Bone loss and soft tissue imbalance are primary concerns after failed total knee arthroplasty (TKA).3,4,10 Both problems must be considered together to fully address the concerns of failed TKA. Careful preservation of ligament attached to the remaining bone stock and careful preservation of the rim of the femoral and tibial diaphyses can establish conditions that allow revision arthroplasty that uses minimally constrained components even with severe bone loss and ligament damage. Pathologic changes in the ligaments and quadriceps mechanism occur gradually as the knee fails (Fig 1), but the soft tissue sleeve remains intact and capable of stabilizing the knee in flexion and extension even with major bone loss (Figs 2, 3). Unfortunately, exposure of the knee after failed TKA often is done with extensive soft tissue stripping to gain access to the knee, and bone loss is worsened by excessive resection of bone stock and repeated use of cement in damaged bone stock.
A conservative approach to the knee (Fig 4) that meticulously spares the remaining bone stock and the capsular and synovial attachments to the bone leaves a reconstructable combination of metaphyseal rim and soft tissue sleeve that can be used in almost every case to achieve stability and flexibility of the knee. This can be done without resorting to bone block allografting techniques or fixed hinges that have high failure rates and result in even greater loss of bone stock and ligament structure of the knee.11
Although the patella can be a major concern in revision TKA, it is not a good landmark for positioning the level of the joint in flexion and extension. The patellar tendon and quadriceps mechanism can experience extensive pathologic change and the patella can be either abnormally high or low. Ligament balancing techniques must focus on achieving correct stability in flexion and extension and the patella must be adjusted to fit this new joint line position. One of the most important issues in patellar kinematics is the position of the patellar groove. The femorotibial joint and the patellofemoral joint must be stabilized to achieve a favorable outcome in revision TKA, and rotational alignment of the femoral and tibial components that optimizes stability and flexibility of the knee may not be optimal for patellar tracking.
Techniques for preserving bone stock and ligament attachments, positioning the joint in flexion and extension to achieve a balanced knee throughout its arc of flexion, and the basic rules for releasing the knee ligaments to address soft tissue contracture in flexion and extension, are outlined. Finally, the techniques for positioning the patellar tendon attachment and trimming the patellar remnant are discussed. A longitudinal consecutive series of revision TKAs is presented to test the hypothesis that bone-conserving and ligament-balancing techniques constitute a reliable regimen for managing the failed TKA with severe bone loss and ligament imbalance.
MATERIALS AND METHODS
Between January 1989 and June 1997, all cases of revision TKA with major bone loss in the femur and tibia (Class III based on the Engh and Ammeen classification3), were entered in a computerized database as a longitudinal study to evaluate ligament balancing and bone reconstruction techniques. Data from 88 patients who had the diagnosis of failed TKA attributable to loosening (89 knees) were retrieved from the database. All cases were treated with the same bone- and soft tissue-sparing techniques, rim seating of the implants, and stem stabilization in the diaphyseal medullary canal of the femur and tibia. All knees but one were stabilized using the spacer effect of the implants to tension the soft tissue envelope and achieve stability in flexion and extension. In one case, a rotating hinge prosthesis was used because of failure of a previous ligament balancing effort at another hospital. This knee is not included in the study. The patients ranged in age from 43–87 years. Seventeen patients died during the followup, but none was lost to followup for other reasons. The mean followup was 96.3 months. Fifty-one patients were women and 35 patients were men. The mean flexion contracture before surgery was 3° ± 5°, and the mean flexion was 95° ± 18.5°. The mean varus laxity was 4.5° ± 6.5° and mean valgus laxity was 7.2° ± 3.5°. The extension lag was 3° ± 5°. Mean anterior drawer laxity was 8.3 mm ± 4.5 mm. The mean posterior laxity was 5.6 mm ± 4.5 mm. The mean Knee Society score was 42 ± 15. Nine patients (nine knees) had lateral patellar subluxation, six patients (six knees) had patella alta with the inferior pole of the patella above the upper edge of the femoral component in full extension, and seven patients (seven knees) had patella infera with the inferior pole of the patella at or below the distal surface of the femur in full extension.
All patients had an anterior incision with medial parapatellar approach to the knee, splitting the interval between the vastus intermedius and medialis. The patella was turned laterally with minimal or no stripping of capsular or synovial attachments from the medial or lateral tibial flare, or from the femoral epicondylar areas. If the quadriceps could not easily be turned laterally, then a tibial tubercle osteotomy was done. Nineteen of these knees (18 patients) required tibial tubercle osteotomy for adequate exposure. No patient had transection of either the quadriceps tendon or the vastus lateralis muscle, and none had extensive stripping of capsular attachments from the tibial or femoral metaphysis.
After implant removal and debridement of the metaphyseal surfaces, the femur and tibia were reamed with reamers of increasing size until tight fixation of the reamer could be achieved in the diaphyseal medullary canal of both bones. Tight fit, defined as underreamed by 1 mm diameter over a length equivalent to at least the length of ½ the slotted portion of the stem, was achieved with either a 150-mm or 200-mm stem. The stems varied in diameter from 10 mm to 22 mm. The reamer was left in place in the tibia to act as a cutting guide, and the tibial surface was resected perpendicular to the long axis of the tibia, removing minimal bone and leaving a rim that often was less than ¼ the circumference of the original proximal tibia. In 27 cases the fibular head was part of the tibial component support, and the tibiofibular joint was grafted in these cases with local bone and allograft.
The size of the tibial component was estimated by comparing the remaining upper tibia with the available tibial trial components, and the femoral size was estimated by comparing the mediolateral distance of the epicondylar area of the remaining femur with the available trial components. Anteroposterior distance also was estimated from a comparison between the remaining AP bone stock height and the inner dimensions of the femoral component. A component was chosen that would span the mediolateral dimension of the distal femur and restore posterior offset to the knee. Femoral buildups were chosen to ensure that posterior contact was made between the surface of the implant and bone. In cases that were between sizes, the larger size was chosen to avoid resection of posterior bone stock and to maximize stability in flexion. If severe distal and posterior bone loss were present on the femoral surface, then a distal and posterior femoral buildup was chosen initially. All the defects in these knees were classified as uncontained defects, and were spanned by seating the femoral component on the rim of the defect. The trials were inserted with trial stems for stabilization. The femoral trial was aligned rotationally so that the epicondylar axis mark on the femoral trial aligned with the epicondylar axis of the remaining femoral bone. Care was taken to position the anterior flange of the femoral component flush with the anterior cortex of the femur, and the trial component was driven firmly onto the end of the femur. In most cases no femoral bone was resected, but instead the femoral bone stock was compacted under the femoral trial to create a flat surface distally. The posterior surfaces of the femur were shaved off with the sharp posterior edges of the femoral trial to ensure posterior seating of the implant against bone. The criteria of distal and posterior seating were achieved using buildup implants when necessary. After insertion of the femoral and tibial metal components, the joint space was estimated in flexion and extension by distracting the joint. A trial tibial polyethylene component then was inserted with the knee in flexion. Knee stability was examined in flexion and extension, and ligament release was done as necessary to achieve correct ligament balance. A previously published protocol for ligament balancing in TKA was followed.7,9,15,17,18,20,21 In knees that were tight laterally in flexion and extension the popliteus tendon and lateral collateral ligament attachments to the femur were released. Knees that were tight laterally in extension only had the iliotibial band and, if necessary, the posterior-lateral capsule released. Knees that were tight medially in flexion only had release of the anterior portion of the medial collateral ligament. Knees that were tight medially in extension only had release of the posteromedial oblique portion of the medial collateral ligament and, if necessary, the posterior medial capsule. Knees that had major medial contracture in flexion and extension had the anterior and posterior portions of the medial collateral ligament released. The pes anserinus attachment to the tibia was released if the knee remained tight medially in extension.
In 33 knees (33 patients) the anterior portion of the medial collateral ligament was released, in 10 knees (10 patients) the entire medial collateral ligament was released, and in three knees (three patients) the medial collateral ligament, posterior capsule, and pes anserinus were released. In five knees (five patients) the lateral collateral ligament and popliteus tendon were released, and in 15 knees (15 patients) the lateral collateral ligament, popliteus tendon, and iliotibial band were released.
After achieving ligament balance in flexion and extension, thicker tibial trial spacers were inserted until good stability was achieved in flexion (Fig 5). Then the knee was extended and, if it extended fully without hyperextension and was correctly stable in flexion and extension, the tibial polyethylene component that achieved this stability was chosen for the final component. If the knee hyperextended (Fig 6), additional distal femoral buildup was applied (Fig 7), and if the knee failed to extend fully, the joint surface was elevated by removing distal femoral buildup material.
After mediolateral stability was achieved in flexion and extension, the tibial trial component was rotated to match the rotational position of the femoral trial component in full extension, therefore maximizing tibial articular surface contact in full extension. All the knees were stable in varus-valgus testing in flexion and extension, but AP stability was augmented by articular surface constraint. Fifty-six knees received a posterior-stabilized revision implant (Advantim, Wright Medial Technology, Arlington, TN), and 31 knees received a deep-dish conforming tibial polyethylene component (Profix, Smith and Nephew, Memphis, TN). No knees were stabilized with an implant with built-in mediolateral stability. All the femoral components were porous-coated, CoCr alloy, and the tibial components were porous-coated, Ti alloy.
The diaphysis of the femur and tibia were underreamed by 1 mm, and the stems were fluted with a slot distally to enhance mechanical interdigitation of the stem with the bone, and to provide toggle control and resistance to torsional loads. As the implants were inserted, the stems became tight in the medullary canal for a distance greater than 3 cm in both bones. The final position of the implant was achieved when the component rested on the remaining metaphyseal rim. This technique created large cavitary defects under the femoral and tibial components. Morselized cancellous allograft and reamings from the medullary canal were mixed, then the mixture was packed loosely in the cavitary defects.
The patella then was placed in its normal position and flexion-extension stability was evaluated. If necessary, lateral patellar release was done outside the synovial membrane and just deep to the deep fascia. If the patella still failed to track, then a distal realignment procedure was done to decrease the Q angle. Seven patients (seven knees) required a modified Roux Goldthwait procedure and three knees (three patients) required medial tibial tubercle transfer to achieve correct patellar tracking.16 Five knees (five patients) required medial transfer of the medial ½ of the patellar tendon because the medial ½ had torn from the tibial tubercle in the process of exposing the knee.16 All patients achieved central patellar tracking without significant patellar tilt after distal realignment.
The patella was low in eight patients (eight knees). The inferior pole of the patella was trimmed in cases in which impingement occurred. Four patients (four knees) had proximal advancement of the tibial tubercle, patella, and patellar tendon through the tibial tubercle osteotomy.
Weightbearing restriction was determined at the time of surgery and followed strictly during the postoperative period. Weightbearing was restricted only in knees with fragile femoral or tibial bone support or compromised quadriceps mechanism. Twenty-two patients (22 knees) were restricted to 25–50 lb weightbearing for 3 months. Eleven patients (11 knees) had splinting in extension during weightbearing activities because of compromise of the quadriceps mechanism. Active knee flexion was begun the day of surgery in all knees and continued during the rehabilitation process. Bracing was not necessary for varus or valgus laxity.
The patients were evaluated postoperatively at 1 month, 3 months, and then at yearly intervals to assess knee function, stability, and pain. Range of motion and valgus angle of the knees were estimated with the aid of a goniometer. Varus and valgus laxity measurements were estimated at full extension, and AP laxity was estimated at 90° flexion. Anteroposterior and lateral radiographs were taken during each followup and were evaluated for radiolucent lines, bone hypertrophy, bone atrophy, and migration of implants. Current radiographs were compared with all previous radiographs.
None of the patients was lost to followup other than 17 patients (17 knees) who died. Of the 72 knees (71 patients) at risk for failure at 5 years postoperatively, one knee failed because of loosening of the tibial component, giving a 5-year survival rate of 98.6% (95% confidence interval, ± 4.2%). None of the knees was revised because of infection. Of 32 knees (31 patients) at risk for failure at 10 years postoperatively, one knee failed because of loosening, giving a 10-year survival rate of 96.9% (95% confidence interval, ± 6.1%). The mean Knee Society scores were 89.2 ± 4.5 at 5 years postoperatively and 84.5 ± 3.6 at 10 years postoperatively. Both scores were higher (p < 0.01, p < 0.02, respectively) than the preoperative scores.
Complications included one fracture of the femur, one traumatic avulsion of the wound, and one patellar fracture. Two knees (two patients) with posterior-stabilized implants had fracture of the polyethylene post at 100 and 113 months postoperatively and required revision of the polyethylene component to restore stability to the knee. No patient had significant recurvatum postoperatively. Postoperative mean flexion contracture at 5 years was 2.0° ± 0.5° and at 10 years was 1.1° ± 0.9°. The mean extension lags were 2.1° ± 1.5° at 5 years and 2.6° ± 1.7° at 10 years postoperatively. The mean flexion 5 years after surgery was 101° ± 13°.
Varus and valgus and AP stability remained unchanged during the followup. None of the patients had symptomatic instability, recurvatum, or flexion contracture greater than 10°. Mean valgus laxity at 5 years postoperatively was 3° ± 2.5° and at 10 years was 4° ± 3.2°. Varus laxity at 5 years postoperatively was 4.5° ± 3.3° and at 10 years was 4.6° ± 3.6°. Anterior laxity at 5 years postoperatively was 6.9° ± 4.5° and at 10 years was 7.2° ± 3.8°. No significant difference occurred between preoperative and postoperative values for laxity or alignment.
Patellar tracking was judged to be stable and midline during physical examination of all knees, but lateral patellar tilt was seen on radiographs in three knees (4%) at 5 years postoperatively, and in one knee (3%) at 10 years postoperatively.
The radiographic evaluation of the implants postoperatively revealed radiolucent lines in 60 knees (60 patients, 84.5%) under the tibial component 5 years postoperatively, and in 17 knees (17 patients, 53%) under the tibial component 10 years postoperatively. All radiolucent lines measured less than 1 mm, and none of the knees had progressive widening of radiolucent lines. There were no signs of migration of the femoral or tibial components. Fifty-seven knees (57 patients, 80%) had less than a 1-mm radiolucent line around the tibial stem 5 years postoperatively, and 25 knees (25 patients, 78%) had similar radiolucency 10 years postoperatively. Fifty-eight knees (58 patients, 82%) had a 1-mm or less radiolucent line around the femoral stem 5 years postoperatively, and 27 knees (27 patients, 84%) had radiolucent lines 10 years postoperatively. None of the knees had progressive stem radiolucent lines or evidence of migration of the stem medially, laterally, anteriorly, or posteriorly. No implant breakage, loosening, or migration occurred during the followup.
Revision TKA often is accompanied by a high failure rate because of implant loosening and ligament imbalance.12 Surgical approaches that jeopardize ligament attachments and blood supply to the bone, and cementing techniques that fail to achieve long-term fixation of the implants to bone, result in a combination of soft tissue and bone deficiencies that lead to early failure, bone destruction, and additional damage to soft tissue stabilizing structures. A successful revision TKA requires preservation of bone stock and ligament attachments, and long-term success requires restoration of lost bone stock and proper tensioning of ligaments. Techniques that involve massive soft tissue stripping, cementing into deficient bone, and stabilized implants that fail to correctly tension the soft tissues all result in high failure rates in knees.2,11,13 Recent improvements in revision TKA have included modular designs that allow the soft tissues to be tensioned correctly and bone to be restored so that constrained implants may be used along with ligament balancing and bone reconstruction techniques.1,5,6,14 Cementless bone reconstruction and ligament balancing techniques, as described in this paper, have been shown to be reliable and effective in maintaining good knee function in long-term followup and without the use of linked hinges or varus and valgus stabilizing mechanisms that are built into the implants.
A simple surgical procedure is necessary for management of the complex deformities and tissue deficits that are present in massively failed TKA. Intramedullary alignment techniques using easily identified landmarks for rotational position of the implants provide a reliable and reproducible technique for correct placement of the implants in relation to one another for alignment and stability in flexion and extension.
The ligament and soft tissue capsular structures and the quadriceps mechanism and patellar tendon invariably experience pathologic changes during failure of the TKA. These soft tissue and musculoskeletal changes are unpredictable, and soft tissue tension often varies in flexion and extension. Therefore, using bony landmarks for preoperative planning to develop a reliable plan for positioning and sizing the implants preoperatively is not feasible. Instead the plan must be developed during the operation. The surgeon must evaluate the soft tissue stabilizing characteristics of the knee after the correct sizes of trial implants have been chosen and the trial implants have been placed on the best available bone stock. The surgeon is restricted in choices for sizes of implants on the femoral and tibial side. The mediolateral dimension of the distal femur determines the size of the implant, and the anterior femoral cortex determines its position. A small amount of leeway is available to increase or decrease the size of the implant to accommodate differences in stability in flexion and extension, but major changes in these dimensions are not realistic. Buildups on the femoral component to move the component distally and to achieve bone contact posteriorly are necessary features of revision arthroplasty systems. Likewise, buildups on the tibial side and thick polyethylene inserts are necessary to tension the remaining soft tissue envelope to achieve overall stability. Although the femoral implant often migrates into the femoral bone stock and changes the size of implant that can be accommodated, some bone stock almost always is present in either the medial or lateral epicondylar area. The intramedullary stem places the implant in the centralized position, and the edge of the implant should engage the bone near the edge of the remaining bone stock. Sizing in this regard is empirical, but easily done. Likewise, distal and posterior buildups are empirically placed and removed to achieve correct ligament balance in flexion and extension.
Numerous pitfalls exist in the decision-making process in revision TKA. In some cases it may seem beneficial to distalize the femoral component more than 1–2 cm to achieve stability in full extension so that taller tibial components will not be necessary. Although this allows the surgeon to place the joint space distal to the inferior pole of the patella, this technique results in destabilization of the knee in flexion because the center of rotation of the tibia about the femur is moved distally, leaving the new center of rotation unattached to soft tissue structures that are attached to the tibia. Therefore the knee is unstable in flexion in these cases whereas it may be stable in extension because of the effect of the posterior capsule. A much wiser choice is to stabilize the knee in flexion first, then extend the knee and adjust the joint space distally or proximally to allow the knee to come to full extension and stabilize.
To achieve correct varus-valgus alignment in flexion, the femoral component is aligned rotationally with the epicondylar axis. This results in a knee that is aligned correctly in flexion and extension with the ligaments attached to the epicondylar area of the femur, which can stabilize the knee throughout the flexion-extension arc.
The patella then must be managed separately. If it subluxes or dislocates laterally, then lateral patellar release and occasionally distal realignment of the patellar mechanism is necessary. If the patella is too low, then its surface must be adjusted so that it does not impinge against the upper lip of the polyethylene. This often involves removal of the patellar component and trimming of the bone structure to assume the shape of a flat disc attached to the quadriceps and patellar tendons.8 Attempts to adjust the joint position in extension to avoid impingement of the patella against the tibial polyethylene result in unsolvable problems in flexion, and often lead the surgeon to choose inappropriately stable implants. The technique used in this series has proven to be simple and reproducible, and to provide long-term reliability.
I thank Diane J. Morton, MS, for editorial assistance during manuscript preparation, also William C. Andrea, CMI, for preparing the illustrations.
1. Barrack RL, Lyons TR, Ingraham RQ, Johnson JC: The use of a modular rotating hinge component in salvage revision total knee arthroplasty. J Arthroplasty 15:858–866, 2000.
2. Bonnin M, Deschamps G, Neyret P, Chambat P: Revision in non-infected total knee arthroplasty: An analysis of 69 consecutive cases. Rev Chir Orthop Reparatrice Appar Mot 86:694–706, 2000. In French.
3. Engh GA, Ammeen DJ: Bone loss with revision total knee arthroplasty: Defect classification and alternatives for reconstruction. Instr Course Lect 48:167–175, 1999.
4. Engh GA, Ammeen DJ: Classification and preoperative radiographic evaluation. Knee. Orthop Clin North Am 29:205–217, 1998.
5. Hube R, Reichel H: Modular revision systems in total knee arthroplasty: Possibilities and techniques. Orthopade 32:506–515, 2003. In German.
6. Jones RE, Skedros JG, Chan AJ, Beauchamp DH, Harkins PC: Total knee arthroplasty using the S-ROM mobile-bearing hinge prosthesis. J Arthroplasty 16:279–287, 2001.
7. Kanamiya T, Whiteside LA, Nakamura T, et al: Ranawat Award Paper: Effect of selective lateral ligament release on stability in knee arthroplasty. Clin Orthop 404:24–31, 2002.
8. Parvizi J, Seel MJ, Hanssen AD, Berry DJ, Morrey BF: Patellar component resection arthroplasty for the severely compromised patella. Clin Orthop 397:356–361, 2002.
9. Saeki K, Mihalko WM, Patel V, et al: Stability after medial collateral ligament release in total knee arthroplasty. Clin Orthop 392:184–189, 2001.
10. Scuderi GR: Revision total knee arthroplasty: How much constraint is enough? Clin Orthop 392:300–305, 2001.
11. Shaw JA, Balcom W, Greer III RB: Total knee arthroplasty using the kinematic rotating hinge prosthesis. Orthopedics 12:647–654, 1989.
12. Sierra R, Pagnano M, Trousdale R: Reoperations after 3200 revision TKRs: Rate, etiology and lessons learned. Clin Orthop, 2004.(In Press)
13. Springer BD, Hanssen AD, Sim FH, Lewallen DG: The kinematic rotating hinge prosthesis for complex knee arthroplasty. Clin Orthop 293:283–291, 2001.
14. Walker PS, Manktelow AR: Comparison between a constrained condylar and a rotating hinge in revision knee surgery. Knee 8:269–279, 2001.
15. Whiteside LA: Cementless revision total knee arthroplasty. Clin Orthop 286:160–167, 1993.
16. Whiteside LA: Distal realignment of the patellar tendon to correct patellar tracking abnormalities in total knee arthroplasty. Clin Orthop 344:284–289, 1997.
17. Whiteside LA: Selective ligament release in total knee arthroplasty of the knee in valgus. Clin Orthop 367:130–140, 1999.
18. Whiteside LA: Positioning the femoral component: The effect of proper ligament balance. Am J Knee Surg 13:173–180, 2000.
19. Whiteside LA, Bicalho PS: Radiologic and histologic analysis of morselized allograft in revision total knee replacement. Clin Orthop 357:149–156, 1998.
20. Whiteside LA, Mihalko WM: Surgical procedure for flexion contracture and recurvatum in total knee arthroplasty. Clin Orthop 404:189–195, 2002.
21. Whiteside LA, Saeki K, Mihalko WM: Functional medial ligament balancing in total knee arthroplasty. Clin Orthop 380:45–57, 2000.