This copy is for personal use only. To order printed copies, contact Purpose: Materials and Methods: Results: Conclusion:

Transcription

1 This copy is for personal use only. To order printed copies, contact Diagnostic Accuracy of a Fluidattenuated Inversion-Recovery Sequence with Fat Suppression for Assessment of Peripatellar Synovitis: Preliminary Results and Comparison with Contrast-enhanced MR Imaging 1 Original Research n Musculoskeletal Imaging Hye Jin Yoo, MD, PhD Sung Hwan Hong, MD, PhD Hye Yeon Oh, MD Ja-Young Choi, MD, PhD Hee Dong Chae, MD Joong Mo Ahn, MD, PhD Heung Sik Kang, MD, PhD 1 From the Department of Radiology, Seoul National University Hospital, Seoul, South Korea (H.J.Y., S.H.H., H.Y.O., J.Y.C., H.D.C.); Department of Radiology, Seoul National University College of Medicine, 103 Daehak-ro, Jongno-gu, Seoul , South Korea (H.J.Y., S.H.H., J.Y.C., H.D.C., J.M.A., H.S.K.); Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, South Korea (S.H.H., H.S.K.); and Department of Radiology, Seoul National University Bundang Hospital, Seongnam, South Korea (J.M.A., H.S.K.). Received January 20, 2016; revision requested March 28; revision received July 4; accepted August 3; final version accepted August 11. Address correspondence to S.H.H. ( drhong@snu.ac.kr). q RSNA, 2016 Purpose: Materials and Methods: Results: Conclusion: To determine the agreement of fat-suppressed (FS) fluid-attenuated inversion-recovery (FLAIR) magnetic resonance (MR) imaging and contrast material enhanced (CE) T1-weighted MR imaging for the assessment of peripatellar synovitis. The institutional review board approved this retrospective study and waived the requirement for patient informed consent. Thirty-three patients with knee pain underwent 3-T MR imaging. The protocol consisted of routine clinical sequences followed by the FLAIR FS sequence (inversion time, 2200 msec) and CE T1- weighted imaging. Visibility of the synovium, synovial thickness, and severity of synovitis in five peripatellar regions were assessed with both sequences. Hoffa synovitis on unenhanced MR images was also analyzed. Then, correlations and agreements between FLAIR FS and CE T1-weighted imaging were evaluated. Diagnostic performance statistics of FLAIR FS and Hoffa synovitis were calculated by using CE T1-weighted imaging as the reference standard. Peripatellar synovitis on FLAIR FS images was found in 14 patients by reader 1 and in 17 patients by reader 2. Strong correlations were found between FLAIR FS and CE T1-weighted imaging in the assessment of peripatellar synovitis by both readers (correlation coefficient, ). With CE T1-weighted imaging as the reference standard, FLAIR FS showed relatively good diagnostic performance for the detection of synovitis of any severity (accuracy of 92.1% 93.9% at the site level and 90.9% at the patient level), while Hoffa synovitis on unenhanced MR images showed moderate sensitivity (78.9% 87.5%) and low specificity (47.1% 64.3%) for the detection of peripatellar synovitis. There was good agreement between the two readers for the synovial visibility (weighted k = ) and synovitis assessments (intraclass correlation coefficient = 0.95, weighted k = ) on FLAIR FS and CE T1-weighted images. Our preliminary study shows that FLAIR FS imaging can potentially enable evaluation of inflamed synovium with high sensitivity and specificity, without the injection of a contrast agent. q RSNA, 2016 Online supplemental material is available for this article. Radiology: Volume 283: Number 3 June 2017 n radiology.rsna.org 769

2 Synovitis is inflammation of the synovial membrane and has been implicated in many joint diseases, such as rheumatoid arthritis and inflammatory osteoarthritis (1). It is believed to play an important role in the pathogenesis of inflammatory osteoarthritis and is regarded as a potential target for novel treatment strategies (2,3). It manifests as synovial membrane thickening or joint effusion, which is optimally imaged by using contrast material enhanced (CE) magnetic resonance (MR) imaging (1,4 9). CE MR imaging generally improves tissue visualization and can allow synovial inflammation to be distinguished from synovial fluid and adjacent articular structures (4 7,10,11). Previous studies have shown that CE MR imaging is more sensitive and specific than unenhanced MR imaging in the assessment of synovitis (10,12). In several studies, investigators found that synovitis detected on CE MR images was correlated with histologic findings (5,13). However, there are obvious drawbacks of intravenous gadolinium-based contrast material administration, including prolonged imaging time, increased cost, possible allergic reactions, and a risk of nephrogenic systemic fibrosis (1). Therefore, CE MR imaging has not been used routinely in large epidemiological studies or clinical trials of therapies for osteoarthritis (14). Commonly, unenhanced fluid-sensitive fat-suppressed (FS) sequences are used in semiquantitative MR imaging assessment of osteoarthritic synovitis Advance in Knowledge nn A fluid-attenuated inversionrecovery (FLAIR) fat-suppressed (FS) sequence seems to enable visualization of the synovium without the injection of an intravenous contrast agent; FLAIR FS imaging yields accuracies of 92.1% 93.9% at the joint level and accuracy of 90.9% at the patient level for the detection of synovitis of any severity when contrast material enhanced (CE) T1-weighted MR images are used as the reference standard. (4,14). With these unenhanced fluidsensitive FS sequences, including T2- weighted and proton density (PD) weighted sequences, the synovium cannot be differentiated from joint effusion. Instead, signal intensity alterations (areas of high intensity) in the Hoffa fat pad (HFP) are scored as a synovitis surrogate and are associated with pain severity (15,16). These signal intensity changes are referred to as Hoffa synovitis with the MR imaging Osteoarthritis Knee Score, which is the MR imaging semiquantitative scoring method for knee osteoarthritis (17). However, it has been shown that these signal intensity changes are a sensitive but nonspecific surrogate for the assessment of synovitis when CE MR imaging is used as the standard of reference (9,14). Therefore, it is necessary to develop new, unenhanced MR imaging sequences to image synovitis directly by improving the image contrast between synovium and joint fluid. The fluid-attenuated inversion-recovery (FLAIR) FS sequence is a combination of FLAIR and FS sequences. FLAIR is an inversion-recovery technique that nulls fluid signals. It is used widely in neuroimaging to suppress cerebrospinal fluid signals, thereby bringing out hyperintense periventricular lesions. By exploiting FLAIR for unenhanced fluidsensitive FS sequences, differentiation between the synovium and joint fluid might be achieved. The purpose of our study was to determine the agreement of unenhanced FLAIR FS images and FS CE MR images for the assessment of peripatellar synovitis. Materials and Methods A retrospective study was performed from November 2014 to August Implication for Patient Care nn A FLAIR FS sequence can potentially provide an alternative method for assessment of knee synovitis, with lower costs and risks when compared with CE sequences. Our institutional review board approved this cross-sectional study and waived the requirement for informed patient consent. Patients Forty-three consecutive patients underwent CE MR imaging of the knee joint with an assigned 3-T MR imaging system between January 2015 and October Among them, 39 patients who underwent both CE T1-weighted imaging and FLAIR FS imaging were included in this retrospective study. Exclusion criteria were as follows: severe image distortion due to metallic implant (n = 2), poor image quality due to motion artifact (n = 1), and different imaging parameters (n = 3) (Fig 1). Finally, 33 patients (mean age, 45.8 years; range, years) were included in the study according to the outcome of power analysis (see the Statistical Analysis section), including 15 men (mean age, 37.9 years; range, years) and 18 women (mean age, 52.4 years; range, years). Men were significantly younger than women (P =.01). The reasons for MR imaging were as follows: follow-up imaging after tumor excision (n = 17), infection (n = 6), osteoarthritis (n = 6), meniscal Published online before print /radiol Content codes: Radiology 2017; 283: Abbreviations: CE = contrast material enhanced CI = confidence interval FLAIR = fluid-attenuated inversion recovery FS = fat suppressed HFP = Hoffa fat pad PD = proton density Author contributions: Guarantors of integrity of entire study, H.J.Y., S.H.H., H.Y.O.; 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; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, H.J.Y., S.H.H., H.Y.O., H.D.C., J.M.A., H.S.K.; clinical studies, H.J.Y., S.H.H., H.Y.O., J.Y.C.; experimental studies, H.Y.O.; statistical analysis, H.J.Y., S.H.H.; and manuscript editing, H.J.Y., S.H.H., H.Y.O., J.Y.C., J.M.A., H.S.K. Conflicts of interest are listed at the end of this article. 770 radiology.rsna.org n Radiology: Volume 283: Number 3 June 2017

3 Figure 1 Figure 1: tear (n = 2), acute calcific periarthritis (n = 1), and osteonecrosis (n = 1). The histologic findings of the tumors were pleomorphic liposarcoma (distal thigh, n = 1), malignant melanoma (distal thigh, n = 1), dermatofibrosarcoma protuberans (distal thigh, n = 1), fibromatosis (popliteal fossa, n = 2), clear cell sarcoma (distal thigh, n = 1), lipoma arborescens (knee joint, n = 1), infrapatellar fat pad chondromatosis (knee, n = 1), synovial sarcoma (popliteal fossa, n = 1), neurogenic tumor (distal thigh, n = 1), hemangioma (distal thigh, n = 1), giant cell tumor (distal femur, n = 5), and enchondromatosis (proximal tibia, n = 1). The mean time interval between surgery and MR imaging was 67.6 months (range, 20 days to 67.6 months). MR Imaging Acquisition MR imaging was performed by using a 3-T MR imaging unit (Skyra; Siemens Healthcare, Erlangen, Germany) with a dedicated 15-channel knee coil. The protocol consisted of routine clinical sequences followed by the CE T1-weighted sequence and the FLAIR FS sequence. Routine knee MR imaging included turbo spin-echo PD-weighted images Flowchart of patient selection. T1WI = axial FS T1-weighted imaging. in three planes (axial, FS sagittal, and coronal planes), turbo spin-echo T1- weighted imaging in the sagittal plane, and turbo spin-echo T2-weighted imaging in three planes (sagittal, FS coronal, and oblique coronal planes). After completion of routine MR examination, FLAIR FS imaging was performed. Before a 90 excitation radiofrequency pulse was applied, an 180 inversion pulse was applied for FLAIR imaging, and then a saturation pulse was applied for chemical shift selective fat suppression. The imaging parameters of the FLAIR FS sequence were as follows: axial acquisition; repetition time (msec)/echo time (msec), 9000/85 (n = 32) or 900/100 (n = 1); echo train length, 23; inversion time, 2200 msec; flip angle, 120 ; bandwidth, 220 Hz per pixel; matrix, ; field of view, 16 cm; section thickness, 3 mm; intersection gap, 0.15 mm; number of signals acquired, one; number of sections, 32; and acquisition time, 4 minutes 50 seconds. The inversion time is the time interval from the first radiofrequency 180 inversion pulse, preceding the acquisition of the turbo spin-echo image, to the 90 excitation pulse. Because an inversion time of 2000 msec was routinely applied for brain imaging at our institution, we tested several inversion times ( msec) to obtain images while maximizing the magnetization of the synovial tissues and finally chose an inversion time of 2200 msec. Axial CE T1-weighted imaging was then performed about 4 minutes after administration of intravenous gadolinium-based contrast agent (gadoterate meglumine, Dotarem; Guerbet, Aulnay-sous-Bois, France) at a dose of 0.2 ml (0.1 mmol per kilogram of body weight). The imaging parameters for CE T1-weighted imaging were /13; matrix, ; field of view, 16 cm; section thickness, 3 mm; intersection gap, 0.15 mm; number of signals acquired, two; number of sections, 32; and acquisition time, 3 minutes 50 seconds. Chemical shift selective fat suppression was also used for the FS technique. We selected a weak mode of fat suppression for both FLAIR FS and CE T1-weighted imaging sequences. MR Image Interpretation All MR images were analyzed independently by two musculoskeletal radiologists (H.J.Y. and H.Y.O., with 5 years or 1 year of experience in musculoskeletal MR image interpretation, respectively). Visibility of the synovium on FLAIR FS and CE T1-weighted images. The hyperintense line around the periphery of the suppressed joint fluid was considered to indicate the synovial lining on FLAIR FS images. The visibility of the synovium on FLAIR FS and CE T1- weighted images was graded by using a four-point scale based on tissue contrast between the synovium and the surrounding structures (eg, intra-articular fluid and pericapsular soft tissues): 1, synovium not visible; 2, synovium only partially visible; 3, synovium fully visible with low tissue contrast; and 4, good depiction of synovium with good tissue contrast. Assessment of synovial thickening on FLAIR FS and CE T1-weighted images. The synovium was assessed in five peripatellar regions on two axial sections. These regions included three Radiology: Volume 283: Number 3 June 2017 n radiology.rsna.org 771

4 Figure 2 Figure 2: Axial FS T1-weighted MR images obtained after intravenous gadolinium-based contrast agent injection demonstrated synovial enhancement. The synovium was assessed in five peripatellar regions. (a) The medial (SPm), central (SPc), and lateral (SPl) parts of the suprapatellar recesses are seen immediately above the patella. (b) The medial (FGm) and lateral (FGl) femoral gutters are just above or lateral to the upper part of the trochlear groove. suprapatellar regions (medial, central, and lateral parts of the suprapatellar recesses immediately above the patella) and one each in the medial and lateral femoral gutters just above or lateral to the upper part of the trochlear groove (Fig 2) (4,9). Synovial thickness was measured in each of the five regions of the knee and scored semiquantitatively by using a whole-joint semiquantitative scoring system for the assessment of synovitis, based on the maximum thickness at each site, as follows: grade 0, less than 2 mm; grade 1, 2 4 mm; and grade 2, more than 4 mm (6). On CE MR images, synovitis was defined as enhanced thickening of the synovium (.2 mm) (4,6,9). To assess peripatellar synovitis, the synovitis scores of the five sites were summed for each patient, giving a summed synovitis score from 0 to 10. After completing the independent image analysis session, the synovial thickness values from the two readers were averaged to compare FLAIR FS images with CE T1-weighted images. Assessment of Hoffa synovitis on unenhanced MR images. On unenhanced sagittal FS PD-weighted images, areas of high intensity in the infrapatellar region of the HFP were assessed and semiquantitatively scored from 0 to 3 as a synovitis surrogate, as described previously (9,14,15,18). The extent of signal intensity alteration in the infrapatellar region of the HFP was evaluated subjectively and scored as follows: 0, normal; 1, mild; 2, moderate; and 3, severe. Hoffa synovitis on unenhanced sagittal FS PD-weighted images was defined as any signal intensity alteration detected in the area of interest of the HFP (grade 1) (9,17). Statistical Analysis Sample size was calculated to detect a 0.9 correlation coefficient in synovial thickness between the two imaging sequences with a power of 0.8 and an a type I error of.05. A minimum sample size of seven was required. For statistical analysis, we treated synovial thickness as a parametric measure and the rest of measures (visibility grade of the synovium, synovitis grade in each joint site, summed synovitis score, and grade of Hoffa synovitis) as nonparametric measures. Interobserver agreement was assessed by using weighted k statistics with linear weights (for nonparametric measures) or intraclass correlation coefficients (for parametric measures). Intraclass correlation coefficient or weighted k coefficient of 0 indicated poor agreement; , slight agreement; , fair agreement; , moderate agreement; , good agreement; and , excellent agreement. Agreement regarding the synovitis grade for each joint site between FLAIR FS and CE T1-weighted imaging was evaluated by using the weighted k value. Differences in data (visibility grade, synovial thickness, and summed synovitis score) between the two MR imaging sequences were assessed by using the Wilcoxon signed rank test or the paired t test. Pearson correlation coefficients (r values), Spearman correlation coefficients (r values), and Bland-Altman analyses were used to determine the agreement of CE T1- weighted imaging and FLAIR FS imaging for the assessment of synovial thickness and summed synovitis score. An r value lower than 0.3 indicated little or no association; , moderate association; and higher than 0.7, strong association. To test the diagnostic performance, we calculated the sensitivity, specificity, and accuracy of peripatellar synovitis detected on FLAIR FS images and unenhanced FS PD-weighted images by using CE T1-weighted images as a reference standard. 772 radiology.rsna.org n Radiology: Volume 283: Number 3 June 2017

5 Table 1 Comparison of FLAIR FS and CE T1-weighted Images for the Assessment of Synovitis Parameter and Reader CE T1-weighted Imaging FLAIR FS Imaging P Value Visibility grade (grade 1 4) Reader Reader Synovial thickness (mm) Reader Reader Summed synovitis score (score of 0 10) Reader Reader Note. Data are means 6 standard deviations, unless indicated otherwise. Visibility of the synovium was assessed on a scale from 1 (not visible) to 4 (good depiction of synovium). The statistical analyses were performed by using the statistical packages SPSS version 21 (SPSS, Chicago, Ill) and MedCalc version 15.8 (MedCalc Software, Ostend, Belgium). For all studies, a difference with a P value less than.05 was considered to indicate a significant difference. Results Visibility of the Synovium on CE T1- weighted Images and FLAIR FS MR Images The synovium was seen readily on images obtained with both sequences, with a mean rating of 3.67 and 3.33 for CE T1-weighted images and FLAIR FS images, respectively (Table 1). Even in patients without apparent synovitis (n = 17), the synovium was well visualized as thin, hyperintense lines on FLAIR FS images. The interobserver agreement of the grading of the visibility of the synovium on CE T1-weighted images and FLAIR FS images was excellent, with weighted k values of 0.88 (95% confidence interval [CI]: 0.72, 1.00) and 0.81 (95% CI: 0.67, 0.96), respectively (Table 2). Interobserver Reliability in the Assessment of Synovitis There was excellent agreement between the two readers in the measurement of synovial thickness (intraclass correlation coefficient = 0.96 [95% CI: 0.94, 0.97] for CE T1-weighted imaging; intraclass correlation coefficient = 0.95 [95% CI: 0.93, 0.96] for FLAIR FS imaging) and the assessment of whole peripatellar synovitis (weighted k = 0.95 [95% CI: 0.90, 0.98] for both MR imaging sequences). There was also good agreement between readers in the grading of synovitis at each of the five joint sites, with a weighted k value of 0.79 (95% CI: 0.69, 0.89) for CE T1- weighted images and 0.72 (95% CI: 0.60, 0.83) for FLAIR FS images. Moderate agreement of the two readers was observed with a weighted k value of 0.66 (95% CI: 0.47, 0.85) for the grading of Hoffa synovitis on unenhanced FS PD-weighted images (Table 2). Comparison of FLAIR FS and CE T1- weighted Images in the Assessment of Synovitis Table 1 shows the agreement of synovitis assessments by using FLAIR FS versus CE T1-weighted imaging. There were excellent linear correlations for the synovial thicknesses (Pearson correlation coefficient, for reader 1 and for reader 2; each P,.001) and summed synovitis scores (Spearman rank correlation coefficient, for reader 1 and for reader 2; each P,.001) between CE T1-weighted imaging and FLAIR FS imaging (Figs E1 and E2 [online]). There was no significant difference in synovial thickness (P =.572 for reader 1 and P =.226 for reader 2) or summed synovitis score (P =.166 for reader 1 and P =.782 Table 2 Interobserver Agreement in the Assessment of Synovitis Parameter Intraclass Coefficient or k Value Visibility CE T1-weighted 0.88 (0.72, 1.00) imaging FLAIR FS imaging 0.81 (0.67, 0.96) Synovial thickness CE T1-weighted 0.96 (0.94, 0.97) imaging FLAIR FS imaging 0.95 (0.93, 0.96) Synovitis grade CE T1-weighted 0.79 (0.69, 0.89) imaging FLAIR FS imaging 0.72 (0.60, 0.83) Summed synovitis score CE T1-weighted 0.95 (0.90, 0.98) imaging FLAIR FS imaging 0.95 (0.90, 0.98) Grading of Hoffa synovitis Unenhanced PD (0.47, 0.85) weighted imaging Note. Numbers in parentheses are 95% CIs. Visibility of synovium was assessed by two readers on a scale from 1 (not visible) to 4 (good depiction of synovium). for reader 2) between the two MR imaging sequences. Bland-Altman plots of synovial thickness on CE T1-weighted images and FLAIR FS images showed a mean bias of 0.02 for reader 1, with narrow limits of agreement ranging from 20.71% to 0.74%, and a mean bias of for reader 2, with narrow limits of agreement ranging from 20.78% to 0.71% (Fig E1 [online]). A mean bias with the lower and upper limits of agreement in the corresponding Bland-Altman analysis for the summed synovitis score by using CE T1-weighted imaging and FLAIR FS imaging were 0.15 (range, 21.06% to 1.36%) for reader 1 and 0.03 (range, 21.22% to 1.28%) for reader 2 (Fig E2 [online]). Excellent agreement between CE T1-weighted images and FLAIR FS images was observed for grading synovitis at each joint site, with a weighted k value of (95% CI: 0.776, 0.943) for reader 1 and (95% CI: 0.756, 0.919) for reader 2 (Table 3). Radiology: Volume 283: Number 3 June 2017 n radiology.rsna.org 773

6 There are three notable differences in the imaging characteristics of FLAIR FS and CE MR imaging. First, the most superficial layer of articular cartilage was hyperintense on FLAIR FS images, and it consequently appeared as if the synovial lining was continuous over the articular cartilage (Fig 3). Second, all synovial tissues were demonstrated to have hyperintensity on FLAIR FS images, even in the nonenhancing synovial tissues of CE MR images (Fig 4). Third, we often observed the thick enhancing synovium on CE T1-weighted images as two layers on FLAIR FS images (Fig 3). Diagnostic Performance of FLAIR FS Images Abnormal synovial thickening (grade 1) in at least one peripatellar region on FLAIR FS images was found in 14 patients by reader 1 and in 17 patients by reader 2. By using CE T1-weighted imaging as the reference standard, FLAIR FS imaging showed good diagnostic performance for the detection of synovitis of any severity at both site and patient levels. The accuracy of FLAIR FS for readers 1 and 2 was (a) 93.9% (155 of 165 patients) and 92.1% (152 of 165 patients), respectively, at the site level and (b) 90.9% (30 of 33 patients) for both readers at the patient level. The sensitivity, specificity, false-positive rate, and false-negative rate are summarized in Table 4. Diagnostic Performance for Hoffa Synovitis on Unenhanced MR Images Signal intensity alterations (grade 1) in the HFP on unenhanced FS PDweighted images were detected in 14 patients by reader 1 and in 15 patients by reader 2. The diagnostic performance for Hoffa synovitis showed moderate sensitivity (87.5% [14 of 16 patients] for reader 1 and 78.9% [15 of 19 patients] for reader 2) but low specificity (47.0% [eight of 17 patients] for reader 1 and 64.3% [nine of 14] for reader 2) for the detection of synovitis, where abnormal synovial thickening (grade 1) in at least one peripatellar region on CE T1-weighted images was used as a reference standard (Table 4). Table 3 Comparison between the Two MR Sequences Studied in the Grading of Synovitis in Each Joint Site CE T1-weighted Imaging Grade and Reader FLAIR FS Grade 0 FLAIR FS Grade 1 FLAIR FS Grade 2 Total Reader 1 Grade Grade Grade Total Reader 2 Grade Grade Grade Total Note. Data are numbers of joint sites. Synovitis was scored semiquantitatively by using a whole-joint semiquantitative scoring system as follows: grade 0 if,2 mm, grade 1 if 2 4 mm, and grade 2 if.4 mm, based on the maximal thickness of the synovium. Figure 3 Figure 3: Axial CE knee MR images obtained in a 56-year-old woman with osteoarthritis. (a) FS T1-weighted image obtained after intravenous gadolinium-based contrast material injection demonstrates enhancing synovium with excellent tissue contrast. Note the thick enhancing synovium in the medial and lateral femoral gutters (arrowheads). (b) FLAIR FS image shows the synovium separated into two layers (arrowheads). In addition, the most superficial layer of articular cartilage of the patella is hyperintense on FLAIR FS images (arrows) and consequently appears as if the synovial lining is continuous over the articular cartilage of the patella. Discussion FLAIR FS MR imaging can be used to suppress signals from both fluid and fat, improving the conspicuity of the synovium in the intra-articular joint space. To show the ability of FLAIR FS acquisition to image the synovium, we compared FLAIR FS MR imaging with CE MR imaging in the assessment of synovitis. FLAIR FS MR imaging substantially improved the conspicuity of the synovium in the peripheral area with joint effusion, which was comparable with CE T1-weighted imaging. The synovial thickness and semiquantitative synovitis score assessed on FLAIR FS MR images correlated well with those on CE T1-weighted images, with excellent agreement. In contrast, signal intensity 774 radiology.rsna.org n Radiology: Volume 283: Number 3 June 2017

7 Figure 4 Figure 4: Axial CE knee MR images in a 22-year-old man with a history of lipoma arborescens excision in the left knee. (a) FS PD-weighted image shows a large joint effusion with prominent synovial proliferation in the lateral part of the suprapatellar recess (arrows). (b) FS T1-weighted image obtained after intravenous gadolinium-based contrast material injection demonstrates the enhancing synovium with excellent tissue contrast (arrow), while the nonenhancing synovial tissue regarded as chronic synovial inflammation with fibrosis in the lateral part of the suprapatellar recess was not seen. (c) FLAIR FS image also demonstrates the synovium, showing both the synovium (black arrow) and chronic synovial hypertrophy (white arrows). alterations in the HFP on unenhanced MR images, as a surrogate of synovitis, showed moderate interobserver reliability, relative high sensitivity, and low specificity. Our results showed the potential of FLAIR FS MR imaging as a new synovial imaging method to detect the inflamed synovium without the injection of a contrast agent. There have been many attempts to develop and validate a noninvasive method of assessing synovitis reliably (4,5,10,14,16,18). Pelletier et al used combined T2-weighted (synovial fluid) and gradient-echo (synovial membrane) MR imaging to assess synovial thickness in the knees of patients with osteoarthritis (19). They suggested that their new MR imaging method was accurate and reproducible. However, they did not directly compare their new imaging method with another reference standard, such as CE MR imaging or histologic findings, so validation of the new imaging method was incomplete. An effusion or signal intensity alteration in the HFP has also been used as a surrogate to identify synovial inflammation on unenhanced MR images. Effusions are usually identified on unenhanced T2-weighted or PD weighted FS images (14,15,18). However, Roemer et al showed that fluid-sensitive MR imaging sequences could not properly allow joint effusion to be distinguished from synovial thickening, and therefore, they typically overreported the presence of effusion (10). Loeuille et al also demonstrated that T2-weighted sequences were adequate for assessing effusion volume, while only CE MR imaging could be used to detect synovitis on the basis of a synovial biopsy reference (4). Regarding the signal intensity changes in the HFP detected on unenhanced MR images, several studies showed high sensitivity (71% 97%) but low specificity (10% 55%) (9,14). They assumed that signal intensity changes seen in the HFP do not always reflect peripatellar synovitis, and such changes may reflect the late sequelae of active synovial inflammation (9,20). In addition, no significant associations were found Radiology: Volume 283: Number 3 June 2017 n radiology.rsna.org 775

8 Table 4 Diagnostic Performance of FLAIR FS Imaging and Signal Intensity Alteration in the HFP Parameter and Reader FLAIR FS Imaging Joint Site Level (n = 165) Patient Level (n = 33) between HFP changes on unenhanced MR images and pain in their study (9). Our study showed a similar result that signal intensity changes in the HFP were a sensitive but nonspecific surrogate for the assessment of synovitis. Recently, newly developed MR imaging technologies were introduced for imaging synovitis without the use of contrast agents. Oei et al suggested that diffusion-weighted imaging by using a modified three-dimensional double-echo steady-state sequence with a high diffusion gradient is a promising MR imaging technique for the diagnosis and grading of synovitis without contrast agents (21). Double-echo steadystate imaging showed good to excellent diagnostic performance for the detection of synovitis of any severity at both site and patient levels (21). However, their study was limited by the small number of subjects, as only nine patients were included. Jahng et al investigated the application of a double inversion-recovery sequence for imaging joint effusion regions in the knee (22). They suggested that image contrast Hoffa Synovitis on Unenhanced PD-weighted Images (n = 33) Sensitivity (%) Reader (33/41) [65.1, 91.2] 87.5 (14/16) [61.7, 98.4] 87.5 (14/16) [61.7, 98.4] Reader (40/48) [69.8, 92.5] 89.5 (17/19) [66.9, 98.7] 78.9 (15/19) [54.4, 93.5] Specificity (%) Reader (122/124) [94.3, 99.8] 94.1 (16/17) [71.3, 99.8] 47.0 (8/17) [23.0, 72.2] Reader (112/117) [90.3, 98.6] 92.8 (13/14) [66.1, 99.8] 64.3 (9/14) [35.1, 87.2] Accuracy (%) Reader (155/165) [88.2, 97.3] 90.9 (30/33) [76.4, 98.2] 66.7 (22/33) [49.2, 83.3] Reader (152/165) [86.5, 96.8] 90.9 (30/33) [76.4, 98.2] 72.7 (24/33) [52.9, 86.4] False-positive rate (%) Reader (2/124) 5.9 (1/17) 52.9 (9/17) Reader (5/117) 7.1 (1/14) 35.7 (5/14) False-negative rate (%) Reader (8/41) 12.5 (2/16) 12.5 (2/16) Reader (8/48) 10.5 (2/19) 21.0 (4/19) Note. Numbers in parentheses are the data used to calculate percentages. Numbers in brackets are 95% CIs. Visibility of synovium was assessed on a four-point scale from 1 (not visible) to 4 (good depiction of synovium). obtained noninvasively by using the double inversion-recovery sequence was similar to that of a contrast-enhanced T1-weighted sequence for assessing knee synovitis (22). However, the study was performed to optimize the double inversion-recovery sequence and needs to be validated in a largescale clinical study. In our study, we investigated the feasibility of unenhanced synovial imaging by combining FLAIR and spectral fat-saturation techniques. Each method is already used widely in routine clinical MR imaging, and the new synovial images are easily created by combining the two MR imaging techniques. In several studies, investigators have reported that FLAIR FS sequences are a reliable method for demonstrating brain tumors or optic neuritis (23 26). However, FLAIR FS sequences applied to the musculoskeletal system have not been reported. Although the low signal-to-noise ratio of the images is a disadvantage of both inversion-recovery and spectral FS sequences, we demonstrated the depiction of the synovium within an acceptable range as compared with CE T1-weighted imaging. Jackson et al suggested that one reason for the superior performance of the FLAIR FS sequence is that the absence of fluid signal intensity on FLAIR FS images increases the dynamic range for the remaining tissues on the images and allows routine use of narrower display windows (23). In our study, the FLAIR FS sequence showed not only high sensitivity but also high specificity in the assessment of synovitis. The FLAIR FS sequence showed several different imaging characteristics in comparison with CE MR imaging. First, the most superficial layer of articular cartilage seems like a continuation with synovial lining. This feature of FLAIR FS imaging might be confusing in the assessment of synovitis, but in practice, it is not difficult to discriminate the cartilage surface from the synovial membrane in a clinical setting. Second, FLAIR FS imaging showed nonenhancing synovial tissues, which are regarded as chronic synovial inflammation with fibrosis (27 30). This property of FLAIR FS imaging can be either advantageous or unfavorable in the evaluation of synovial disease when compared with CE MR imaging. By showing chronic synovial hypertrophy, FLAIR FS imaging depicts the overall extent of the synovial abnormality. Alternatively, it may be difficult to differentiate between active synovitis and synovial fibrosis on FLAIR FS images. Third, we often observed the thick enhancing synovium on CE T1-weighted images as two layers on FLAIR FS images. When considering reports that CE T1-weighted imaging can lead to overestimation of synovitis owing to the diffusion of contrast material into the joint cavity, FLAIR FS imaging might be applied in clinical synovial imaging to show the actual width of the synovium accurately (10,22). There are several limitations to our study. First, the relatively small number of patients was one of the major limitations. Second, all analyses at the site level in our study should have accounted for clustering 776 radiology.rsna.org n Radiology: Volume 283: Number 3 June 2017

9 of within-patient data because there were five joint sites to be assessed in each patient. Third, the whole-joint semiquantitative scoring system of CE images should be performed for nine joint sites on both sagittal and axial images (6). However, we assessed synovitis in only five peripatellar sites on axial images because FLAIR FS images were obtained in only the axial plane. Our results should be applied in the assessment of synovitis in peripatellar regions, not posterior joint sites. Fourth, patients with nodular synovitis, such as those with rheumatoid arthritis or pigmented villonodular synovitis, were not included in our study. A patient with lipoma arborescens was the only one who showed irregular and frondlike synovial proliferation. Fifth, our study included patients without apparent synovitis, whose synovium was very thin. In those patients, the measurement of the synovial thickness may not be accurate by considering the image resolution and potential partial volume effect. Therefore, a high-spatial-resolution three-dimensional MR imaging sequence could be considered for accurate measurement of synovium in future studies. However, this should have little effect on our results, because it was problematic for both FLAIR FS imaging and CE T1-weighted imaging. Sixth, FLAIR FS MR imaging has several limitations, including the incomplete suppression of fat or water, a low signal-to-noise ratio, and a relatively long imaging time. Although the use of long echo times could enhance the effects of T2 and shorten the imaging time, it results in blurring and loss of image contrast. These make FLAIR FS images look blurred when compared with CE MR images. Seventh, a study in which investigators try to evaluate the diagnostic accuracy of an MR imaging sequence should exclude all surgically treated patients to avoid bias, but we included some follow-up MR imaging acquisitions after tumor excision if the image quality was good enough to assess the synovitis, since we had a small sample size. Two patients with metallic implants were excluded owing to severe artifact, but 17 patients were included on the basis of the consensus of two experts in musculoskeletal MR image interpretation. Finally, there was no histologic correlation in our study. Although previous studies of synovitis demonstrated that CE MR imaging is an accurate tool for the assessment of synovitis, there is a possibility that CE MR imaging is inaccurate when compared with histologically diagnosed synovitis because delayed image acquisition could cause more diffusion of contrast material into the synovial fluid (5,31). In summary, our preliminary study shows that FLAIR FS imaging can potentially enable evaluation of inflamed synovium with relative high sensitivity and specificity without the injection of a contrast agent. These results indicate that FLAIR FS imaging may be a feasible method for assessing knee synovitis, with lower costs and risks when compared with CE sequences. Disclosures of Conflicts of Interest: H.J.Y. disclosed no relevant relationships. S.H.H. disclosed no relevant relationships. H.Y.O. disclosed no relevant relationships. J.Y.C. disclosed no relevant relationships. H.D.C. disclosed no relevant relationships. J.M.A. disclosed no relevant relationships. H.S.K. disclosed no relevant relationships. References 1. Braun HJ, Dragoo JL, Hargreaves BA, Levenston ME, Gold GE. Application of advanced magnetic resonance imaging techniques in evaluation of the lower extremity. Radiol Clin North Am 2013;51(3): de Lange-Brokaar BJ, Ioan-Facsinay A, van Osch GJ, et al. Synovial inflammation, immune cells and their cytokines in osteoarthritis: a review. Osteoarthritis Cartilage 2012;20(12): Attur M, Samuels J, Krasnokutsky S, Abramson SB. Targeting the synovial tissue for treating osteoarthritis (OA): where is the evidence? Best Pract Res Clin Rheumatol 2010;24(1): Loeuille D, Sauliere N, Champigneulle J, Rat AC, Blum A, Chary-Valckenaere I. Comparing non-enhanced and enhanced sequences in the assessment of effusion and synovitis in knee OA: associations with clinical, macroscopic and microscopic features. Osteoarthritis Cartilage 2011;19(12): Loeuille D, Rat AC, Goebel JC, et al. Magnetic resonance imaging in osteoarthritis: which method best reflects synovial membrane inflammation? Correlations with clinical, macroscopic and microscopic features. Osteoarthritis Cartilage 2009;17(9): Guermazi A, Roemer FW, Hayashi D, et al. Assessment of synovitis with contrast-enhanced MRI using a whole-joint semiquantitative scoring system in people with, or at high risk of, knee osteoarthritis: the MOST study. Ann Rheum Dis 2011;70(5): Guermazi A, Hayashi D, Roemer FW, et al. Synovitis in knee osteoarthritis assessed by contrast-enhanced magnetic resonance imaging (MRI) is associated with radiographic tibiofemoral osteoarthritis and MRI-detected widespread cartilage damage: the MOST study. J Rheumatol 2014;41(3): de Lange-Brokaar BJ, Ioan-Facsinay A, Yusuf E, et al. Degree of synovitis on MRI by comprehensive whole knee semi-quantitative scoring method correlates with histologic and macroscopic features of synovial tissue inflammation in knee osteoarthritis. Osteoarthritis Cartilage 2014;22(10): Crema MD, Felson DT, Roemer FW, et al. Peripatellar synovitis: comparison between non contrast-enhanced and contrastenhanced MRI and association with pain. The MOST study. Osteoarthritis Cartilage 2013;21(3): Roemer FW, Kassim Javaid M, Guermazi A, et al. Anatomical distribution of synovitis in knee osteoarthritis and its association with joint effusion assessed on non-enhanced and contrast-enhanced MRI. Osteoarthritis Cartilage 2010;18(10): Baker K, Grainger A, Niu J, et al. Relation of synovitis to knee pain using contrast-enhanced MRIs. Ann Rheum Dis 2010;69(10): Song IH, Althoff CE, Hermann KG, et al. Knee osteoarthritis. Efficacy of a new method of contrast-enhanced musculoskeletal ultrasonography in detection of synovitis in patients with knee osteoarthritis in comparison with magnetic resonance imaging. Ann Rheum Dis 2008;67(1): Loeuille D, Chary-Valckenaere I, Champigneulle J, et al. Macroscopic and microscopic features of synovial membrane inflammation in the osteoarthritic knee: correlating magnetic resonance imaging findings with disease severity. Arthritis Rheum 2005;52 (11): Roemer FW, Guermazi A, Zhang Y, et al. Hoffa s fat pad: evaluation on unenhanced MR images as a measure of patellofemoral synovitis in osteoarthritis. AJR Am J Roentgenol 2009;192(6): Radiology: Volume 283: Number 3 June 2017 n radiology.rsna.org 777

10 15. Hunter DJ, Lo GH, Gale D, Grainger AJ, Guermazi A, Conaghan PG. The reliability of a new scoring system for knee osteoarthritis MRI and the validity of bone marrow lesion assessment: BLOKS (Boston Leeds Osteoarthritis Knee Score). Ann Rheum Dis 2008;67(2): Hill CL, Hunter DJ, Niu J, et al. Synovitis detected on magnetic resonance imaging and its relation to pain and cartilage loss in knee osteoarthritis. Ann Rheum Dis 2007;66(12): Hunter DJ, Guermazi A, Lo GH, et al. Evolution of semi-quantitative whole joint assessment of knee OA: MOAKS (MRI Osteoarthritis Knee Score). Osteoarthritis Cartilage 2011;19(8): Kornaat PR, Ceulemans RY, Kroon HM, et al. MRI assessment of knee osteoarthritis: Knee Osteoarthritis Scoring System (KOSS) inter-observer and intra-observer reproducibility of a compartment-based scoring system. Skeletal Radiol 2005;34(2): Pelletier JP, Raynauld JP, Abram F, Haraoui B, Choquette D, Martel-Pelletier J. A new non-invasive method to assess synovitis severity in relation to symptoms and cartilage volume loss in knee osteoarthritis patients using MRI. Osteoarthritis Cartilage 2008;16(Suppl 3):S8 S Saddik D, McNally EG, Richardson M. MRI of Hoffa s fat pad. Skeletal Radiol 2004; 33(8): Oei EH, McWalter EJ, Sveinsson B, Alley MT, Hargreaves BA, Gold GE. Non-contrast diffusion weighted imaging for the assessment of knee synovitis: a comparative study against contrast-enhanced MRI. Osteoarthritis Cartilage 2014;22(Suppl):S Jahng GH, Jin W, Yang DM, Ryu KN. Optimization of a double inversion recovery sequence for noninvasive synovium imaging of joint effusion in the knee. Med Phys 2011;38(5): Jackson A, Sheppard S, Johnson AC, Annesley D, Laitt RD, Kassner A. Combined fat- and water-suppressed MR imaging of orbital tumors. AJNR Am J Neuroradiol 1999;20(10): McKinney A, Palmer C, Short J, Lucato L, Truwit C. Utility of fat-suppressed FLAIR and subtraction imaging in detecting meningeal abnormalities. Neuroradiology 2006; 48(12): Lavdas E, Mavroidis P, Vassiou K, Roka V, Fezoulidis IV, Vlychou M. Elimination of chemical shift artifacts of thoracic spine with contrast-enhanced FLAIR imaging with fat suppression at 3.0 T. Magn Reson Imaging 2010;28(10): McKinney AM, Lohman BD, Sarikaya B, Benson M, Lee MS, Benson MT. Accuracy of routine fat-suppressed FLAIR and diffusion-weighted images in detecting clinically evident acute optic neuritis. Acta Radiol 2013;54(4): König H, Sieper J, Wolf KJ. Rheumatoid arthritis: evaluation of hypervascular and fibrous pannus with dynamic MR imaging enhanced with Gd-DTPA. Radiology 1990;176(2): Carotti M, Salaffi F, Manganelli P, Salera D, Simonetti B, Grassi W. Power Doppler sonography in the assessment of synovial tissue of the knee joint in rheumatoid arthritis: a preliminary experience. Ann Rheum Dis 2002;61(10): Klauser A, Frauscher F, Schirmer M, et al. The value of contrast-enhanced color Doppler ultrasound in the detection of vascularization of finger joints in patients with rheumatoid arthritis. Arthritis Rheum 2002; 46(3): Sommer OJ, Kladosek A, Weiler V, Czembirek H, Boeck M, Stiskal M. Rheumatoid arthritis: a practical guide to state-of-the-art imaging, image interpretation, and clinical implications. RadioGraphics 2005;25 (2): Østergaard M, Peterfy C, Conaghan P, et al. OMERACT rheumatoid arthritis magnetic resonance imaging studies. Core set of MRI acquisitions, joint pathology definitions, and the OMERACT RA-MRI scoring system. J Rheumatol 2003;30(6): radiology.rsna.org n Radiology: Volume 283: Number 3 June 2017