History and Physical Examination

A good history identifying the specific presentation, duration, and localization of symptoms, predisposing factors, and response to treatments may render an early diagnosis.

No universal criteria exist to determine and differentiate infection from aseptic loosening and osteolysis, and the entire process is based on the integration of clinical presentation, laboratory data, and imaging modalities.

PPI may present with nonspecific symptoms, including minor hip pain. Administration of antibiotics may ameliorate the symptoms but leave a nidus of infection that ultimately leads to chronicity and premature implant failure. Therefore a rigorous algorithm consisting of preoperative and intraoperative tests is required to confirm or refute septic failures. The presence of signs and symptoms such as a draining sinus tract ( Fig. 31-1 ), fever, chills, and/or a history of persistent wound drainage with concomitant painful range of motion can aid in the diagnosis of infection but often are absent. Some infections may be chronic or subacute in nature with insidious presentation that may overlap with aseptic failure of arthroplasty and hence go undetected. In fact, a recent study showed that, using sonication, bacteria could be isolated from the surface of 8.3% of implants of patients who were assumed to have an aseptic etiology for failure.

External opening of a sinus tract overlying a periprosthetic hip infection.

Radiographs

After the initial history and examination that may be revealing, a battery of tests often is performed. Radiographs are an essential part of the assessment. Radiographs are useful to determine the cause of implant failure and reveal problems such as wear, osteolysis, or fracture. Radiographs, besides providing important information about the local environment around the prosthesis, may, in cases of overt infection, reveal periosteal reaction, bone cysts, and focal resorption indicative of PPI. Other imaging tests such as computerized tomography or magnetic resonance imaging have little role in diagnosis of infection.

Certain changes such as focal areas of osteolysis, osteopenia, and endosteal or periosteal reaction ( Fig. 31-2 ) are consistent with PPI, whereas early loosening of the implant should alert the surgeon to the possibility of an underlying dormant infection. Although these characteristics can be used to determine PPI, they rarely manifest in infected joint arthroplasties; therefore the role of radiographs lies in ruling out the presence of other aseptic etiologies.

Periosteal reaction (arrows) suggestive of infection around a hip replacement.

Radionuclide Modalities

Bone scans, particularly when combined with white blood cell labeling, may be a useful preoperative test. The technetium-99m ( ^99m Tc) isotope bone scan that detects areas of increased uptake often is used as a part of initial workup for PPI. Three-phase bone imaging by itself caries a sensitivity of 33% and specificity of 86%, with poor predictive value. This enables the bone scan to play an important role in screening and ruling out the presence of infection. It is, however, rare for the technetium bone scan to be used alone in diagnosis of PPI. A dual-tracer technique such as the use of indium-111 ( ¹¹¹ In)-labeled leukocyte scan usually is performed simultaneously with a ^99m Tc diphosphonate scan. The combination of ^99m Tc and ¹¹¹ In decreases the rate of false-positive results and improves specificity.

In theory, labeled leukocyte tracers such as ¹¹¹ In do not accumulate at sites of increased bone turnover in the absence of infection. The sulfur colloid scan combined with leukocyte scan also relies on a similar principle. Both the sulfur colloid and white blood cell tracers accumulate in normal bone marrow, but sulfur colloid does not accumulate in white cells present in the areas of infection. The diagnosis of infection therefore is based on the relative difference in the uptake of two tracers. The reported accuracy of this technique ranges from 89% to 98%.

Technically the combined scan is labor intensive and associated with risks such as transfusion reactions. The technetium tracer is first injected into the patient and allowed to accumulate in areas of high metabolic activity and increased blood flow. Leukocytes are then obtained from the patient, labeled with ¹¹¹ In, and reinjected into the patient.

In recent years positron emission tomography with fluorodeoxyglucose imaging (FDG-PET) has been shown to have a promising role for diagnosis of PPI. FDG-PET has been shown to carry 95% sensitivity and 93% specificity for diagnosis of infection around a hip prosthesis. This imaging modality relies on glucose metabolism and uptake of FDG by inflammatory and metabolically active cells. The FDG, once taken up by the cells, is phosphorylated to deoxyglucose-6-phosphate and trapped in the cell long enough to be imaged by PET. However, PET generally is nonspecific and targets all areas of inflammation; thus inflammatory conditions must be ruled out to have any predictive value. Some centers are currently implementing FDG-PET as part of the preoperative evaluation with possible PPI.

Love et al noted that the combined ^99m Tc/ ¹¹¹ In scans had a greater specificity than FDG-PET, but other centers reported superior results in terms of sensitivity and specificity with PET scan. A recent prospective study by Pill et al compared FDG-PET scan to the combined ^99m Tc/ ¹¹¹ In and concluded that FDG-PET scan has a far greater sensitivity (95% vs. 50%) than the ^99m Tc/ ¹¹¹ In scans and is a useful diagnostic tool with promising ability in differentiating PPI from aseptic failures of the hip. However, false-positive results still plague FDG-PET scans, especially in areas of particle-induced inflammation, where macrophages accumulate.

Overall, radionuclide imaging is an exciting test but requires extensive resource utilization that is not easily available at most centers. Other cheaper and cost-effective tests, namely blood tests, still remain the workhorse of diagnosis of PPI.

Serologic Tests

The erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are traditional serologic markers used as a part of a diagnostic workup for infection. ESR greater than an arbitrary cutoff of 30 mm/hr and CRP greater than 1 mg/dL are considered abnormal. However, elevated ESR and CRP levels may occur in conditions such as rheumatoid arthritis, lupus, polymyalgia rheumatica, and temporal arteritis as well as metastatic disease. The ESR also tends to be high in patients with anemia, renal failure, or obesity and during pregnancy. On the other hand, a low ESR concomitant with a red blood cell disorder or an abnormality in plasma protein production can mask PPI.

CRP is one of many acute-phase reactants or proteins produced by hepatocytes in response to various pathologic states. Because plasma levels of CRP in healthy individuals are present in trace amounts undetected by standard laboratory techniques, they often are reported as less than 0.5 mg/L. After total joint arthroplasty CRP begins to rise, reaching a peak at 48 hours postoperatively. It then begins to normalize and reaches a normal level within 3 weeks. On the other hand, ESR reaches a maximal level at 5 to 7 days postoperatively and may not return to normal up to 3 months postoperatively. In the absence of the confounding factors previously listed, persistent elevations of ESR for more than 3 months and CRP for more than 3 weeks postoperatively are alarming signs for possible joint infection.

A prospective study of revision hip arthroplasty that excluded patients with inflammatory arthropathy was conducted by Spangehl et al. The authors concluded that an ESR of greater than 30 mm/hr has a sensitivity of 82% and a specificity of 85% in determining infection. However, a C-reactive protein level greater than 1 mg/dL was a better prognostic indicator of infection than the ESR and achieved a sensitivity of 96% and a specificity of 92%. Although the ESR and CRP are not diagnostic of infection when used individually, Spangehl et al concluded that when the ESR and CRP are below their respective cutoff values, PPI can reliably be excluded from the differential in the majority of cases.

Another serologic marker that has shown promising results is interleukin-6 (IL-6). This cytokine is produced by monocytes and macrophages and induces the production of acute-phase proteins, including CRP. IL-6 achieves its peak level during the first 12 hours postoperatively and returns to its preoperative baseline value within 3 days. Therefore serum IL-6 levels can be used to detect early postoperative infections and monitor the patient’s response to treatment, which the other commonly used serologic markers cannot do. DiCesare et al showed that a cutoff value of 12 pg/mL has adequately high sensitivity, specificity, and negative predictive value and acceptable positive predictive value to diagnose PPI.

Joint Aspiration, Fluid Analysis, and Culture

Joint aspiration remains as one of the most important diagnostic tests for PPI. Aspiration can confirm the diagnosis and identify the infecting organism, which in turn guides antibiotic therapy and the surgical management. Joint aspiration may be considered in patients with abnormal ESR and CRP. Once aspirated, the joint fluid should be sent for cell count and neutrophil differential count as well as culture for aerobic and anaerobic bacteria and fungi.

Joint aspiration does not, however, have absolute accuracy. False-negative aspirations occur, particularly in patients who are taking antibiotics. Hence aspiration of the joint should be deferred for at least 2 weeks in patients taking antibiotics. In addition, repeat aspirations may be considered in patients with suspected infection in whom initial aspiration failed to isolate an organism. Aspiration should be performed only in patients with a strong suspicion for infection because the rate of false-positive results also can be unacceptably high.

The joint aspirate should be analyzed for absolute white cell count and neutrophil differential. Cell count analysis of the joint aspirate is demonstrated to have a high accuracy (90%) for diagnosis of PPI. The neutrophil percentage is more efficient in excluding infection but could be inundated with a high false-positive rate because of the inclusion of bloody or clotted aspirates. Trampuz et al, in a prospective study of total knee arthroplasties, defined a cutoff value of fluid cell count of 1700 cells/μL and neutrophil percentage of 65%. These authors concluded that the neutrophil percentage was a significantly better diagnostic modality than the leukocyte count. In a recent study performed by Parvizi et al, similar cutoff values for fluid cell count (1760 cells/μL) and a neutrophil percentage (73%) were conceived. The fluid cell count had a slightly higher positive predictive value and a slightly lower negative value compared with neutrophil percentage. From a practical aspect, a leukocyte count greater than 2000 cells/μL and neutrophil percentage greater than 70% can be used to assess for the presence of infection in patients with artificial hip or knee joints.

The isolation of an organism from joint fluid or periprosthetic tissue obtained intraoperatively is considered the gold standard of diagnosing PPI. Although this test possesses high specificity (97% to 100%) and a near absolute positive predictive value, limitations exist. False-positive cultures may occur in 6% of cases, and an organism may not be isolated in 10% to 12% of the cases with “confirmed” PPI. Another drawback of relying on intraoperative culture to diagnose PPI is that the result of the culture usually is not available for 3 to 4 days after the surgery.

The exact number of samples that must be obtained to confirm the presence of PPI is debated. Some investigators advocate obtaining five or more samples for accurate diagnosis of PPI. However, for practical purposes and cost effectiveness most surgeons obtain two to three culture samples.

Intraoperative Tests

A few intraoperative tests may aid the surgeon in reaching or refuting the diagnosis of PPI. Analysis of tissue samples for the presence of organisms (Gram stain) and the presence of leukocytes (more than five per high power field) are some of those tests. Although the latter test has near absolute specificity, the sensitivity is exceedingly low at 30% to 40%. Part of the problem is that the histologic criteria and neutrophil cutoff values used for the diagnosis of infection vary from center to center.

Gram stain of periprosthetic tissue samples has demonstrated consistently poor sensitivity. The sensitivity has ranged from 15% to 30% depending on the criteria used as standard for diagnosis of PPI and therefore lacks the ability to detect infection in a consistent manner. On the contrary, the specificity and positive predictive value of this test is high at 98% to 100%, which enables the surgeon to confirm the presence of PPI when a positive smear is encountered. At the present time the Gram stain is an ineffective tool for the diagnosis of PPI and further investigation is required to reveal its importance.

No concrete data have been presented to date concerning the indications and timing for performing frozen sections. Some investigators have conducted prospective studies in which frozen sections were obtained from every patient undergoing revision surgery, and others have analyzed frozen sections when a high clinical suspicion for infection was present. Two early studies conducted by Mirra et al began the process of clarifying the histologic criteria used to diagnose infection. These authors documented the presence of more than five neutrophils per high power field (hpf) in five separate fields in samples taken from sites of acute inflammation with confirmed positive cultures. However, in both original articles histologic analysis was performed under a magnification of ×500, which may influence results when applying the criteria to the more commonly used ×400 microscopes. Other investigators, including Lonner et al, attempted to validate these criteria in a prospective study of 175 consecutive patients. They reported a sensitivity and specificity of 84% and 96%, respectively, when implementing the above recommended criteria using ×400 magnifications. However, the specificity improved to 98% when using more stringent criteria of more than 10 neutrophils/hpf in more than five fields. In another prospective study of 106 total knee and hip revision arthroplasties, Athanasou et al compared the following histologic criteria (more than 5 cells/hpf, including neutrophils, lymphocytes, and plasma cells in more than 10 fields) to the gold standard of intraoperative culture and yielded adequate sensitivity (90%) and specificity (96%). On the other hand, Pandey et al performed a large retrospective study of 617 revision total joint arthroplasties in which they considered the presence of 1 cell/hpf in at least 10 fields to be consistent with infection in 97.8% of the cases of septic failures.

Molecular Techniques

Novel molecular techniques for diagnosing PPI have previously been described. During recent years molecular techniques based on selective DNA amplification by polymerase chain reaction (PCR) have been introduced. PCR analysis has been reported by some investigators to have better specificity and sensitivity than the standard culture methods. The technique involves the identification and amplification of the bacterial ribosomal 16S fragment. A drawback of this technique is that it does not differentiate between live and dead bacteria, and a high number of false-positive results may occur. Although PCR is a highly sensitive test that can detect the presence of bacteria even in small quantities, its role in routine clinical investigation remains unestablished. A recent study that applied PCR technology to dry reagent dipsticks showed it was able to detect different pathogens within a few hours.

During recent years the role of other molecular techniques, such as microarray, also has been tested. Deirmengian et al demonstrated that white blood cells found in synovial fluid of patients with PPI express a “signature” gene. Among the genes found to be differentially expressed were IL-1, chemokine ligands CCL3 and CCL4, and intercellular adhesion ligand ICAM1.