Fig. 17.1
A small sinus present around the incision several years after TKA should raise the concern regarding the likelihood of a substantial chronic late infection
Weight-bearing radiographs should be obtained on all patients presenting with a painful total knee arthroplasty. Radiographic signs of loosening are unlikely in the acute postoperative period or in late hematogenous infections that present acutely. TKAs with chronic long-standing infections may have evidence of loosening of the implants, but they are usually indistinguishable from those failures that occur for noninfectious reasons. Subtle findings, however, may be present with chronic infection, particularly when there is osteomyelitis, namely, endosteal erosion, reactive periosteal bone, and occasionally heterotopic ossification (Fig. 17.2).
Fig. 17.2
(a–c) Anteroposterior radiograph of a knee after removal of an infected revision TKA and implantation of antibiotic-impregnated spacer blocks . Note the periosteal reaction of the medial and lateral metaphyseal flares of the distal femur. Intraoperative biopsies of these bony sites showed chronic osteomyelitis. Distal femoral resection was necessary to eradicate the extensive osteomyelitis of the distal femur, and eventually a hinged knee arthroplasty with distal femoral modular augments was necessary (From Lotke PA, Lonner JH, eds. Master Techniques in Orthopedic Surgery: Knee Arthroplasty. Philadelphia: Lippincott Williams & Wilkins, 2002, Fig. 22–29 A–C, with permission)
While a variety of diagnostic tests have been advocated and used to evaluate the painful TKA, the frustrating reality is that many are inaccurate and cannot be relied on in isolation to clearly establish whether or not an infection is present. Nonetheless, when taken in concert, several studies can be helpful. Considering the cost of treating the infected TKA, which may be 3–4 times that of a primary knee arthroplasty, it is important that unnecessary diagnostic tests be avoided when evaluating the knee for deep infection [7]. Despite our best intentions, approximately 7–12% of deep infections after total joint arthroplasty are undetected by standard preoperative diagnostic tools [22].
Some serologic studies are more useful than others. The peripheral white blood cell count is rarely elevated in the setting of infections after total knee arthroplasty unless there is clear bacteremia. Windsor et al. [23] reported that only 28% of cases had peripheral white blood cell counts greater than 11,000 in the presence of deep knee infection. Erythrocyte sedimentation rate (ESR ) peaks 5–7 days after surgery, normalizing gradually by approximately 3 months, whereas C-reactive protein (CRP) peaks 2–3 days after surgery and normalizes within 3 weeks [24, 25]. In a retrospective review of revision TKA patients, Bare et al. reported that ESR had a specificity of 63% and sensitivity of 55% for infection, whereas CRP had a specificity of 60% and sensitivity of 63% [25]. Austin et al. found that using ESR and CRP together made for an excellent screening test, with high sensitivity and negative predictive value as well as low cost in ruling out PJI after TKA [26].
Radioisotope scanning has been used in the evaluation of the painful joint arthroplasty, with variable results. Technetium scans have proven ineffective in the majority of cases, with a sensitivity of 60% and specificity of 65% [27]. In general, technetium scans are unnecessary in the evaluation of the failed total knee arthroplasty, because they are ineffective in distinguishing between mechanical and septic loosening. Technetium scans, however, may be effective in identifying occult loosening of a painful total knee arthroplasty, and a total body technetium scan might be considered in the evaluation of other joint arthroplasties to rule out metachronous polyarticular infection. Indium scans are moderately more accurate than technetium scans. One study by Rand and Brown, which evaluated 38 total knee arthroplasties, found that indium scans had a sensitivity of 83%, specificity of 85%, and accuracy of 84% [28]. Scher et al. subsequently reported on 153 indium scans that were done to evaluate painful total hip, knee, or resection arthroplasties. In that series there were 41 total knee arthroplasties evaluated, and the indium scans had a sensitivity of 88%, specificity of 78%, positive predictive value of 75%, negative predictive value of 90%, and accuracy of 83% [29]. This study showed a high-percentage false-positive indium scan results in knees that were loose but not infected. It is not clear whether the tendency for false-positive scan results is related to the indiscriminant labeling of both acute and chronic inflammatory white cells, which may be present in infection or chronic inflammation of osteolysis, respectively, ongoing postsurgical inflammation, persistent joint inflammatory disease, or a combination of these factors [29]. The accuracy of indium scans may be enhanced by combining this study with a technetium-99 m sulfur colloid scan [30, 31]. The technetium-99 m sulfur colloid scan can detect increased density of bone marrow elements, which in the case of PJI are replaced by inflammatory mediators, including leukocytes, that inhibit the uptake of the technetium sulfur colloid . Matched areas on the indium and sulfur colloid scans are indicative of marrow packing and the absence of infection, thereby reducing the number of false-positive scan results [30]. In a study by Joseph et al., the combined indium/colloid scan was found to have 100% specificity, 46% sensitivity, 100% positive predictive value, 84% negative predictive value, and 88% accuracy. Sensitivity improved to 66%, negative predictive value to 89%, and accuracy to 90%, and specificity was reduced to 98% and positive predictive value to 91% when blood pooling and flow phase data were included [30]. The low sensitivity of these combined studies makes their routine use in the evaluation of the potentially infected total knee arthroplasty imprudent.
Aspiration of the knee is probably the most valuable diagnostic tool in determining the presence of deep knee infection [32, 33]. Aspirated fluid can be sent for cell count and culture. Historically, fluid and tissue cultures have been considered to be the gold standard but have since been found to result in a relatively high rate of both false-positive and false-negative results in 5–37% of cases and 2–18% of cases, respectively [34]. Recent efforts have been focused on synovial fluid analysis as a more accurate means of diagnosing potential PJI, namely, quantifying the synovial white blood cell count (WBC ) and polymorphonuclear neutrophil (PMN) percentage (PMN%). Reported values have ranged from WBC counts of 1100–3000 cells/μL and PMN% of 60–73% to be used as thresholds for defining chronic infection, with an accuracy of up to 99% [34]. In the acute postoperative period, however, the threshold has been observed to be higher, with a synovial WBC count of 10,700 cells/μL and PMN% of 89% providing 92% accuracy in diagnosing acute PJI [35]. Barrack et al. observed improved the sensitivity, specificity, and accuracy of synovial aspiration if the initial aspiration was delayed at least 2 weeks after discontinuing antibiotics [36]. Clearly then, the aspiration should be delayed at least 2 weeks after discontinuing antibiotics to avoid the potential effect of suppression. The gross appearance of the fluid should be assessed as well, although turbid fluid can be found in total knee arthroplasties affected by noninfectious processes such as gout or calcium pyrophosphate disease. Also, intraoperative purulence per se was unreliable in diagnosing PJI, with a sensitivity of 82%, specificity of 32%, positive predictive value of 91%, and negative predictive value of 17% [37].
In situations where the diagnosis of infection is unclear, molecular diagnostic techniques or intraoperative frozen section histoanalysis can provide further clues regarding the presence of infection. Polymerase chain reaction (PCR ) has been used to detect bacterial pathogens within synovial fluid after total knee arthroplasty. PCR amplifies bacterial DNA but unfortunately is extremely sensitive and suffers from a high rate of false-positive results [38]. Bergin et al. found that ribosomal RNA (rRNA)-based PCR helped overcome the aforementioned limitations of existing PCR techniques, demonstrating 100% specificity and positive predictive value, with a sensitivity equivalent to that of intraoperative culture [39]. Furthermore, this novel technique detected bacterial rRNA 7 days after sterilization, potentially allowing it to identify infection even after antibiotic administration. Evolving methods to enhance the specificity of PCR and other molecular techniques will potentially make these important diagnostic tools in the future.
Histological analysis of intraoperative frozen sections can be helpful in a number of patients in whom the presence of infection is equivocal or uncertain. This method, however, has been limited by variations in histological criteria and reference standards employed to diagnose infection. The reported data are further confounded by whether the intraoperative frozen sections were obtained as a screening or confirmatory test. Histological criteria have ranged from one “inflammatory cell” per high-power field (HPF) in ≥10 HPFs [40] to >10 PMNs per HPF in ≥5 HPFs [41]. Using less stringent histological criteria would maximize sensitivity at the cost of greater false-positive results and decreased specificity, whereas more stringent criteria involving a greater number of PMN per HPF would improve specificity at the expense of sensitivity. Moreover, numeric criteria are complicated by variations in the visual field size of different microscopes, the location of the neutrophils relative to capillaries that compose granulation tissue, and the different baseline histologic appearance in patients with underlying inflammatory arthropathy [42].
Synovial biomarkers represent the most recent advancement in the ongoing efforts to more accurately and consistently diagnose PJI. Deirmengian et al. evaluated the diagnostic characteristics of 16 synovial fluid biomarkers and reported that five, in particular α-defensin 1–3, neutrophil elastase 2, bactericidal/permeability-increasing protein, neutrophil gelatinase-associated lipocalin, and lactoferrin, demonstrated 100% sensitivity and 100% specificity in diagnosing PJI as defined by the latest Musculoskeletal Infection Society definition [43]. Of these, α-defensin has garnered ongoing attention as the leading synovial biomarker in the detection of PJI. α-Defensin immunoassay outperformed the leukocyte esterase strip, whose interpretation in several samples was limited by blood interference [44]. In this study, α-defensin immunoassay demonstrated both a sensitivity and specificity of 100% for PJI versus a sensitivity of 69% and specificity of 100% for the leukocyte esterase test strip. The robustness of α-defensin was independently validated through a prospective study. Bonanzinga et al. performed the α-defensin assay on intraoperative synovial fluid samples obtained from 156 patients undergoing revision total joint arthroplasty. The results of the assay were compared to intraoperative tissue samples sent for cultures and histologic evaluation. The authors found α-defensin to be a reliable test, with a sensitivity and specificity of 97%, positive predictive value of 88%, and negative predictive value of 99% [45].
No single test, however, can identify infection in all painful or failing total knee arthroplasties. It is important that a careful history be taken in all cases and a high index of suspicion maintained. ESR, CRP, and aspiration can be invaluable in many patients.
A comprehensive and systematic approach is essential to diagnosing PJI. The American Academy of Orthopaedic Surgeons’ Clinical Practice Guideline Summary contains a helpful algorithm that illustrates this principle and incorporates many of the aforementioned diagnostic tests in evaluating patients with suspected PJI [46].
Definition of Infection
Despite the myriad of diagnostic tools that have been developed and analyzed to detect PJI, a unified definition of PJI has remained elusive. This has made it difficult to standardize the diagnosis and reporting of PJI. To remedy this, the Musculoskeletal Infection Society (MSIS ) convened a workgroup in 2011 to issue diagnostic criteria for PJI [47]. The MSIS definition included two major criteria, one of which would indicate PJI, and six minor criteria, four or more of which would indicate PJI. Subsequently in 2013, a consensus group convened at the International Consensus Meeting on PJI endorsed the existing MSIS definition and further refined it by adding leukocyte esterase test as a minor criteria (Table 17.1) and defining the thresholds for each of the minor diagnostic criteria (Table 17.2) [48]. This modified MSIS definition of PJI has since been adapted by the Centers for Disease Control and Prevention, along with 130 societies and organizations.
Table 17.1
Definition of PJI
Major criteria |
Two positive periprosthetic cultures with phenotypically identical organism, or |
A sinus tract communicating with the joint, or |
Minor criteria |
Elevated serum CRP and ESR |
Elevated synovial fluid WBC count or ++ change on leukocyte esterase test strip |
Elevated synovial fluid PMN% |
Positive histological analysis of periprosthetic tissue |
A single positive culture |
Table 17.2
Threshold for minor diagnostic criteria
Acute PJI (<90 days) | Chronic PJI (>90 days) | |
---|---|---|
ESR (mm/hr) | — | 30 |
CRP (mg/L) | 100 | 10 |
Synovial WBC (cells/μl) | 10,000 | 3000 |
Synovial PMN% | 90 | 80 |
Leukocyte esterase | + or ++ | + or ++ |
Histological analysis of tissue | >5 PMN per HPF in 5 HPFs (at x400 magnification) | Same as acute |
Classification of Infection
Classifying infection after TKA based on symptom duration and the interval from surgery is important because it puts into perspective the potential treatment options. Acute postoperative or late hematogenous infections with acute onset are often treated with methods that attempt to retain the components, while more chronic infections frequently require component removal. In an effort to classify the clinical presentation of an infected TKA, three main categories have been described [6, 21] (Table 17.3).
Table 17.3
Classification of prosthetic joint infection
Positive intraoperative culture |
Early postoperative infection |
Superficial |
Deep |
Acute hematogenous |
Late chronic |
Early postoperative infections become evident within 4 weeks after index TKA. They may have started at the time of surgery or by hematogenous means. Aspiration should be done to rule out hematoma, the most common alternate diagnosis. CRP and ESR will likely still be elevated, as a result of the surgery, but very high values should raise concern. Gram’s stain and culture are sent to identify the presence of organisms. Do not assume that bacterial growth in broth only is a contaminant; when in doubt, reaspirate.
Acute hematogenous infections are those that present with a short duration of acute symptoms in a previously well-functioning knee. These may occur after invasive procedures, such as dental or genitourinary interventions, after abrasions or lacerations, or after remote or unrelated infections, but often there is no identifiable source of infection. While an acute hematogenous infection with 4 weeks or less of symptoms is often considered amenable to open debridement and retention of components, the results are optimized when patients present within 1 week of the onset of the infection.
Late chronic infections present with greater than 4 weeks of symptoms and may be associated with osteomyelitis, sinus tracts, and loose components. These patients often have a long insidious course of pain, swelling, and stiffness. Patients with chronic infections often present with a history of antibiotic use that decreases the sensitivity of cultures, making accurate diagnosis difficult, or limiting the identification of all organisms, in the case of polymicrobial infections [49]. Chronic infections involve organisms that have penetrated interfaces and tissues. They have often been subjected to a number of antibiotics and may have formed biofilms that resist nonoperative treatments . Therefore, these infections almost always require debridement with component removal and at least 4–6 weeks of intravenous antibiotics for complete eradication.
Treatments
Antibiotic Suppression
Antibiotic treatment alone will fail to eradicate infection from a surrounding total joint arthroplasty. However, in specific clinical scenarios, antibiotics may be used to suppress an infection (Table 17.4). Antibiotic suppression may be appropriate for patients who are poor candidates for surgical intervention. These patients are usually at a high risk of local or systemic complications and often have other medical issues that preclude an operative procedure. For successful antibiotic suppression, the organism must have low virulence and demonstrate susceptibility to an orally available and tolerable antibiotic. The success rate of antibiotic suppression alone is about 20% [50]. Patients with signs of advanced infection, such as loosening and sinuses, are unlikely to respond well to antibiotic suppression [1, 2, 51]. Attempting to suppress a deep prosthetic infection in the presence of other joint arthroplasties or artificial implants (e.g., heart valves) puts the patient at risk for metastatic implant infection and should be avoided if possible.
Table 17.4
Criteria necessary for successful antibiotic suppression
Surgical intervention contraindicated (patient health) |
Low-virulence organism |
Organism sensitive to antibiotics |
Patient can tolerate antibiotic |
No component loosening |
Patients treated with antibiotic suppression should be routinely followed for signs of advancing infection. Failed treatment may manifest with either acute or insidious symptoms, such as increased pain, swelling, drainage, and erythema. Constitutional signs of bacteremia are a clear indication of failure of suppression.
Open Debridement with Component Retention
Open debridement of acute TKA infections is an attractive option, given the possibility of retaining a stable implant, avoiding revision, and preserving a functional limb. The currently accepted indications for this treatment option include acute postoperative or hematogenous TKA infections that are identified within weeks from the onset of symptoms (Table 17.5). The presence of loosening, sinus tracts, or osteomyelitis suggests more chronic infection and is associated with a high rate of failed debridement. This option is less desirable when other joint implants are present, unless performed within 1–2 weeks of the onset of symptoms.
Table 17.5
Criteria necessary for successful open debridement with component retention
Low-virulence organism |
Organism sensitive to antibiotics |
Acute infection (<4 weeks) |
No component loosening |
No osteomyelitis |
No sinus tracts |
An open arthrotomy and a complete synovectomy are performed to remove the proliferative, inflamed, and sometimes necrotic tissue. A polyethylene insert exchange provides access to interfaces and also assists with exposure of the posterior capsule. Four to six liters of saline, with antibiotics, are then used to irrigate the knee, and a standard closure using a heavy deep monofilament suture is completed over drains. Multiple intraoperative tissue and fluid samples are sent for the identification of infecting organisms. Four to six weeks of appropriately directed intravenous antibiotics is administered, followed by chronic oral antibiotics in select cases. Multiple debridements may enhance the outcome.
Numerous published series have evaluated the capability of early debridement at eradicating infection [14, 51, 52]. Despite the use of various methodologies, these reports reveal common themes that provide guidelines for the debridement of infected TKA (Table 17.5). An evaluation of more than 20 published articles on this topic revealed a success rate ranging from 19% to 83%, with most studies reporting success rates less than 60% [24]. A 2002 meta-analysis of 530 patients treated with open irrigation and debridement for acute PJI showed an overall success rate of 33.6% [14].
The most important factor determining its success is the timing of debridement after the onset of infection [53–55]. Retrospective case series have demonstrated a statistically significant difference in outcome when comparing patients debrided soon after symptoms from those patients debrided after prolonged symptoms [54–57]. Marculescu et al. found the risk of treatment failure to be twice as high with symptom duration greater than 8 days [56]. Hsieh et al. identified short duration of symptoms (<5 days) as the only factor associated with success of irrigation and debridement in patients with gram-negative PJI [57]. It is likely that prolonged infections establish deeper penetration within tissues and interfaces and are more difficult to successfully debride. The development of protective mechanisms such as biofilms may be generated by the organism and contribute to failure [16]. Evidence of chronic infection such as sinuses, loosening, or osteomyelitis is generally considered contraindications to attempting component retention. In general, the literature supports component retention if debridement is done within 2–4 weeks after the onset of symptoms, but it is best done within days.
Patients who are young and healthy with an infection after primary knee arthroplasty are also more likely to have a successful debridement [58, 59]. Some authors have suggested that patients with multiple medical problems or immunocompromise are more difficult to treat with debridement and component retention [58]. Additionally, although exceptions have been reported, generally poor results have been found after debridement of hinged and multiply revised components [58].
Debridement is more likely to succeed with less virulent organisms such as streptococcal species and Staphylococcus epidermidis, whereas failed debridement has been associated with more virulent organisms such as Staphylococcus aureus and gram-negative organisms and in the setting of antibiotic resistance [54, 58, 59]. A statistically significant difference was found in outcome after the debridement of S. aureus infections versus infection with other gram-positive organisms [59]. Only 1 of 13 TKAs infected with S. aureus were successfully debrided in that series, compared with 10 of 18 successful debridements in patients infected with other gram-positive organisms. Likewise, Choi et al. observed that though initial infection control rate was much lower with prosthesis retention compared to removal, retention with polyethylene exchange can be selectively considered for patients with non-S. aureus infection [60]. In contrast, methicillin-resistant S. aureus infection is more difficult to treat with isolated irrigation and debridement, with one series demonstrating an 84% failure rate at a minimum 2-year follow-up [61].
Arthroscopic debridement generally has unacceptably poor results and should be avoided. It permits limited examination of the joint, precludes polyethylene exchange, and limits the ability to perform a complete and thorough synovectomy [24]. Waldman et al. reported on the arthroscopic irrigation and debridement of 16 infected TKAs [62]. Despite a strict definition of acute infection (≤7 days of knee symptoms), only six infected knees (38%) were successfully treated using this method. Similarly, Dixon et al. described an infection eradication of 60% in 15 patients treated with arthroscopic irrigation and debridement at a mean follow-up of 55 months [63]. Arthroscopic treatment for the acutely infected TKA should be limited to patients who are medically unstable or anticoagulated.
Exchange Arthroplasty
Exchange arthroplasty involves removal of the infected TKA, thorough debridement, and reimplantation. Direct exchange (one-stage) arthroplasty involves open debridement of the infected TKA followed by immediate revision. Two-stage reimplantation involves open debridement, removal of the infected prosthesis, and delayed reimplantation, with an intervening time for antibiotic therapy.
Exchange arthroplasty is preferred for infections present for greater than 2–4 weeks or persistent infections that could not be eradicated with debridement alone. In order to successfully use exchange arthroplasty, the patient should be medically stable for multiple operative procedures , with an intact immune system that will aid in eradicating the infection. Furthermore, the inherent elements of the knee, such as bone stock, extensor mechanism, and soft tissue envelope, should be amenable to eventual TKA function.
Direct exchange arthroplasty with primary reimplantation involves prosthesis removal and thorough irrigation and debridement, followed by reimplantation of a new prosthesis in a single surgery. Goksan and Freeman [64] described a technique comprised of irrigation with saline, packing with iodine-soaked sponges, and a one-layer wound closure, followed by deflation of the tourniquet to allow for antibiotic perfusion for 30 min. After a complete replacement of all gowns, drapes, and gloves, the knee is prepared again with sterile technique, and the components are reimplanted with antibiotic-impregnated cement. With this technique, Goksan and Freedman reported successful eradication of infection in 16 of 18 patients treated with direct exchange arthroplasty, but clinical follow-up was short. A more recent prospective study compared the outcomes of one-stage and two-stage revisions in 28 and 74 patients, respectively [65]. Patients who underwent single-stage revision were carefully selected to ensure healthy soft tissues, known organism with known sensitivities to available antibiotic treatments, absence of immunocompromise, and good bone stock. At an average follow-up of 6.5 years, the single-stage patients had no reinfection and also had higher Knee Society Scores than the two-stage patients. The largest study to date on single-stage TKA revision studied 63 patients without methicillin-resistant organisms who underwent a one-stage revision for PJI of TKA. At an average follow-up of 36 months, the patients demonstrated an infection control rate of 95% and higher knee scores than two-stage revision patients [66]. Zahar et al. obtained the longest clinical outcomes on single-stage TKA revision patients, with an average follow-up of 10 years [67]. In this series, the 10-year infection-free survival rate was 93% in 11 patients who had undergone aggressive debridement of the collateral ligaments and posterior capsule with implantation of a rotating hinge construct. Even in patients with chronically infected TKA, one-stage revision can lead to infection control rate of 91% at 3 years [68].
The relative ease and seemingly encouraging outcomes of one-stage revision, however, are tempered by the absence of high-level evidence and limited outcome data based on studies with small cohorts. With proper patient selection and meticulous surgical technique, direct exchange has been associated with a rate of success comparable to two-stage exchange arthroplasty, even in chronically infected TKA [14, 65, 68]. This is particularly reassuring in those patients who undergo revision arthroplasty in the setting of previously undetected infection and highlights the importance of using antibiotic-impregnated polymethyl methacrylate (PMMA) cement in all revision TKAs, with or without known infection.
The two-stage approach, first described by Insall et al. [69], is considered to be the gold standard for definitive treatment of TKA PJI, especially with long-standing or late TKA infections , with reported success rates greater than 85–90% [55, 69–73]. A recent retrospective study of 253 patients found two-stage revision for infected TKA yielded an infection-free survivorship of 85% at 5 years and 78% at 10 years [74].
At the time of implant removal, a complete debridement must be performed to provide an optimal environment for eventual reimplantation. This includes not only an extensive synovectomy but also removal of necrotic and infected bone. The previous incision can almost always be used. Sinuse s should be excised and muscle flaps used if coverage is a potential problem. On entering the joint, several samples of synovial fluid should be sent for culture and analysis. Synovial tissue, interface tissues, and tissue from the canals (when removing stemmed components) should also be sent for culture and pathologic analysis. When removing components, it is critical to preserve maximal bone stock. However, one must be sure to remove all fragments of cement and necrotic bone in an effort to reduce the interfaces available to organisms. Irrigation of the joint with several liters of antibiotic saline is performed, and a spacer is implanted. The capsular closure is performed over drains using a running monofilament suture.