ABSTRACT
Introduction Anterior cruciate ligament (ACL) reconstruction is one of the most commonly performed sports medicine procedures. A variety of grafts are currently used for reconstruction, including both allograft and autograft. Despite numerous meta-analyses, there exists no high-quality quantitative synthesis of all randomized controlled trial (RCT) data on graft choice.
Objective To identify the optimal graft choice for ACL reconstruction by performing the first systematic review and network meta-analysis (NMA) to include both functional outcomes and complications.
Methods Multiple digital databases including MEDLINE, Embase, and CENTRAL will be searched independently and in duplicate for RCTs randomizing graft choice in ACL reconstruction in skeletally mature patients. A Bayesian framework with a random-effects model will be used for NMA. Surface under the cumulative ranking curve (SUCRA) values will be used to generate a rank list for each outcome. Results will be reported as mean differences (MD) (or standardized mean difference, if necessary) or relative risk (RR) with 95% credible intervals (CI). Comparisons will be inferred to be statistically significant if the 95% CI of MD does not cross zero or if the 95% CI of relative risk does not cross one. Studies will be assessed for quality using the Cochrane risk of bias assessment tool. Quality of evidence for each network comparison will be determined as per the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) approach for network meta-analyses. This NMA will be reported according to the PRISMA extension statement for network meta-analyses
Outcomes of interest Functional outcomes of interest including range of motion, return to activity/sport, and IKDC, Lysholm, Tegner, ACL-QOL, and KOOS scores. Persistent laxity as measured by Lachman, Pivot-shift, side-to-side, and measured laxity (e.g. KT-1000) will also be analyzed. Complications (e.g. infection, graft failure, donor site pain), tunnel osteolysis, and failure (including but not limited to graft rupture and/or persistent laxity) will be compared between grafts.
Relevance/Impact This NMA will be the first high-quality syntheses of all randomized evidence regarding graft choice in ACL reconstruction. As the first analysis to compare all major graft choices independently, it will be used to inform surgeon-patient decision making. It has the reasonable possibility of changing clinical practice.
Introduction
Anterior cruciate ligament (ACL) reconstruction is one of the most commonly performed sports medicine procedures and the incidence of this procedure continues to increase.1,2 A variety of grafts have been used for reconstruction, including both allograft and autograft.3,4 Commonly used graft choices include bone-patellar tendon-bone autograft (BPTB), hamstrings autograft (HT), quadriceps tendon autograft (QT), and tibial tendon allograft (TT), while hybrid grafts— autograft augmented with allograft—have received recent attention.2,3,5 Each graft is associated with a unique functional and complication profile, often leaving surgeons and patients with the task of choosing grafts on an individualized basis.4 In fact, modern graft preference has remained fairly consistent with the exception of QT, which has seen a recent increase.3,5 Despite numerous meta-analyses and large prospective knee ligament registries, there is no clearly preferred graft, demonstrated by the frequent use of numerous different grafts.3 Currently, there is no high-quality quantitative synthesis of all randomized controlled trial (RCT) data on graft choice. Conventional meta-analyses are limited to comparison of two groups, necessitating either exclusion of commonly used grafts or grouping of different graft types. Network meta-analysis (NMA) generates multiple simultaneous comparisons utilizing both direct and indirect evidence.6,7 Accordingly, the objective of this study is to identify the optimal graft choice for ACL reconstruction by performing the first systematic review and NMA to include both functional outcomes and complications.
Methods
This systematic review and network meta-analysis will be performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) for network meta-analyses8,9 and the Cochrane Handbook for Systematic Reviews of Interventions.10 A PRISMA checklist for the reporting of network meta-analyses will be included as a supplementary material in the final publication.
Literature search
A comprehensive literature search will be conducted using the following databases: Medline, Embase, CINAHL, Web of Science, Cochrane Central Register of Controlled Trials (CENTRAL), Cochrane Library, WHO International Clinical Trials Registry Platform (ICTRP), and ClinicalTrials.gov. This search will be limited to randomized controlled trials and will be conducted without language restrictions. An example search strategy for MEDLINE is provided in Table 2. The database search will be supplemented by manually screening of references from included articles and previous systematic reviews.
Inclusion criteria and limits
Search strategy for the Medline electronic database using Ovid interface for MEDLINE from 1946 to present.
Cochrane risk of bias evaluation criteria
PRISMA Systematic Review and Network Meta-Analysis Reporting Checklist
Screening
Articles will be screened by two investigators, independently and in duplicate. Disagreements between reviewers will be resolved by consensus, and if necessary, consultation with a senior reviewer. For studies published more than once, only the article with the most complete data will be included. For articles published in a non-English language, a medical translator fluent in the language of the study will translate and assist with article screening if possible.
Exclusion criteria will include trials randomizing interventions other than graft choice (e.g. rehabilitation protocol, graft fixation method), observational studies, case series, case reports, biomechanical cadaver studies, review articles, letters, basic science studies, correspondences or comments. Trials examining revision reconstruction or those including skeletally immature patients will be excluded.
Data Extraction
Articles produced from the literature search will be downloaded as complete reference files. Two reviewers will independently extract data in parallel from all included articles into a piloted, standard extraction document (Microsoft Excel 16.2, Redmond, United States) designed a priori. Article authors will be contacted in instances where additional information or clarity is required. The following study details will be extracted: year of publication, study design, recruitment period, country of recruitment, trial funder, role of trial funder, number of patients, number of knees, patient age, sex, and activity level, graft fixation method, concurrent procedures at the time of ACL reconstruction, post-operative protocol, and follow-up duration.
Outcomes of Interest
Outcomes of interest include range of motion, return to activity/sport, subjective and objective International Knee Documentation Committee (IKDC)11,12, Lysholm13,14, ACL-Quality of Life (ACL-QOL)2, Knee Injury and Osteoarthritis Outcome (KOOS)15, and Tegner scores14,16, results of Lachman and Pivot-shift tests, side-to-side, measured laxity (e.g. KT-1000), complications (e.g. infection, graft failure, donor site pain), tunnel osteolysis, and failure (including but not limited to graft rupture and/or persistent laxity).
If range of motion is reported only as a percentage of the uninjured knee, a degree measurement will be calculated using published normal values.17,18 Means and standard deviations (SDs) will be collected; medians will be used in lieu of means if mean values are not reported.19 If a 95% confidence interval was reported as the measure of variability, standard deviation will be approximated.20 When no measure of variance is reported, standard deviation will be imputed using a p value (if reported exactly) or a weighted average of variances observed in other included studies.20,21
Statistical Analysis
Statistical analysis was performed using R 3.4.2 (Open Access Online) with BUGSnet (Lighthouse Outcomes, Toronto, Canada) and CINeMA.22 Heterogeneity between studies will be calculated using the I2 statistic; if heterogeneity is high, a Bayesian framework with a random-effects model and non-informative priors will be used. A graphical framework of all trials comparing different interventions will be created for each outcome. Ranking diagrams and forest plots will also be created for each outcome. Furthermore, surface under the cumulative ranking curve (SUCRA) values will be reported for each study. The SURCRA score represents the likelihood that a given treatment will rank first in a specific category; a score closer to one indicates that treatment is more likely to represent the best treatment. Treatment rank orders will be generated and presented using SUCRA values. Incoherence—inconsistency between direct and indirect evidence—will be assessed globally using the design-by-treatment interaction test and individually using the Separating Indirect from Direct Evidence (SIDE) and node-splitting methods.23,24 If appropriate, a nested analysis will be conducted; the first level of analysis will compare autograft to allograft while the second level of analysis will look at each graft independently.
Results of the network for functional outcomes will be reported as mean differences (MD) (or standardized mean difference, if necessary) with 95% credible intervals (CI). Complications will be presented using relative risk (RR), 95% credible intervals, and number needed to treat, as appropriate. Credible intervals are derived using the posterior distribution of the outcome in question and can be thought of as the Bayesian equivalent of confidence intervals. Comparisons will be inferred as statistically significant if the 95% CI of MD does not cross zero or if the 95% CI of relative risk does not cross one.
Quality Assessment
The quality of each included study will be evaluated in duplicate using the Cochrane Risk of Bias tool, while the Cochrane CINeMA tool, designed around the GRADE framework, will be used for risk of bias assessment specific to outcomes from network meta-analyses.22,25 Included RCTs will be assessed for quality by two independent reviewers using the Cochrane risk of bias assessment tool.26 Disagreement will be resolved through consultation with a third reviewer. Overall quality of evidence for each network comparison will be determined and ranked based on within-study bias, reporting bias, indirectness, imprecision, heterogeneity, and incoherence, as per the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) approach for network meta-analyses.25,27
Data Availability
Not applicable.
ETHICS AND DISSEMINATION
Approval from a research ethics board is not required as this study will synthesize data from conducted studies. Results from this study are expected to comprehensively summarize and inform clinical practice. The results from this review will be submitted to a peer-reviewed journal for publication and will be presented at conferences.
AUTHORS’ CONTRIBUTIONS
TW, DA, and DdS conceived and designed the protocol. AZ and JH designed the search strategy and piloted it across all relevant databases. TW, DA, and DdS developed the review protocol, selection criteria, and risk of bias assessment. TW and DA designed the data management and synthesis methodology. AZ and JH performed data abstraction. All authors critically reviewed and collaborated in the discussion of the intellectual content of the protocol. All authors provided final approval of the protocol.
FUNDING STATEMENT
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
COMPETING INTERESTS STATEMENT
DdS is a member of the International Quadriceps Tendon Interest Group and is also a lead investigator for an ongoing RCT comparing soft-tissue quadriceps and hamstrings autografts in pediatric ACL reconstruction. No other authors have conflicts of interest to disclose.
Footnotes
This work was performed at the Division of Orthopaedic Surgery, McMaster University, Hamilton, ON, Canada.
T Woolnough, Michael G DeGroote School of Medicine, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S4L8, Taylor.woolnough{at}medportal.ca
References
