Lamellated Hyperintense Synovitis: Potential MR Imaging Sign of an Infected Knee Arthroplasty 1

Transcription

1 Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at Original Research n Musculoskeletal Imaging Andrew J. Plodkowski, MD Catherine L. Hayter, MBBS Theodore T. Miller, MD Joseph T. Nguyen, MPH Hollis G. Potter, MD Lamellated Hyperintense Synovitis: Potential MR Imaging Sign of an Infected Knee Arthroplasty 1 Purpose: Materials and Methods: To determine the sensitivity and specificity of lamellated hyperintense synovitis for infection following knee arthroplasty and to determine the inter- and intraobserver variability of this sign at magnetic resonance (MR) imaging. The purpose of the retrospective case control study was approved by the hospital s institutional review board. MR images from 28 patients with proved infected total knee arthroplasty and 28 patients with noninfected arthroplasty were reviewed by two musculoskeletal radiologists for the presence of lamellated hyperintense synovitis. Cases were rereviewed 2 weeks later by each reader. The sensitivity and specificity were calculated with the initial reads. The k statistic was used to assess inter- and intraobserver reliability. Results: Conclusion: The sensitivity of lamellated hyperintense synovitis for infection was (95% confidence interval [CI]: 0.75, 0.97) and the specificity was (95% CI: 0.74, 0.94). There was almost perfect interobserver agreement (k = 0.82; 95% CI: 0.72, 0.93; P,.001) and intraobserver agreement (for reader 1, k = 0.89 [95% CI: 0.78, 1.00; P,.001] and for reader 2, k = 0.89 [95% CI: 0.77, 1.00; P,.001]) in the classification of the synovial pattern. In this selected series of patients, the presence of lamellated hyperintense synovitis at MR imaging of knee arthroplasty had a high sensitivity and specificity for infection. This sign had high inter- and intraobserver reliability. q RSNA, From the Department of Radiology and Imaging, Hospital for Special Surgery, 535 E 70th St, New York, NY Received January 16, 2012; revision requested March 19; final revision received May 30; accepted June 14; final version accepted July 11. Address correspondence to A.J.P. ( Plodkowski@yahoo.com). q RSNA, radiology.rsna.org n Radiology: Volume 266: Number 1 January 2013

2 Magnetic resonance (MR) imaging is an increasingly used modality to investigate patients presenting with pain following arthroplasty (1,2). Some studies indicate that MR imaging is the most accurate imaging modality to detect and quantify osteolysis (3,4) and that MR imaging findings may be used to distinguish different patterns of wear-induced synovitis (5). Infection, which has been reported in less than 1% of arthroplasties (6), may be a cause of synovitis. We have previously reported anecdotally that a lamellated hyperintense synovium is suggestive of an infected arthroplasty (7). Thus the purpose of this retrospective case control study was to determine the sensitivity and specificity of lamellated hyperintense synovitis as an MR sign of an infected knee arthroplasty and to determine the inter- and intraobserver reliability of this sign. Materials and Methods Subject Cohort The project was approved by the institutional review board at the Hospital for Special Surgery (New York, NY). A total of 1425 patients with painful total knee arthroplasty who underwent MR imaging at our institution between January 2008 and July 2011 were identified through the picture archiving and communication system. The clinical and laboratory notes were reviewed to determine which patients underwent knee joint aspiration or revision surgery within 3 months of the MR imaging. A total of 213 patients met these inclusion criteria. Of the 213 patients, there were 29 (cohort 1) who were ultimately clinically considered to have an infection, and infection was defined as a culture positive for bacterial organisms at aspiration Advance in Knowledge nn A lamellated hyperintense synovial pattern in the setting of a total knee arthroplasty is suggestive of infection. and/or histopathologic findings positive for bacterial infection (8) at revision surgery, with subsequent long-term intravenous antibiotic treatment. We then arbitrarily selected 29 patients from the remaining 184 patients who showed no evidence of infection at either aspiration or histopathologic examination to serve as controls (cohort 2). One patient from each cohort was excluded from analysis because of motion artifact that degraded the diagnostic quality of the MR images. Therefore, each cohort consisted of 28 patients. Data, including patient age, sex, side of arthroplasty, duration of knee arthroplasty placement, indication for performing the MR imaging, and type of bacterial infection, were recorded. The control cohort was not matched for body mass index or sex. MR Image Acquisition Imaging was performed with 1.5-T clinical imaging units (GE Healthcare, Waukesha, Wis) by using eight-channel phased-array transmit-receive knee coils (In Vivo, Orlando, Fla). Images were optimized to minimize susceptibility artifact by using commercially available protocols (7). Two-dimensional fast spin-echo (SE) imaging was performed in three orthogonal planes with the following parameters: repetition time msec/echo time msec, /24 34; bandwidth, 6 83 to 100 khz ( Hz/pixel); field of view, cm; number of signals acquired, four to six; acquisition matrix, 512 (frequency) (phase); section thickness, mm; echo train length, A short-tau fast inversion recovery sequence was obtained in the sagittal plane with the parameters: repetition time msec/echo time msec/inversion time msec, 6000/17/150; receiver bandwidth, 6 83 to 100 khz ( Hz/pixel); field of view, 18 cm; number of Implication for Patient Care nn The finding of lamellated hyperintense synovitis at MR imaging of the knee arthroplasty should alert the radiologist to the possibility of infection. signals acquired, two; acquisition matrix, 256 (frequency) (phase); section thickness, 4 mm; echo train length, 12. Total acquisition time was 30 minutes. MR Image Analysis The two-dimensional fast SE images in the axial plane were used for the assessment of the synovium. The images were evaluated independently by two musculoskeletal radiologists who were blinded to the indication for performing the MR imaging and the results of subsequent aspiration and histopathologic examination. One of the readers was a senior musculoskeletal radiologist with 12 years of experience in MR imaging of arthroplasty (T.T.M., reader 1); the other was a dedicated musculoskeletal MR imaging fellow with 18 months of experience in MR imaging of arthroplasty (C.L.H., reader 2). Images were assessed for the presence of lamellated hyperintense synovitis. Lamellated synovium was defined as a thickened synovium composed of multiple layers. The synovium was characterized as hyperintense if the signal intensity of the synovium was higher than that of skeletal muscle on the intermediate echo time fast SE images (Fig 1). Any other synovial pattern was designated as not lamellated (Fig 2). No training session was given. The images were not assessed for secondary signs of infection such as extracapsular soft-tissue edema, extracapsular collections, bone destruction, Published online before print /radiol Content code: Radiology 2013; 266: Abbreviations: CI = confidence interval SE = spin echo Author contributions: Guarantors of integrity of entire study, all authors; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, A.J.P., C.L.H., H.G.P.; clinical studies, A.J.P., C.L.H., T.T.M., H.G.P.; statistical analysis, A.J.P., J.T.N., H.G.P.; and manuscript editing, all authors Conflicts of interest are listed at the end of this article. Radiology: Volume 266: Number 1 January 2013 n radiology.rsna.org 257

3 or reactive lymphadenopathy. Because the presence of these secondary signs of infection might have subconsciously biased the reader, only the non fatsuppressed axial images were used in this study, because marrow edema and soft-tissue collections are not as conspicuous in these sequences. The readers classified findings on MR images in each case as lamellated or not lamellated. The data from this first read were used to calculate the sensitivity, specificity, positive predictive value, and negative predictive value of the sign. Blinded to the results of the original read, the observers reread images in each case 2 weeks later to determine intraobserver reliability. To further assess interobserver reproducibility, a musculoskeletal fellow with limited experience in MR imaging of arthroplasty was given a training session, including 20 knee arthroplasty cases, some of which were from the patient cohorts. After a 1-week wait, this fellow then read images in each case in the study for the presence of lamellated hyperintense synovitis by using the criteria above. Statistical Analysis The k statistic was used to assess interrater reliability between the readers. Additional reliability analyses were conducted for each reader to evaluate intraobserver reliability. Overall interpretations of the k statistic were based on criteria described by Landis and Koch (9), with calculated 95% confidence intervals (CIs) used to assess the precision of our estimates. The level of agreement was defined as follows: k statistic of was considered to represent almost perfect agreement; k statistic of , substantial agreement; k statistic of , moderate agreement; k statistic of , fair agreement; and k statistic of , slight agreement. All P values reported were results from two-sided tests, with significance set as a of.05. All statistical tests were conducted by using statistical analysis software (SAS software, version 9.2; SAS Institute, Cary, NC). Figure 1 Figure 1: Axial intermediate echo-time fast SE MR image demonstrates the criteria used to denote the presence of lamellated hyperintense synovitis. The lamellated synovium is thickened and composed of multiple layers (white arrows). The synovium is hyperintense when compared with skeletal muscle (black arrow). The synovium was read as lamellated by both readers. Subsequent cultures revealed a Staphylococcus aureus infection. Results Cohort 1 (infected) comprised 12 women (mean age, 64.2 years [standard deviation]) and 16 men (mean age, 67.1 years ), with a combined mean age of 65.8 years The left knee was imaged in 13 patients, and the right knee in 15 patients. The mean duration of knee arthroplasty placement was 80.7 months, with a range of months. In 11 patients, the arthroplasty was revision surgery. The indication for performing the MR imaging was that a patient was suspected of having infection (17 cases), loosening (three cases), quadriceps tendon rupture (two cases), neuroma, Baker cyst, polyethylene wear, component position, patella-backing integrity, and hemarthrosis. The bacteria isolated after culture for these patients consisted of Staphylococcus epidermidis (seven cases), Staph aureus (five cases), Enterococcus faecalis (two cases), methicillin-resistant Staph aureus (two cases), Streptococcus viridans (two cases), Propionibacterium acnes (two cases), Staphylococcus haemolyticus, beta-hemolytic Streptococcus group G, coagulase-negative Staphylococcus species, Figure 2 Figure 2: Axial intermediate echo-time fast SE MR image demonstrates a nonlamellated synovitis. The synovium is thickened and is of low signal intensity (white arrows) when compared with skeletal muscle (black arrow). The synovium was read as not lamellated by both readers. This patient eventually was determined to have wear-induced synovitis. Peptostreptococcus micros, Escherichia coli, Streptococcus intermedius, Bacteroides fragilis, Enterobacter cloacae, and Diphtheroids species. In four patients, two isolated bacteria were observed, and in two patients, no bacteria were identified but infection was confirmed at histologic examination. Overall, infection was confirmed in 20 patients with histologic examination and culture, in six patients with culture only, and in two patients with histologic examination only. Cohort 2 (control) comprised 15 women (mean age, 66.4 years 6 8.0) and 13 men (mean age, 64.8 years ), with a combined mean age of 65.7 years The left knee was imaged in 15 patients, and the right knee in 13 patients. The mean duration of knee arthroplasty placement was 67 months, with a range of months. In three patients, the arthroplasty was revision surgery. The indication for performing the MR imaging was pain (nine cases), the patient was suspected of having infection (four cases), loosening (four cases), polyethylene wear (two cases), osteolysis (two cases), scar tissue (two cases), popliteal aneurysm (one case), patellar fracture (one case), instability (one case), stiffness (one case), and alignment (one case). 258 radiology.rsna.org n Radiology: Volume 266: Number 1 January 2013

4 Figure 3 Statistical Analysis Results for MR Interpretation of Lamellated Hyperintense Synovitis as Sign of Infection Statistic Reader 1 Reader 2 Sensitivity 0.86 (0.75, 0.93) 0.92 (0.82, 0.97) Specificity 0.87 (0.76, 0.94) 0.85 (0.74, 0.92) Positive predictive value 0.88 (0.79, 0.96) 0.84 (0.74, 0.94) Negative predictive value 0.86 (0.77, 0.95) 0.93 (0.86, 0.99) Note. Data in parentheses are the 95% CIs. Figure 4 Figure 3: Axial intermediate echo-time fast SE MR image in a 67-year-old woman referred for MR imaging to evaluate for patellar component loosening. The synovium is mildly thickened and hyperintense but does not appear lamellated (arrow). The synovium was read as not lamellated by both readers. Subsequent culture revealed methicillinresistant Staph aureus infection that was confirmed at histopathologic examination after surgery for removal of the arthroplasty. Figure 4: A, B, Axial intermediate echo-time fast SE MR images in a 69-year-old woman referred for MR imaging to evaluate for infection. A, The synovium is lamellated and hyperintense to skeletal muscle (arrows). The synovium was read as lamellated by both readers. B, Subsequent diagnostic arthroscopy was performed, and a piece of extruded cement was removed. Cultures were negative for infection, and the patient s symptoms resolved following arthroscopy. In retrospect, the piece of extruded cement was visible in the lateral aspect of the suprapatellar pouch (arrow). intraobserver agreement for the musculoskeletal fellow between the two reads (k = 0.75; 95% CI: 0.58, 0.92; P,.001). Using the musculoskeletal fellow s first set of data after the training session, the sensitivity for infection for the musculoskeletal fellow s read was 0.78 (95% CI: 0.66, 0.87). The specificity for infection for the musculoskeletal fellow s read was 0.81 (95% CI: 0.69, 0.89). The results of analysis for readers 1 and 2 are outlined in the Table. The sensitivity of lamellated hyperintense synovitis for infection was (95% CI: 0.75, 0.97). The specificity of lamellated hyperintense synovitis for infection was (95% CI: 0.74, 0.94). In the infected cohort, images in three cases were read as not lamellated by both readers (Fig 3). In the noninfected cohort, the image in one case was read as lamellated by both readers. This patient underwent diagnostic arthroscopy following MR imaging, and a piece of extruded cement was removed (Fig 4). There was almost perfect interobserver agreement between the two readers in the classification of the synovial pattern (k = 0.82; 95% CI: 0.72, 0.93; P,.001). There was almost perfect intraobserver agreement for each reader in their classification of the synovial pattern between the two reads. For reader 1, the k statistic was 0.89 (95% CI: 0.78, 1.00; P,.001). For reader 2, the k statistic was 0.89 (95% CI: 0.77, 1.00; P,.001). There was moderate to substantial interobserver agreement between the first read of the musculoskeletal fellow and the first read of readers 1 and 2 in the classification of the synovial pattern (k = ; 95% CI: 0.50, 0.90; P,.001). There was moderate Discussion While the best diagnostic algorithm for diagnosis of infection relies on histologic and microbiological studies of several tissue specimens (10), other studies such as amplification of the 16S rrna gene have been recently introduced to improve detection of infection (11). Secondary clinical parameters such as erythrocyte sedimentation rate and C-reactive protein level are nonspecific and have limited predictive values in certain patient cohorts (12). Radionuclide bone scanning is sensitive but not specific, (13) and indium 111 white blood cell scans are not universally available and may present technical challenges with labeling (14,15). Radiology: Volume 266: Number 1 January 2013 n radiology.rsna.org 259

5 We previously reported, using a different cohort of patients, that the positive predictive value of lamellated hyperintense synovitis alone for infection was 0.47, which increased to 0.83 when it was combined with extracapsular softtissue edema, extracapsular collections, or reactive lymphadenopathy (16). In this current study we found a high positive predictive value of lamellated synovitis alone, along with a high sensitivity and specificity for infection. The finding of lamellated hyperintense synovitis should therefore alert the radiologist to the possibility of infection. We have also anecdotally seen lamellated hyperintense synovitis in infected native knees but not in inflammatory arthritides such as rheumatoid arthritis. There was one case in which the image was read as lamellated by both readers and the knee was, in fact, not infected. This patient subsequently underwent arthroscopy, and a piece of extruded cement was removed. Following arthroscopy, the patient s symptoms resolved. It is possible that the mechanical irritation of the intraarticular body generated a synovial inflammation pattern similar to that of infection, resulting in the lamellated hyperintense pattern observed at MR imaging. There were some limitations to this study. There were a small number of patients in each cohort. This, however, reflects the fact that infection is an uncommon complication following arthroplasty. We found 29 cases of confirmed infection of 1425 MR images obtained during a 3-year period; this finding represents an infection rate of approximately 2% in this select group and an overall infection rate of less than 1% at our institution (17). Both readers in this study were musculoskeletal radiologists who were experienced with reading MR images of arthroplasty. This may have also affected our results. For this reason, we undertook a brief training session with a musculoskeletal fellow with little experience in MR imaging of arthroplasty. Although inter- and intraobserver reproducibility was lower for the musculoskeletal fellow, using the results of the first read, he also had a high sensitivity and specificity for determining infection by using the MR imaging sign of lamellated hyperintense synovitis. Finally, because of the retrospective nature of this study, we do not have a histologic explanation of what is the lamellated synovium. We speculate that the lamellated hyperintense appearance of the synovium is caused by a combination of inflammatory edema and cellular infiltration. In conclusion, the presence of lamellated hyperintense synovitis at MR imaging of knee arthroplasty is highly suggestive of infection. This sign is reproducible, with high inter- and intraobserver reliability. Disclosures of Conflicts of Interest: A.J.P. No relevant conflicts of interest to disclose. C.L.H. No relevant conflicts of interest to disclose. T.T.M. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: received royalties from Amirsys Publishers for an anatomy chapter. Other relationships: none to disclose. J.T.N. No relevant conflicts of interest to disclose. H.G.P. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: receive payment for advisory board membership from Kensey Nash; received consultancy fees from Biomet, BioMimetic Therapeutics, and Smith & Nephew; author and institution received institutional research support from GE Healthcare. Other relationships: none to disclose. References 1. Kosy JD, Eyres KS, Toms AD. The value of magnetic resonance imaging in investigating a painful total knee arthroplasty. J Arthroplasty 2011;26(6):977.e9 977.e S Published October 29, Accessed December 1, Mallo GC, Stanat SJ, Jones JA, Capozzi JD, Luchs JS. Catastrophic polyethylene failure diagnosed with magnetic resonance imaging in a painful total knee arthroplasty. J Arthroplasty 2011;26(3):505.e e15. S %2810% /abstract. Published March 23, Accessed December 1, Walde TA, Weiland DE, Leung SB, et al. Comparison of CT, MRI, and radiographs in assessing pelvic osteolysis: a cadaveric study. Clin Orthop Relat Res 2005;(437): Weiland DE, Walde TA, Leung SB, et al. Magnetic resonance imaging in the evaluation of periprosthetic acetabular osteolysis: a cadaveric study. J Orthop Res 2005;23(4): Hayter CL, Koff MF, Potter HG. Magnetic resonance imaging of the postoperative hip. J Magn Reson Imaging 2012;35(5): Meehan JP, Danielsen B, Tancredi DJ, Kim S, Jamali AA, White RH. A population-based comparison of the incidence of adverse outcomes after simultaneous-bilateral and stagedbilateral total knee arthroplasty. J Bone Joint Surg Am 2011;93(23): Potter HG, Foo LF. Magnetic resonance imaging of joint arthroplasty. Orthop Clin North Am 2006;37(3): Morawietz L, Classen RA, Schröder JH, et al. Proposal for a histopathological consensus classification of the periprosthetic interface membrane. J Clin Pathol 2006;59(6): Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977;33(2): Zimmerli W, Ochsner PE. Management of infection associated with prosthetic joints. Infection 2003;31(2): Ince A, Rupp J, Frommelt L, Katzer A, Gille J, Löhr JF. Is aseptic loosening of the prosthetic cup after total hip replacement due to nonculturable bacterial pathogens in patients with low-grade infection? Clin Infect Dis 2004;39(11): Accessed December 1, Spangehl MJ, Masri BA, O Connell JX, Duncan CP. Prospective analysis of preoperative and intraoperative investigations for the diagnosis of infection at the sites of two hundred and two revision total hip arthroplasties. J Bone Joint Surg Am 1999;81(5): Love C, Marwin SE, Palestro CJ. Nuclear medicine and the infected joint replacement. Semin Nucl Med 2009;39(1): Love C, Tomas MB, Marwin SE, Pugliese PV, Palestro CJ. Role of nuclear medicine in diagnosis of the infected joint replacement. RadioGraphics 2001;21(5): Love C, Palestro CJ. Radionuclide imaging of infection. J Nucl Med Technol 2004;32(2):47 57; quiz Hayter CL, Miller TT, Bogner EA, Potter HG. MR imaging findings of infection following knee arthroplasty: the positive predictive value of a lamellated synovitis [abstr]. Skeletal Radiol 2011;40(8): Ranawat AS. Revision total knee replacement: frequently asked questions (FAQs). Hospital for Special Surgery Web site. Published April 14, Accessed April 9, radiology.rsna.org n Radiology: Volume 266: Number 1 January 2013