➢ Failure of a total knee arthroplasty could be the result of multiple intrinsic and extrinsic factors.
➢ Understanding the causes of failure in study populations is essential for guiding future strategies to optimize safety and outcome as well as the utilization of health-care resources.
➢ A systematic evaluation is crucial for identifying the mechanism of failure and for establishing a standardized treatment plan.
Total knee arthroplasty is one of the most successful interventions for reducing knee pain and for improving function and quality of life. Advances in implant materials and design as well as surgical technique have increased the longevity of prostheses and have decreased the rate of revision surgery to <5% within ten years1. In light of the rapidly growing demand for knee arthroplasty, especially in younger and more active patients, the number of revision procedures is projected to grow sevenfold, from 38,000 in 2005 to 270,000 by 20302,3. Understanding the causes of failure is essential for guiding future strategies to optimize safety and outcome as well as the utilization of healthcare resources.
The failure of total knee arthroplasty presents clinical challenges to orthopaedic surgeons and their patients. A systematic evaluation is crucial for identifying the mechanism of failure and for developing an appropriate treatment protocol. Once the diagnosis has been established, the treatment can then be divided into operative and nonoperative options. The failure of total knee arthroplasty does not necessarily imply the need for revision surgery. It is important to avoid surgical intervention before a diagnosis is made, even in cases of pain with no clear etiology4,5. The modes of failure of total knee replacement can be classified as intra-articular or extra-articular, as biological or mechanical, as early or late, or according to the predominant symptom, namely, pain, instability, or stiffness. This article presents a current overview of the multiple etiologies of failure of total knee arthroplasty.
Modes of Failure
Infection remains the most devastating and costly complication of knee replacement as well as one of the most common causes of revision surgery, accounting for a quarter of revision procedures6. It must be considered in the evaluation of every patient with pain at the site of a total knee arthroplasty. The reported rates of infection following primary total knee arthroplasty have varied widely in the literature, from 0.51% to 1.55%7-10. Risk factors include any condition that impairs host defense mechanisms or that prolongs wound-healing, namely, advanced age, malnutrition, diabetes mellitus, prolonged hospital stay1,2, and rheumatoid arthritis11,12. Patients with hepatitis C have higher complication rates, longer hospital stays, and increased susceptibility to infection and bleeding13,14.
Surgical risk factors include prolonged operative time (>150 minutes), blood transfusion, and simultaneous bilateral arthroplasty15,16. The implanted prosthesis is believed to decrease the size of the bacterial inoculum needed for infection by >100,000-fold17 because of the degranulation of granulocytes around it, leading to decreased superoxide dismutase and loss of defense capacity, particularly against Staphylococcus aureus. The polysaccharide glycocalyx layer seals off the infection and protects the bacteria from the host response, making them 1000 to 1500 times more resistant to antimicrobials18.
Periprosthetic infections are typically classified according to the time of clinical manifestation after the procedure as well as the route by which the infectious organism gains access to the joint space. Early postoperative infections present acutely within the first three weeks after surgery and are attributed to seeding at the time of implantation or to wound contamination. Such infections are commonly caused by Staphylococcus aureus or gram-negative bacilli and typically present with acute joint pain, effusion, erythema, warmth, discharge, and fever. More indolent organisms such as coagulase-negative staphylococci present as chronic postoperative infections more than three weeks postoperatively and may manifest as implant loosening, stiffness, limited range of motion, or persistent pain. A high index of suspicion is crucial, especially in the absence of other classic symptoms. The third type of periprosthetic infection includes those that are due to hematogenous seeding of a previously well-functioning prosthesis from skin, soft-tissue, urinary, respiratory, gastrointestinal tract, or dental infections15,19-21. Guidelines from the American Academy of Orthopaedic Surgeons (AAOS) and the American Dental Association have evolved from recommending lifelong prophylaxis in 2012 to advising a change in such practice because of limited evidence showing an association between dental procedures and prosthetic joint infections. In such cases, it is crucial to ascertain the onset and duration of symptoms to determine treatment options. An onset of symptoms of less than six weeks in duration is considered acute and might still be amenable to irrigation and debridement without explantation22.
The AAOS developed a clinical practice guideline in 2010 to standardize the evaluation of suspected periprosthetic joint infections and set forth the following criteria for the diagnosis of infection16: (1) the presence of a sinus tract communicating with the prosthesis, or (2) isolation of a pathogen on culture of two separate tissue or fluid samples obtained from the affected prosthetic joint, or (3) the presence of at least four specific findings, including an elevated serum erythrocyte sedimentation rate (ESR) or serum C-reactive protein (CRP) concentration, an elevated synovial white blood-cell (WBC) count, an elevated synovial neutrophil percentage (PMN%), purulence in the joint, isolation of a microorganism on culture of one sample of periprosthetic tissue or fluid, or the presence of more than five neutrophils per high-powered field in five fields observed on histologic analysis of periprosthetic tissue at 400-times magnification.
The workup for any suspected periprosthetic infection begins with evaluation of the serum ESR and CRP levels. When both are within the normal range, an infection is unlikely (sensitivity, >90%)16. The ESR remains elevated for as long as a year after surgery but begins to decline after the first month. The CRP level usually normalizes within two to four weeks after surgery. Different threshold values have been considered for each test, ranging from 15 to 50 mm/hr for the ESR and from 0.5 to 3.2 mg/L for the level of CRP. Threshold values of 22.5 mm/hr for the ESR and 1.35 mg/L for the level of CRP have been associated with a sensitivity and specificity of 96% and 77%, respectively, if one is positive and of 89% and 93%, respectively, if both are positive23. The serum interleukin-6 (IL-6) level can aid in the workup, although IL-6 as a diagnostic marker has not yet been included in AAOS guidelines for the diagnosis of infection. An elevated IL-6 level of >10 pg/mL has demonstrated a sensitivity and specificity of 100% and 95%, respectively24.
Abnormal results on serum testing warrant joint aspiration, microbiologic analysis, and a synovial WBC count with differential. A synovial WBC count in the range of 1100 to 3300 cells/μL or a neutrophil count of >65% (range, 64% to 80%) are highly suggestive of chronic periprosthetic infection. However, this is based on the notion that these values return to baseline levels three months postoperatively. In the acute postoperative phase (within six weeks after the index arthroplasty), the levels indicative of infection are not as clear-cut16 as they may be elevated even in the absence of infection25. A synovial WBC count of approximately 27,800 cells/mL with 89% neutrophils has been shown to have a sensitivity of 84% and specificity of 99% for detecting infection25. With regard to the time frame between the early and late phases, the synovial WBC count and neutrophil percentage have been found to decrease at different rates26. The total neutrophil count reflects both parameters and may be a more reliable indicator of infection. A total neutrophil count threshold of >25,000 cells/mL during the first six weeks postoperatively, >6500 cells/mL from six weeks to three months, and >2000 cells/mL thereafter is highly suggestive of infection26.
Synovial fluid should be sent for aerobic, anaerobic, and fungal cultures to identify the causative agent and its sensitivity to antibiotics. Gram staining of synovial fluid has shown a sensitivity of 27% and a specificity of 99.9%27, and the results of gram staining are affected by recent antibiotic use. Overall, Gram staining has been surpassed by other, superior tests for the accurate diagnosis of infection16. Novel markers have been investigated as potential diagnostic indicators in serum as well as synovial fluid. Elevated levels of synovial human β-defensin 3 and cathelicidin LL-37 may prove to be reliable for distinguishing septic from aseptic loosening28. Additional investigation is required to validate their role in diagnosing periprosthetic infections.
Imaging is an important part of any workup for infection at the site of a total knee replacement. Radiographs may show a radiolucent line between the prosthesis and the cement, but such lines may be seen in the absence of infection, and infection may present with normal radiographic findings29. Signs that are suggestive of infection include irregular or scalloped borders on the endosteal surface of the cortex, marked periosteal reaction, or new bone formation30-32. As computed tomography (CT) and magnetic resonance imaging (MRI) can be hindered by hardware-induced artifact, nuclear radionuclide imaging is routinely used to assess implant failure. However, bone scintigraphy and gallium scans are limited by the inability to distinguish septic from aseptic loosening33. A bone scan can be used as a screening test and, in combination with a gallium scan, can reach a sensitivity of up to 65% to 80% for the diagnosis of infection. Unfortunately, increased uptake persists in areas of bone remodeling for as long as one year postoperatively34-36. Therefore, nuclear scans have limited specificity and are better for ruling out infection than they are for detecting infection37. Fluorodeoxyglucose positron emission tomography (FDG-PET) scanning has been explored as an alternative, with variable and inconclusive results38-41. Currently, the greatest sensitivity and specificity can be achieved in association with combined leukocyte and marrow imaging33. The AAOS guidelines recommend nuclear imaging for patients for whom the diagnosis of periprosthetic infection has not been established and who are not scheduled for reoperation16.
Acute infections may be treated with irrigation and debridement with or without polyethylene exchange42. When the infection is believed to have invaded the prosthetic-bone interface, revision arthroplasty is required. Single-stage exchange arthroplasty, during which the prosthesis is removed and replaced in the same procedure, has limited indications. Single-stage exchange can be successful in 73% to 100% of patients with good soft tissue and no severe comorbidities who are infected by a single pathogen that is nonresistant and nonvirulent43-45. Second-stage arthroplasty is the gold standard of treatment46 and consists of first-stage debridement and antibiotic treatment for an interval period. A new prosthesis is implanted after four to six weeks if serological markers are within the normal range. This approach is successful for the treatment of >90% of periprosthetic infections15. Cases involving unknown or resistant organisms, sinus tracts, or sepsis should always be treated with second-stage revision arthroplasty47.
Loosening and Component Failure
Osteolysis is the predominant mechanism of late failure of total knee arthroplasty48,49. The Knee Society defines osteolysis as “an expansile lytic lesion adjacent to one of the implants greater than or equal to 1 cm in any one dimension or increasing in size on serial radiographs/CT scans.”50 Osteolysis usually precedes aseptic loosening, which is confirmed intraoperatively or identified radiographically as a change in implant position or a progressive radiolucent line at the bone-cement or bone-implant interface50. It is important to distinguish between focal osteolytic lesions and linear osteolysis at the bone-cement or cement-implant interface, both of which may be caused by wear debris. Focal osteolytic lesions can occur around well-fixed implants without implant loosening or migration. The possibility of focal osteolytic lesions occurring around well-fixed implants without loosening leads to progressive bone loss, which may jeopardize the structural strength of the construct and result in periprosthetic fractures. Reported rates of aseptic loosening have ranged from 6% to 30% after total knee arthroplasty without cement and from 0% to 16% after total knee arthroplasty with cement51.
The rate of polyethylene wear and the development of osteolysis are functions of implant design, surgical technique, and patient-related factors51,52. The average time until the radiographic appearance of osteolysis is approximately two and one-half to three years postoperatively53,54. Incongruent articulations, thin polyethylene liners, sterilization with gamma irradiation in air, and tibial base plates with fixation screws are risk factors for osteolysis55. Polyethylene wear after total knee arthroplasty is a major determinant of the longevity of the implants. Linear wear rates of >0.1 mm/year have been associated with osteolysis and subsequent component loosening56. Polyethylene thickness of <8 mm leads to localized load transmission that exceeds the inherent yield strength of polyethylene and increases the risk of catastrophic failure57.
A major concern related to periprosthetic osteolysis is that patients may be completely asymptomatic even with substantial bone loss58. Most patients remain asymptomatic or report mild to moderate pain with activity. A triad of effusion, pain, and change in coronal alignment, usually into varus, is strongly suggestive of accelerated polyethylene wear55. Osteolysis is diagnosed incidentally in most patients, manifesting as radiolucent areas adjacent to the component, sometimes with sclerotic borders or thinning of the cortex. Radiographs tend to underestimate the extent of osteolysis, and CT59 or MRI with metal suppression60 may be needed to appreciate the full extent of the process. A technetium-99m-labeled bone scan is useful to rule out loosening in problematic cases, but it is not recommended as a primary diagnostic modality61.
Once an osteolytic lesion is found, the priority is to prevent component failure or periprosthetic fracture. Small areas of limited osteolysis should be followed yearly for progression. Pain or discomfort, expansion of the osteolytic lesions, cortical disruption, component loosening or instability, and periprosthetic fractures are indications for revision replacement surgery62.
Instability describes symptoms rather than a diagnosis and encompasses abnormal and excessive displacement of the articular elements. Patients with preoperative deformities or neuromuscular abnormalities are more prone to the development of instability63. Identifying the underlying etiology is the most important step in planning for surgery64-67.
The classic form is varus-valgus instability, which is most commonly identified in extension68. Such instability is attributed to improper soft-tissue balancing, bone resection, or component malalignment. Patients may present with pain or buckling and may walk with a stiff-legged gait to decrease discomfort. The diagnosis is confirmed on the basis of stress radiographs. Instability in extension may be symmetric or asymmetric. Asymmetric extension instability is more common and usually is due to persistent ligamentous asymmetry in knees with preoperative angular deformities, improper soft-tissue balancing, or malalignment of the femoral or tibial component in the coronal plane. Revision surgery with correct soft-tissue balancing or the use of a constrained implant may be necessary. Isolated symmetric extension instability is due to excessive bone removal from the distal part of the femur and can be treated with the addition of distal femoral augments. Excessive proximal tibial resection leads to concurrent flexion instability and is corrected with a thicker tibial insert69,70.
Flexion or anteroposterior instability stems from the failure to balance the flexion and extension gaps in the sagittal plane. It also may be due to inadequate restoration of the joint line or posterior condylar offset. Symptoms range from recurrent effusions to pain and instability when going up and down stairs, standing from a seated position, or walking on uneven surfaces63,69. This type of instability is the result of excessive strain on the secondary knee stabilizers such as the popliteus, biceps tendons, and extensor mechanism64,65. A rare but serious possible manifestation is knee dislocation71, which can occur in association with a combination of varus or valgus stress and posterior translation as, for example, when the ankle is placed on the contralateral knee in order to put on shoes or socks. The first episode of dislocation should be treated with closed reduction and bracing. Recurrent dislocations should be addressed with the insertion of a thicker polyethylene insert or with conversion to a constrained condylar implant72. Flexion instability is also evident in cruciate-retaining knees with inadequate posterior cruciate ligament (PCL) balancing, late rupture, or insufficiency. Presentation usually occurs months after the procedure because the rest of the knee structures provide a temporary protective effect73. Assessment of the knee in 90° of flexion should be part of the routine physical examination of any patient with knee pain, regardless of the type of prosthesis69. Common findings include tenderness over the pes anserinus bursa and a positive posterior drawer or sag test. The role of nonoperative treatment is limited in such cases. Revision might be required to reduce and balance the flexion-extension gaps or to increase constraint.
Global instability is a combination of loose flexion and extension gaps. Recurvatum or hyperextension deformity is usually seen in patients with neuromuscular conditions and marked quadriceps weakness or rheumatoid arthritis. Primary constrained or rotating-hinge prostheses are particularly beneficial for patients with global or recurvatum instability as well as for those with Charcot arthropathy73,74.
Achieving an acceptable range of motion remains imperative to the success of total knee arthroplasty. The prevalence of stiffness following total knee arthroplasty has ranged from 1.3% to 5.3%75-77. Stiffness is defined as a limited range of motion, reported by the patient and verified during a physical examination, with extension limited to 15° short of full extension or flexion of <90°50. Postoperative stiffness can be the result of patient-related, technical, or postoperative factors.
Preoperative range of motion remains the most potent predictor of postoperative range of motion75,78-80. Ideal component positioning is important for the prevention of stiffness. The rotational position of the femoral component should be parallel to the epicondylar axis in order to maximize range of motion79,80. Controversy exists about the role of the posterior tibial slope with regard to the postoperative maximum flexion angle81-87. With PCL-retaining prostheses, posterior condylar offset plays a critical role in the flexion angle and must be decreased to achieve maximum flexion84,88. In patients with PCL-sacrificing prostheses, the degree of posterior condylar offset does not significantly affect postoperative flexion86,88-90.
Overstuffing the patellofemoral joint can lead to a loss of extension79,80,91. Particular attention needs to be given to the elevation of the joint line as it can lead to a lack of flexion79. The location of the joint line needs to be maintained when the flexion and extension gaps are being balanced. In addition, failure to create balanced flexion and extension gaps can lead to stiffness79,80,87. All osteophytes must be removed78,80,92, and the PCL must be adequately released and balanced to avoid flexion limitation79,93. Postoperative rehabilitation is key in order to ensure optimal range of motion. Pain, infection, delayed wound-healing, hemarthrosis, arthrofibrosis, component failure, or periprosthetic fractures can substantially limit postoperative range of motion78-80.
Patellofemoral joint problems and extensor insufficiency are frequent causes of pain after total knee arthroplasty with or without patellar resurfacing. The most common complications include extensor mechanism disruption, soft-tissue impingement, patellar malalignment and maltracking, loosening, osteonecrosis, and fracture71,94. The typical presentation consists of anterior knee pain that is exacerbated by activities such as descending stairs, prolonged sitting, and squatting. Although it has been reported that nonresurfaced patellae carry higher risks of anterior knee pain and secondary resurfacing95,96, there is no conclusive evidence to support resurfacing in terms of anterior knee pain, outcome scores, and reoperation risk71,97.
It is important to differentiate mechanical pain from functional pain related to the extensor mechanism98. Extensor insufficiency presents with pain during eccentric loading as, for example, when descending stairs. Patellar maltracking or subluxation occurs as a consequence of extensor mechanism imbalance or component malpositioning and causes pain, crepitus, loosening, and fracture. Factors that increase the quadriceps angle increase the lateral tilt and displacement force as well as contact pressure99. Rotational malalignment of the femoral or tibial components, excessive resection or filling of the patellofemoral space, alteration of the joint line, osteophytes, and asymmetrical patellar resection all can contribute to patellar maltracking100,101. Pain approximately 5 cm distal to the knee joint in the anteromedial aspect of the tibia may be due to pes anserine bursitis, whereas pain along the patellar tendon and tibial tubercle may be due to patellar tendinitis. A painful snapping or audible popping sensation in the posterolateral aspect of the knee raises suspicion of impingement of the popliteus tendon over a retained lateral femoral osteophyte or the edge of the prosthesis, which might require arthroscopic release102,103.
The diagnosis of patellar instability usually can be made on the basis of physical examination and Merchant radiographs. CT may be needed to assess the rotational alignment of the components. Treatment of patellar subluxation relies on aggressive quadriceps rehabilitation and patellofemoral bracing. Revision of malrotated components or additional soft-tissue procedures might be indicated. Anterior knee pain with localized tenderness over the patella could indicate ischemia, necrosis, or fracture. Ischemia appears to be more prevalent after lateral release with or without preservation of the superior lateral genicular artery104. Patellar fracture is an infrequent complication after total knee arthroplasty, with a reported prevalence of 0.68% in a series of 12,464 consecutive knees105. Patellar clunk is sometimes observed in patients with posterior-stabilized total knee prostheses as a result of a painful fibrous nodule or synovial proliferation in the intercondylar fossa and might require arthroscopic or open resection106.
Pain may occur as a result of the formation of a neuroma after surgical laceration of the infrapatellar branch of the saphenous nerve or any other nerve with a cutaneous distribution that might be disrupted in surgical approaches to the knee107. The response to division of a peripheral nerve involves degeneration of the distal axons and sprouting of the proximal nerve fibers, followed by the regeneration of the proximal axon sprouts distally along a chemotactic gradient108.
Neuromas should be suspected in patients who have had persistent pain for at least six months after the exclusion of other etiologies of knee pain such as infection or mechanical causes. Diagnosis can be confirmed on the basis of a positive Tinel sign or selective anesthetic injections55. Initial treatment has largely relied on conservative modalities in light of the fact that resection and denervation have been associated with recurrence109. Wrapping the nerve in collagen conduits or autologous fat grafts has shown promising results in terms of pain resolution, with no reported recurrence110,111.
Other Etiologies of Pain
Etiologies of pain after total knee arthroplasty include heterotopic ossification, bursitis, chronic regional pain syndrome, hemarthrosis, pseudomeniscus, and metal hypersensitivity.
Heterotopic ossification could cause pain and limited range of motion after total knee arthroplasty, although most cases are usually asymptomatic73. Heterotopic ossification in the knee usually occurs in the quadriceps expansion and may resolve in a few months112. Predisposing factors include a history of heterotopic ossification, trauma, previous operations, and infection. Prophylaxis with a single fraction of radiation to the knee has been found to be effective in high-risk patients113.
Complex regional pain syndrome after total knee arthroplasty is observed in patients with relatively smooth postoperative courses who cease to improve and fall short of the expected recovery. These patients have an exaggerated physiological response and report skin, muscle, and joint pain out of proportion to physical findings; sensory disturbances; and pain at rest5,114. The most favorable prognostic factor is the institution of treatment within the first six months. Treatment consists of physical therapy, pharmacologic agents, and sympathetic blockade, which can confirm the diagnosis115.
Recurrent hemarthrosis can cause pain and restrict the range of motion after any total knee arthroplasty116 and must be immediately investigated if suspected117. The entrapment of proliferative synovial tissue between the prosthetic articulations has been implicated and leads to fibrosis and hemosiderin deposition116,118. If conservative treatment fails, arthroscopic synovectomy119, synoviorthesis120, and surgical embolization are effective121-123.
Pseudomeniscus is a rare condition that causes a painful impingement of soft tissue between the femoral and tibial components4. This condition could be due to a retained meniscal fragment and/or regeneration of fibrocartilaginous tissue triggered by exposing mesenchymal-derived cells to compressive forces124,125. Patients may describe a sharp pain under weight-bearing conditions and knee flexion as, for example, during walking on stairs or rising from a chair. Similar symptoms also have been described in association with retained femoral osteophytes126. Knee arthroscopy is used for diagnosis and allows for removal of the retained or regenerated fragment124,127.
Hypersensitivity to metals (nickel, chromium, cobalt) affects 10% to 17% of the general population128-131. Metal allergy is a delayed-type hypersensitivity and should be considered in patients with knee pain, effusion, itching on different body parts, and functional impairment, especially if they have a positive history of allergic diseases39. The best outcome is achieved by revising the implant with a nonallergenic implant with alternative bearings or coating such as oxidized zirconium or ceramic132.
In certain cases, the etiology of pain or dissatisfaction cannot be pinpointed despite extensive and comprehensive evaluation. It has been suggested that psychological factors influence the outcomes of knee replacement. Depression, anxiety, or a tendency to somatize have been identified as predictive factors for poorer outcomes after total knee arthroplasty and should be investigated in such cases133. In addition, undiagnosed hip or spine abnormality should always be considered, especially when the pain constellation is exactly the same before and after surgery.
Failure of total knee arthroplasty could be a result of multiple intrinsic and extrinsic causes. The first differential diagnosis that should be excluded is periprosthetic infection. Successful treatment with or without a revision procedure requires an accurate understanding of the cause of failure.
Source of Funding: No external funding of any kind was received for this manuscript.
Investigation performed at the Department of Orthopedic Surgery, University of Illinois at Chicago, Chicago, Illinois
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. None of the authors, or their institution(s), have had any financial relationship, in the thirty-six months prior to submission of this work, with any entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, no author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2014 by The Journal of Bone and Joint Surgery, Incorporated