Anterior cruciate ligament tears are common among ath-letes, and although their true natural history remains unclear, these injuries are functionally disabling; they pre-dispose the knee to subsequent injuries and the early onset of osteoarthritis. This article, the second of a 2-part series, was initiated with the use of the PubMed database (http://www.nlm.nih.gov) and a comprehensive search of publications that appeared between January 1994 and the present, with anterior cruciate ligament as keywords. A total of 3810 citations were identified and reviewed to determine the current state of knowledge regarding the treatment of ACL injuries. Part 1 focused on studies that pertained to the biomechanical behavior of the ACL, the prevalence of and risk factors for ACL injuries, the natural history of the ACL-deficient knee, injuries associated with ACL disruption, indications for treatment of ACL injuries, and nonoperative and operative treatments. Part 2 includes technical aspects of ACL surgery, bone tunnel widening, graft healing, rehabilitation after ACL recon-struction, and the effects of sex, age, and activity level on the outcome of ACL surgery. Our approach was to build on
prior outstanding reviews41,59,102 and to provide an overview of the literature for each of the aforementioned areas of study by summarizing the highest level of scien-tific evidence available. For the areas that required a descriptive approach to research, we focused on the prospective studies that were available; for the areas that required an experimental approach, we focused on the prospective, randomized, controlled trials (RCTs) and when otherwise necessary, the highest level of evidence available on the subjects presented. As with part 1, we were pleasantly surprised to learn that considerable advances were made during the past decade with regard to the treatment of this devastating injury.
TECHNICAL ASPECTS OF ACL RECONSTRUCTION
Graft Position (Single-Tunnel Technique)
The ACL has a complex, 3-dimensional attachment to bone. The femoral insertion of the ACL does not insert on a flat area that is aligned in an anatomical plane, as many publications have suggested; rather, it is located on a curved surface, with the wall of the femoral notch becom-ing the roof of the femoral notch. Our review of the litera-ture revealed that different approaches for characterizing the location of tunnel position in relation to anatomical landmarks (eg, the location of bone tunnels, or an ACL graft fixed within bone tunnels, in relation to anatomical landmarks) were a source of confusion. These different
Treatment of Anterior Cruciate Ligament
Injuries, Part 2
Bruce D. Beynnon,*
†PhD, Robert J. Johnson,
†MD, Joseph A. Abate,
†MD,
Braden C. Fleming,
‡PhD, and Claude E. Nichols,
†MD
From the
†Department of Orthopaedics and Rehabilitation, McClure Musculoskeletal Research
Center, University of Vermont, Burlington, Vermont, and the
‡Department of Orthopaedic
Research, Brown Medical School, Providence, Rhode Island
Anterior cruciate ligament tears, common among athletes, are functionally disabling; they predispose the knee to subsequent injuries and the early onset of osteoarthritis. A total of 3810 studies published between January 1994 and the present were iden-tified and reviewed to determine the current state of knowledge regarding the treatment of anterior cruciate ligament injuries. Part 1 of this article focused on studies pertaining to the biomechanical behavior of the anterior cruciate ligament, the preva-lence of and risk factors for injuries related to it, the natural history of the ligament-deficient knee, injuries associated with ante-rior cruciate ligament disruption, indications for the treatment of anteante-rior cruciate ligament injuries, as well as nonoperative and operative treatments. Part 2 includes technical aspects of anterior cruciate ligament surgery, bone tunnel widening, graft heal-ing, rehabilitation after anterior cruciate ligament reconstruction, and the effects of sex, age, and activity level on the outcome of such reconstructive surgery.
Keywords:anterior cruciate ligament (ACL); knee; reconstruction
1751
*Address correspondence to Bruce D. Beynnon, PhD, University of Vermont College of Medicine, Department of Orthopaedics and Rehabilitation, Stafford Hall, Room 438A, Burlington, VT 05405-0084 (e-mail: bruce.beynnon@uvm.edu).
No potential conflict of interest declared.
The American Journal of Sports Medicine, Vol. 33, No. 11 DOI: 10.1177/0363546505279922
approaches stem in part from the fact that arthroscopic visualization of the knee occurs when it is flexed, whereas standard anatomical nomenclature is referenced to the fully extended knee.7
Although most orthopaedic surgeons would agree that graft position is a critical variable affecting the outcome of ACL reconstruction, our review found no prospective RCTs that studied the superiority of one position as opposed to another. This finding is not surprising because it would be unethical to perform a superiority or equivalency trial involving the placement of a graft in any location other than where it is considered optimal. Consequently, the effect of graft position on clinical outcome has been, and probably will remain, based on prospective studies that are descriptive in design.
Our review of the literature revealed 2 prospective stud-ies that evaluated the relationship between ACL graft insertion site and outcome.45,66In the well-designed study by Good et al45of bone–patellar tendon–bone (BPTB) ACL reconstruction, the center of the femoral tunnel was meas-ured from a lateral radiograph of the knee, was expressed as a percentage of the total condylar depth measured along a line constructed parallel to the Blumensaat line, and was then referenced to the center of the anatomical insertion of the ACL.44Two years after surgery, subjects with an ACL graft insertion greater than 2 mm anterior to the center of the anatomical femoral insertion had signifi-cantly greater anterior-posterior (A-P) knee laxity values than subjects with centrally or posteriorly positioned femoral tunnels. Khalfayan et al66built on the earlier work of Good et al45by measuring the tibial and femoral tunnel positions of central third BPTB grafts with the use of lat-eral and anteroposterior view radiographs; they also reported that graft tunnel position has a direct effect on clinical outcome. A femoral tunnel placed too far anteriorly was found to limit knee flexion or produce graft attrition. Position of the tibial tunnel is also important. Anterior placement causes graft impingement against the roof of the femoral notch when the knee is extended, with per-sistent flexion contracture or graft attrition and subse-quent failure. It is important for us to point out that most of what is known about the effect of graft impingement on outcome has been derived from retrospective studies, and this issue is a concern because when making a diagnosis of graft impingement after failure of ACL reconstruction, one must consider that anterior subluxation of the tibia rela-tive to the femur may give a false impression of impinge-ment.5 Howell and Clark53 performed a retrospective study with a 6-month follow-up period of BPTB grafts and reported that knee stability and extension were signifi-cantly better when the center of the tibial tunnel was 2 to 3 mm posterior to the center of the normal ACL insertion on the tibia. In a subsequent retrospective study, Howell et al54reported that a tibial tunnel angle of 75°or more in the coronal plane was associated with greater loss of flex-ion and increased anterior knee laxity. A tibial tunnel angle between 65°and 70°was recommended.
Our group evaluated the relationship between the elon-gation behavior of central third BPTB grafts produced by flexion-extension of the knee at the time of surgery (a
measurement that is dependent on the location of the graft’s insertion to bone) and clinical outcome.16 Graft elongation values produced by knee flexion-extension motion, at the time of surgery, outside the 95% confidence limits of the normal ACL resulted in significant increases in anterior knee laxity at 5-year follow-up, whereas grafts with elongation values similar to the normal ACL had A-P knee laxity values similar to the normal knee.19This find-ing suggests that it is not only important to restore A-P laxity to within normal limits at the time of surgery, it is also important to consider the biomechanical behavior of the graft at the time of surgery.
Arthroscopic visualization, combined with modern drill guides, allows experienced surgeons to identify where they want to locate bone tunnels relative to landmarks such as the ACL “footprint,” the “over-the-top” position, and the clock face (eg, 2 o’clock) positions. Even with the use of these advanced tools, placing the tibial and femoral bone tunnels in desired locations without creating impingement of the graft against the femoral notch is a challenging task. This point is well emphasized by the work of Kohn et al,68who studied a series of cadaveric knees after endo-scopic ACL reconstruction with a BPTB graft. Only 17% of the knees were considered to have correct tunnel ment without graft impingement. Femoral tunnel place-ment was considered excellent, acceptable, or unaccept-able in 17%, 33%, and 50% of the knees, respectively. Similarly, tibial tunnel placement was classified as excel-lent, acceptable, or unacceptable in 42%, 33%, and 25% of the knees, correspondingly.
The position of an ACL graft is the most critical surgical variable because it has a direct effect on knee biomechan-ics and, ultimately, on clinical outcome. Our review of the literature revealed that the position of an ACL graft has been measured with 2-dimensional, radiographically based approaches. When the radiographic technique is obtained in a standardized manner—for example, with the posterior aspects of the femoral condyles superimposed in a lateral view—and measurements are made relative to reproducible bone-fixed coordinate systems, this approach has the advantage of characterizing graft position in a manner that is clinically relevant and that can be repro-duced in biomechanical studies. However, an accurate descrip-tion of an ACL graft’s posidescrip-tion requires a 3-dimensional technique that is applied to make measurements relative to standardized bone-fixed coordinate systems. Although such technology has recently become available in the form of electromagnetic and video-based position tracking sys-tems, there has been no clinical study reported that has determined the relationship between the 3-dimensional position of an ACL graft and clinical outcome. Currently, there are limited data available from prospective studies that identify the optimal intra-articular position of an ACL graft on the femur and tibia. There is, however, con-siderable lower-level evidence that derives from retrospec-tive studies. Our assessment of the literature is that the center of the femoral attachment of an ACL graft should be located along a line parallel to the Blumensaat line just posterior to the center of the normal ACL’s insertion to bone and at either the 2 o’clock position (left knee) or the
10 o’clock position (right knee) when observed through the femoral notch. The tibial tunnel should be placed to avoid impingement of the graft against the roof of the femoral notch as the knee is brought into extension.
Graft Tension at the Time of Fixation
Most surgeons would agree that the initial tension applied to an ACL graft at the time of fixation has a direct effect on outcome; an undertensioned graft may result in abnor-mal knee laxity and an unstable knee, and an overten-sioned graft may lead to graft failure, fixation failure, or restricted range of knee motion. Our review revealed 4 prospective RCTs that evaluated the effect of this critical surgical variable on clinical and functional outcomes.
In an investigation by Yasuda et al,114 subjects were randomized into 3 groups with initial graft tensions of 20, 40, and 80 N applied to a 4-strand hamstring graft connected in series with polyester tape and fixed extra-articularly with double staples. Follow-up measurements were made at a mean interval of 2.5 years (range, 2.0-3.5 years) and included 94% of the subjects. The 6% of patients who were lost to follow-up were evenly divided between the 20-N and 40-N tension groups; all of the subjects in the 80-N initial tension group returned for follow-up. A significant correla-tion was found between the initial tension applied to the graft at the time of fixation and A-P knee laxity measure-ments after healing was complete. Subjects in the high-tension group had anterior laxity values closer to normal compared with similar values in the low-tension group. There were, however, no differences between the initial tension groups with regard to knee motion and clinical outcome (Noyes scale85). The same surgeon performed all reconstruction procedures, yet the method of randomiza-tion was not described, A-P knee laxity was not measured immediately after graft fixation to establish baseline laxi-ty values for each treatment group, and there was no description of whether the examiner responsible for mak-ing the follow-up measurements was blinded to the treat-ment groups.114
A prospective RCT was performed by van Kampen et al107to determine the effect of tensioning BPTB grafts at 20 and 40 N with the knee in 20°of flexion. The details of how the randomization was performed were not pre-sented. Surgery was performed by 2 surgeons using the same single-incision procedure, and all subjects followed the same rehabilitation program, which included immedi-ate full range of motion, weightbearing as tolerimmedi-ated, and return to sport 9 months after surgery. Follow-up meas-urements were made at 1 year and included all study par-ticipants; there was, however, no description of whether the patients and the researcher making the follow-up measurements were blinded to the treatment groups. Immediately after surgery (baseline) and at the 1-year follow-up interval, there was no difference in A-P laxity between the 2 treatment groups. It may be that the differ-ence in the 2 tension levels was not of sufficient magnitude to create differences in knee laxity at baseline or that a large proportion of the tension applied to the bone block was lost at the bone block–tunnel interface. From this
per-spective, the observation of similar knee laxity between the 2 initial graft tension groups at the 1-year follow-up may be attributed, at least in part, to the fact that the ini-tial tension groups had similar knee laxity values at base-line.107
Yoshiya et al115 performed a similar prospective RCT comparing 2 groups in which initial tensions of 25 and 50 N were applied to BPTB grafts with the knee in full exten-sion. Follow-ups were performed immediately after sur-gery (KT-1000 arthrometer measurement) and at 3, 6, 12, and 24 months. Among the 50 subjects enrolled in the study, 88% of those in the 25-N initial tension group and 84% of those in the 50-N initial tension group returned for all follow-up visits. Immediately after surgery and at the 2-year follow-up, there was no difference in A-P knee laxity between the 2 groups. Similarly, no differences were found between the treatment groups in terms of knee motion, isokinetic thigh muscle strength, and International Knee Documentation Committee (IKDC) rating. Two surgeons performed the procedures; however, the method used to perform the randomization was not presented, and there was no description of whether the patients and examiners were blinded to the treatment groups. As with the earlier study by van Kampen et al,107 the observation of similar knee laxity values between the 2 initial tension groups after healing may be explained by the fact that laxity values for the 2 groups were the same at baseline.115
Nicholas et al82 performed a prospective, randomized, double-blind clinical trail that compared 2 groups in which initial tension loads of 45 and 90 N were applied by the same surgeon to central third BPTB grafts with the knee in full extension. Knee motion and A-P knee laxity were measured before surgery and at 1 week and 20 months after surgery. A total of 49 subjects enrolled in the study; 100% of those in the 45-N initial tension group and 82% of those in the 90-N initial tension group returned for all follow-up visits. Subjects receiving the 45-N tension load had increased anterior knee laxity, whereas patients receiving the 90-N tension load had laxity values similar to normal. At follow-up, 23% of the subjects in the 45-N ini-tial tension group had side-to-side differences in knee lax-ity greater than 5 mm, whereas none of the subjects in the 90-N initial tension group showed such abnormal changes. Subjects in both groups had similar range of motion.
Our review revealed that 1 RCT of ACL reconstruction with a BPTB graft compared the effect of preloading (application of a 39-N tensile load for 10 minutes before graft implantation) to no preloading before graft implan-tation.32Two years after surgery, there were no differences between the treatment groups with regard to activity level, clinical outcome (IKDC grade), and A-P knee laxity.
Graft tensioning at the time of fixation involves consid-eration of the knee’s position during the tensioning proce-dure (eg, the flexion angle and internal-external rotational position of the tibia relative to the femur) and, of course, the magnitude of tension applied to the graft at the time of its fixation to bone. Both variables interact and have a direct effect on knee biomechanics. It was difficult to com-pare the RCTs that were reviewed and to develop a
con-sensus because different tensioning procedures and graft materials were used. If a BPTB graft is used and the ten-sioning procedure is performed with the knee in extension, application of high tension (90 N) appears to produce more normal A-P knee laxity values compared to application of low tension (45 N). Similarly, if a 4-strand hamstring graft is used, application of high tension (80 N) appears to pro-duce A-P knee laxity values similar to normal, whereas low tension (40 N and less) results in increased anterior knee laxity. The effect that graft tensioning at the time of fixation has on the contact stress distribution of healing articular cartilage about the tibiofemoral joint is currently unknown and requires study.
Graft Fixation
Substitutes for the ACL can be prepared as either bone-tendon-bone grafts or tendon grafts. Depending on the graft material that is selected for ACL reconstruction, the bony or soft tissue portions of these constructs can be fixed either within bone tunnels or externally to cortical bone. From a biomechanical perspective, fixation represents the weak link during the early stages of healing. The long-term goal is to obtain biological incorporation of the graft at the anatomical attachment site of the ACL and to restore the transition from soft tissue to fibrocartilage, to calcified fibrocartilage, and to bone.
This review did not focus on the techniques and associ-ated biomechanical characteristics of the wide array of fix-ation devices used in cruciate ligament surgery; this topic has been covered in outstanding review articles by Brand et al22and Wilson et al.111Instead, we focused on the 3 prospec-tive RCTs of ACL graft fixation that have been published. Aglietti et al,3performing an RCT of ACL reconstruction with a central third BPTB graft, compared 2 different types of tibial fixation. Graft fixations in the tibial tunnel were performed on 2 groups of subjects, with an interfer-ence screw placed either at the level of the tibial plateau (aperture fixation) or distal to the plateau. The same fixa-tion device was used in the femoral tunnels for both groups. The ACL reconstruction was performed by the same surgeon, randomization was performed using an alternating scheme, all subjects underwent the same reha-bilitation program, and follow-up was performed by an independent examiner at a mean interval of 18 months. There were no differences between the treatments in range of motion, A-P laxity, symptoms, and subjective eval-uation of outcome. Tibial tunnel enlargement was less fre-quent in the group that underwent graft fixation at the level of the tibial plateau (23% vs 43%); however, tibial tuberosity pain, attributed to harvesting a longer tibial bone block, was more frequent in this group.
Presenting an RCT of ACL reconstruction with a central third BPTB graft, Fink and colleagues39compared 2 dif-ferent types of femoral fixation. Subjects were randomized to undergo graft fixation in the femoral tunnel with either a titanium interference screw or a polyglyconate (a copoly-mer of polyglycolic acid and trimethylene carbonate) bioabsorbable interference screw. The same metal interfer-ence fit fixation was used in the tibial tunnels of both
groups. Two surgeons performed the single-incision proce-dures, all subjects observed the same rehabilitation proto-col, and follow-up evaluations were made over a 2-year period. Graft fixation with the bioabsorbable screw pro-duced the same clinical outcome and A-P knee laxity val-ues compared to graft fixation with the metal screw. Complete degradation of the bioabsorbable screw was apparent at 1 year, and replacement of the screw with bone occurred by 3 years. A similar investigation by the same group revealed that fixation of BPTB grafts with a polyglyconate interference screw was safe and effective compared to fixation with a titanium interference screw.15 Our review of the literature on ACL graft fixation revealed that the fixation of BPTB grafts with interference screws is adequate for the loads produced by current reha-bilitation programs. In contrast, the frequent publication of new techniques to fix free-tendon grafts, such as the 4-strand hamstring graft, suggests that an optimal tech-nique of fixation has yet to be identified, particularly when rehabilitation includes immediate weightbearing, early use of the quadriceps muscles with the knee near exten-sion, and early return to sport.
BONE TUNNEL WIDENING
Jansson et al57performed an RCT that studied ACL recon-struction using a 4-strand hamstring graft fixed proximally with an EndoButton and distally with a screw and a spiked washer, compared to reconstruction using a central third BPTB graft fixed with interference screws in the tibia and femur. Subjects were randomized by birth year, given the same rehabilitation protocol, and followed up over 2 years. No details were presented with regard to the number of surgeons who performed the procedures or whether the follow-up measurements were made by an independent examiner. No differences were observed between the groups in A-P knee laxity values, clinical find-ings, and knee scores. At the 2-year follow-up, however, the femoral and tibial tunnel diameters detected on antero-posterior view radiographs for subjects who underwent reconstruction with the 4-strand hamstring graft had increased by means of 33% and 23%, respectively. Although the increases in tunnel diameters were consid-erable, the 2 graft materials and corresponding fixation methods were reported to produce similar outcomes.57
Webster et al108carried out an RCT, over a 2-year inter-val, that focused on whether tunnel enlargement in ACL reconstruction differed when performed with a central third BPTB or 4-strand hamstring graft. Subjects were randomized via a computer-based random-number gener-ator to undergo ACL reconstruction with either a central third BPTB graft (fixed proximally to the femur with an EndoButton and in the tibial tunnels with a metal inter-ference screw) or a 4-strand hamstring graft (fixed proxi-mally to the femur with an EndoButton and to the tibia with suture tied to a fixation post). All subjects had the same single-incision procedure performed by the same surgeon, they all observed the same postoperative rehabil-itation program, and 94% of the subjects were followed up at 4 months, 1 year, and 2 years. The clinical outcome was
similar between the treatments; however, bone tunnel enlargement was more common and greater with the 4-strand hamstring graft. Eleven percent of the subjects who underwent ACL reconstruction with a BPTB graft had tunnel widening greater than 25%, compared to 94% of subjects who received a hamstring graft.108
Although it is clear that bone tunnel enlargement occurs more frequently and is greater after ACL reconstruction with hamstring grafts compared to BPTB grafts, the cause and clinical relevance of tunnel widening remain unclear. One explanation for the increased tunnel widening associ-ated with hamstring grafts secured with EndoButton fixa-tion has been offered by Jorgensen and Thomsen,61 who observed movement of the graft in the proximal two thirds of the tibial tunnel with cinematic MRI. Alternatively, L’Insalata et al71reported that the tunnel expansion asso-ciated with hamstring grafts may be produced by the greater distance between fixation points, in comparison to BPTB grafts, and the corresponding larger force-moment produced during graft cycling. To date, we are unaware of any study confirming that tunnel widening has an adverse effect on the outcome of ACL reconstruction. Perhaps such an effect will be found as the number of cases identified with this condition and the length of follow-up periods increase. Significantly enlarged bone tunnels, however, make revision ACL reconstruction more difficult.
GRAFT HEALING
Although it has been reported that BPTB autografts used to reconstruct the ACL in the rabbit model undergo bio-logical remodeling and incorporation after implantation (a process termed ligamentization6), the fully incorporated graft never replicates the normal ACL, and it appears to function as a check rein of organized scar tissue.28Oaks and associates87 have performed quantitative ultrastruc-tural morphometric analysis of collagen fibril populations in the ACL and patellar tendon grafts using the goat model. The graft remodeling process was found to change the ultrastructural profile of the original patellar tendon at the time of harvest to one containing a larger number of small-diameter fibrils (<100 nm). A rapid decrease in the number of large-diameter collagen fibers (>100 nm) was found after 12 weeks of healing. Remodeling was found to begin from the outside of the graft and then move toward the graft center as the remodeling progressed over time. Oaks et al87found this remodeling behavior to be consis-tent with revascularization of the graft from synovium, demonstrating the importance of not only investigating the surface and central portions of the graft but also studying different regions along the length of the graft. Remodeling was found to continue for as long as 52 weeks after reconstruction. Arnoczky et al9evaluated the tempo-ral revascularization behavior of the patellar tendon auto-graft using a canine model. They demonstrated that revas-cularization of the patellar tendon graft came from the bone tunnels and progressed from the proximal and distal regions to the central portion of the graft. After 5 months of healing, revascularization was complete.
The choice of graft material and the method of fixation affect the healing and remodeling response at the graft–bone tunnel interface, and this relationship has been studied in various animal models. Rodeo et al94used the canine model to study the tensile failure properties of digital extensor tendons fixed in bone tunnels. After 2, 4, and 8 weeks, graft failure occurred by pullout from the bone tunnels; after 12 and 26 weeks, the graft failed at its midsubstance. Grana et al46studied the healing response of a hamstring autograft used to reconstruct the ACL in rabbits. They found that hamstring grafts healed by for-mation of a fibrous insertion to bone, and the fixation strength of the bone-graft composite in the bone tunnel exceeded the intra-articular portion of the autograft strength early in the postoperative period. At 3 weeks, failure of the bone–hamstring graft–bone construct occurred at the intra-articular portion of the graft and not by pullout from the bone tunnels.46In a subsequent study of hamstring graft insertion-site healing in rabbits reported by the same group, the formation of the fibrous insertion was found to be complete after 26 weeks of healing.21 Tomita and colleagues105 compared intraosseous graft healing between a doubled flexor tendon graft and a BPTB graft used to reconstruct the ACL in canines. Tensile fail-ure testing revealed that the weakest site of the doubled flexor tendon graft was the graft–tunnel wall interface at 3 weeks and the intraosseously grafted tendon at 6 weeks. For the BPTB graft, the weakest site was the graft–tunnel wall interface at 3 weeks and the proximal site in the bone plug at 6 weeks.105 After 3 weeks, the ultimate failure strength of the doubled flexor tendon graft was 45% of the BPTB graft, and this value had increased to 85% at 6 weeks of healing.105
It is important to point out that experimental studies using the animal knee joint as a model are limited with respect to the lack of similarity to the human knee joint and that animals have an uncontrolled postoperative rehabilitation regimen. Although animal investigations have provided insight into graft remodeling and biome-chanical behavior during graft healing, direct application of these results to clinical practice must be made with cau-tion. For example, BPTB grafts used to reconstruct the ACL may not undergo the dramatic decrease in structural properties that have been reported from animal studies.17 This opinion is supported by a case study of a patient who underwent ACL reconstruction with a central third BPTB autograft.17 Eight months after surgery, the linear stiff-ness and ultimate failure load values of the graft approached those of the contralateral, normal ACL, whereas laxity of the injured knee was greater than the normal knee. More recently, Delay et al30described a case study of a central third BPTB autograft after 18 months of healing. They found complete osseous union of both tibial and femoral bone blocks, although deep areas of necrosis asso-ciated with bone graft remodeling were observed. The deep and superficial regions of the proximal half of the soft tis-sue graft had become revascularized, whereas the deep portion of the distal half of the graft had persistent regions of necrosis that were acellular and avascular. These find-ings are supported by the work of Rougraff and
Shelbourne,96who performed second-look arthroscopy and biopsy of patellar tendon grafts on 9 subjects after 3 to 8 weeks of healing. Although all specimens showed regions of acellularity and degeneration, the researchers observed that graft vascularity was present at 3 weeks and contin-ued to increase over the 8-week sample interval. Petersen and Laprell90 obtained biopsy specimens of BPTB and hamstring grafts from patients undergoing revision sur-gery. Both graft materials healed within the femoral and tibial tunnels, but the insertions were different. The inser-tion of the BPTB grafts to the bone tunnels healed by bone plug incorporation and resembled the chondral insertion of the normal ACL, whereas the hamstring grafts healed by the fibrils of the graft penetrating the bone directly and resulted in a fibrous insertion of the tendon, not the nor-mal chondral insertion of the ACL to the tibia or femur.
The temporal change in the structural and material properties of different autografts used for ACL reconstruc-tion has been investigated using primate, canine, goat, sheep, and rabbit models.81 Animal studies that have investigated the iliotibial tract (ITT) autograft for as long as 1 year indicate that the ultimate failure load values of this graft ranged between 23% and 40% of the control ACL, whereas the graft stiffness was 45% of the average uninjured ACL.81Studies in animals that have investigated the healing response of patellar tendon autografts a year or more after reconstruction have reported ultimate fail-ure load values ranging between 30% and 45% of the con-trol ACL, whereas the stiffness has been reported to range between 35% and 57% of the normal ACL.81Hunt et al55 used the superficial digital flexor tendon as an autograft to simulate hamstring reconstruction of the ACL in sheep. After 1 year, grafts fixed anatomically with interference screws in the tibia and femur had an ultimate failure load value that was 45% of the normal ACL.
It is not enough to evaluate the structural and material properties of an ACL graft; in addition to having adequate strength and stiffness, an ACL graft must also control anterior translation of the tibia relative to the femur and, to a lesser degree, internal-external and varus-valgus lax-ity of the knee joint. Butler25 used the canine model to investigate the A-P displacement response of the knee joint at 4 time intervals after a combined ACL reconstruc-tion using the fascia lata and lateral one third of the patel-lar tendon. At implantation, the A-P laxity of the operated knee was 154% of the control limb; after 4 weeks of heal-ing, this ratio increased to 306%. After 12 and 26 weeks of healing, the A-P laxity had decreased to 209% and 153% of the control limb, respectively.
Animal studies of ACL allografts have shown a slower rate of biological incorporation, a greater decrease in struc-tural properties, and a prolonged inflammatory response compared to ACL autografts.56 Human retrieval studies have shown that remodeling of ACL allografts is slow.52,74 After 2 years of transplantation, the central portions of the allografts were found to remain acellular. Complete remodeling and cellular replacement of the entire graft were seen in grafts studied 3 years after transplantation.
The healing of BPTB and hamstring grafts is different both temporally and histologically. Thus, aggressive
reha-bilitation within the first 6 weeks after ACL reconstruc-tion with a hamstring graft may cause greater anterior knee laxity compared to reconstruction with a BPTB graft. Attempts to improve healing by treating ACL autografts with growth factors and gene therapy are in the develop-ment stages, and although these approaches hold great promise,109,113there are no proven clinical applications for these procedures at the present time.
REHABILITATION AFTER ACL RECONSTRUCTION
Although most clinicians would agree that the strains applied to an ACL graft by body weight, muscle activity, and joint motion affect its healing response, there is little consensus on how these factors influence the biomechani-cal behavior of the healing graft and, in turn, how this behavior modulates the healing response of the graft, car-tilage, and knee.Our review of the literature did not identify a consensus regarding which variables should be used to characterize a rehabilitation program, nor did it reveal how this infor-mation should be used to compare different programs. Although it is clear that rehabilitation after ACL recon-struction includes characteristics such as the series of activities (or restrictions) that a subject is directed to per-form, the time when the activities are recommended, the duration of the activities (eg, the number or sets and rep-etitions per day, week, month), the overall length of the program, and the time when a subject is advised to return to sport-specific training and subsequently to sport, it is unclear how a single term such as “accelerated rehabilita-tion” can be used to provide insight into these characteris-tics. As a consequence, there is little consensus in the lit-erature about what composes an accelerated versus a more conservative rehabilitation program or an aggressive versus a nonaggressive approach to rehabilitation. Our review identified a substantial number of RCTs that have focused on rehabilitation after ACL reconstruction, but very few of these reports described the rehabilitation pro-tocol adequately or provided details with regard to subject compliance. These concerns made it difficult to arrive at a consensus with regard to optimal rehabilitation programs after ACL reconstruction.
Our review of RCTs of rehabilitation programs after ACL reconstruction was categorized according to the use of the following approaches: cold therapy, immediate versus delayed motion, immediate versus delayed weightbearing, closed versus open kinetic chain exercises, bracing, home-versus clinic-based rehabilitation, neuromuscular electri-cal stimulation versus voluntary muscle contraction, spe-cific exercise programs, and intensity and duration of rehabilitation.
Use of Cold Therapy Immediately After ACL
Reconstruction
We identified a prospective RCT by Konrath et al69 that focused on the effectiveness of postoperative cold therapy in patients undergoing ACL reconstruction. After ACL reconstruction with a BPTB graft, patients were
random-ized to receive 1 of 4 treatments: a cooling pad filled with water ranging between 40°F and 50°F, a cooling pad filled with water ranging between 70°F and 80°F, a bag of crushed ice, or no cold therapy. Both the cooling pad and crushed ice treatments were found to produce a significant decrease in knee temperature; however, there were no dif-ferences among the 4 treatments regarding duration of hospital stay, range of knee motion at the time of dis-charge, or the use of intramuscular and oral pain medica-tion.
Immediate Versus Delayed Motion
Our review identified 5 RCTs comparing immediate to delayed knee motion during the initial stages of rehabili-tation, and there appears to be reasonable consensus that immediate motion is beneficial for the healing ACL graft and soft tissue structures that span the knee.47,50,86,92,95
Haggmark and Eriksson were among the first to per-form a prospective RCT of rehabilitation after ACL recon-struction with a patellar tendon graft.35,47 Patients were treated with a dorsal plaster splint during the first week after surgery and were then randomly assigned to continue rehabilitation during the following 4 weeks while wearing either a hinged cast that allowed knee motion or an ordi-nary cylinder cast that prevented knee motion. All of the patients were followed up during a 1-year interval; those treated with standard cast immobilization had significant atrophy of the slow-twitch muscle fibers of the vastus lat-eralis, whereas those treated with the hinged cast and early motion demonstrated no changes in the cross-sec-tional area of the slow- or fast-twitch fibers. Haggmark and Eriksson47(p55)noted that “there appeared to be no dif-ference in the end result of the surgical procedure” and that treatment with the hinged cast “facilitated an early return to sports.”
A prospective RCT that compared immediate to delayed range of motion after ACL reconstruction was carried out by Noyes et al.86 Subjects in the immediate motion pro-gram began continuous passive motion of the knee on the second postoperative day, whereas those in the delayed motion group had their knees placed in a brace at 10°of flexion and began continuous passive motion on the sev-enth postoperative day. Subjects in both rehabilitation pro-grams reported similar rates of joint effusion, hemarthro-sis, soft tissue swelling, flexion and extension limits of the knee, use of pain medications, and time of stay in the hos-pital. Continuous passive knee motion immediately after ACL reconstruction did not lead to an increase in anterior knee laxity during healing.
Rosen et al95 carried out a prospective RCT of rehabili-tation after arthroscopically assisted ACL reconstruction with a central third BPTB autograft performed by the same surgeon. After surgery, subjects were randomized via a lottery system to 1 of 3 programs: early active motion, continuous passive motion, or a combination of both. This work extended the research of Noyes et al86by showing that continuous passive motion during the first month after ACL reconstruction, compared with early active
motion, produced similar range of joint motion and KT-1000 arthrometer measurements of A-P knee laxity.
Richmond et al92 reported the results of a prospective RCT that compared the effects of continuous passive knee motion for 4 to 14 days after arthroscopically assisted ACL reconstruction with a BPTB autograft. They found similar values for knee range of motion and lower limb girth between treatment groups.
More recently, Henriksson et al50described a prospective RCT of rehabilitation after ACL reconstruction with a BPTB graft performed by 1 of 4 surgeons using the same technique. After surgery, subjects were randomly assigned to rehabilitation protocols consisting of cast immobiliza-tion or early range of moimmobiliza-tion training with a brace. Subjects in both groups underwent similar supervised rehabilitation, and during the first 5 weeks, all rehabilita-tion exercises, with the exceprehabilita-tion of range of morehabilita-tion exer-cises, were the same for both treatments. Follow-up meas-urements made after 2 years included 88% and 92% of subjects in the brace and plaster cast treatment groups, respectively. The researchers found that rehabilitation with the use of a brace and early range of motion training after ACL reconstruction produced equivalent knee laxity, knee motion, subjective knee function, and activity level in comparison to rehabilitation with plaster cast immobiliza-tion for 5 weeks. There were, however, differences in terms of strength. At 2-year follow-up, subjects in the brace group had a larger strength deficit of the knee flexors (5.9% loss compared to the contralateral, normal side) in comparison to subjects in the plaster cast group (0.9% loss). As well, there was a strong trend for subjects in the brace group to have a strength deficit of the knee exten-sors (11.1% decrease compared to the contralateral side) in comparison to patients in the plaster cast group (3.8% decrease).
Of the 5 RCTs reviewed above, only Rosen et al95 ade-quately described their method of randomization, and only Haggmark and Eriksson47and Henriksson et al50had min-imal loss of patients at follow-up; no author stated whether the investigators were blinded at follow-up.
After ACL reconstruction, it is clear that extended immo-bilization of the knee, or limited motion without muscle activity, is detrimental (inferior structural and material properties) to the structures that surround the knee (liga-ments, cartilage, bone, and musculature).4,10,62-65,70,84,112 There is little doubt that early joint motion after ACL reconstruction is beneficial; it leads to a reduction in pain, lessens adverse changes in articular cartilage, and helps prevent the formation of scar and capsular contractions that have the potential to limit joint motion.24,65
Immediate Versus Delayed Weightbearing
Two prospective RCTs have compared immediate versus delayed weightbearing rehabilitation programs after ACL reconstruction, and both have reported that immedi-ate weightbearing programs produce similar clinical, patient, and functional outcomes to delayed weightbearing programs.60,106
Jorgensen et al60performed a prospective RCT to evalu-ate the effect of weightbearing on the results of ACL recon-struction with the iliotibial band graft. After surgery, sub-jects were randomized to undergo rehabilitation with either immediate weightbearing or nonweightbearing for 5 weeks followed by a gradual return to full weightbearing during the first 9 weeks of healing. Evaluation 2 years after surgery revealed no differences between the groups with regard to A-P knee laxity and patient activity level (evaluated with the Tegner and International Knee Documentation Committee [IKDC] scores).
In a subsequent prospective RCT of ACL reconstruction with a central third BPTB autograft, Tyler et al106 com-pared rehabilitation with immediate weightbearing to delayed weightbearing for 2 weeks. Only 2 subjects in each treatment group were lost to up. At a mean follow-up of 7.3 months, there were no differences between the treatments with regard to knee range of motion, vastus medialis oblique function, and A-P knee laxity (clinical examination and KT-1000 arthrometer measurement). However, patients treated with immediate weightbearing had a decreased incidence of anterior knee pain.
Authors of these RCTs did not describe their method of randomization, and they did not mention if the subjects or investigators responsible for the follow-up measurements were blinded to the treatments that were studied.
The findings from these investigations indicate that immediate weightbearing after ACL reconstruction does not produce excessive loads that permanently deform the graft or its fixation and suggest that immediate weight-bearing may be beneficial because it lowers the incidence of anterior knee pain. After ACL injury and reconstruc-tion, the effect of weightbearing on the healing response of injured articular cartilage or meniscus repair is currently unknown.
Closed Versus Open Kinetic Chain Rehabilitation
Our review revealed 3 randomized trials comparing the use of closed versus open kinetic chain exercises during ACL rehabilitation.Bynum et al26 performed a prospective RCT comparing open versus closed kinetic chain rehabilitation after ACL reconstruction with a central third BPTB autograft. Immediately after surgery, patients’ knees were placed in a rehabilitation brace that was adjusted to allow 0° through 90° of motion, and continuous passive motion from 0° to 60° of flexion was started. Rehabilitation was begun on the first postoperative day and for all patients included passive and active motions of the knee without external resistance. Partial weightbearing with the use of crutches was permitted, and subjects progressed to full weightbearing as tolerated. Patients were then random-ized via a computer-generated list of random numbers to either open or closed kinetic chain rehabilitation groups. Subjective and objective follow-up measurements were taken on 66% of patients 1 year after surgery; the examiner was blinded to the patients’ rehabilitation program. The subjects in the closed kinetic chain group had KT-1000
arthrometer measurements that were closer to normal, in addition to less anterior knee pain, earlier return to nor-mal daily activities, and greater satisfaction compared to the subjects in the open kinetic chain group. The authors attributed the decreased anterior knee pain in patients from the closed kinetic chain group to reduced patellofemoral reaction forces associated with exercises performed with the knee near extension, in contrast to the open kinetic chain treatment exercises, which were per-formed with the knee in a more flexed position.26
Mikkelsen et al77 reported the results of an RCT that compared closed kinetic chain to combined closed and open kinetic chain rehabilitation programs initiated 6 weeks after single-incision BPTB reconstruction of the ACL per-formed by 1 of 3 surgeons. The randomization procedure appeared to be designed to match patients with regard to age, gender, and type and level of physical activity, although the details of how this design was accomplished were not presented, and it is unclear whether the follow-up measurements were made with the investigator blinded to the treatment groups. Assessment at 6 months after surgery revealed that the addition of open kinetic chain exercises produced a significant improvement in quadri-ceps strength (evidenced by moderate improvements in extension torque), an earlier return to sport at the prein-jury level, and no effect on KT-1000 arthrometer measure-ments of A-P knee laxity.
Hooper et al51reported the results of a prospective RCT comparing closed to open kinetic chain rehabilitation exer-cises during the early phase of healing. Reconstruction was performed by 3 surgeons: 1 surgeon reconstructed the ACL with a ligament augmentation device, and the other 2 surgeons used a central third BPTB graft. Patients were assigned to the treatment groups using a block random-ization scheme; there were no details given with regard to whether the follow-up measurements were made with blinding of the examiner to the treatment groups. Two weeks after ACL surgery, the patients underwent a base-line gait analysis and were then randomly assigned to either closed or open kinetic chain rehabilitation exercises administered 3 times per week during a 4-week interval. At the 6-week follow-up, the patients underwent a second gait analysis. At this early stage of healing, no differences were found between the closed and open kinetic chain pro-grams in the gait variables associated with level walking, ascending stairs, and descending stairs; however, subjects in both groups had functional deficits in the involved side compared to their contralateral, normal side. Subsequent investigation of the same patients revealed no differences in anterior knee pain or knee laxity between treatment groups.79,80
In the 3 RCTs reviewed above, different open and closed kinetic chain rehabilitation programs were compared over different times, and it is therefore impossible to come to a consensus regarding the effectiveness of one approach compared to another. The report by Bynum and col-leagues26 had the longest follow-up interval. This study indicated that rehabilitation with closed kinetic chain exercises is more effective in terms of patient satisfaction,
reduced anterior knee pain, and earlier return to daily activities compared to programs that include open kinetic chain exercises.
Rehabilitation Braces
Three prospective RCTs have compared rehabilitation with and without the use of a rehabilitation brace.23,48,78
A prospective RCT of rehabilitation after ACL recon-struction with a central third BPTB graft was performed by Harilainen et al.48Subjects were randomized by their birth year to either the braced group (rehabilitation with a gradual increase in weightbearing and the use of a brace for 12 weeks postoperatively) or the unbraced group (reha-bilitation with crutch use for 2 weeks postoperatively fol-lowed by weightbearing as tolerated but without use of a brace). The number of surgeons involved with the investi-gation and the specific details of the rehabilitation pro-gram were not presented. The 1- and 2-year follow-up intervals included 100% and 93% of subjects enrolled, respectively, and these follow-ups revealed no differences between the treatments in activity level, joint laxity, and isokinetic thigh muscle strength.
Moller et al78performed a prospective RCT that focused on the use of a rehabilitation brace during the first 6 weeks after ACL reconstruction with a BPTB graft. The rehabilitation program included early weightbearing and lasted 6 months. Four surgeons were involved in the study (2 performed a single-incision procedure, 2 performed a 2-incision procedure). Subjects were randomized to undergo rehabilitation without a brace or with a brace during the initial 6 weeks of healing (the first 2 weeks during the day and night, the subsequent 4 weeks during the day). Ninety-eight percent of the subjects were evaluated at 6-month follow-up, and those who did not receive a rehabili-tation brace had a better Tegner activity score. However, at the 2-year follow-up, which included 90% of subjects enrolled, there were no differences between the treat-ments in subjective outcome (Lysholm score and visual analog scale measurements of pain, discomfort, and insta-bility), range of knee motion, functional performance (1-legged hop test), isokinetic strength, and A-P knee laxity.
A prospective RCT focusing on the use of a rehabilita-tion brace was also performed by Brandsson et al.23After single-incision ACL reconstruction with a central third BPTB graft, patients were randomized to undergo 6 months of rehabilitation without a brace or the same 6-month program with a rehabilitation brace during the ini-tial 3 weeks of healing. Follow-up measurements were per-formed by an independent investigator through a 2-year follow-up interval; the results included 92% and 80% of subjects in the braced and unbraced groups, respectively. Rehabilitation with a brace resulted in fewer problems with swelling, a lower prevalence of hemarthrosis and wound drainage, and less pain throughout the early recovery period compared to rehabilitation without a brace. At the 2-year follow-up, no differences were found between the treat-ments in terms of activity level (Tegner scale), IKDC rat-ing, function (1-legged hop test and isokinetic strength), and knee laxity (KT-1000 arthrometer measurement).
In the 3 randomized trials described above, only Harilainen et al48 described their method of randomiza-tion; only Brandsson et al23indicated that follow-up meas-urements were performed by an independent investigator, and it was unclear if this examiner was blinded to the treatment groups. All studies revealed a minimal loss of patients to follow-up.
There appears to be a consensus among investigators that, during the early phase of recovery, the use of a reha-bilitation brace results in fewer problems with swelling, lower prevalence of hemarthrosis and wound drainage, and less pain compared to rehabilitation without a brace; however, at longer-term follow-up, rehabilitation bracing does not appear to have an effect on clinical outcome, range of knee motion, subjective outcome, A-P knee laxity, activity level, or function (thigh muscle strength and 1-legged hop test). It should be mentioned that one of the primary reasons for using rehabilitation braces is to help prevent flexion contractures by maintaining full knee extension during the early phase of healing.
Functional Braces
The effect of a functional brace on the knee and ACL graft is determined by the brace attachment technique, brace design parameters, the brace-limb attachment interface, and the loading environment to which the braced knee is exposed. Our review revealed 2 RCTs that studied the effect of functional bracing on healing after ACL recon-struction with a BPTB graft.
McDevitt and associates75presented a prospective RCT comparing rehabilitation using functional bracing for 1 year to rehabilitation without bracing after ACL recon-struction with a BPTB graft. The patients were random-ized by random numbers or a coin toss. Both groups of 50 patients were treated for the first 3 weeks after surgery with a rehabilitation brace locked in extension. The brace was removed 2 to 3 times per day for range of motion activ-ities. In the functional brace group, the knee was mobilized gradually from 3 to 6 weeks with the rehabilitation brace used intermittently. Then, the patient was fitted for a func-tional brace at 6 weeks and was allowed full range of motion. The brace was worn full-time for the following 6 months and thereafter during all rigorous activities until 1 year after surgery. In the nonbraced group, bracing was discontinued after 3 weeks. Other details of the rehabili-tation protocol were not provided. Ninety-five percent of the patients were followed up for a minimum of 2 years (mean, 29 months). At the time of final follow-up, no dif-ferences were revealed between the groups in terms of A-P knee laxity, 1-legged hop distance, IKDC and Lysholm scores, range of motion, and isokinetic strength. Two braced subjects and 3 nonbraced subjects sustained rein-juries to their ACL graft. The authors concluded that there were no significant differences between the braced and nonbraced treatment groups.
In a prospective RCT, Risberg et al93compared rehabili-tation with functional bracing to rehabilirehabili-tation without bracing after ACL reconstruction with BPTB grafts in 60 patients. The number of surgeons was not provided; the
patients were randomized into braced and nonbraced groups by an unspecified method. The braced group was protected by a rehabilitation brace for 2 weeks, and a func-tional brace was used nearly full-time for the following 10 weeks. Thereafter, the functional brace was used as needed for sports. The nonbraced group had no brace at any time postoperatively. Otherwise, both groups followed the same postoperative rehabilitation protocol, which was described in detail. Ninety-three percent of the patients were fol-lowed up for 2 years. The authors found no evidence that bracing had an effect on knee joint laxity, range of motion, strength, functional knee tests, patient satisfaction, or pain at final follow-up. No evidence that bracing reduced the risk of new injury was observed.
The 2 studies reviewed in this section do not identify a compelling reason to use functional braces after ACL reconstruction.
Home- Versus Clinic-Based Rehabilitation
Three RCTs have compared home-based rehabilitation (limited supervision by a physical therapist during heal-ing) to clinic-based rehabilitation (supervision by a physi-cal therapist throughout the entire rehabilitation pro-gram), and there appears to be reasonable consensus that home-based programs produce similar outcomes compared to clinic-based programs.14,40,97
Schenck et al97 studied rehabilitation after 2-incision ACL reconstruction with a central third BPTB graft per-formed by the same surgeon. Subjects were randomized after surgery via a lottery drawing to rehabilitation with either a clinic-based program (mean of 14.2 visits to phys-ical therapy) or a home-based program that was monitored by a physical therapist (mean of 2.9 visits to physical ther-apy). All patients were examined at a minimum of 1 year after surgery. There were no differences in functional or subjective outcomes between the clinic- and home-based rehabilitation programs.
Fischer and colleagues40performed an RCT of rehabili-tation after ACL reconstruction with a BPTB graft per-formed with a 2-incision technique. Patients were ran-domized to undergo rehabilitation with either a home-based program that included a mean of 5 physical therapy visits (range, 3-7 visits) or a clinic-based program that included 20 physical therapy visits (range, 10-28 visits). Compliance with the rehabilitation programs was docu-mented with a training log. Details were not presented regarding the specific activities and restrictions associated with both programs; instead a goal-based approach was used, with the following temporal sequence of activities: restoration of range of motion, beginning strengthening, advanced strengthening, and improvement of agility and speed. All of the subjects in the home-based program and 96% of those in the clinic-based program (1 patient was excluded because of unanticipated foot surgery) were fol-lowed up for 6 months. There were no differences between the treatments with regard to clinical examinations (range of motion, thigh atrophy, knee laxity, and pivot-shift examinations), 1-legged hop test, and the health status questionnaire.
More recently, Beard and Dodd14reported the results of rehabilitation after ACL reconstruction with a central third BPTB graft performed by the same surgeon. During the first month after surgery, all patients participated in the same regimen. They were subsequently randomized to either home-based rehabilitation (attending physical ther-apy only for education, assessment, and monitoring of the treatment plan) or the same home-based program with supervision (attending physical therapy twice weekly). Randomization was performed with a computer-based random-number generator, the study subjects and investi-gator responsible for making the follow-up measurements were blinded to the treatment groups, and follow-up at 12 and 24 weeks was performed on 86% and 81% of subjects in the home and supervised programs, correspondingly. No differences were found between the treatments in terms of activity level, IKDC rating, function, muscle strength, and knee laxity. These findings led the authors to conclude that “supervised exercises, in addition to a home program, has minimal extra benefit for patients that have undergone ACL reconstruction.”14(p134)
In the 3 randomized trials reviewed above, both Schenck et al97and Beard and Dodd14described their methods of randomization; only Beard and Dodd14 stated that the study subjects and investigators were blinded to the treat-ments that were studied. However, all 3 studies revealed a minimal loss of subjects at follow-up.
It is important to point out that the studies we reviewed compared rehabilitation programs with different amounts of supervision and that none of the programs was com-pletely unsupervised. These reports suggest that rehabili-tation after ACL reconstruction need not be monitored by a physical therapist in a continuous manner, but attending physical therapy for the purpose of education, assessment, and monitoring of the treatment plan remains a critical aspect of a safe and effective rehabilitation program.
Neuromuscular Electrical Stimulation Versus
Voluntary Muscle Contraction
Snyder-Mackler and colleagues104 performed an RCT of rehabilitation after ACL reconstruction with either a semitendinosus tendon combined with a ligament aug-mentation device or a central third BPTB preparation. After surgery, patients were randomized to undergo reha-bilitation with neuromuscular electrical stimulation and volitional exercises or with volitional exercises alone. All subjects documented compliance with the use of a training log; follow-up measurements of gait and thigh muscle strength were made at 8 weeks. Patients who underwent rehabilitation incorporating combined volitional exercises and neuromuscular electrical stimulation had more nor-mal gait parameters and stronger quadriceps muscles compared to patients who underwent volitional exercises alone. Subsequent to this effort, Snyder-Mackler et al103 reported the results of a multicenter RCT of rehabilitation after ACL reconstruction with a variety of graft materials and surgical techniques. All patients participated in the same rehabilitation program 3 times per week for the first 6 weeks and were then randomized to have additional
treatments of either high-intensity neuromuscular electri-cal stimulation, high-level volitional exercises, low-intensity neuromuscular electrical stimulation, or combined high-and low-intensity neuromuscular electrical stimulation. Rehabilitation was monitored with training logs, and the investigators making the follow-up measurements were blinded to the treatment groups. The 4-week follow-up revealed that high-intensity neuromuscular electrical stimulation combined with volitional exercises was better at restoring extensor strength compared to volitional exer-cises alone.
The methods used to perform the randomization were not presented, and the proportion of subjects who were fol-lowed up was not described in either of the above-mentioned studies. In the more recent study,103the investigators mak-ing the follow-up measurements were blinded to the treat-ment groups that the subjects were assigned to, whereas there was no mention of whether this procedure was fol-lowed in the earlier study.104There appears to be consen-sus that rehabilitation with volitional exercises combined with neuromuscular electrical stimulation results in more normal gait parameters and better restoration of extensor strength compared to rehabilitation with volitional exer-cises alone.
Specific Exercise Programs
There have been 4 prospective RCTs focusing on the effect of specific exercises on thigh muscle strength after ACL reconstruction.20,49,72,76
The effect of adding isokinetic strength training to reha-bilitation programs after augmented repair or reconstruc-tion of the ACL was studied by Hehl et al.49Significant improvements in muscle strength were found with the addition of isokinetic strength training between the sev-enth and ninth weeks after ACL reconstruction; at the 6-month follow-up, there was no difference in knee joint laxity. Blanpied and associates20carried out an RCT of reha-bilitation after ACL reconstruction with a central third BPTB graft performed by the same surgeon. After surgery, subjects were randomized to a home-based rehabilitation program that included lateral slide exercises or the same home-based program without lateral slide exercises. All of the study subjects were followed up at 8- and 14-week intervals; results revealed that the group rehabilitated with the lateral slide exercises showed significant improvement in knee extension strength compared to patients rehabilitated without lateral slide exercises.
Meyers and colleagues76performed an RCT of rehabili-tation after ACL reconstruction with a central third BPTB graft performed by the same surgeon. All patients followed the same aggressive rehabilitation program for 4 weeks after surgery, at which time patients were randomized to undergo either 8 weeks of rehabilitation with stair climb-ing or 8 weeks of rehabilitation with cyclclimb-ing. Twelve weeks after surgery, there were no differences between the reha-bilitation programs with regard to isokinetic thigh muscle strength.
An RCT of rehabilitation after ACL reconstruction with a semitendinosus tendon was presented by Liu-Ambrose
et al.72 After surgery, patients underwent 12 weeks of rehabilitation with either isotonic strength training or proprioceptive training. Compliance with these programs was monitored through direct supervision and the use of training logs, and follow-up measurements were made over the same time interval. Follow-up included all 5 study participants in each group, a sample size that was chosen a priori based on a 10% difference in time to create peak torque between the treatments. At the end of the 12-week training period, subjects in both programs experienced similar improvements in function and subjective scores. Patients who underwent proprioceptive training experi-enced a greater increase in isokinetic torque compared to those who underwent strength training.
In the 4 randomized studies reviewed above, no authors adequately described their method of randomization, no authors stated whether the investigators were blinded at follow-up, and only Blanpied et al20 and Liu-Ambrose et al72 revealed that they had minimal loss of study patients at follow-up.
Intensity and Duration of Rehabilitation
Shelbourne and colleagues were one of the first groups to report that rehabilitation with immediate walking and full weightbearing, combined with early return to sport, was effective and safe.99,100,101 Shelbourne et al100studied the effect of rehabilitation on subjects undergoing ACL recon-struction with a BPTB graft. Follow-up at a mean interval of 4 years (range, 2-9 years) included 81% of patients treated. Patients were able to return to sport-specific activities at a mean of 6.2 weeks (range, 1-13 weeks) and to athletic competition at full capacity at a mean of 6.2 months (range, 2-18 months). This achievement was accomplished while only 2.6% of the subjects retore their ACL graft at a mean interval of 2.5 years postoperatively (range, 4-78 months).
The duration and intensity of rehabilitation after ACL reconstruction have been evaluated in 3 RCTs.18,34,42
Ekstrand34performed a prospective RCT of soccer ath-letes after ACL reconstruction and compared standard rehabilitation (a 6-month program) to extended rehabilita-tion (an 8-month program). At baseline and 1-year follow-up, there were no differences between the groups. There was, however, a trend for subjects in the extended treat-ment group to have more normal knee laxity compared to those in the standard group.34 Using the criteria of 90% quadriceps strength in comparison to the contralateral side and full range of joint motion, the subjects who par-ticipated in the standard program were allowed to return to sports at 6 months, whereas those in the extended pro-gram returned at 9 months.
Frosch et al42carried out an RCT comparing prolonged rehabilitation (2.5-hour sessions performed 3-5 times/wk) to a standard rehabilitation program (30-minute sessions performed 2-3 times/wk). Baseline data were not reported, and at the 1-year follow-up, subjects receiving the pro-longed program had better joint position sense and Lysholm scores and returned to work earlier than those receiving the standard program.
Recently, our group reported the results of a prospective RCT comparing accelerated versus delayed rehabilita-tion.18Patients who had ACL reconstruction with a BPTB graft were randomized to either accelerated rehabilitation (a 19-week program that allowed unrestricted weightbear-ing after 1 week, no brace use after 2 weeks, open kinetic chain exercises involving contraction of the quadriceps muscle group with the knee near extension [0°-45°] after 4 weeks, and return to preinjury activity at 24 weeks) or nonaccelerated rehabilitation (a 32-week program that included the same exercises prescribed over a delayed time interval, including unrestricted weightbearing after 3 weeks, no brace use after 4 weeks, open kinetic chain exer-cises involving contraction of the quadriceps muscles [0° -45°] at 12 weeks, and return to preinjury activity at 32 weeks). Compliance with the rehabilitation programs was monitored with training logs. At the time of surgery, and then 3, 6, 12, and 24 months later, measurements of A-P knee laxity (KT-1000 arthrometer), clinical assessment (IKDC evaluation), patient satisfaction (Knee Osteoarthritis Outcome Score), function (1-legged hop test), and cartilage metabolism (synovial fluid–based bio-markers of synthesis and cleavage of type II collagen, and turnover of aggrecan) were completed. At 2-year follow-up, there was no difference in the increase of anterior knee laxity between the 2 groups (a 2.2-mm vs 1.8-mm increase relative to the normal knee for the nonaccelerated and accelerated programs, respectively). The treatments were also similar in terms of clinical assessment, patient satis-faction, activity level, function, and the response of the synovial fluid biomarkers of articular cartilage metabo-lism. There was concern that the biomarkers from subjects in both groups remained elevated over periods consider-ably longer than modern rehabilitation programs and sub-stantially greater than the interval after which most peo-ple attempt to return to preinjury activities. Soon after injury and just before surgery, the levels of cleavage and synthesis of type II collagen and turnover of aggrecan were elevated compared to normal values. After 12 months of healing, cleavage of type II collagen returned to normal values, whereas synthesis of collagen and turnover of aggrecan remained elevated. Synthesis of type II collagen remained elevated at 24-month follow-up, whereas the turnover of aggrecan approached normal limits.
In the 3 RCTs reviewed above, only Beynnon et al18 ade-quately described their method of randomization, both Beynnon et al18and Frosch et al42stated that the investi-gators were blinded at follow-up, and all authors revealed a minimal loss of patients to follow-up.
Our review of the ACL rehabilitation literature revealed several concerns with the quality of the studies. Although most studies claimed to be based on prospective RCT designs, many reports did not describe how the random-ization was performed, it was often unclear if the assessors and study subjects were blinded to the treatment groups, in most reports the follow-up intervals were quite short, and in some studies the proportion of subjects lost to follow-up was not presented. Most of the randomized stud-ies that were reviewed clearly described the activitstud-ies (and restrictions) a subject was advised to perform and the time
that the activities were recommended. Little information, however, was presented regarding the frequency and dura-tion of the activities and how well subjects complied with the prescribed program. Many of the articles delineated when subjects were allowed to return to sports, but few reports provided data describing whether subjects actually returned to sport and if so, at what level. Furthermore, no consensus was given on what primary outcome measure should be used to determine if a rehabilitation program is both safe and effective. A common outcome for most of the investigations we reviewed was A-P knee laxity. Although most orthopaedic surgeons would agree that an increase in anterior laxity of more than 3 mm in the index knee com-pared to the normal knee is a concern from a biomechani-cal perspective,29it remains unclear what magnitude of an increase in A-P laxity is a concern from a biological per-spective. It may be that increases of knee laxity that are within certain limits of normal do not result in altered metabolism of the articular cartilage, damage to the meniscus, or additional intra-articular injury. However, we do not know the relationship between increased anterior knee laxity and metabolism of the menisci and articular cartilage, and therefore, any increased laxity has the potential to lead to adverse changes within the joint over time. There also appears to be a place for determining the control of rotational laxity by modern reconstruction pro-cedures. Single-bundle procedures, especially involving femoral tunnel placement high in the notch (11 o’clock to 12 o’clock), may well be unable to prevent abnormal rota-tional kinematics of the tibia relative to the femur.
Our interpretation of the ACL rehabilitation literature is that there is some information available from prospec-tive RCTs regarding how much loading and motion a knee with a healing BPTB graft can sustain without perma-nently stretching the graft (as evidenced by abnormal increases of anterior knee laxity), disrupting the graft, or creating failure of graft fixation. In contrast, there is very little information available about the effect of rehabilita-tion on other graft materials, such as the 4-strand ham-string graft. As well, there is little information available about the effect of rehabilitation on the healing response of an ACL graft with combined injuries to the articular cartilage and meniscus.
EFFECTS OF SEX, AGE, AND ACTIVITY LEVEL ON
THE OUTCOME OF ACL RECONSTRUCTION
There are many studies on the outcome of ACL recon-struction in the literature. Unfortunately, few of these articles have addressed the potential differences between male and female patients. There are no prospective, con-trolled studies directly comparing the outcomes on men and women after ACL reconstruction. Barber-Westin and colleagues12 published one of the first studies comparing the outcome of ACL reconstruction with BPTB grafts between male and female patients. Their report showed that sex alone should not be the basis for selection regard-ing surgical intervention. Failure rates were low, and out-comes were similar between male and female subjects.
