A Systematic Review and Meta-Analysis of Strength Recovery Measured by Isokinetic Dynamometer Technology after Anterior Cruciate Ligament Reconstruction Using Quadriceps Tendon Autografts vs. Hamstring Tendon Autografts or Patellar Tendon Autografts

Background: This systematic review and meta-analysis compared the isokinetic strength of the muscular knee joint between quadriceps tendon autografts (QTAs) and hamstring tendon autografts (HTAs) or patellar tendon autografts (PTAs) after anterior cruciate ligament (ACL) reconstruction by determining the isokinetic angular velocity and follow-up time points. The functional outcomes and knee stability at the same time points were also compared using isokinetic technology. Methods: Two independent reviewers searched the Medline (via PubMed search engine), Scopus, Web of Science and Cochrane Library databases to include full text comparative studies that assessed isokinetic strength test following ACL reconstruction. The DerSimonian and Laird method was used. Results: In total, ten studies were included; seven compared studies QTAs vs. HTAs, and three compared QTAs vs. PTAs. Five studies were included in the meta-analysis. Isokinetic strength data were reported 3, 6, 12 and 24 months after ACL reconstruction. Conclusions: The QTAs showed better and significant results with knee flexion compared with HTAs, similar results to PTAs at 6 and 12 months. While HTAs showed better and significant results with knee extension at 6 months and similar results at 12 months compared to QTAs. Furthermore, a standardized isokinetic strength test must be followed to achieve a more specific conclusion and better clinical comparison among participants.


Introduction
Anterior cruciate ligament (ACL) injury is a common knee injury with an incidence of between 32 and 80 cases per 100,000 inhabitants every year worldwide [1][2][3][4] and approximately 25,000 injuries per year in the United States [5]. Reconstruction of the ACL is the standard surgical method that aims to repair knee stability [6], improve both clinical and functional outcomes, achieve a rapid return to sport (RTS) [7][8][9] and reduce the potential risk of knee osteoarthritis [6,10]. Quadriceps tendon autografts (QTAs) have become more popular in the last 20 years because of their advantages over knee stability and muscle strength recovery [11][12][13][14][15]. Patellar tendon autografts (PTAs) and hamstring 2 of 18 tendon autografts (HTAs) are the most commonly used autografts [6, 16,17]. HTAs show good quadriceps recovery and less donor site morbidity but are associated with hamstring muscle deficits and slower rehabilitation processes [18,19]. PTAs offer the advantage of good hamstring recovery and a stable knee [20]. By contrast, PTAs have been associated with anterior knee pain and quadriceps strength deficiency [20,21]. Therefore, choosing ACL reconstruction autografts remains controversial because of their advantages and disadvantages.
Rehabilitation protocols play a significant role in RTS. Pre-surgery protocols comprise one phase and aim to achieve a 90% quadriceps limb symmetry index [22]. The postsurgery protocol comprises four to five phases over 6 months [22][23][24], forming the standard rehabilitation protocol [22][23][24]. However, evidence supports that the type of graft used in reconstruction alters the above phases. For example, reconstructed patients with HTAs delay their resisted hamstring exercise [19,22]. Therefore, studying the impact of each type of graft on rehabilitation protocols is warranted.
Rapid RTS is the desire for all athletes [25] and one of the most commonly used endpoints to evaluate the effectiveness of surgical techniques. In this sense, a systematic review and meta-analysis by Ardern et al. [26] reported that between 62% and 81% of injured athletes returned to their pre-injury level after surgery. Additionally, 44% returned to their competitive level [26]. Furthermore, RTS criteria are multifactorial and include time after surgery, muscle strength, and functional outcomes [9,22,27]. Functional outcomes are considered subjective methods to evaluate patient progress after rehabilitation protocols (e.g., the Lysholm, Tegner, International Knee Documentation Committee IDKC, and Cincinnati scores) [9,22,28]. Among all RTS criteria, muscle strength is the most important considered criterion [9,29], and the most commonly used methods for its evaluation are manual muscle testing [30], isometric strength tests [30] and isokinetic strength tests [31]. Furthermore, the isokinetic technology is considered "the gold standard" method for evaluating muscle strength, allowing the quantification of muscle strength through the determined angular velocity [31,32].
Zemach et al. [33] and Ypici et al. [34] reported that the use of different angular velocities led to different results [33,34]. Accordingly, Undheim et al. mentioned that no clear standardized isokinetic testing protocol was used in the published articles, limiting the quantitative comparison among published data [35]. Therefore, the homogeneity of the patient evaluation time points, isokinetic strength test protocol and instrument used is complex.
The current literature has revealed many studies comparing HTAs vs. PTAs. Kurz et al. [16] included 17 meta-analyses in their study comparing HTAs vs. PTAs regarding muscle strength, functional outcomes, and knee stability, among others [16]. Adam et al. [22] mentioned no differences in time to RTS between HTAs and PTAs. Additionally, previous systematic reviews compared QTAs vs. HTAs and PTAs and reported similar functional outcomes and better knee stability results with the QTA group. The current literature has revealed no evidence regarding RTS and strength recovery with QTAs vs. HTAs or PTAs. In this respect, a recent systematic review and metaanalysis by Johnston et al. [36] compared isokinetic and isometric tests between QTAs and PTAs or HTAs using the categorical angular velocity (low: 60 • /s-90 • /s and moderate: 160 • /s-180 • /s) and categorical follow-up periods (5-8, 9-15, 24, and 36-60 months). However, they did not compare a determined angular velocity or specific follow-up time points. Furthermore, no functional outcomes or knee stability were reported in the mentioned meta-analysis [36].
Therefore, given the substantial scientific evidence, as well as the difficulties posed by the meta-analyses mentioned above, we hypothesized that: (1) there would be a statistically significant differences between QTAs and HTAs or PTAs in the isokinetic strength test after ACL reconstruction; and (2) the current literature would show a difference between QTAs and HTAs or PTAs regarding knee stability and functional outcomes at the same follow-up points. Thus, this systematic review and meta-analysis aimed primarily to compare the isokinetic strength test of the quadriceps and hamstring muscles between QTAs and HTAs or PTAs after ACL reconstruction using the isokinetic angular velocity and follow-up time points. Additionally, as a secondary objective, we aimed to compare functional outcomes and knee stability at the same time points with isokinetic strength tests.

Protocol and Registration
This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Supplementary File S1) [37]. A detailed protocol for the systematic review was registered in the International Prospective Registry of Systematic Reviews (PROSPERO). It can be accessed with the code CRD42020191849. According to the PRISMA guidelines, the specific question posed for this review was, "which tendon autograft for anterior cruciate ligament reconstruction is better for strength recovery in athletes?".

Study Eligibility
Studies were selected for inclusion based on the following criteria: (1) comparative studies; (2) participants aged between 16 and 45 years who had undergone ACL reconstruction surgery with a tendon autograft; (3) strength assessment using the isokinetic strength test; (4) accessible online full text (in any case, consideration was given to contacting the authors if access to the full text online was not available); and (5) studies published in English or Spanish. Studies such as reviews, case reports, monographs, guidelines, surveys, commentaries, conference papers and/or unpublished data were excluded, as well as studies performed on animals or in vitro.

Literature Search
The comprehensive search occurred between January and March 2021 in the Medline (via PubMed search engine), Scopus, Web of Science and Cochrane Library databases using the following search terms ((ACL reconstruction OR ACLR) AND (Quadriceps autograft OR quadriceps tendon OR QT) AND (isokinetic dynamometer OR isokinetic test)). To select information, the descriptors used were obtained from the Medical Subjects Heading (MeSH) database. The information was filtered using terms and keywords related to ACL reconstruction and rehabilitation procedures, combined with Boolean operators and search techniques adapted to each database. Additionally, the reference lists of retrieved reports were manually searched for additional references. The search equation was developed and replicated by two independent researchers (F.H. and C.F.-L.) autonomously and independently to ensure the reliability of the results.

Study Selection and Data Abstraction
This systematic review was developed independently by two authors (F.H. and C.F.-L.), who screened by title and abstract first and then by full text. Studies were evaluated in both phases according to the eligibility criteria mentioned above. If disagreement occurred between the reviewers, a third external reviewer (M.L.-L.) participated to decide whether to include or exclude the article. When completing both screenings, the search strategy was re-executed if additional studies were added to the literature and were retrieved for inclusion (latest search released on 1 March 2021).
The data abstraction process was performed by two researchers (F.H. and C.F.-L.). One first selected the data, and then the other verified this selection for accuracy. If any disagreement occurred, a third researcher (M.L.-L.) was asked to make a final decision. The collected data items were as follows: (1) first author; (2) year of publication; (3) study design; (4) clinical entity responsible for the study; (5) sample size; (6) type of intervention(s); (7) if applicable, details of control or comparison groups; and (8) main findings.

Risk of Bias Assessment
The risk of bias of the included studies in this systematic review was determined by two independent reviewers (F.H. and M.L.-L.) and was evaluated using specific scales depending on the type of study, following the instructions given by the Cochrane Handbook for Systematic Reviews of Intervention [38] and the National Institutes of Health (NIH) [39].
Randomized controlled trials (RCTs) were evaluated using the Revised Cochrane riskof-bias tool for randomized trials (RoB 2) [40]. This common tool is used for randomized trials and has been updated in the last year. It assesses bias in five distinct domains (e.g., randomization process, intended interventions, missing data, measurements, and results). Observational studies were evaluated using the Cochrane's tool Risk Of Bias In Non-Randomized Studies-of Interventions (ROBINS-I) with five level judgment criteria (low, moderate, serious, critical, and no information) for each domain. ROBINS-I tool assessed seven distinct domains (confounding, selection of participants, classification of interventions, deviations from intended, missing data, measurement of outcomes and selection of the reported results).

Data Analysis
To pool the results quantitatively and develop the proposed meta-analysis, as well as to generate corresponding forest plot graphs, STATA software was used (StataCorp. 2019; Stata Statistical Software: Release 16; StataCorp LLC, College Station, TX, USA). Only studies comparing the use of HTAs vs. QTAs or PTAs vs. QTAs and reporting valid isokinetic strength data obtained using an isokinetic dynamometer were included in this quantitative combination. Thus, a total of five studies were included in the metaanalysis [41][42][43][44][45]. The data were obtained from the tables or text of the articles, extracting the means and standard deviations (SD) of the follow-up values at 6 months, 12 months or 24 months. Where these data were not reported, the mean and SD were calculated from the available data based on the protocol previously published by Wan et al. [46]. The included studies were combined according to the follow-up, speed used (60 • /s or 180 • /s) and movement employed (knee flexion or knee extension). A random effects model of the DerSimonian and Laird method, which considers variations within and between studies, was used. Forest plots were developed to visualize individual study summaries and pooled estimates. Cochran's Q statistic and the I 2 value were used to study heterogeneity between studies. Cohen's D was calculated for each of the original studies and an overall estimator, and a two-sided p value < 0.05 was considered statistically significant. Because of the low number of studies (<10), a more in-depth study of publication bias was not possible.

Risk of Bias
Two independent authors evaluated the risk of bias using the Cochrane risk-of-bias tool for randomized trials (RoB 2) for three randomized control trials [40] and Risk of Bias In Non-Randomized Studies of Interventions (ROBINS-I) for seven nonrandomized studies [50]. Two of three studies showed a high risk of bias, and the highest risk was in the "deviation from intended intervention" domain ( Figure 2). However, most of the nonrandomized studies had a serious risk of bias, and the highest risk of bias was found in the "bias due to confounding" domain (

Risk of Bias
Two independent authors evaluated the risk of bias using the Cochrane risk-oftool for randomized trials (RoB 2) for three randomized control trials [40] and Risk of In Non-Randomized Studies of Interventions (ROBINS-I) for seven nonrandomized s ies [50]. Two of three studies showed a high risk of bias, and the highest risk was in "deviation from intended intervention" domain ( Figure 2). However, most of the non domized studies had a serious risk of bias, and the highest risk of bias was found in "bias due to confounding" domain ( Table 2).

Combined Outcomes
All the included investigations reported patient post-surgery outcomes with sim follow-up time points (one trial post 3 months [44], five studies post 6 months [14

Three Months
Martin-Alguacil et al. [44] compared QTAs vs. HTAs 3 months after surgery and showed a significant increase in the quadriceps isokinetic strength (QIS) test in favor of the QTA group. No significant differences were reported between the groups in the hamstring isokinetic strength (HIS) test, functional outcome, or anteroposterior laxity.
Regarding functional outcomes, two studies compared QTAs vs. HTAs. Martin-Alguacil et al. [44] showed no significant differences between the QTA and HTA groups. By contrast, Cavaignac et al. [48] showed significant differences between the QTA and HTA groups. No study has compared QTAs vs. PTAs regarding functional outcomes. Concerning knee stability, two studies compared QTAs vs. HTAs [44,48]. Cavaignac et al. reported better and significant results within the QTA group. However, Martin-Alguacil et al. [44] reported no differences between the groups. Additionally, only the study of Pigozzi et al. [14] reported knee stability between QTAs and HTAs and showed no significant differences between the autografts.

Twenty-Four Months
The fourth evaluation point was approximately 24 months after surgery [15,43]. Only Lee et al. [15] compared QTAs vs. HTAs and reported no significant differences in the QIS test and significant differences in the HIS test for the QTA group. Similarly, Han et al. [43] compared QTA vs. PTA and mentioned no significant differences in the QIS test. Finally,

Twenty-Four Months
The fourth evaluation point was approximately 24 months after surgery [15,43]. Only Lee et al. [15] compared QTAs vs. HTAs and reported no significant differences in the QIS test and significant differences in the HIS test for the QTA group. Similarly, Han et al. [43] compared QTA vs. PTA and mentioned no significant differences in the QIS test. Finally, no significant differences were found in functional outcomes or knee stability (1,3) between QTAs and HTAs [15,44] or between QTAs and PTAs [43].

Return to Sport and Rehabilitation Protocols
Return to sport evaluation was different in the reviewed articles. Seven of ten studies described their rehabilitation protocol and return to sport criteria [14,15,[42][43][44][45]47]. Post rehabilitation timing was mentioned in four studies [14,15,44]. Three studies had a 6-month accelerated rehabilitation program [14,15,44]. And one project had a 12-month nonaccelerated rehabilitation program [47]. However, 80% to 90% quadriceps strength recovery is considered a criterion to recover full activity and return to sport [14,43,47].

Discussion
The main purpose of this systematic review and meta-analysis was to compare isokinetic strength tests, functional outcomes, and knee stability between QTAs and HTAs or PTAs after ACL reconstruction. Furthermore, this systematic review and meta-analysis added further quantitative analysis to previous systematic reviews [51,52] and included more studies than previously published studies [51,52]. Overall, 754 patients were evaluated from ten studies, and five of ten studies were included in the meta-analysis. The results suggest that ACL reconstructed patients with QTA showed better isokinetic strength results in the short term (e.g., 3 and 6 months). Additionally, they showed similar isokinetic strength results in the long term (e.g., 12 and 24 months) to HTAs and PTAs. Finally, our results showed similar results in functional outcomes and knee stability during short-and long-term evaluations between QTAs and HTAs or PTAs.
Comparing the isokinetic strength test between QTAs and HTAs or PTA, our results were similar to previous systematic reviews [36,51,52]. Additionally, our results were similar to a previous meta-analysis by Johnston et al. [36], where QTAs showed better isokinetic strength results during the short-term evaluation and similar results during the long-term evaluation. However, Johnston et al. [36] compared the isokinetic strength test using the categorical angular velocity (low: 60 • /s-90 • /s; moderate: 160 • /s-180 • /s) and categorical follow-up periods (5-8, 9-15, 24, and 36-60 months). We compared a determined angular velocity (60 • /s or 180 • /s) and determined follow-up time points (3, 6, 12 and 24 months). Furthermore, we could not compare the heterogeneity between the mentioned meta-analysis and our study because it was not reported. Additionally, they also compared only the peak torque of the LSI and did not compare that of the injured limb. In our study, we compared the peak torque from the injured limb and that of the LSI, revealing that the peak torque results for the uninjured limb contrast the peak torque results of the LSI [36]. Furthermore, the studies from Martin-Alguacil et al. [44] and Undheim et al. [35] have shown that the use of different angular velocity lead to statistical different results, which were not considered by the author of the previous meta-analysis [36]. Moreover, the mentioned meta-analysis has used downs and black scale to evaluate the risk of bias of the selected studies. This tool has been considered numerical quality assessment scale and, in our study, we have used RoB 2 and ROBINS-I from Cochrane Handbook for Systematic Reviews of Intervention. Five studies were excluded in the meta-analysis [14,43,[47][48][49]. The main reasons were that Pigozzi et al. [14] did not report the isokinetic test angular velocity and Cavaignac et al. [48] reported the isokinetic angular velocity at 90 • /s, preventing the formation of a meta-analysis group with other studies. Three studies were excluded because their follow-up time points did not form any meta-analysis group [43,47,49]. Finally, no meta-analysis subgroup comparing QTAs vs. PTAs was introduced because of the variations in the testing protocols or follow-up time points.
Regarding functional outcomes, similar to our results, a systematic review and metaanalysis by Hurly et al. [51] showed no significant differences between QTAs and HTAs or PTA. Additionally, Hurly et al. [51] reported functional outcomes with a mean of 24 months for HTA and 36 months for PTA, our results included the first 24 months post-surgery. However, this systematic review and meta-analysis did not match Ajrawat et al. [53] or Belk et al. [11], who reported better functional outcomes with QTAs vs. HTAs. This difference may be because both reviews included mostly nonrandomized or cohort studies. Finally, the functional outcome scores were similar among QTAs, HTAs and PTAs, but drawing a strong conclusion might be difficult because the data were reported using different functional outcomes (Lysholm, Tegner, IDKC, and Cincinnati scores) and different time points (6, 12, 26, and 36 months) [15,[43][44][45]47,48].
Restoring knee stability is considered an important purpose of ACL reconstruction [54]. QTAs showed better knee stability than HTAs in a previous systematic review by Belk et al. [11]. However, the current systematic review and meta-analysis showed no difference in knee stability between QTAs and HTAs or PTAs. The systematic review by Belk et al. included eight studies (1 Level II, 7 Level III), none of which had a randomized controlled clinical trial RCT design. We showed similar results to the previous systematic review and meta-analysis by Mouarbes et al. [55] that analyzed 12 studies; of those, seven compared QTAs vs. HTAs, and five compared QTAs vs. HTAs. They reported no significant differences among the reconstructed autografts [55]. Additionally, knee stability might be affected by several factors in addition to the autograph type, such as the screw type, surgical procedure and aggressive postoperative rehabilitation [54]. Similar to the functional outcomes, a meta-analysis was not performed because of few matched studies [14,15,43,44,48].
This systematic review and meta-analysis showed some limitations. First, only five studies were included in the meta-analysis because of their methodological differences. Second, all the studies were included in the review despite their methodological characteristics. Third, the search, limited to English and Spanish languages, led to a potential publication bias. Fourth, some ACL reconstruction outcomes were not analyzed, such as the one-legged hope test or graft failure, because of the high variation among the studies. However, this study showed some strength points, reported according to the PRISMA guidelines. A risk of bias assessment was included, and a meta-analysis with low statistical heterogeneity was obtained because of the inclusion of determined subgroups.
We propose a standardized isokinetic strength test to ensure the comparison between further studies, as previous authors have recommended [35]. Such tests included five repetitions for knee flexion and another five for knee extension with one minute of rest between each test. Two angular velocities should be applied starting at 60 • /s and then 180 • /s. Additionally, patients should be seated with 85 degrees of hip flexion and 90 degrees of knee flexion. Furthermore, to ensure the comparison between testing protocols, standardized time points for evaluation (6, 12, and 24 months after surgery) may be useful. Indeed, all the tests may be applied to injured and uninjured limbs, allowing the examiner to report data on one limb and LSI.

Conclusions
This systematic review and meta-analysis adds further quantitative data analysis to previously published systematic reviews. The QTAs showed better and significant results in HIS compared with HTAs and similar results to PTAs at 3, 6 and 12 months. While HTAs showed a better and significant result in QIS at 6 months and similar results at 12 months compared to QTAs. This review showed similar results between QTAs and HTAs or PTA in functional outcomes and knee stability. Furthermore, a standardized isokinetic strength test must be followed to achieve a more specific conclusion and better clinical comparison among participants.  Data Availability Statement: Raw data from data analysis are available upon reasonable request by contacting the corresponding author.