| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Nuclear Medicine |
1 From the Department of Medical Radiology, Division of Nuclear Medicine, University Hospital, CH-8091 Zurich, Switzerland (K.D.M.S., E.M.K., T.F.H., G.W.G., G.K.v.S.); and Departments of Radiology (M.Z., J.H.) and Orthopaedic Surgery (H.P.N.), Orthopedic University Hospital Balgrist, Zurich, Switzerland. From the 2002 RSNA scientific assembly. Received December 11, 2002; revision requested February 7, 2003; final revision received August 5; accepted September 29. Address correspondence to K.D.M.S. (e-mail: katrin.stumpe@dmr.usz.ch).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Thirty-five patients with painful total hip replacements and possible septic prosthetic loosening were examined with FDG PET, conventional radiography, and three-phase bone scintigraphy. PET, radiographic, and scintigraphic images were each evaluated by two independent observers in a blinded fashion. For 32 of 35 patients, serial conventional radiographs were available. Results of microbiologic examinations of surgical specimens represented the standard of reference in 26 patients, and results of joint aspiration plus clinical follow-up of at least 6 months represented the standard of reference in the remaining nine patients. Sensitivity, specificity, accuracy, and interobserver variability (
) values were calculated. The imaging modalities were compared in terms of diagnostic confidence by using the sign test.
RESULTS: Nine patients had septic and 21 patients had aseptic loosening. In five patients, neither loosening nor infection was confirmed. For diagnosing infection with FDG PET, conventional radiography, and bone scintigraphy, respectively, sensitivity values for reader 1 and reader 2 were 33% and 22%, 89% and 78%, and 56% and 44%, while specificity values were 81% and 85%, 50% and 65%, and 88% and 92% and accuracy values were 69% for both readers, 60% and 69%, and 80% for both readers. PET was significantly more specific (P = .035) but less sensitive (P = .016) than conventional radiography for the diagnosis of infection.
CONCLUSION: In a study population of patients suspected of having infected total hip replacements, FDG PET performed similarly to three-phase bone scintigraphy. FDG PET was more specific but less sensitive than conventional radiography for the diagnosis of infection.
© RSNA, 2004
Index terms: Hip, infection, 442.201, 443.201 • Hip, prostheses, 442.454 • Hip, radiography, 44.11 • Hip, radionuclide studies, 44.12172 • Positron emission tomography (PET), comparative studies, 44.12163
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The diagnosis of prosthetic joint infection is usually rendered on the basis of laboratory abnormalities, radiologic findings, and results of conventional nuclear medicine studies (6,7), although the value of these diagnostic methods has been questioned. Laboratory parameters of inflammation (eg, erythrocyte sedimentation rate and C-reactive protein levels) may be elevated in both loosening and infection, but elevations in these parameters are often not conclusive (8–10).
In the past, joint aspiration has been advocated for all patients with failed hip replacement. However, the role of aspiration as a sole diagnostic test has been debated, in that reported sensitivities of the procedure range from 50% to 93% and reported specificities range from 82% to 97% (11–13).
Conventional radiographs have been reported to be of limited value in the diagnosis of infection. Several radiographic findings, including periprosthetic areas of radiolucency, osteolysis, and migration, may be present in both infection and aseptic loosening (14). However, analysis of sequential radiographs contributes to differentiation because changes occur more quickly in the presence of infection (15). Computed tomography and magnetic resonance imaging are limited in this setting by beam hardening and susceptibility artifacts, respectively, even when specifically modified imaging parameters are used (16–18).
Radionuclide studies represent the current imaging method of choice in patients with metallic implants. Because of the high negative predictive value of its results, conventional bone scintigraphy is useful as an initial screening test (19). Levitsky et al (20) reported a sensitivity of 33%, a specificity of 86%, a positive predictive value of 30%, and a negative predictive value of 88% in 72 total joint replacements examined with bone scintigraphy. The use of white blood cells labeled with indium 111 (111In), in combination with technetium 99m (99mTc) sulfur colloid marrow imaging, currently provides the highest sensitivity and specificity and has become the method of choice in the assessment of infection in total joint replacements (21).
Fluorine 18 (18F) fluorodeoxyglucose (FDG) positron emission tomography (PET) has proved to be an encouraging method in infection imaging and may replace other scintigraphic imaging methods in the future. However, data about the usefulness of FDG PET in the presence of total hip or knee replacements are still controversial.
The purpose of this investigation was to compare the diagnostic efficacy of FDG PET with that of conventional radiography and three-phase bone scintigraphy in patients suspected of having infection in their total hip replacements.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The patients were referred to the radiology department for conventional radiography of the pelvis and fluoroscopically guided joint aspiration according to our standard diagnostic protocol in cases of suspected infection. Of the original 39 eligible patients, four refused to participate in the additional investigations required for study inclusion for personal reasons. Thirty-five patients (18 of whom had a cemented and 17 of whom had an uncemented prosthesis) were evaluated. There were 23 women (age range, 47–89 years; mean age, 71 years) and 12 men (age range, 46–81 years; mean age, 64 years). According to results of the Mann-Whitney test, there was no significant difference (P = .09) between the mean age of the men and that of the women. The mean time between the last surgical intervention at the site of the involved total hip prosthesis and study inclusion was 71 months (range, 12–260 months). Twenty-six patients who had undergone primary implantation and nine patients who had undergone revision arthroplasty of the hip were included.
FDG PET and three-phase bone scintigraphy with 99mTc dicarboxidiphosphonate were performed within 3 days after conventional radiography and fluoroscopically guided joint aspiration. The study was approved by the institutional review board of University Hospital. All patients were fully informed about the study purpose and any potential risks and gave informed consent.
Conventional Radiography
The radiographs were acquired by using a Siemens Multix U unit (Siemens Medical Solutions, Erlangen, Germany) with an AGFA ADC phosphor plate system (AGFA-Gevaert, Mortsel, Belgium) and consisted of conventional anteroposterior views of the pelvis (available for all of the 35 patients) and cross-table lateral views of the symptomatic hip (available for 26 of the patients). Conventional radiographs were evaluated by two independent and experienced (>10 years) musculoskeletal radiologists (J.H., M.Z.) who were blinded to the results of the other imaging studies. In all cases except three, serial radiographs obtained during a 6–235-month follow-up period (mean follow-up period, 60.2 months) were available for review. In those three cases, only one set of conventional radiographs was available.
The following findings in the acetabular and femoral regions were recorded (14,15,22,23): the location and width of any periprosthetic areas of lucency (at the metal-bone, metal-cement, and cement-bone interfaces); the shape of any areas of lysis (smooth or irregular); the appearance of any periosteal bone reaction (absent, solid, or nonsolid); and the presence or absence of granuloma, migration, and additional findings such as pathologic fractures and cement fractures. Periprosthetic areas of lucency were classified on the conventional radiographs according to the method of DeLee and Charnley (22) for the acetabulum (zones I-III) and according to Gruen et al (23) for the proximal femur (regions 1–7).
The diagnosis of infection (14,15) was rendered when at least one of the following criteria was present: rapid prosthetic migration (of at least 2 mm within 6–12 months), rapidly progressive periprosthetic osteolysis, and/or irregular periprosthetic osteolysis. The diagnosis of loosening (without infection) (15,23) was rendered when at least one of the following criteria was present: migration (of less than 2 mm within 6–12 months), periprosthetic lucency (in a smooth area 2 mm or greater in diameter), periosteal reaction (of the solid type), and/or cement fracture. In the three patients with only one set of radiographs, the diagnosis of infection or loosening was based on the same morphologic criteria, without the information regarding migration.
Bone Scintigraphy
Three-phase bone scintigraphy was performed with a gamma camera (Body Scan; Siemens, Hoffman Estates, Ill) equipped with a low-energy, high-spatial-resolution, parallel-hole collimator by using a 20% window centered on the 140-keV photopeak after intravenous injection of 700 MBq of 99mTc dicarboxidiphosphonate (Teceos; CIS bio international, Gif-Sur-Yvette, France). In the dynamic phase, images of the hip were acquired in a 3-second exposure up to 2 minutes after the injection. The blood pool phase consisted of acquisition of two static spot images of the hip (anterior and posterior) 2 minutes after intravenous tracer injection. The third phase (ie, the bone study) was performed 2–4 hours after injection, and anterior and posterior spot views of the hip were again obtained.
The images were analyzed by two independent board-certified nuclear physicians (E.M.K., G.W.G.) who each had more than 5 years of experience. The reviewers were blinded to the results of the other imaging studies. Analysis of the three-phase bone scintigrams was performed with a digital viewing system (Extended Viewing Station; GE Medical Systems, Waukesha, Wis). For image interpretation, hip prostheses were divided into the following four zones according to the system of Palestro et al (21): the head (including the acetabulum), the trochanter, the shaft, and the tip. The appearance (focal or diffuse) of the increased radionuclide uptake was described (15,20,24,25). Focally or diffusely increased periprosthetic uptake in all three phases of bone scintigraphy was considered to be diagnostic for infection. Increased radionuclide uptake limited to the third phase was considered to indicate a loosened but not infected total hip replacement.
FDG PET
FDG PET studies were performed with a PET scanner (Advance; GE Medical Systems, Waukesha, Wis). Several data sets with 35 two-dimensional sections of 4.25-mm thickness, each with a transverse field of view of 14.6 cm, were acquired to cover the body from the pelvic floor to the feet. Patients were asked to fast for at least 4 hours before this examination. Thirty to 40 minutes before scanning, the patients received an intravenous injection of 300–400 MBq of 18F FDG, which was produced in house by using a 17.8-MeV cyclotron (PET Trace 2000; GE Medical Systems, Uppsala, Sweden) and an automated FDG synthesis module (PET Tracer Synthesizer; GE Nuclear Interface, Muenster, Germany). Attenuation-corrected and uncorrected transverse images were acquired. Attenuation correction was performed by using the rotating germanium 68 sources built into the PET scanner. A multiplicative iterative reconstruction algorithm for improvement of image quality and reduction of computation time was used (26). Additionally, coronal and sagittal reformations were obtained.
Image analysis of PET scans was performed with a digital viewing system (Extended Viewing Station; GE Medical Systems, Waukesha, Wis). The images were analyzed by two independent board-certified experienced nuclear physicians (G.K.v.S., T.F.H.) who each had more than 4 years of experience and were blinded to the results of the other imaging studies. The readers were different from those involved in the evaluation of the bone scintigrams. They used the same image evaluation zones used for bone scintigraphy (21). In addition, the intensity of FDG uptake in each zone was graded on a five-point scale as follows: A score of 0 indicated that FDG uptake was similar to that in the background; a score of 1, that FDG uptake was low and comparable to that in inactive muscles and fat; a score of 2, that FDG uptake was moderate, clearly noticeable, and distinctly higher than the uptake in inactive muscles and fat; a score of 3, that FDG uptake was strong but was distinctly less than the physiologic uptake in the bladder; and a score of 4, that FDG uptake was very strong and was comparable to physiologic urinary uptake in the bladder.
A description of a similar grading system has been published by Stumpe et al (27). However, contrary to the definition used in that study for grade 4 uptake, in which uptake in the cerebral cortex was used as a reference, we used uptake in the bladder as a reference because only partial-body PET scans from the pelvis to the feet were obtained in this investigation. In the previous investigation of Stumpe et al (27), results of a receiver operating characteristic analysis had shown that classifying lesions with grade 3 or 4 uptake as infection yielded the best discrimination between abnormalities with infection and those without. Therefore, diffusely increased FDG uptake of grade 3 or 4 along the interface between the prosthesis and bone was used as the criterion for infection in our study. To make sure that the focus of infection was not an artifact associated with misregistration between attenuation-corrected and uncorrected FDG PET images or caused by the dense and solid material of the prosthesis itself, we evaluated the uncorrected images separately. An artifact was considered to be present if grade 1–4 uptake was observed on the attenuation-corrected images but no FDG uptake was observed on the uncorrected images. A hip prosthesis was classified as normal if no periprosthetic FDG uptake (grade 0) was observed.
The readers first graded the lesions on the attenuation-corrected FDG PET images. Then they evaluated the uncorrected images and graded them similarly. Infection was diagnosed only if diffusely increased uptake of grade 3 or 4 was also clearly visible on the uncorrected images. Focally increased periprosthetic uptake of grade 1 or 2 on attenuation-corrected and uncorrected images was considered to indicate a loosened total hip replacement.
Standard of Reference
In 26 of 35 patients, the final diagnosis was rendered at microbiologic evaluation of the surgical specimens and on the basis of intraoperative findings. Our criterion for infection was based on the detection of microorganisms in cultures. If no microorganisms were found, the detection of local abscess formation and the presence of neutrophilic granulocytes were also considered, for the purposes of this study, to indicate that infection was definitely present. For eight of the 26 patients, results of analysis of additional histologic specimens of capsular tissue and bone were available. In the remaining nine of the 35 patients, the diagnosis of infection was based on results of joint aspiration together with clinical findings after a minimum follow-up duration of 6 months (range, 6–14 months; mean, 8 months). Fluoroscopically guided joint aspiration with standard microbacterial examination was performed in all 35 patients. No infection was assumed to be present in patients who had negative microbiologic results after joint aspiration; a normal erythryocyte sedimentation rate, C-reactive protein level, and white blood cell count; and improvement in their clinical symptoms for more than 6 months.
Statistical Evaluation
Sensitivity, specificity, and accuracy values for conventional radiography, bone scintigraphy, and FDG PET in the diagnosis of infection in a total hip replacement were calculated separately for each reader. The interobserver agreement for each imaging modality was determined with statistics. A value of less than 0.40 was considered to indicate poor agreement; a value between 0.40 and 0.75, fair to good agreement; and a value of greater than 0.75, excellent agreement (28). Diagnostic confidence was classified according to the following system: A score of 0 indicated that there was a high likelihood that no infection was present (ie, both readers diagnosed no infection); a score of 1, that the presence of infection was uncertain (ie, the readers’ diagnoses were discordant); and a score of 2, that there was a high likelihood of infection (ie, both readers diagnosed infection). Diagnostic confidence was calculated separately for sensitivity and specificity values and was compared with the sign test. P values less than .05 were considered to indicate statistically significant differences.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Seventeen of the 26 patients who underwent surgery had no infection on the basis of results of bacteriologic culture (n = 17), histologic findings (n = 3), and/or findings at joint aspiration (n = 17). Ten of these 17 patients had a foreign body reaction to polyethylene debris with granuloma formation; four patients, aseptic loosening of the entire prosthesis; and two patients, secondary acetabular component malposition. At surgery, one patient proved to have a femoral fatigue fracture that had not been prospectively diagnosed at conventional radiography.
For all nine patients who did not undergo surgery, no diagnosis of infection was rendered according to our standard of reference. Of these nine patients, one had rheumatoid arthritis, three had weakness of the abductor muscles, and one had persistent pain relating to the iliopsoas muscle at the site of the total hip replacement. In four patients, no reason for their hip pain was found. Signs of loosening (eg, the presence of periprosthetic areas of lucency > 2 mm in width) were seen on conventional radiographs in these patients. However, after a follow-up of 6 months, the hip pain of these four patients disappeared and no progression of loosening occurred.
Conventional Radiography
Findings on sequential conventional radiographs were true-positive for the first reader in eight of nine patients with infection and were true-positive for the second reader in seven patients: Rapid osteolysis was observed in three patients by the first reader and in four patients by the second, extensive migration was observed in two patients by the first reader and in one patient by the second, and irregular osteolysis during a short period of time (6–12 months) was observed in three patients by the first reader and in two patients by the second (Figs 1, 2). False-positive findings were seen in 13 cases by the first reader and in nine cases by the second reader. The first reader saw only one, and the second reader only two, false-negative findings on conventional radiographs in patients for whom no follow-up radiographs were available. In one case, when results of three-phase bone scintigraphy and PET were false-negative, the correct diagnosis could only be made with serial conventional radiographs (Fig 2).
|
|
|
FDG PET
With use of diffusely increased grade 3 or 4 FDG uptake as a criterion for infection, results of PET were true-positive in three of nine patients with infection for reader 1 and in two of nine patients for reader 2. In five (reader 1) and four (reader 2) of 10 patients in whom a foreign body reaction with granuloma and a hyperplastic capsule were seen at surgery, results of PET were false-positive. For three of these patients, histopathologic analysis revealed inflammatory cells—including histiocytes, giant cells, and fibrovascular tissue—associated with a foreign body reaction (Fig 3).
|
Additional inclusion of diffusely increased grade 2 FDG uptake for the diagnosis of infection would have resulted in an increase in sensitivity values and a decrease in specificity and accuracy values. PET results would have been true-positive in seven of the nine cases of infection for the first reader and in five cases for the second, false-positive in 11 cases for the first reader and in 14 cases for the second, true-negative in 15 cases for the first reader and in 12 cases for the second, and false-negative in two cases for the first reader and in four cases for the second. Sensitivity values would have increased from 33% to 78% for the first reader and from 22% to 56% for the second. Specificity would have decreased from 81% to 58% for the first reader and from 85% to 46% for the second, while accuracy would have decreased from 69% to 63% for the first reader and from 69% to 49% for the second.
Conventional radiography and three-phase bone scintigraphy had poor interobserver agreement, with bone scintigraphy having a value of 0.11, while PET, with a value of 0.47, had fair to good interobserver agreement (Table 1). Calculating the diagnostic confidence for sensitivity and specificity values separately revealed that PET (P = .035) and three-phase bone scintigraphy (P = .004) had greater specificity than conventional radiography for excluding an infection. Conversely, conventional radiography appeared to be more sensitive than PET (P = .016) and three-phase bone scintigraphy (P = .031) for the diagnosis of an infected total hip replacement (Table 2). Presumably because of the small number of infections, the difference in sensitivity values between three-phase bone scintigraphy and PET was not significant (P = .22).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In aseptic loosening, polyethylene particles of wear debris (composed of polymethylmethacrylate, polyethylene, or metal) lead to macrophage activation, which in turn releases bone-resorbing products like prostaglandin E2 (32,33). This results in areas of progressive bone loss and granulomatous pseudotumors, which may simulate abnormalities relating to infection. The time course of bone resorption, however, may represent an important tool in the differentiation of aseptic from septic loosening (11,34).
Lieberman et al (35) reported that serial radiography is an effective method of detecting loosening and that bone scintigraphy was useful only when radiographic results were inconclusive with regard to loosening or infection. In our investigation, the sensitivity of conventional radiography in the diagnosis of hip infection compared favorably with that of bone scintigraphy and FDG PET. This may be explained by the fact that in a large proportion of the patients in our study (ie, 32 of 35), serial radiographs (at least one early postoperative and one follow-up radiograph) were available. In one patient in our study, the correct diagnosis of infection could only be made with serial radiographs; PET and bone scintigraphy yielded false-negative findings.
At bone scintigraphy, focally increased radionuclide uptake around the prosthesis is commonly considered to represent loosening, while diffusely increased uptake is commonly considered to represent infection (36,37). However, infection may also be present in prostheses with focal uptake patterns (24,25).
In contrast to Levitsky et al (20), who described a sensitivity of 33% and a specificity of 86% for bone scintigraphy in diagnosing infection, Itasaka et al (38) described a sensitivity of 83%, a specificity of 79%, and an accuracy of 79%. Bone scintigraphy reveals increased activity in the first weeks postoperatively; this activity gradually decreases over a period of 6–12 months. Bone scintigraphy is of limited value in the diagnosis of loosening and infection in patients who undergo it within 12 months after an arthroplasty procedure. According to our study protocol, such patients were excluded from the investigation. The inclusion of these patients probably would have affected the specificity of bone scintigraphy.
Aliabadi et al (25) reported that even the use of combined radiography and bone scintigraphy did not enable differentiation between loosening and infection: Combining the results of conventional radiography and bone scintigraphy resulted in a sensitivity of 84% and a specificity of 92% for the diagnosis of loosening, infection, or both. In the same study, sensitivity and specificity values, respectively, were 54% and 96% for radiography alone and 73% and 96% for bone scintigraphy alone.
The use of 111In-labeled white blood cells in combination with 99mTc sulfur colloid marrow imaging allows correction for the abnormal leukocyte accumulation that may occur due to either displacement of marrow at the time of surgical implantation of the prosthesis or stimulation of the reconversion of yellow to red marrow (39). A noncongruent pattern with absent bone marrow uptake and accumulation of leukocytes is highly accurate for diagnosing hip prosthesis infection. Sensitivity, specificity, and accuracy with this technique were 100%, 97%, and 98%, respectively (21). However, the preparation of 111In-labeled white blood cells is laborious, and the procedure is time consuming for the patient. In addition, the procedure is substantially more expensive than bone scintigraphy, although it is not as expensive as PET scanning in most countries. Finally, the aim of the present study was to compare FDG PET with the most widely available imaging methods used to detect loosening and infection.
Several investigations of FDG PET for the evaluation of suspected infection have had promising results (40–42). Zhuang et al (40) reported a sensitivity of 90%, a specificity of 89%, and an accuracy of 89% for FDG PET in diagnosing infected lower limb prostheses. FDG PET was more accurate for detecting infections in patients with hip replacements than for detecting infections in patients with knee prostheses. Zhuang et al (40) found that increased tracer uptake along the interface between bone and the prosthesis is associated with infection, whereas increased uptake limited to the head or the neck of the prosthesis indicates prosthetic loosening.
However, a more recent report by Zhuang et al (41) revised the statement that increased uptake around the head or neck of the prosthesis is limited to loosening. Postsurgical inflammatory changes may be indistinguishable from infection and may persist for years. In our investigation, localization of uptake was not used for the differentiation of infection from loosening. However, we agree with Zhuang et al (40) that using increased uptake as the sole criterion for diagnosing infection would result in an increased false-positive rate because uptake in loosening can be as high as or even higher than uptake in infection. In our study, five patients with extensive granuloma caused by a foreign body reaction had highly increased FDG uptake (of grade 3 or 4) around the entire prosthesis.
Prior to our investigation, Love et al (43) expressed doubts that FDG PET can help differentiate aseptic from septic loosening. Our data appear to support this statement. This probably relates to the remarkably similar histopathologic morphologic features of aseptic and septic loosening. Aseptic loosening is mostly a result of an inflammatory immune reaction to the prosthetic material. In the study of Love et al, histopathologic analysis of specimens revealed the presence of histiocytes in 95% of patients and giant cells in 80%, as well as lymphocytes and plasma cells in 25%. The important difference in these patients compared with those with infection was that neutrophils were present in less than 10% of patients with aseptic loosening associated with nonspecific inflammation (43). Neutrophilic granulocytes are typically present in acute infection (44). Mononuclear white cells exhibit a so-called respiratory burst when fighting inflammation. When exposed to certain stimuli, phagocytes (neutrophils, eosinophils, and mononuclear phagocytes) start metabolizing large quantities of glucose by way of the hexose monophosphate shunt, and their rates of oxygen uptake increase greatly, sometimes more than 50-fold (45). This change of the resting cells to activated phagocytes is known as the respiratory burst. The partial histologic overlap between infection and aseptic loosening may explain the diagnostic difficulties encountered in the present study.
de Winter et al (46) examined 17 patients suspected of having periprosthetic infection from a group of 60 patients suspected of having musculoskeletal infection. They had difficulties in differentiating infection from aseptic loosening with PET. In one false-positive finding they observed cell-rich vascular areas between the implant and bone and a pseudocapsule around an aseptically loose implant that contained more activated macrophages and proliferating fibroblast-like cells than did the tissue around well-fixed implants.
Five of the false-positive findings in our study proved at surgery to represent a loosened prosthesis; at histologic analysis, a foreign body reaction with the presence of macrophages and giant cells, both of which accumulate FDG, was observed for three of these five prostheses.
The diagnostic accuracy of FDG PET in patients with prosthetic replacements is also affected by attenuation-correction–induced artifacts that mimic increased FDG uptake around the prostheses and, therefore, a disease process adjacent to the prostheses (47,48). However, in our study, PET artifacts relating to the material of the prostheses (eg, cobalt nickel) could easily be distinguished from active disease adjacent to a hip prosthesis by comparing the apparently increased uptake seen on the attenuation-corrected images with the uptake on the uncorrected images, in which fewer reconstruction artifacts were present (47). Indeed, owing to the relatively high photon absorption of the solid prosthetic material itself, we observed apparently increased FDG uptake around the neck of the hip prosthesis in regions where the prostheses were not surrounded by bone on attenuation-corrected images obtained in two patients. In a series of patients who had sustained trauma, Schiesser et al (49) were unable to identify any metallic artifacts and therefore any false-positive results due to the material. This may relate to the fact that patients who have been treated for trauma commonly have slender plates rather than the massive implants seen in patients who have undergone arthroplasty (49).
Our study had limitations. The prevalence of infection was low—this was related to the prospective nature of the study and to the relatively low rate of infection in prosthetic joint surgery—and, therefore, sensitivity was low and there was a wide range of confidence for sensitivities. In the past, we were often too sensitive in classifying a lesion as positive for infection at PET, with the result of low specificity values. Therefore, we based our diagnosis at PET on degrees of positivity according to a receiver operating characteristic analysis performed by Stumpe et al (27), who reported that the best cutoff point for the diagnosis of infection was when grade 3 or 4 uptake was considered to indicate infection. Also including grade 2 uptake for the diagnosis of infection results in an increase in the sensitivity values and a decrease in the specificity and accuracy values. Therefore, the operating point of the observers on the receiver operating characteristic curves seemed to be adequately chosen in our study. The low values in our study may not only reflect the uncertainty of the various readers in diagnosing an infection in prosthetic joint replacements with the current imaging modalities. Moreover, values are strongly influenced by low disease prevalence.
In addition, for obvious reasons, surgery with microbiologic evaluation is normally not performed in all patients with a painful total joint replacement, and clinical follow-up sometimes has to be used as the standard of reference; this was the case for nine patients in our study. To some degree, the decision to obtain surgical proof was affected by the low prevalence of infection in prosthetic joint replacements and by the results of conventional radiography and bone scintigraphy. However, PET results were not included in the decision-making process because data are controversial so far.
Our data suggest that FDG PET as an infection imaging modality offers no benefit in addition to three-phase bone scintigraphy in patients with prosthetic joint replacement. In our study population, PET performed similarly to three-phase bone scintigraphy. PET was more specific but less sensitive than conventional radiography for the diagnosis of infection.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, K.D.M.S., J.H., H.P.N.; study concepts and design, K.D.M.S., H.P.N., J.H.; literature research, K.D.M.S., M.Z.; clinical studies, K.D.M.S., H.P.N., J.H.; data acquisition, K.D.M.S., H.P.N.; data analysis/interpretation, K.D.M.S., M.Z., E.M.K., T.F.H., G.W.G., G.K.v.S., J.H.; statistical analysis, K.D.M.S.; manuscript preparation, definition of intellectual content, editing, and final version approval, K.D.M.S., M.Z., J.H., G.K.v.S.; manuscript revision/review, M.Z., J.H., G.K.v.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Nagoya, M. Kaya, M. Sasaki, K. Tateda, and T. Yamashita Diagnosis of peri-prosthetic infection at the hip using triple-phase bone scintigraphy J Bone Joint Surg Br, February 1, 2008; 90-B(2): 140 - 144. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Kumar, S. Basu, D. Torigian, V. Anand, H. Zhuang, and A. Alavi Role of Modern Imaging Techniques for Diagnosis of Infection in the Era of 18F-Fluorodeoxyglucose Positron Emission Tomography Clin. Microbiol. Rev., January 1, 2008; 21(1): 209 - 224. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H Marmery and S Ostlere Imaging of prosthetic joints Imaging, September 1, 2007; 19(3): 299 - 309. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. W. Bauer, J. Parvizi, N. Kobayashi, and V. Krebs Diagnosis of Periprosthetic Infection J. Bone Joint Surg. Am., April 1, 2006; 88(4): 869 - 882. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. K. von Schulthess, H. C. Steinert, and T. F. Hany Integrated PET/CT: Current Applications and Future Directions Radiology, February 1, 2006; 238(2): 405 - 422. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Ohashi, G. Y. El-Khoury, D. L. Bennett, J. M. Restrepo, and K. S. Berbaum Orthopedic Hardware Complications Diagnosed with Multi-Detector Row CT Radiology, November 1, 2005; 237(2): 570 - 577. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Love, M. B. Tomas, G. G. Tronco, and C. J. Palestro FDG PET of Infection and Inflammation RadioGraphics, September 1, 2005; 25(5): 1357 - 1368. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Love, S. E. Marwin, M. B. Tomas, E. S. Krauss, G. G. Tronco, K. K. Bhargava, K. J. Nichols, and C. J. Palestro Diagnosing Infection in the Failed Joint Replacement: A Comparison of Coincidence Detection 18F-FDG and 111In-Labeled Leukocyte/99mTc-Sulfur Colloid Marrow Imaging J. Nucl. Med., November 1, 2004; 45(11): 1864 - 1871. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
