Vol 7, No 3, May/June 1999 189 Anterior cruciate ligament (ACL) reconstruction was first described by Hey-Groves in 1917. Since then, sig- nificant improvements have been made in performing intra-articular reconstructions. With current tech- niques and accelerated rehabilitation protocols, the morbidity has been greatly decreased. Reconstruction is now widely accepted as the treat- ment of choice for individuals with functional instability due to an ACL- deficient knee. It has been estimated that between 60,000 and 75,000 ACL reconstructions are performed annu- ally in the United States. 1 Restoration of stability and re- turn to activity can generally be expected after ACL reconstruction, with reported long-term success rates of between 75% and 95%. 2 However, these results indicate that a subset of patients will have ACL grafts that do not adequately stabi- lize the knee. It has been reported that recurrent instability and graft failure will develop in as many as 8% of patients who undergo prima- ry ACL reconstruction. 2 This small group of patients may be candidates for revision ACL reconstruction. In this article, we will discuss the various reasons for failure of ACL reconstruction, among them arthrofibrosis, traumatic arthrosis, extensor mechanism dysfunction, and recurrent instability, 3 with a focus primarily on the latter. We will also outline the most impor- tant factors to be considered in planning revision ACL reconstruc- tion and will review the literature with regard to results. Causes of Recurrent Instability It is important that the treating physician initially evaluate patients with recurrent instability to deter- mine the cause of failure. Failures can be divided into four groups: failures due to technical errors, fail- ures due to biologic factors, failures due to trauma, and failures due to laxity in the secondary restraints (Table 1). 4 Some authors have cho- sen to combine secondary restraint failures with technical failures 1,2,5 ; however, the mechanism of a sec- ondary restraint failure is different from that of a failure due to a poor- ly placed tunnel or an improperly tensioned graft. Recognition of secondary instability patterns is critical to the prevention of failure of a revision; therefore, primary graft failure should be considered a separate category. Technical Errors Technical shortcomings are the most common cause of failure in patients who come to revision ACL reconstruction. 2,4,6,7 In one series, 4 Dr. Getelman is Attending Surgeon, Southern California Orthopedic Institute, Van Nuys, California. Dr. Friedman is Attending Surgeon, Southern California Orthopedic Institute, and Assistant Clinical Professor of Orthopaedics, UCLA School of Medicine, Los Angeles. Reprint requests: Dr. Getelman, 6815 Noble Avenue, Van Nuys, CA 91405. Copyright 1999 by the American Academy of Orthopaedic Surgeons. Abstract Revision anterior cruciate ligament (ACL) reconstruction is indicated for select- ed patients with recurrent instability after a failed primary procedure. The cause of the failure must be carefully identified to avoid pitfalls that may cause the revision to fail as well. Associated instability patterns must be recognized and corrected to achieve a successful result. The choice of graft, the problem of retained hardware, and tunnel placement are the major challenges of revision ACL reconstruction. The patient must have reasonable expectations and under- stand that the primary goal of surgery is restoration of the ability to perform activities of daily living, rather than a return to competitive athletics. The results of revision ACL reconstructions are not as good as those after primary reconstructions; however, the procedure appears to be beneficial for most patients. J Am Acad Orthop Surg 1999;7:189-198 Revision Anterior Cruciate Ligament Reconstruction Surgery Mark H. Getelman, MD, and Marc J. Friedman, MD technical failures, such as nonana- tomic tunnel placement, inade- quate notchplasty, improper graft tensioning, inadequate graft fixa- tion, or insufficient graft material, were implicated in 77% of the revi- sion cases. Nonanatomic Tunnel Placement Tunnel location dictates the isometry of the graft over a range of motion, and poorly placed tun- nels can lead to increased graft ten- sion. It has been estimated that 70% to 80% of technical failures are due to malpositioned tunnels. 2 The ACL graft can withstand only a small amount of strain before de- forming. Malpositioned grafts incur excessive tension and may be impinged or become lax, leading to failure. The most common error is im- proper positioning of the femoral tunnel. 1,2,4,8 A femoral tunnel placed too far anterior and ten- sioned in extension will lead to ex- cessive strain during flexion (Fig. 1). This results in overconstraint of the knee with loss of flexion or stretch- ing of the graft. If the same anteri- orly placed graft is tensioned in flexion, the joint will not be con- strained, but there will be unac- ceptable laxity in extension and failure. Similarly, a posteriorly positioned graft will result in laxity in flexion if the graft was tensioned in extension or will cause loss of extension or excessive strain in extension if the graft was tensioned with the knee in flexion. With endoscopic techniques, the graft may also be placed too centrally (12-oÕclock position). In this case, the anteroposterior excursion may be controlled; however, the rota- tional component of the instability can remain, resulting in a persis- tent pivot shift. 9 The tibial tunnel is more forgiv- ing and has less influence on fiber- length changes. Nevertheless, place- ment is critical, as an anterior tibial tunnel results in graft impingement in extension and leads to excessive graft tension with flexion. Howell and Taylor 10 have carefully docu- mented the importance of proper tunnel placement. 10 They state that the tibial tunnel should be inclined posterior to BlumensaatÕs line (line indicating the roof of the intercon- dylar notch, as visualized on a later- al radiograph) with the knee in full extension in order to prevent notch impingement (Fig. 2). Posterior placement will result in excessive laxity in flexion. Medial or lateral placement leads to impingement on the walls or roof, chronic synovitis, and increased laxity. 11 Inadequate Notchplasty The intercondylar notch must be large enough to allow full range of motion after reconstruction. The graft is often larger than the native ACL, and thus more space needs to be created. Impingement on the notch can lead to loss of extension and the formation of a cyclops lesion. Impingement on the lateral Revision Anterior Cruciate Ligament Reconstruction Surgery Journal of the American Academy of Orthopaedic Surgeons 190 Table 1 Causes of Failure Technical Nonanatomic tunnel placement Inadequate notchplasty Improper tensioning Graft fixation Insufficient graft material Biologic Failed ligamentization Infection Arthrofibrosis Infrapatellar contracture syndrome Traumatic Early (before graft incorporation) Late (after incorporation) Failure due to secondary instability Rotatory instability Skeletal malalignment Varus/valgus instability Fig. 1 Anteroposterior (A) and lateral (B) radiographs demonstrate anterior femoral tun- nel position. Note the double density of the interference screw in the femur and the two screws in the tibia. The patient had undergone both primary and revision ACL reconstruc- tions. Both failed because the femoral tunnel was too anterior. A B wall should be avoided. It has been recommended that the notchplasty include removal of enough of the medial portion of the lateral femoral condyle so that the posterior horn of the lateral meniscus can be visual- ized. 12 With cyclical motion, there will be repetitive impingement on the graft, which can affect the blood supply and cellular ingrowth and eventually result in failure. Magnetic resonance (MR) imag- ing has been used extensively to evaluate grafts for impingement. 2,9,13 Grafts that are free of impingement will have homogeneous low signal intensity, similar in appearance to a tendon. Impinged grafts will dem- onstrate irregular increased signal, representing narrowing of the mid- substance (Fig. 3). 13 Improper Tensioning Proper graft tensioning is critical to achieving a successful reconstruc- tion. The ideal tension is dependent on several variables, including the length, stiffness, and viscoelasticity of the graft; the tension applied; and the position of the leg at the time of fixation. 12 Grafts that are underten- sioned at the time of fixation are too loose, resulting in immediate resid- ual laxity. However, too much ten- sion is problematic as well; a graft fixed with excessive tension may constrain the knee, affect graft incorporation, limit graft strength, and lead to failure. In a dog model, Yoshiya et al 14 studied the effect of various graft loads. With greater tensioning force, there was de- creased vascularization and delayed graft incorporation, with subse- quent myxoid degeneration of the graft substitute. Recommendations about proper tensioning 12 have been made, but these have not been studied experi- mentally. For boneÐpatellar tendonÐ bone (BPTB) grafts, it is suggested that 5 to 10 lb of tension be applied with the graft fixed with the knee in 10 to 15 degrees of flexion. Ham- string grafts should be fixed with slightly greater force (10 to 15 lb) with the knee in 20 to 30 degrees of flexion. It is also suggested that the knee be cycled through a range of motion to preload the graft before fixation. This may help reduce laxity from stress relaxation; however, ten- sion must be maintained until fixa- tion. Seventy-five percent of the viscoelasticity will return to the graft tissue if the tension is allowed to drop for 1 minute. 15 Graft Fixation Failure Initial graft fixation is essential to the early success of ACL recon- structions. With present rehabilita- tion protocols stressing immediate knee motion, it is critical that the fixation maintain the graft position and proper tension. Many types of fixation failure can occur. With bone-tendon-bone grafts, failures can result from bone- block advancement or loss of fixa- tion when tibial interference screws are used and from screw divergence when endoscopic femoral fixation is used. Hamstring-graft fixation fail- ure can also occur with improper positioning of an endoscopic fas- tener (Endobutton [Smith Nephew, Andover, Mass]) or poor soft-tissue interference fixation. It takes 6 to 12 weeks for the graft to incorporate (bone-to-bone healing occurs earlier than soft tissueÐto-bone healing). Several methods of fixation are available, and the strengths have been evalu- ated. It is important that the cho- sen technique provide the neces- sary fixation strength to protect the reconstruction during the early postoperative period. Biologic Failure Any tissue chosen as an ACL substitute, whether autograft or allograft, is biologically different from the native ligament. A col- lagenous substitute is placed into the knee, which must then undergo remodeling to become incorporat- ed as organized scar tissue that can function as a checkrein against instability. 16 These changes have been referred to as Òligamentiza- tionÓ 17 ; however, this is something of a misnomer, as a new ligament is not created. The native ligament is able to provide stability with dif- ferent loads over a range of motion. It has large- and small-diameter collagen fibers of different lengths. The collagen fibers of the graft are uniform in length, small in diame- ter, and parallel in orientation, unlike those of the native liga- ment. 16,18 The inability of the graft to fully duplicate the native ACL can contribute to graft failure. Biologic failure should be con- sidered when a patient presents with instability without a history of Mark H. Getelman, MD, and Marc J. Friedman, MD Vol 7, No 3, May/June 1999 191 Fig. 2 Lateral radiograph in extension demonstrates slightly anterior placement of the tibial tunnel with reference to the roof of the intercondylar notch, creating minimal impingement (20%). (Reproduced with permission from Howell SM, Taylor MA: Failure of reconstruction of the ante- rior cruciate ligament due to impingement by the intercondylar roof. J Bone Joint Surg Am 1993;75:1044-1055.) Roof of intercondylar notch Tibial tunnel trauma and without an identifiable technical error. 5 Possible causes of biologic failure include avasculari- ty, immunologic reaction, and stress shielding. Successful biologic incorporation requires vasculariza- tion to allow cellular repopulation and subsequent matrix remod- eling. 16 This is influenced by the biomechanical loads affecting the graft, the graft position and ten- sion, and the biologic response of the host. If these factors all work in concert, a successful outcome can be expected. The ligamentization process can be delayed and can be less uniform when allografts are used. 12 Fresh patellar-tendon allografts have been shown to incite a marked inflam- matory response, with plasma-cell and lymphocyte infiltration similar to that seen in other rejection re- sponses. 18 With deep-freezing, the inflammatory response can be mini- mized; however, the potential for immunogenicity of the graft re- mains. Harner has documented the expression of immunoglobulin G antibodies to donor human leuko- cyte antigens, which does not occur when autografts are used. Fur- thermore, with allograft use, several authors have noted postoperative bone-tunnel enlargement not usual- ly seen with autografts. 12,19 Pro- cesses used for sterilization, such as gamma irradiation, freeze-drying, and treatment with ethylene oxide, can weaken the graft and delay or impede ligamentization. Infection and arthrofibrosis should be considered in the catego- ry of biologic failure, although tech- nical factors may be contributory. Infection is rare in ACL surgery, but it can be devastating. Once infec- tion has been recognized, the joint should be immediately irrigated and debrided, and broad-spectrum antibiotic therapy should be begun pending the results of cultures. The decision whether to remove the graft must be individualized. The deciding factors include the extent of the infection, the causative organ- ism, the type of graft, and the type of fixation. It is generally agreed that prosthetic ligaments must be removed in the presence of infec- tion. Revision may be considered after 6 weeks provided the clinical examination findings and laboratory values are normal. 20 Arthrofibrosis and infrapatellar contracture syndrome are perhaps the most debilitating causes of fail- ure. The natural history of the lat- ter is worse than that of an ACL- deficient knee. 21 The fibrosis can lead to marked limitation in the range of motion. The goal of treat- ment is restoration of motion, which often requires manipulation, debridement, and removal of the graft. In these cases, patients must be aware that the primary goal is improved joint function and must understand that instability may recur. Traumatic Failure There are two types of traumatic failure: those that occur early, be- fore graft incorporation, and those that occur late, after resumption of normal activities. Aggressive phys- ical therapy may play a role in early failures. 2,22 Since accelerated rehabilitation protocols have been implemented, there has been an increased risk of graft loosening, necessitating secure initial fixation. The graft is at its weakest during the early rehabilitation period, and patients must understand the po- Revision Anterior Cruciate Ligament Reconstruction Surgery Journal of the American Academy of Orthopaedic Surgeons 192 A B Fig. 3 Sagittal MR images of two patients. A, An intact ACL graft with homogeneous low signal intensity. B, An impinged graft with bright signal within the intra-articular graft. tential consequences of overzealous rehabilitation or returning to activi- ty too soon. Late reruptures seem to occur infrequently in patients with technically well-done recon- structions. The true incidence of traumatic reruptures is not known, in part, perhaps, because there can be multi- ple contributing factors to these injuries. It has been reported that there is a 5% to 10% incidence of traumatic rerupture in athletic indi- viduals. 1,8 Others have reported that up to 43% of their revision cases were due to traumatic events. 2 However, there were no traumatic failures in the series reported by Schepsis and co-workers. 4 Although failure may at first appear to be due to reinjury, careful evaluation will often reveal another cause for recur- rent instability. Failures Due to Secondary Instability In chronically ACL-deficient knees, the secondary restraints to anterior translation can become attenuated. Reconstruction of the ACL alone in this situation will often result in failure. The ACL graft will provide early anterior restraint for the first 6 months; how- ever, increased-demand activities will lead to gradual recurrence of the instability. 3 In a study of the results of primary ACL reconstruc- tion in 80 knees, OÕBrien et al 23 found that all patients who had postoperative clinical instability with giving way demonstrated evi- dence of associated ligamentous instability, which had not been appreciated or addressed at the time of the primary surgery. Similarly, Schepsis and co-workers 4 reported that 4 (15%) of their 26 revision cases were due to failure to address associated laxities, such as postero- lateral rotatory instability and anteromedial rotatory instability. Although other authors have noted the importance of this phe- nomenon, they have chosen to include failures due to secondary instability among their technical failures. Therefore, the true inci- dence in those series is more diffi- cult to establish. Nonetheless, the importance of these laxity patterns remains significant, and the treat- ing surgeon must recognize and address pathologic instability in order to ensure a successful out- come and avoid late failure. Considerations in Revision ACL Reconstruction Surgery Revision surgery is complex and technically challenging. The choice of graft, the problem of retained hardware, and tunnel placement are important considerations in planning revision ACL reconstruc- tion. Graft Selection There are three options in graft selection: synthetic grafts, auto- grafts, and allografts. Synthetic grafts have several potential advan- tages. The grafts are strong and do not entail harvest morbidity or risk of disease transmission. However, with permanent grafts, stents, and ingrowth grafts, there have been unacceptably high rates of compli- cations and failures. Therefore, they are not recommended for rou- tine use. The decision to be made is really between autograft and allograft. Autografts, such as hamstring, quadriceps-patella, iliotibial-band, and BPTB grafts, offer shortened incorporation times without the potential for disease transmission or immunologic reaction. However, associated harvest morbidity is a concern, and in the revision situa- tion, lack of availability may prevent or limit the use of autologous tissue. Contralateral BPTB grafts have been used by Rubinstein at al 24 with no major complications re- ported. Repeated ipsilateral BPTB harvesting has also been reported and is more commonly used in Europe. Magnetic resonance imag- ing has been used to document reconstitution of the central third of the BPTB donor site at 1 year. 1,5,8 Karns et al 25 have reported a revi- sion case in which a reharvested patellar tendon was successfully used 4 years after the primary reconstruction. While this sounds enticing, there are no biomechani- cal studies documenting the me- chanical properties and tensile strength of the reharvested graft. Furthermore, the histologic compo- sition at the tendon-bone interface of the reconstituted ligament is scar tissue rather than ligament. Thus, at present, the routine use of a reharvested BPTB graft is not rec- ommended. CalcaneusÐAchilles tendon, BPTB, and iliotibial-band allografts have been used successfully for both pri- mary and revision ACL reconstruc- tions. There are several advantages to their use, such as shorter operative times, smaller incisions, and no po- tential for donor-site morbidity. Furthermore, there is no size limita- tion. In revision cases, there may be a need for larger tunnel diameters, which can more easily be accommo- dated with larger allograft bone plugs. There are disadvantages inher- ent to the use of allograft tissue, such as longer incorporation times, the possibility of immunologic reactions, and higher cost. There is also the potential risk of viral dis- ease transmission, as preparation techniques will not necessarily eradicate all viruses. It has been estimated that the risk of disease transmission with allograft use is less than 1 per 1,000,000 uses, as- suming that the current American Association of Tissue Banks guide- lines are followed. With use of those guidelines, there has never Mark H. Getelman, MD, and Marc J. Friedman, MD Vol 7, No 3, May/June 1999 193 been a reported case of viral trans- mission from allograft connective tissue. 26 Radiation kills viruses, but the necessary dosage alters the collagen and reduces the graft strength. 26,27 Thus, graft selection must be individualized. The only absolute requirement is that the chosen graft have adequate strength to provide long-term stability. Beyond that, the ultimate decision is between the patient and the treating physi- cian. Preoperative Evaluation The preoperative evaluation of patients with failed ACL recon- structions is important for many reasons. It is at this time that the surgeon must counsel the patient about realistic expectations. Results after revision ACL reconstruction do not compare favorably with those after primary procedures. It is important to emphasize that the surgery is a salvage operation. False expectations can lead to a sub- jective failure despite a technically successful procedure. 20 The goal is to allow the patient to perform activities of daily living without instability. The patient must under- stand that a return to sports may not be possible. The preoperative evaluation should include a careful history, assessing aspects such as the prima- ry procedure, postoperative reha- bilitation, return to activity, and the time at which recurrent instability was first noted. Instability, rather than pain, should be identified before considering revision. A painful knee that demonstrates some laxity on examination is quite different from a knee that demon- strates recurrent instability. Pain can be due to factors that are unre- lated to the laxity and often will not improve even if the ACL recon- struction is revised. The physical examination should be comprehensive. Other causes for pain should be sought. Any sign of sepsis must be fully investi- gated. The patientÕs gait, align- ment, and stability should be evalu- ated, and range of motion and crepitus should be noted. The physician must carefully address skeletal alignment and secondary restraints in addition to anterior instability. Unrecognized laxity may be the difference between suc- cess and failure of the revision reconstruction. Weight-bearing anteroposterior and 45-degree flexion posteroanteri- or radiographs are helpful in deter- mining alignment and assessing arthrosis. Lateral radiographs in full extension allow evaluation of the tunnel position and size. Axial views of the patella and full-length standing films are also useful. If there is concern about tunnel osteol- ysis on the basis of the radiographic appearance, computed tomography (CT) or MR imaging can be per- formed to obtain a better under- standing of the defect (Fig. 4). 28 Al- though CT offers better bone defini- tion, MR imaging provides more information with regard to meniscal injury and the status of the articular surfaces. The previous records, especially the original operative report, can provide valuable information that will elucidate the prognosis and facilitate the procedure. The status of the knee and the condition of the menisci and articular surfaces at the time of the primary reconstruc- tion can be determined. The type Revision Anterior Cruciate Ligament Reconstruction Surgery Journal of the American Academy of Orthopaedic Surgeons 194 Fig. 4 Coronal CT scan (A) and MR image (B) demonstrate tibial tunnel enlargement. (Reproduced with permission from Seemann MD, Steadman JR: Tibial osteolysis associated with Gore-Tex grafts. Am J Knee Surg 1993;6:31-38.) A B of graft, the technique, and the hardware used should be noted so that the necessary equipment can be obtained for future removal. Staging If there is tunnel enlargement, staged reconstruction may be con- sidered. The first procedure would involve graft removal, tunnel curet- tage, and bone grafting. The revi- sion can then be performed 6 to 12 weeks later, once the bone graft has incorporated. If the patient has substantial loss of motion (lacking more than 5 degrees of extension or 20 degrees of flexion), the first goal should be restoration of motion. In such a case, the revision procedure may have to be staged. 2 Skin Incisions Skin incisions should be carefully planned to avoid complications. Whenever possible, previous inci- sions should be reused if they will allow access for proper tunnel place- ment and hardware removal. Skin bridges measuring less than 7 cm should be avoided. 2 The periosteum should be handled gently and closed when possible to avoid complica- tions. Allografts usually require smaller incisions and should be con- sidered if the skin is of particular concern. Hardware Removal With present ACL fixation tech- niques, there is a myriad of different types of hardware. Occasionally, the hardware does not interfere with the revision procedure. If that is the case, it can be left alone. However, the surgeon should always be pre- pared for hardware removal. If the exact type of hardware can be deter- mined, the proper equipment can be made available. Removal of hardware can be dif- ficult. Soft tissue or bone that obscures the device may also have to be removed with use of curettes, a burr, or coring reamers. To success- fully remove interference screws, the proper screwdriver must be com- pletely seated, and the angle of knee flexion at the time of screw insertion must be duplicated. A chest tube can be used over the screwdriver to prevent loss of the screw into the joint during attempted removal. If the screw is stripped, it can be drilled out. This may require addi- tional bone removal, necessitating a staged procedure. If staples are used, the appropriate extractor or osteotome should also be used. The surgeon should be familiar with several techniques for revi- sion. Changing from an endoscopic to a two-incision technique, or vice versa, can often avoid the need for hardware removal. This may allow preservation of bone stock and reduce operative time. Removal of the Prosthetic Ligament If a failed synthetic ligament reconstruction (popular in the 1980s) is encountered, the prosthetic ligament must be removed at the time of revision. It is ideal to re- move the prosthesis en bloc. A gouge can be used to loosen the bone attachments before removal. If particulate debris is left behind, there is the potential for a marked inflammatory response, with possi- ble bone tunnel and cartilage de- struction. If the synthetic ligament demonstrates intra-articular dam- age, it may be necessary to perform a synovectomy to eliminate the par- ticulate debris. Staged reconstruc- tions are often necessary when there is marked tunnel osteolysis. Revision Notchplasty Most revision cases will require repeat notchplasty. There is natural regrowth by as much as 1 cm after an ACL reconstruction. 20 In cases in which failure was due to graft impingement, revision notchplasty is imperative. The roof and lateral wall can both cause graft impinge- ment, which must be addressed at the time of revision. It is necessary to remove enough bone to prevent impingement and to allow visual- ization for proper tunnel place- ment. The over-the-top position and the previous tunnels must be visualized before creating any new tunnels. Excessive bone removal must be avoided, however, to ensure that adequate bone stock is available for graft fixation. Bone Tunnel Placement Bone tunnel placement is the most important and challenging aspect of a revision ACL recon- struction. Malpositioned bone tun- nels are the most frequent cause of graft failure. When the original tunnels are placed improperly, new tunnels must be created. On the femoral side, the most commonly encountered error is placement of the original tunnel too far anterior (Fig. 1). A new tunnel can often be drilled posterior to the original one, and existing hardware does not need to be removed. If the tunnel is only slightly anterior, there are several options. The tunnel can be expanded posteriorly, and a larger bone plug (allograft) can be utilized. For autograft use, an alternative technique, such as fixation with stacked interference screws, may be required. Alternatively, a two- incision technique can be used to move the tunnel to a more posterior position, allowing fixation within the tunnel. When tunnel placement is too posterior, the posterior wall is often deficient, which will necessi- tate conversion to a two-incision or over-the-top technique to get secure fixation. If cystic changes prevent femoral tunnel fixation, the graft can be placed over the top and secured with a staple or a post and washer in the lateral femoral condyle. Rosenberg has described a bi- socket technique in which two small femoral tunnels are created posterior to the malpositioned tun- Mark H. Getelman, MD, and Marc J. Friedman, MD Vol 7, No 3, May/June 1999 195 nel. A hamstring graft is divided in half, and each half is placed into a 6-mm tunnel and fixed with sepa- rate Endobuttons (Fig. 5). In those cases in which the fe- moral tunnel is appropriately locat- ed, revision interference fixation may be difficult. It may be neces- sary to change the fixation tech- nique or convert from an endo- scopic to a two-incision technique to reorient the tunnel so that inter- ference fixation can be used. It is possible to apply similar principles to the tibial tunnel. If the tunnel is too anterior, the graft will impinge and fail. Often, the exist- ing tunnel and hardware can be ignored, and a new tunnel can be created in a more posterior location. If the tunnel is slightly malposi- tioned such that the existing tunnel must be incorporated, the tunnel can be expanded anteriorly or pos- teriorly, and a larger bone block can be used. For capacious tunnels, the surgeon either can use bone graft simultaneously with the new ACL graft or can stage the reconstruc- tion. Most commonly, capacious tunnels are encountered with failed synthetic-ligament or allograft reconstructions. Graduated tunnel expanders are helpful when fashioning new tun- nels. They allow the surgeon to create a small-diameter hole initial- ly, which can then be enlarged in a controlled, progressive manner. Thus, there is less risk of cortical perforation. These expanders also allow bone to be compacted rather than removed, which can facilitate graft fixation. 20 Graft Fixation Graft fixation is critical in the early postoperative period after revision ACL reconstruction. If the tunnels are created without diffi- culty or significant bone loss, rou- tine fixation techniques can be employed. However, if there is any concern about fixation strength with these methods, secondary techniques should also be used. Post-and-washer fixation or staple fixation can easily be added as a backup, particularly on the tibial side. It is recommended that the treating surgeon become familiar with the various options for fixa- tion. Associated Instability Patterns Laxity of the secondary re- straints can develop in patients with chronic ACL insufficiency. Failure to address the associated instability will ultimately result in failure of the revision ACL recon- struction. Underlying problems, such as bone malalignment, menis- cal loss, and rotatory instability, must be identified preoperatively and addressed at the time of recon- struction. In cases of anteromedial rotatory instability, the posteromedial cor- ner can often be treated by reefing the lax tissues. However, if there is severe anteromedial instability and the patient previously underwent medial meniscectomy, a medial meniscal transplant may also be considered. This will allow restora- tion of the Òbreakstop mechanismÓ of the posterior medial meniscal horn and may assist in controlling instability that cannot be corrected with capsular reefing alone. 20 To date, there are no reports from con- trolled clinical studies of long-term success with meniscal transplanta- tion for anteromedial rotatory insta- bility. Chronic posterolateral rotatory instability is more difficult to con- trol. Several treatment options are available, and selection of the pre- ferred method is based on the extent of the instability. For severe laxity, reconstruction with auto- graft or allograft is often required. Marked varus alignment of the lower extremity can further devel- op in patients with long-standing ACL insufficiency and posterolat- eral instability. In these cases, it is also necessary to perform a valgus osteotomy to correct the alignment and limit tension on the recon- structed posterolateral corner. An osteotomy should also be consid- ered for patients with advanced articular cartilage changes and malalignment coexistent with the instability. Rehabilitation Rehabilitation after revision ACL reconstruction is different from, and more conservative than, the aggressive protocols used for pri- mary ACL reconstruction. Patients must be reminded that the results are less predictable and that there- fore they must not exceed the limits placed on them. Each rehabilita- tion protocol is individualized and is based on the type of reconstruc- tion, the strength of fixation, and Revision Anterior Cruciate Ligament Reconstruction Surgery Journal of the American Academy of Orthopaedic Surgeons 196 Fig. 5 Bisocket femoral tunnel technique described by Rosenberg. any associated reconstructions that were performed. Weight bearing is protected for up to 6 weeks, and return to activities is delayed. Results of Revision ACL Reconstruction There are few studies that critically evaluate the results of revision ACL reconstruction. 2,4,5,29 A recent sym- posium presented the results from three studies. 6,7,30 The graft choices in the three series differed: in a series of 25 patients, Johnson et al 7 used only fresh-frozen irradiated allografts; Uribe et al 6 and Noyes and Barber-Westin 30 used both autograft and allograft in their patients (54 and 85, respectively). Therefore, it is difficult to compare their results directly; nevertheless, there are some important conclu- sions that can be drawn. All three studies demonstrated improve- ments in function compared with the prerevision status (however, there were no comparisons with the results after primary reconstruc- tions). One of the major determining factors in ultimate outcome ap- peared to be the status of the articu- lar cartilage. Patients with normal or minimally damaged cartilage had fewer symptoms and were more likely to return to sports and strenu- ous occupations. The overall results for the allografts appeared similar in all three studies. When both graft sources were compared, there appeared to be greater objective stability with the autograft tissue; however, the differ- ence was not statistically significant in either study. 6,30 Functionally, there was no difference between the allograft and autograft groups. Many of the autografts used were obtained through contralateral har- vest with limited morbidity, sup- porting the earlier findings by Rubinstein et al. 24 Similar findings were noted in a study by Schepsis and co-workers. 4 In that study, 14 fresh-frozen BPTB allografts and 12 BPTB autografts were used in 26 patients. The pri- mary reconstruction had failed be- cause of technical reasons in 77% of cases and because of secondary restraint instability in 15%. After revision, the autografts showed greater stability, as indicated by objective measurements (e.g., KT- 1000 arthrometry, pivot-shift and Lachman tests), but the difference compared with the results in the allograft patients was not statistical- ly significant. When the patients in that study were evaluated with the Interna- tional Knee Documentation Com- mittee (IKDC) system, there was a statistically significant (P<0.03) dif- ference between the autograft and allograft groups. In that system, the status of the articular cartilage and the amount of arthrosis present are important criteria, and the worst rating in any one category deter- mines the final rating. The patients treated with allografts had more prevalent patellofemoral crepitus and joint-space narrowing, which therefore resulted in lower ratings. These degenerative findings had been noted preoperatively and were significant in choosing allografts for those patients. By the IKDC criteria, 29% of the allograft group and 67% of the autograft group had satisfac- tory results. Subjectively, however, 86% of the allograft patients and 92% of the autograft patients were satisfied and would have under- gone surgery again. Autogenous tissue may offer more predictable results; however, to date no study has demonstrated a functional difference between allo- graft revision and autograft revision. For patients with limited articular cartilage damage, it may be prudent to encourage the use of autografts, as it appears that they provide the greatest likelihood of a successful return to sports. Patients with more advanced articular damage are less likely to return to strenuous activi- ties. Thus, it seems reasonable to encourage allograft use so as to limit morbidity and expedite the return to activities of daily living. Summary Revision ACL reconstruction is rec- ommended for patients who have instability with objective sympto- matic laxity after a failed primary procedure. The cause of failure must be carefully identified to avoid pitfalls that may cause fail- ure of the revision as well. Graft choice, hardware removal, and tun- nel placement are the major chal- lenges of revision ACL reconstruc- tion. Meticulous preoperative planning is essential. Associated instability patterns must be recog- nized and corrected to achieve an acceptable result. Revision patients often have undergone multiple operations and yet remain disabled and frustrated. The patient must have reasonable expectations and understand that revision is a sal- vage procedure, with the primary goal being a return to activities of daily living rather than a return to competitive athletics. An Òexcel- lentÓ outcome should not be antici- pated in most cases. Although revision results are not as good as those for primary reconstructions, the procedure appears to be benefi- cial for most patients. Acknowledgment: The authors thank Anthony A. Schepsis, MD, for allowing his results to be incorporated into this article. Mark H. Getelman, MD, and Marc J. Friedman, MD Vol 7, No 3, May/June 1999 197 References 1.Johnson DL, Harner CD, Maday MG, Fu FH: Revision anterior cruciate liga- ment surgery, in Fu FH, Harner CD, Vince KG (eds): Knee Surgery. Balti- more: Williams & Wilkins, 1994, vol 1, pp 877-895. 2.Wetzler MJ, Bartolozzi AR, Gillespie MJ, Rubenstein DL, Ciccotti MG, Miller LS: Revision anterior cruciate ligament reconstruction. Operative Techniques Orthop1996;6:181-189. 3.Newhouse KE, Paulos LE: Compli- cations of knee ligament surgery, in Nicholas JA, Hershman EB (eds): The Lower Extremity & Spine in Sports Medicine. St Louis: Mosby, 1995, pp 901-908. 4.Getelman MH, Schepsis AA, Zimmer J: Revision ACL reconstruction: Auto- graft versus allograft [abstract]. Arthroscopy1995;11:378. 5.Maday MG, Harner CD, Fu FH: Revision ACL surgery: Evaluation and treatment, in Feagin JA Jr (ed): The Crucial Ligaments: Diagnosis and Treatment of Ligamentous Injuries About the Knee, 2nd ed. New York: Churchill Livingstone, 1994, pp 711-723. 6.Uribe JW, Hechtman KS, Zvijac JE, Tjin- A-Tsoi EW: Revision anterior cruciate ligament surgery: Experience from Miami. Clin Orthop1996;325:91-99. 7.Johnson DL, Swenson TM, Irrgang JJ, Fu FH, Harner CD: Revision anterior cruciate ligament surgery: Experience from Pittsburgh. Clin Orthop1996;325: 100-109. 8.Greis PE, Johnson DL, Fu FH: Revision anterior cruciate ligament surgery: Causes of graft failure and technical considerations of revision surgery. Clin Sports Med1993;12:839-852. 9.Carson EW, Simonian PT, Wickiewicz TL, Warren RF: Revision anterior cru- ciate ligament reconstruction. Instr Course Lect1998;47:361-368. 10.Howell SM, Taylor MA: Failure of reconstruction of the anterior cruciate ligament due to impingement by the intercondylar roof. J Bone Joint Surg Am1993;75:1044-1055. 11.Muneta T, Yamamoto H, Ishibashi T, Asahina S, Murakami S, Furuya K: The effects of tibial tunnel placement and roofplasty on reconstructed anteri- or cruciate ligament knees. Arthroscopy 1995;11:57-62. 12.Jaureguito JW, Paulos LE: Why grafts fail. Clin Orthop1996;325:25-41. 13.Howell SM, Clark JA, Farley TE: Serial magnetic resonance study as- sessing the effects of impingement on the MR image of the patellar tendon graft. Arthroscopy1992;8:350-358. 14.Yoshiya S, Andrish JT, Manley MT, Bauer TW: Graft tension in anterior cruciate ligament reconstruction: An in vivo study in dogs. Am J Sports Med 1987;15:464-470. 15.Brown CH, Jr, Steiner ME: Anterior cruciate ligament injuries, in Siliski JM (ed): Traumatic Disorders of the Knee. New York: Springer-Verlag, 1994, pp 193-284. 16.Corsetti JR, Jackson DW: Failure of anterior cruciate ligament reconstruc- tion: The biologic basis. Clin Orthop 1996;325:42-49. 17.Amiel D, Kleiner JB, Akeson WH: The natural history of the anterior cruciate ligament autograft of patellar tendon origin. Am J Sports Med1986;14: 449-462. 18.Arnoczky SP: Biology of ACL recon- structions: What happens to the graft? Instr Course Lect1996;45:229-233. 19.Fahey M, Indelicato PA: Bone tunnel enlargement after anterior cruciate lig- ament replacement. Am J Sports Med 1994;22:410-414. 20.Safran MR, Harner CD: Technical con- siderations of revision anterior cruci- ate ligament surgery. Clin Orthop 1996;323:50-64. 21.Paulos LE, Wnorowski DC, Green- wald AE: Infrapatellar contracture syndrome: Diagnosis, treatment, and long-term follow-up. Am J Sports Med 1994;22:440-449. 22.Graf B, Uhr F: Complications of intra- articular anterior cruciate reconstruc- tion. Clin Sports Med1988;7:835-848. 23.OÕBrien SJ, Warren RF, Pavlov H, Panariello R, Wickiewicz TL: Re- construction of the chronically insuffi- cient anterior cruciate ligament with the central third of the patellar liga- ment. J Bone Joint Surg Am1991;73: 278-286. 24.Rubinstein RA Jr, Shelbourne KD, VanMeter CD, McCarroll JC, Rettig AC: Isolated autogenous bone-patellar tendon-bone graft site morbidity. Am J Sports Med1994;22:324-327. 25.Karns DJ, Heidt RS Jr, Holladay BR, Colosimo AJ: Case report: Revision anterior cruciate ligament reconstruc- tion. Arthroscopy1994;10:148-151. 26.Ritchie JR, Parker RD: Graft selec- tion in anterior cruciate ligament re- vision surgery. Clin Orthop1996;325: 65-77. 27.Gibbons MJ, Butler DL, Grood ES, Bylski-Austrow DI, Levy MS, Noyes FR: Effects of gamma irradiation on the initial mechanical and material properties of goat bone-patellar ten- don-bone allografts. J Orthop Res1991; 9:209-218. 28.Seemann MD, Steadman JR: Tibial osteolysis associated with Gore-Tex grafts. Am J Knee Surg1993;6:31-38. 29.Noyes FR, Barber-Westin SD, Roberts CS: Use of allografts after failed treat- ment of rupture of the anterior cruci- ate ligament. J Bone Joint Surg Am 1994;76:1019-1031. 30.Noyes FR, Barber-Westin SD: Revi- sion anterior cruciate ligament sur- gery: Experience from Cincinnati. Clin Orthop1996;325:116-129. Revision Anterior Cruciate Ligament Reconstruction Surgery Journal of the American Academy of Orthopaedic Surgeons 198