Tear of the anterior cruciate ligament (ACL) is the most common ligamentous injury of the knee. Reconstructing this ligament is often required to restore functional stability of the knee. Many graft options are available for ACL reconstruction, including different autograft and allograft tissues. Autografts include bone-patellar tendon-bone composites (PT), combined semitendinosus and gracilis hamstring tendons (HT), and quadriceps tendon. Allograft options include the same types of tendons harvested from donors, in addition to Achilles and tibialis tendons. Tissue-engineered anterior cruciate grafts are not yet available for clinical use, but may become a feasible alternative in the future. The purpose of this systematic review is to assess whether one of the popular grafts (PT and HT) is preferable for reconstructing the ACL. For this objective, the authors selected only true level I studies that compared these graft choices in functional clinical outcomes, failure rates, and other objective parameters following reconstruction of the ACL. In addition, this review discusses mechanical considerations related to different allograft tissues.
Tear of the anterior cruciate ligament (ACL) is the most common ligamentous injury of the knee. Reconstructing this ligament is often required to restore functional stability of the knee. Despite the popularity of the procedure, the preferred graft remains controversial. Ideally, the graft should have similar characteristics as the native ACL. Regardless of graft type, the biologic and mechanical properties of the graft material should provide a favorable setting for early biologic incorporation, be amenable to secure fixation, and limit potential morbidity related to donor site.
Many graft options are available for ACL reconstruction, including different autograft and allograft tissues. Autografts include bone-patellar tendon-bone composites (PT), combined semitendinosus and gracilis hamstring tendons (HT), and quadriceps tendon. Allograft options include the same types of tendons harvested from donors, in addition to Achilles and tibialis tendons. Tissue-engineered anterior cruciate grafts are not yet available for clinical use, but may become a feasible alternative in the future.
For the past few decades, PT autograft has been the gold standard for ACL reconstruction. Reasons for this include the strength of the tissue, relative ease of harvest, and bone-to-bone healing with secure fixation. More recently, HT autografts have joined PT in surgeons’ popularity. The recent trend toward increased use of HT resulted from concerns with use of PT relating to a potential negative effect on the knee extensor mechanism and donor site morbidity, including anterior knee pain and risk for patella fracture. Nevertheless, despite their increasing popularity, HT grafts also have potential limitations, including slower soft-tissue graft-tunnel healing compared with bone-to-bone healing with PT grafts, potential for tunnel widening and graft laxity, and functional hamstring weakness resulting from graft harvesting.
There are several randomized controlled trials (RCTs) in the literature comparing the two most popular graft choices, PT and HT, either used as autografts or allografts. Many of the systematic reviews and meta-analyses in the literature that investigate graft choice for ACL reconstruction are biased by their inclusion of inadequately randomized trials that are not true level I studies. Also, functional outcomes, rather than graft failure, tend to be the focus of these reviews. The authors believe, however, that graft failure represents a critically important outcome measure in ACL reconstruction, which has not been given enough attention in previous systematic reviews and meta-analyses. The purpose of this systematic review is to assess whether one of the popular grafts (PT and HT) is preferable for reconstructing the ACL. For this objective the authors selected only true level I studies that compared these graft choices in functional clinical outcomes, failure rates, and other objective parameters following reconstruction of the ACL.
A systematic literature review was performed using the following data sources: MEDLINE with OVID and PubMed (basic search, related articles, clinical queries search), EMBASE, and the Cochrane Central Register of Controlled Trials for relevant articles in the English language. Bibliographies of the identified articles on this topic were also reviewed. In addition, a manual search of recent pertinent hard copy journals from the previous 6 months was undertaken to identify journal articles that may not yet have been included in electronic databases.
Initial inclusion criteria included prospective RCTs, meta-analyses of RCTs, studies comparing PT and HT, either autografts or allografts, for ACL reconstruction, minimum of 2-year follow-up after the reconstruction for RCTs but not for meta-analyses of RCTs, no restrictions on date of publication or publication status. Following this initial search the inclusion criteria were further refined to include, in addition to the above criteria, only properly randomized trials comparing 2-strand HT or 4-strand HT with PT autografts. The criteria for proper randomization were strict to avoid any potential selection bias. Proper randomization techniques included random numbers table, computer-generated randomization, and randomly ordered sealed envelopes. Trials using even and odd birth years/months, patient registration numbers, or another alternating sequence of allocation were excluded because of inadequate randomization and the associated potential bias.
All studies identified in the initial search were screened for duplications by entering them into a computer-based reference management system. All eligible articles were then screened first by title and abstract, followed by an in-depth review of the methodology and outcomes. The results of this search are shown in Tables 1 and 2 , which include studies with proper randomization techniques and those that were quasi-randomized. Following this, the authors limited the review further to studies with appropriate randomization only, as described earlier. A standardized data extraction form was modified and used to retrieve data from each article on study design, population, interventions, and outcomes. Outcomes of particular interest included return to preinjury level of activity, graft failure rate, donor site morbidity, laxity measurements, knee range of motion, isokinetic muscle strength, and standardized knee outcomes scores. The authors defined graft failure rate as either revision ACL reconstruction or a 2-plus positive pivot shift test. KT-1000 measurements were not included as a criterion for failure because of variability in testing and because the pivot shift test is associated with function, whereas the KT-1000 is not.
|Study||Year of Publication||Sample Size (N) (% Follow-up)||Mean Follow-up (Months)||Number of HT Strands||Method of Fixation|
|Mean Age (Years) (Range)||PT||HT|
|Aglietti et al||1994||NA||60 (95)||28||4||IfSc||ScW||ScW||ScW|
|Aglietti et al||2004||25 (16–39)||25 (15–39)||120 (100)||24||4||IfSc||S||ScW||Sc|
|Anderson et al||2001||23.6 (14–44)||21 (14–40)||68 (97)||35||2||St||IfSc||Su||St|
|Beynnon et al||2002||29.2 (18–46)||44 (79)||36||2||IfSc||IfSc||St||St|
|Biau et al c||2007||NA||1263 (NA)||NA||2, 3, 4 or 5||Variable||Variable||Variable||Variable|
|Ejerhed et al||2003||26 (14–49)||29 (15–59)||66 (93)||24||3 or 4||IfSc||IfSc||IfSc||IfSc|
|Eriksson et al||2001||25.7||154 (94)||33||4||IfSc||IfSc||Sc||Eb|
|Feller & Webster||2003||26.3||25.8||57 (88)||36||4||IfSc||Eb||Post||Eb|
|Grontvedt et al b||1996||26 (16–48)||92 (92)||24||0||IfSc||IfSc + St||IfSc + St||IfSc + St|
|Harilainen et al||2006||31||79 (80)||60||4||IfSc||IfSc||ScW||P|
|Ibrahim et al||2005||22.3 (17–34)||85 (77)||81||4||IfSc||Eb||ScW, P + St||P|
|Jansson et al||2003||NA||89 (90)||24||4||IfSc||IfSc||ScW||P|
|Laxdal et al||2005||28 (16–52)||25 (12–41)||118 (88)||26||3 or 4||IfSc||IfSc||IfSc||IfSc|
|Liden et al||2007||28 (14–49)||29 (15–59)||68 (96)||86||3 or 4||IfSc||IfSc||IfSc||IfSc|
|Maletis et al||2007||27.2 (15–42)||27.7 (14–48)||96 (97)||24||4||IfSc||IfSc||2 IfSc||IfSc|
|Marder et al||1991||21.6 (16–35)||23.8 (17–41)||72 (90)||29||4||PW||PW||PW||PW|
|Matsumoto et al||2006||23.7||24.4||72 (90)||87||5||IfSc||IfSc||IfSc||IfSc|
|Moyen et al b||1992||24||24||64 (64)||36||0||St||St||St||St|
|Muren et al b||2003||25 (20–33)||25 (19–44)||40 (100)||84||0||Su||Post + Su||Su||ScW|
|O’Neill||1996||27 (14–56)||125 (98)||42||2||IfSc or St||IfSc||St||St|
|O’Neill||2001||NA||225 (95)||102||2||IfSc or St||IfSc||St||St|
|Sajovic et al||2006||27 (16–46)||24 (14–42)||54 (84)||60||4||IfSc||IfSc||IfSc||IfSc|
|Shaieb et al||2002||32 (14–48)||30 (14–53)||70 (85)||33||4||IfSc||IfSc||IfSc||IfSc|
|Sun et al a||2009||29.7 (16–59)||30.1 (20–63)||65 (96)||31||0||IfSc||IfSc||IfSc||IfSc|
|Sun et al a||2009||31.7 (20–54)||32.8 (19–65)||156 (93)||67||0||IfSc||IfSc||IfSc||IfSc|
|Taylor et al||2009||21.7 (18–37)||22.1 (17–44)||53 (83)||36||4||IfSc + ScW||IfSc + Eb||IfSc + ScW||IfSc + Eb|
|Webster et al||2001||26||27||61 (94)||24||4||IfSc||Eb||Post + Su||Eb|
|Zafagnini et al||2006||30.5 (22–47)||29 (15–49)||75 (100)||60||2 or 4||IfSc||IfSc||IfSc ± St||Eb ± St|
|Study||Randomization Method||Selection Bias||Performance Bias||Detection Bias||Attrition Bias|
|Aglietti et al||Alternating sequence||+||−||+||−|
|Aglietti et al||Alternating sequence||+||−||−||−|
|Anderson et al||Computer-generated||−||−||+||−|
|Beynnon et al||Random numbers table||−||−||+||+|
|Biau et al c||Variable||+||−||−||−|
|Ejerhed et al||Sealed envelopes||+||−||−||−|
|Eriksson et al||NA||+||−||−||−|
|Grontvedt et al b||Sealed envelopes||+||−||+||−|
|Harilainen et al||Even/odd birth year||+||−||+||+|
|Ibrahim et al||Even/odd birth year||+||−||+||+|
|Jansson et al||Even/odd birth year||+||−||+||+|
|Laxdal et al||Sealed envelopes||+||−||−||+|
|Liden et al||Sealed envelopes||+||−||−||+|
|Maletis et al||Computer-generated||+||−||−||−|
|Marder et al||Alternating sequence||+||−||+||−|
|Matsumoto et al||Even/odd birth year||+||−||+||−|
|Moyen et al b||Drawing of lots||−||−||+||+|
|Muren et al b||Random sealed envelopes||−||−||+||−|
|O’Neill||Birth month allocation||+||−||+||−|
|O’Neill||Birth month allocation||+||−||+||−|
|Sajovic et al||Even/odd registration number||−||−||+||−|
|Shaieb et al||Even/odd birth year||+||−||−||−|
|Sun et al a||Computer-generated||−||−||+||−|
|Sun et al a||Computer-generated||−||−||+||−|
|Taylor et al||Random sealed envelopes||−||−||−||−|
|Webster et al||Computer-generated||−||−||+||−|
|Zafagnini et al||Alternating sequence||+||−||+||−|
The quality of the studies, including internal and external validity, was appraised using the items contained in the CONSORT Statement: Revised Recommendations for Improving the Quality of Reports of Parallel-Group Randomized Trials. Furthermore, each study was assessed for the 4 main biases affecting method quality: selection bias, performance bias, detection bias, and attrition bias.
General Description of Studies
Twenty-eight studies published between 1991 and 2009 (27 prospective RCTs and 1 meta-analysis of RCTs) met the initial inclusion criteria (see Table 1 ). The data for each study were collected using a worksheet developed by the authors. The basic details of the studies are shown in Table 1 , including sample sizes, length of follow-up, and methods of fixation of the grafts. Of the 28 studies, 23 prospectively compared PT autografts with 2-, 3-, 4-, or 5-strand semitendinosus and gracilis (HT) composite autografts (including 1 meta-analysis of studies comparing PT autografts with HT autografts of varying sizes). Three studies compared PT autografts with PT autografts augmented by the Kennedy ligament augmentation device (KLAD), and two studies compared PT autografts with PT fresh-frozen allografts, which were γ-irradiated in 1 study and nonirradiated in the other.
Study Design Appraisal
The presence of the 4 main biases affecting study quality, and the treatment allocation methods used in each of the studies, are shown in Table 2 . Each of the studies in the initial stage of this review allocated patients during the same period in a prospective random fashion, either by computer-generated random models or via quasi-randomized allocation methods (ie, birth date, alternating sequence, sealed envelopes that were not randomly ordered).
Detection bias can be minimized by blinding patients and investigators at follow-up evaluations. No patient in any study was blinded to the type of graft they received, but several independent follow-up evaluations were performed by blinded investigators, and the outcomes of the treatment groups in each study were assessed in identical fashion, thereby minimizing detection bias (see Table 2 ).
Attrition bias pertains to loss of patients from treatment groups after allocation, by either late exclusion or lost to follow-up. As shown in Table 2 , several studies excluded patients after treatment allocation, but 1 study had more than 30% lost to follow-up, which has been reported as the threshold for acceptable follow-up, with less than 20% being preferable. Another study had 23% lost to follow-up. Attrition bias was also prevalent in another study, and in its subsequent study with longer follow-up, in which data from 4 graft failures (all in the HT group) were excluded in the final analysis. This finding may represent a systematic exclusion of data that could potentially overestimate the favorable results in the HT group. Graft failures were similarly excluded from final data analysis in another study. As graft failure rate is a critical outcome following ACL reconstruction, methodology that excludes failures from the final analysis limits the value of the conclusions reached in these studies.
To improve the validity of our conclusions, our analysis of functional clinical outcomes, failure rates, and other objective parameters was limited to RCTs that had proper randomization (ie, computer-based, random numbers table, random sealed envelopes), and 80% follow-up at a minimum of 2 years follow-up. Also, as discussed earlier, trials comparing PT with HT were required to use HT composites of 2- or 4-strand quadrupled grafts only. Studies not meeting these strict criteria were excluded from all subsequent analyses, leaving 6 of 28 studies. Of these, 4 studies compared PT autografts with 4-strand HT autografts, and two studies compared PT autografts with 2-strand HT autografts. These 6 studies served as the basis for our analyses and subsequent conclusions. One study had a follow-up rate of 79% and we elected to include it.