with the LSI of the quadriceps peak

cannot return to their pre-injury level of sports after ACLR (5,6,8). One of the factors affecting these problems is insu cient functional recovery after ACLR, likely causing the high incidence of secondary ACL injury, low subjective knee function and low ratio of return to sports (10). Therefore, clinicians should consider the importance of fully restoring knee function after ACLR and post-operative rehabilitation.
Quadriceps femoris dysfunction after ACLR is considered a signi cant problem (11). The rate of torque development (RTD) of the quadriceps has been proposed as an index of functional recovery after ACLR (12)(13)(14)(15)(16)(17). The RTD is calculated from the slope of the torque-time curve, which re ects the ability for instantaneous force production and development (18). The RTD within 100 msec after torque production is affected by the neural drive and ring motor unit frequency (19). However, the RTD after 100 msec is affected by the stiffness of the muscle-tendon complex and the peak torque production potential (19,20).
Recent studies have shown that the RTD of the injured limb is lower than that of the contralateral limb after ACLR and that in healthy controls (12)(13)(14)(15)(16). The limb symmetry index (LSI) of the quadriceps peak torque was reported to be approximately 80% at 6 months after ACLR, while that of the RTD was only 50-70% (14,15). Therefore, the RTD can reveal quadriceps dysfunction after ACLR.
ACLR with bone-patellar-tendon bone (BTB) grafts alters the patellar tendon stiffness (21)(22)(23); thus, the RTD is affected by not only the neural drive but also musculotendinous architectural changes. Furthermore, the difference in the LSI of the RTD and peak torque would be affected by the BTB graft rather than ACLR. Semitendinosus and gracilis tendon (STG) grafts and BTB grafts have been commonly used for ACLR (21,23). ACLR with STG grafts does not invade the patellar tendon and, as such, is less effective on the RTD than ACLR with BTB grafts. After ACLR with STG grafts, the quadriceps peak torque LSI was reported to be 80-90%, and the LSI of the RTD was reported to be 90% (12,17,24). However, no report has investigated the associations of RTD after ACLR with STG grafts and the relationship with the self-reported function. Therefore, this study aimed to compare the LSI of the quadriceps RTD with that of the peak torque after ACLR with STG tendon autografts and to investigate the relationship between the self-reported function and LSI of the torque parameters. We hypothesized that there was no difference in the LSI of the RTD or peak torque and that the LSI of the RTD and peak torque was signi cantly correlated with the IKDC-SKF score.

Participants
Eighteen young female athletes after ACLR with STG tendon autografts (25,26) participated in the present study (age: 16.8 ± 1.3 years; height: 160.6 ± 4.6 cm; body weight: 56.3 ± 6.6 kg; time after surgery: 8.6 ± 0.8 months; pre-injury Tegner activity scales: 8.4 ± 0.9). The inclusion criteria were as follows: age younger than 18 years, unilateral ACL injury and competitive level of sports. The risk of secondary ACL injury to athletes younger than18 years is considered high (27); thus, participants younger than 18 years were included in the study. All participants with a history of any orthopaedic surgery other than ACLR, neurological disorders of the lower limb or complicated ligament injuries requiring additional procedures (posterior cruciate ligament injury and medial collateral ligament injuries) were excluded from this study.
The time from injury to surgery was 3.1 ± 3.5 months. Six participants had sustained ACL injuries to their right side, and 12 participants had injured their left side. Five participants underwent concomitant meniscus repair.
All the participants completed a standardized rehabilitation protocol. Quadriceps strengthening was started with the straight leg raising from 2 days after surgery, squatting with a knee exion range of motion larger than 60° from 1 week after surgery and quadriceps setting from 2 weeks after surgery. Leg extension exercises with larger angles of knee exion (> 60°) were started 6 weeks after surgery. The participants started running at 12 weeks and were allowed to jump and sprint with submaximal effort from 5 months. They were allowed to return to sports approximately 9 months after surgery. The rehabilitation protocol was progressed based on the postoperative period. In this study, all the participants were tested between 8 and 10 months after the surgery. The present study was approved by the Institutional Review Board of the authors' a liated institution (approval number: 18-64). All the participants and their guardians received a written document explaining the study objectives and procedures and were required to provide written informed consent before participating in the research activities.

Procedures
Each participant performed a 5-minute warm-up using a stationary bike with a self-selected speed (28).
After the warm-up, each participant performed 3 practice sets of 5-second maximal voluntary isometric contraction (MVIC) of the knee extensors before the actual measurement. The knee extension torque was recorded using an isokinetic dynamometer (Biodex System 3 isokinetic dynamometer; Biodex Medical Systems, Inc., Shirley, NY). The sampling rate was set at 100 Hz. All trials were performed with the hip at 90° and knee at 70° of exion (28). Straps were rmly fastened around the patient's chest, waist and distal thigh for stabilization (13). A shin pad was placed two nger widths above the lateral malleolus (13). Before the test, the participant was instructed to extend the knee "as fast as and as much as possible" and practiced for familiarization using visual feedback from the Biodex monitor screen (28). The participant then performed each task 3 times for each leg (uninvolved limb rst) with a 1-minute rest interval (28). Verbal encouragement was given to the participants to maximize the torque production during the tests.
Self-reported knee function was evaluated using the IKDC-SKF, which is an index comprising a score from 0 to 100 and scores for knee symptoms, function and sports activities. The IKDC-SKF score is highly relevant and reliable (29).

Data analysis
Custom MATLAB code (The Math Works, Inc., Natick, MA, USA) was used for data processing. The forcetime signal was low-pass ltered at 6 Hz using a second-order Butterworth lter (13). The peak torque, arrival time of the peak torque and RTD were calculated from the torque-time curves (Fig. 1). The onset of muscle contraction was de ned as the time when the knee extension torque exceeded the baseline by 7.5 Nm (18,19). The RTD 100 and RTD 200 were also calculated (19). The RTD 100 during the early phase of muscle contraction (0 to 100 ms) represents the in uence of neural drive. Thus, the RTD 200 during the late phase (100 to 200 ms) represents the in uence of musculotendinous stiffness the and peak torque (19,20). Both the peak torque and RTD were normalized to the body weight. The LSI of the peak torque and RTD was calculated as the percentage in the involved limb compared with that in the uninvolved limb.

Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics 22 software (IBM, Chicago, IL, USA). Oneway analysis of variance (ANOVA) was used to examine the difference in the LSI of the peak torque, RTD 100 and RTD 200 . Post-hoc comparisons were conducted using Bonferroni's correction. The paired ttest was used to examine the between-limb differences in the peak torque, arrival time of the peak torque, RTD 100 and RTD 200 . The Pearson correlation coe cient was used to examine the relationship between the IKDC-SKF score and LSI of the peak torque, RTD 100 and RTD 200 . A correlation (R) of 0.90 to 1.00 was interpreted as extremely large, 0.70 to 0.89 as very large, 0.50 to 0.69 as large, 0.30 to 0.49 as moderate, 0.10 to 0.29 as small, and less than 0.09 as trivial or no relationship (30). Furthermore, the value of d z was calculated as the effect size (31). A d z value greater than 0.80 was interpreted as large, 0.50 to 0.79 as moderate, and 0.20 to 0.49 as small (31). The statistical signi cance level was set at P < 0.05. The sample size was calculated using G*Power 3.1 based on previously published data (32). More than 17 participants were required to detect any between-limb difference in the RTD 100 (80% power; α = 0.05).

Results
One-way ANOVA revealed a signi cant difference in the LSI of the peak torque and RTD (P < 0.001). The LSI of the peak torque was signi cantly lower than that of the RTD 100, as shown by the moderate effect size (P = 0.017; d z = 0.748) (Fig. 2). The LSI of the RTD 200 was also signi cantly lower than that of the RTD 100, with a moderate effect size (P = 0.015; d z = 0.756) (Fig. 2). No other differences in the LSIs were detected by post-hoc testing.
Concerning the inter-limb differences, the peak torque was signi cantly lower in the involved limb than in the uninvolved limb, with a large effect size (P < 0.001; d z = 1.366) (Fig. 3 & Table 1). The RTD 200 was also signi cantly lower in the involved limb than in the uninvolved limb, with a large effect size (P < 0.001; d z = 0.986) ( Table 1). However, no differences were found in the RTD 100 or arrival time of the peak torque between the limbs (Table 1). Table 1 Inter-limb differences in the peak torque, RTD 100 , RTD 200 and arrival time of the peak torque The LSI of the peak torque was positively correlated with the IKDC-SKF score (R = 0.621; P = 0.006) ( Fig. 4a). However, the IKDC-SKF score was not correlated with the LSI of the RTD 100 (R = 0.257; P = 0.304) or RTD 200 (R = 0.322; P = 0.193) ( Fig. 4-b, c).

Discussion
We aimed to compare the LSI of the RTD with that of the peak torque and to determine the inter-limb difference in the peak torque and RTD after ACLR with STG grafts. We investigated the associations of the IKDC-SKF score with the LSI of the RTD and peak torque. Our ndings showed that the RTD 200 and peak torque were signi cantly lower in the involved limb than in the uninvolved limb. The LSI of the peak torque showed a signi cant correlation with the IKDC-SKF score, but there was no correlation with the RTD 100 or RTD 200 . These results partially support our hypothesis.
The LSI of the RTD 100 was signi cantly higher than that of the RTD 200 and peak torque. No signi cant difference was found in the RTD 100 between the involved and uninvolved limbs. The present results indicate that the RTD 100 of the involved limb is comparable to that of the uninvolved limb, supporting our hypothesis. The voluntary RTD 100 of the quadriceps is signi cantly correlated with the non-voluntary RTD induced by electrical stimulation (19); thus, the RTD 100 is considered to re ect neural drive actions such as the ring frequency effects of motor units (18). The recovery of neural function after ACLR may manifest differently depending on the type of tendon graft. The central activation ratio (CAR) measures the potential muscle exertion ability using electrical stimulation during MVIC (33)(34)(35) and is used to assess the neural drive of the quadriceps femoris after ACLR. A recent systematic review revealed that the CAR in the involved limb after ACLR with BTB grafts was lower than that in the uninvolved limb (35). By contrast, the CAR in the involved limb after ACLR with STG grafts was higher than that in the uninvolved limb (35). One possible reason for the different results between the studies was that post-operative pain affected the neural drive. Anterior knee pain after ACLR has been determined in 48% of patients receiving BTB grafts and in 20% of those receiving STG grafts at 6 months after ACLR (21). Another study showed a positive result for anterior knee pain in 73% of patients with BTB grafts and 35% of patients with STG grafts at 8 months after ACLR (21,23). Regarding the RTD 100 after ACLR, the LSI at 6 months after ACLR with BTB grafts was reported to be 49% (14), and the LSI at 11 months after ACLR with BTB or STG grafts was 72% (16). In our study, the RTD 100 was 95.9%, which was higher than that in previous studies, indicating that the RTD 100 recovered after ACLR with STG grafts. Therefore, post-operative pain due to differences in the graft type may have affected neural drive recovery after ACLR.
The RTD 200 and peak torque were signi cantly lower in the involved limb than in the uninvolved limb. No difference was found in the LSI of the peak torque or RTD 200 , indicating that the RTD 200 is an index for detecting between-limb differences as well as the peak torque. The RTD 200 is affected by structural factors with musculotendinous stiffness (20). A previous study reported structural changes, such as an increased cross-sectional area and decreased stiffness in the patellar tendon after ACLR with BTB grafts (22). In an animal study, the duration of patellar tendon stiffness recovery was approximately 1 year (36, 37). The LSI of the RTD 200 at 6 months after ACLR with BTB grafts was 43% (14); even at 4 years after ACLR with BTB or STG grafts, it was up to 78% (12). In our study, the LSI of the RTD 200 was 86.0%, which is higher than that reported previously (12,14). Therefore, the RTD 200 of the quadriceps after ACLR with STG grafts may be less likely to decrease than that after ACLR with BTB grafts. However, the RTD 200 is affected by structural factors and musculotendinous stiffness (20); thus, the RTD 200 after ACLR with BTB grafts was lower than that after ACLR with STG grafts. Additional studies should be conducted to clarify the effects of the graft type on RTD 200 recovery after ACLR.
The LSI of the peak torque was positively correlated with the IKDC-SKF score. Thus, the IKDC-SKF score was not signi cantly correlated with the LSI of the RTD 100 or RTD 200 . These results do not support our hypothesis that the LSI of the RTD and peak torque is signi cantly correlated with the IKDC-SKF score.
These study results are similar to those of other studies showing that the IKDC-SKF score was signi cantly correlated with the LSI of the peak torque (38, 39). However, the LSI of the RTD 100 and RTD 200 was not correlated with the IKDC-SKF score. At 3 months after ACLR, the RTD of the involved limb showed a signi cant positive correlation with the IKDC-SKF score (13). In contrast, there was no correlation between the LSI of the RTD and IKDC-SKF score, even at 4 years after ACLR (12). The cause may be the difference in the IKDC-SKF scores; the IKDC-SKF score was 66.5 points at 3 months after ACLR (13) and 86.8 points at 4 years after ACLR (12). The IKDC-SKF score in this study was 86.3 points; thus, a high IKDC-SKF score may also in uence the results of any correlation between the LSI of the RTD and the IKDC-SKF score. No signi cant correlation was found between the LSI of the RTD and the IKDC-SKF score in the present study, but the peak torque could re ect subjective knee function after ACLR with STG grafts. The peak torque after ACLR was reported to be related to the kinetic asymmetry between the involved and uninvolved leg during landing (40)(41)(42). However, previous studies of the RTD after ACLR have been limited to the association with the kinetics of walking and running, such that its usefulness as a functional screening tool remains unproven (7,43). Therefore, the peak torque of the quadriceps should be considered a more useful index to assess functional recovery after ACLR.
Some limitations of this study should be addressed. First, the participants were limited to young female athletes. Age and sex differences might affect knee extension torque (44,45); thus, the present results might apply to young female athletes only. Second, all the participants in this study had undergone ACLR with STG grafts. Different types of ACLR might lead to different results. Finally, knee function in the present study was assessed based on the subjective knee score and was not a result of functional dynamic tasks.

Conclusions
The present study shows that the LSI of the peak torque and RTD 200 was signi cantly lower than that of the RTD 100 . The IKDC-SKF score was signi cantly correlated with the LSI of the peak torque but not with that of either the RTD 100 or RTD 200 . These results suggest that the peak torque of the quadriceps is appropriate for the functional screening of female athletes after ACLR with STG grafts.