The Promising Role of ¹⁸F-FDG PET in Detecting Infected Lower Limb Prosthesis Implants

MATERIALS AND METHODS

RESULTS

The FDG PET scan correctly revealed 10 of 11 knee prosthesis infections (Fig. 1). However, 7 positive results in 25 knees eventually proved not to be infections. For hip prostheses, the PET scan successfully revealed 9 of 10 hip infections associated with prostheses (Figs. 2A–C) and excluded infection in 25 of 28 without infection (Fig. 3). The sensitivity, specificity, and accuracy of PET for detecting infection associated with knee prostheses were 90.9%, 72.0%, and 77.8%, respectively. The sensitivity, specificity, and accuracy of PET for detecting infection associated with hip prostheses were 90.0%, 89.3%, and 89.5%, respectively. Overall, the sensitivity for the entire population studied was 90.5% and the specificity was 81.1% for detecting infections of lower limb arthroplasties.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

PET images of 60-y-old woman with painful right knee prosthesis proven infected by revision arthroplasty. Bone–prosthesis interface shows areas of intense uptake in both femoral and tibial component of prosthesis (arrowheads).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

(A) Coronal image of 72-y-old woman with hip prosthesis. Periprosthetic infection on right side was identified (arrowheads). (B) Coronal image of 76-y-old woman with bilateral hip prostheses. Both infection (arrowhead) and loosening (arrows) were shown. (C) Coronal image of 78-y-old man with painful left hip prosthesis. Arrowheads indicate periprosthetic infection and osteomyelitis. (D) Coronal image of 76-y-old woman with bilateral hip prostheses. FDG uptake is noted only around neck of prosthesis (arrows). FDG PET diagnosis of loosening was confirmed after revision arthroplasty.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

A 76-year-old woman with right hip prosthesis who presented with right hip pain. No focally increased FDG accumulation is seen around any portion of prosthesis on PET scan. Therefore, both infection and loosening were ruled out. Patient responded to conservative therapy during 16 mo of follow-up.

DISCUSSION

FDG PET has been used extensively for the detection of malignancy. However, on the basis of a large body of data in the literature, FDG uptake is not specific for a malignant process and is the source of low specificity when used in such settings. FDG accumulation has been reported in a variety of infectious or inflammatory sites, including abdominal infection (11), brain abscesses (12), inflammatory pancreatic disease (13), lobar pneumonia (14), sarcoidosis (15), osteomyelitis (16,17), tuberculosis (18), colitis (19), asthma (20), sinusitis (21), myositis (22), mastitis (23), and thyroiditis (24). On the other hand, in a proper setting, increased FDG accumulation may suggest the presence of infection or inflammation. This study clearly shows the feasibility of using FDG PET for discriminating infection as a complication of prosthesis from noninfectious processes in patients who have undergone lower limb arthroplasty.

Differentiation of an infected from a uninfected prosthesis (such as loosening) is essential for optimal and cost-effective management of these patients. Accurate confirmation or exclusion of infection before revision surgery can substantially simplify plans in these circumstances. In addition, the outcome can be significantly influenced by an accurate initial diagnosis. Current tests do not provide a sensitivity or specificity adequate for accomplishing these goals (25). Plain radiography has a limited role for the diagnosis of infection associated with prostheses because findings are common to both septic and aseptic failure. Also, infection and loosening commonly show no finding. Therefore, aspiration biopsy of the hip or knee of the patient is perhaps the most useful investigative tool for preoperative confirmation of infection, with sensitivity and specificity ranging from 50% to 93% and from 82% to 97%, respectively (25–27). However, the sensitivity of even preoperative joint aspiration is not high enough to exclude infection with an acceptable certainty (28). Moreover, the administration of antibiotics before aspiration biopsy can further reduce the sensitivity of this approach (26,27).

Radionuclide imaging has been extensively used for the diagnosis of infection after joint replacement. However, conventional nuclear medicine modalities, in addition to having limited sensitivity and specificity, suffer from two other shortcomings: the time required to complete the procedures and the cost of the examinations. Bone and gallium scans were first used in patients with a painful lower limb prosthesis in the 1970s (29). The sensitivity of combined bone and gallium scans is relatively poor (38%) (30). In only about one fourth to one third of patients with proven prostheses infections was the infection detected with these modalities (31,32). According to most reports, ¹¹¹In-labeled white blood cell scanning, when combined with ^99mTc-sulfur colloid bone marrow imaging (33), may provide superior sensitivity and specificity for detecting infection in this setting compared with other conventional nuclear medicine modalities. The sensitivity, specificity, and accuracy of this dual-tracer modality have been reported to be 86%–100%, 89%–94%, and 89%–96%, respectively (34–36).

A major concern about indium-labeled leukocyte imaging is the separating, labeling, and reinjection of the white blood cells, which offer several significant opportunities for iatrogenic errors (37) and significantly affect the function and viability of the final product. ¹¹¹In- and ^99mTc-labeled antibodies are emerging as a new approach to detect prosthesis infections without the necessity of labeling the cells outside the body and have had some success (38,39). However, other disadvantages of indium-labeled leukocyte imaging, such as the necessity for delayed imaging (24 h or longer), relatively high radiation exposure, and additional bone marrow imaging, remain major shortcomings.

To our knowledge, no major reports have appeared in the literature about the usefulness of FDG PET in the setting of chronic infection of orthopedic implants. FDG PET is often misperceived as an expensive, complicated imaging modality and, as such, requiring limited use. However, compared with traditional nuclear medicine methods, FDG PET may prove to be a single and cost-effective modality for accurately diagnosing infection associated with lower limb arthroplasty, especially in the hips. Based on routine practice in most institutions, the expense of FDG PET is substantially lower than the combined costs of ¹¹¹In-labeled white blood cell scanning, bone marrow studies, and bone imaging, which are performed as a combination in most patients. In addition, tomographic images with FDG PET provide better spatial resolution than do conventional nuclear medicine modalities, significantly improving the performance of the test. Therefore, PET can better detect small and subtle changes in disease activity, which may be missed by other techniques. Figure 4 shows the value of this modality in providing information about a complex process. In the patient illustrated, who had clear clinical evidence of infection, FDG PET revealed the presence of infection in the bone–prosthesis interface, fistulous drainage of pus in the area, and a suggestion of loosening with superimposed infection around the neck. The details shown were unique and exemplified the unusual capability of this modality. Furthermore, FDG PET can be completed within a few hours, compared with 2 d for other nuclear medicine methods. Considering that the average cost of treating an arthroplasty patient with infection ranges from $50,000 to $60,000 (40) and that the success of treating infection associated with a prosthesis largely depends on early, accurate diagnosis, FDG PET may prove useful and cost-effective for optimal patient management of painful lower limb prostheses.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Coronal images of 63-y-old man with painful right hip prosthesis. Loosening (short arrow), infection (arrowheads), and fistula (long arrow) are shown.

Interestingly, the intensity of uptake at the bone–prosthesis interface was not important when making the diagnosis, and only the presence of uptake was an indication of infection. Therefore, increased tracer uptake at the bone–prosthesis interface should be considered suggestive of infection. In the hip, increased tracer uptake around the femoral head portion of the prosthesis was a reliable sign of loosening. If the accumulated FDG was limited around the femoral head or neck and did not extend to the bone–prosthesis interface, the probability of infection was low (Fig. 2D; compare with Figs. 2A and B). The simplicity of the diagnostic criteria proposed should reduce potential interobserver variation and increase the accuracy of the scan.

As a result of the high sensitivity of FDG PET, only two false-negative results were reported in this study. In contrast, 10 false-positive results, mostly among patients with knee prostheses, were reported. Incorrect interpretation of postsurgical changes as positive findings in FDG PET is well known. However, the interval between surgery and FDG PET was more than 1 y in all patients with false-positive results in this study. Therefore, in addition to postsurgical changes, other factors must have contributed to inaccurate results. Further research is required to clearly define the factors that contribute to false-positive results, especially in knee prostheses.

Our data clearly show that the performance of this technique is optimal for diagnosing and excluding the presence of infection in hip and knee prostheses. We have shown that using appropriate criteria for uptake of FDG in uninfected, infected, and loose prostheses achieves high accuracy, particularly for FDG PET of complicated hip prostheses. For reasons unknown, the FDG PET results for knee prostheses are somewhat inferior to those for hip prostheses. The major advantage of FDG PET over conventional nuclear medicine procedures is the simplicity of the approach and the timely availability of results (within 2–3 h).

CONCLUSION

These preliminary data show that the use of FDG PET for detecting infection associated with lower limb prosthesis is feasible but has some limitations. FDG PET is more useful for detecting infections in painful hip prostheses than in painful knee prosthesis.