| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Musculoskeletal Imaging |
1 From the Department of Radiology, Showa University Fujigaoka Hospital, Yokohama, Kanagawa, Japan (H.S., A.F., Y.K., W.Y.); Department of Orthopedics, Showa University Fujigaoka Rehabilitation Hospital, Yokohama, Kanagawa, Japan (I.H., Y.I.); and Department of Science, Tokyo University of Science, Japan (E.M.). Received August 29, 2002; revision requested October 24; final revision received April 25, 2003; accepted May 19. Address correspondence to H.S., Department of Radiology, Jichi Medical School, 3311 Yakushiji, Minamikawachimachi, Kawachi-gun, Tochigi-ken 329-0431, Japan (e-mail: sugimoto@jichi.ac.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: The study group comprised 22 consecutive women who underwent hip arthroplasty (mean age, 62 years; age range, 35–74 years). All patients underwent MR imaging prior to revision surgery. Coronal fast short inversion time inversion-recovery (STIR) images and spin-echo T1-weighted images were obtained with a 0.5-T MR imaging unit before and after administration of contrast material. The periprosthetic region was divided into the seven femoral Gruen zones. Two observers retrospectively analyzed signal intensity patterns. Association of signal intensity patterns with radiographic, surgical, and pathologic findings was determined with 
2 analysis and generalized estimating equations.
RESULTS: Diagnostic-quality images were obtained for 150 zones. Periprosthetic signal intensity was greater than that of bone marrow in the distal femur on the fast STIR images, and no contrast enhancement was seen on the T1-weighted images (type I signal intensity pattern) in 11 zones. Signal intensity was greater than that of bone marrow on the fast STIR images, and contrast enhancement was seen on the T1-weighted images (type II signal intensity pattern) in 45 zones. Signal intensity was less than or equal to that of bone marrow on the fast STIR images, and no contrast enhancement was seen on the T1-weighted images (type III signal intensity pattern) in 94 zones. Type I and II patterns were associated with focal or nonfocal lucency, an unstable stem, and fibrosis or granuloma. A type III pattern was associated with a normal radiographic appearance, a stable stem, and normal bone tissue. Significant association was demonstrated between periprosthetic signal intensity and radiographic (P < .001, 2 test and generalized estimating equations), surgical (P < .05, Mantel-Haenszel 2 test and generalized estimating equations), and pathologic findings (P < .05, 2 test).
CONCLUSION: Low-field-strength MR imaging depicted periprosthetic tissue signal intensity that was significantly associated with radiographic, surgical, and pathologic findings.
© RSNA, 2003
Index terms: Hip, MR, 442.121411 • Hip, prostheses, 442.454 • Magnetic resonance (MR), low-field-strength imaging, 442.121411, 442.121413 • Hip, surgery, 442.458
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Many studies have investigated the usefulness of MR imaging in the determination of the status of metallic implants (7–12), and they have emphasized the value of postoperative MR imaging for conditions such as hip arthroplasty (5,13). To our knowledge, no study has correlated periprosthetic MR signal intensity with radiographic, surgical, or pathologic findings in patients who underwent hip arthroplasty, nor has a study fully investigated the potential of combining low-field-strength MR imaging and MR sequences with metallic artifact reduction to diagnose loosening of the femoral prosthesis. The purpose of the present study was to investigate the association between periprosthetic signal intensity at low-field-strength MR imaging after failed hip arthroplasty and radiographic, surgical, and pathologic findings.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Surgical revision was needed because of loosening of the femoral component (n = 7), osteolysis (n = 6), breakage of components (n = 3), infection (n = 3) or loosening (n = 2) of the acetabular component, or protrusion of the Charnley long stem into the knee joint (n = 1). Osteolysis was defined as an expansile osteolytic lesion around the prosthesis. Loosening of either a femoral or an acetabular component was determined on the basis of clinical findings, radiographic findings, or both. Infection was determined on the basis of serologic signs, clinical signs, or both. Revision surgery had already been performed one time in one patient and three times in another. No patient had an underlying condition such as rheumatoid arthritis that might have affected the periprosthetic signal intensity.
The need for revision surgery was determined on the basis of clinical, radiographic, arthrographic, bone and gallium scintigraphic, and joint aspiration findings. The average interval between arthroplasty and revision surgery was 16.9 years (range, 5.5–27.6 years). Patients’ records were retrospectively analyzed to find any association between MR findings and radiographic, surgical, and pathologic findings. Institutional review board approval and informed consent are not required at our hospital for review of patients’ records or images.
MR Imaging
All MR imaging examinations were performed with a 0.5-T imaging system (Contour; GE Medical Systems, Milwaukee, Wis) and a flexible quadrature coil wrapped around the pelvic girdle. The localizer images were obtained with fast multiplanar spoiled gradient recalled acquisition in the steady state. The MR imaging parameters were as follows: 80/5.2 (repetition time msec/echo time msec), 60° flip angle, 8-mm section thickness with a 2-mm section gap, 40 x 40-cm field of view, 7.81-kHz readout bandwidth, and 256 x 128 matrix. One signal was acquired, and acquisition time was 11 seconds. Oblique coronal imaging planes parallel to the femoral prosthesis were determined by means of oblique sagittal localizer images obtained along the femoral prosthesis. Fast short inversion time inversion-recovery (STIR) and spin-echo (SE) T1-weighted images were obtained for analysis.
The fast STIR imaging parameters were as follows: 3,000/40/100 (repetition time msec/echo time msec/inversion time msec), echo train length of six, 5-mm section thickness with a 0.5-mm section gap, 28 x 22-cm field of view, 32-kHz readout bandwidth, and 256 x 160 matrix. Four signals were acquired, and acquisition time was 4 minutes 12 seconds. The frequency-encoding axis was parallel to the long axis of the prosthesis.
The SE T1-weighted imaging parameters were as follows: 340/15, 5-mm section thickness with a 0.5-mm section gap, 28 x 22-cm field of view, 16-kHz readout bandwidth, and 256 x 224 matrix. Three signals were acquired, and acquisition time was 3 minutes 28 seconds. The frequency-encoding axis was parallel to the long axis of the prosthesis. T1-weighted images were obtained before and after intravenous administration of 0.1 mmol/kg gadopentatate dimeglumine (Magnevist; Schering, Berlin, Germany) via the antecubital vein. The total MR acquisition time was approximately 25 minutes.
MR Image Analysis
The femoral prosthesis was divided into seven zones (Fig 1) according to the Gruen system (14). The signal intensity of each zone was retrospectively evaluated at the cement-bone or stem-bone interface in cemented prostheses and at the stem-bone interface in cementless prostheses for an overall total of 154 zones. When there was expansile osteolysis suggestive of aggressive granulomatosis that showed mixed signal intensity, the signal intensity adjacent to the prosthesis was used for evaluation. When the distal end of the femoral stem was outside the imaging field owing to its length, the periprosthetic region was divided into six zones (zones 1–3 and 5–7) to accommodate the Gruen system.
|
Radiographic Analysis
Anteroposterior radiographs of the pelvis were obtained in all patients before MR imaging. The average interval between radiography and MR imaging was 31 days (range, 0–93 days). Radiographic findings were determined retrospectively for each Gruen zone with consensus of the same two observers. Approximately 2 months (53 days) passed between MR imaging and radiographic analysis. The presence of focal lucency, nonfocal lucency, or diffuse sclerosis with periostitis (15) was recorded for each zone. Focal lucency was defined as a discrete region of bone loss that occurred at the bone-prosthesis or bone-cement interface that did not conform to the shape of the prosthesis. Nonfocal lucency was defined as a linear region of bone loss wider than 2 mm that occurred at the bone-prosthesis or bone-cement interface that outlined and conformed to the shape of the entire prosthesis. Diffuse sclerosis with periostitis was noted when reactive bone formation was observed in association with new periosteal bone formation. When there were no such findings, the zone was considered radiographically normal. Furthermore, an orthopedic surgeon (I.H.) retrospectively assessed whether there were any patients in whom normal radiographic findings and abnormal MR signal intensity were associated with a loose prosthesis at surgery.
Surgical Analysis
The average interval between MR imaging and revision surgery was 28 days (range, 4–193 days). One orthopedic surgeon (I.H.) who was familiar with the MR findings performed all revision surgeries. At surgery, this surgeon classified the femoral prosthesis as either stable or unstable. When forced manipulation of the femoral component did not cause motion between the metallic implant and the cement or between the bone and the cement, the prosthesis was considered stable. Difficulty in extraction of the femoral prosthesis was noted by the orthopedist. Furthermore, whether a stable prosthesis was associated with a particular signal intensity pattern distribution was evaluated.
Pathologic Analysis
Pathologic specimens from the six patients who underwent removal of their femoral prosthesis were not sent to pathologists, because the orthopedic surgeons believed no clinically relevant information would be obtained. Pathologic specimens from 16 patients were examined for any foreign body or debris in the tissue around the prosthesis. In two of these 16 patients, the specimen was obtained only from the acetabular component. Thus, pathologic studies were performed in 14 patients. In two patients who did not undergo revision surgery of the femoral prosthesis, pathologic specimens were obtained from zones 1 and 7 when abnormal tissue in these zones was replaced with bone grafts. In the remaining 12 patients, the pathologic specimen was obtained during revision surgery for the femoral prosthesis. In eight of these 12 patients, the stem was easily extracted without destruction of the periprosthetic tissues. In these patients, the orthopedic surgeon obtained the specimen from the specific zones without difficulty. In the other four patients, the stem was difficult to extract. In two of these patients, a part of the femoral cortex was surgically sawed to extract the stem, and the specimen was removed from zones 1 and 7 and showed osteolysis. In the remaining two patients, the stem was loosened with a bone saw, and the pathologic specimen was removed from tissue that had adhered to the stem. In all instances, when the specimen was sent to the pathologist, the orthopedic surgeon could specify where it was obtained.
One of two senior pathologists who were blinded to the radiologic and surgical findings made each pathologic diagnosis. When abundant foreign bodies were associated with reactive proliferative changes, a pathologic diagnosis of foreign body granuloma was made. When there were fibrohistiocytic tissue and collagen fibers with relatively few foreign bodies, a pathologic diagnosis of fibrosis was made.
Statistical Analysis
2 tests for independence and generalized estimating equations (16) were used to determine whether MR signal intensity patterns were associated with radiographic and pathologic findings. The Mantel-Haenszel 2 test was used to examine differences in the proportion of surgical findings relative to MR signal intensity types. The 2 test was used to determine whether the particular signal intensity patterns in the proximal, middle, or distal two zones were associated with a stable prosthesis. P values of less than .05 indicated statistical significance.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MR Signal Intensity
The overall signal intensity patterns were of three types. A signal intensity greater than that of bone marrow on fast STIR images without contrast enhancement on SE T1-weighted images was considered a type I pattern (Fig 2). A signal intensity greater than that of bone marrow on fast STIR images with contrast enhancement on SE T1-weighted images was considered a type II pattern (Fig 3). A signal intensity equal to or less than that of bone marrow on fast STIR images but without contrast enhancement on SE T1-weighted images was considered a type III pattern (Fig 4). The type I, II, and III signal intensity patterns characterized 11, 45, and 94 zones, respectively.
|
|
|
|
|
|
|
|
|
|
2 test and generalized estimating equations (P < .001).
|
Surgical Association with Signal Intensity
The femoral component was stable in 14 patients. In five of these patients, the femoral prosthesis was rigidly anchored to the femur and was left without revision. The femoral component was unstable in the remaining eight patients. Five of these patients had a prosthesis that was cemented, and three had a prosthesis that was cementless. Relationships between the total number of particular zones in each patient and surgical findings are summarized in Table 2. For statistical analysis, zones with a type I or II pattern were combined, since both intensity patterns were unfavorable in terms of periprosthetic tissue. The Mantel-Haenszel 2 test revealed a significant difference in the proportion of stable surgical findings relative to the signal intensity patterns (P = .045).
|
Pathologic Association with Signal Intensity
Pathologic findings around the femoral component were obtained in 45 zones in 14 patients. The relationship between MR signal intensity and pathologic findings is summarized in Table 3. Foreign body granuloma was found in 11 zones, fibrosis (Fig 3) was found in 29, and apparent bone ingrowth was found in five. Significant association was found between MR signal intensity patterns and pathologic findings with the 2 test (P < .01); however, statistical significance was not reached with generalized estimating equations.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Another approach to obtaining diagnostic-quality MR images in patients with metallic implants is use of the low-field-strength MR imaging unit (7,9). With the use of a 0.02-T unit, healing of a femoral neck fracture fixed with titanium screws was successfully depicted without disturbing artifacts (9). Loosening in the distal part of the femoral prosthesis was diagnosed successfully by depicting demarcation lines at the interfaces on images obtained with a 0.5-T MR unit (7). Although these reports described the advantage of the low-field-strength MR imager for minimizing metallic artifacts, the images obtained were less than satisfactory, probably because of the weak read-out gradient of MR units in use at the time.
In the present study, because our fast STIR sequence protocol fulfilled the criteria proposed by Tormanen et al (10) for minimizing metallic artifacts, diagnostic-quality images were consistently obtained with both fast STIR and SE T1-weighted sequences, unless surgical wires made of stainless steel or a stainless steel tip were included in the imaging field. Even though we did not use multiple refocusing sequences for SE T1-weighted imaging, diagnostic-quality T1-weighted images were obtained for all patients. Thus, the low-field-strength MR unit may be the important factor in obtaining diagnostic-quality images in the presence of metallic implants. Sufficient strength of the readout gradient may contribute to image quality, as well.
Although there was significant association between MR signal intensity and radiographic findings, 25 zones with a normal radiographic appearance showed type I or II patterns that did not appear to correspond to normal bone. Conversely, eight zones with nonfocal or focal lucency on radiographs revealed type III patterns on MR images. These findings corresponded to the result of a previous investigation showing that stress shielding, aggressive granuloma, infection, and aseptic loosening may manifest radiographically as periprosthetic lucency (17) and that radiolucent periprosthetic zones might represent osteoporosis rather than the presence of a fibrous membrane (3).
There was a significant association between the total number of specific MR signal intensity patterns and stability of the femoral prosthesis. The prosthesis tended to be stable when the total number of zones with a type III pattern around the prosthesis was increased. In this regard, the distribution of the type III pattern may be more important than the simple sum of the zones with the type III pattern because the biomechanical contribution to the stability of the stem differs in each zone. When the type III pattern was simultaneously seen in zones 3 and 5, the prosthesis tended to be stable, regardless of the MR signal intensity pattern of the remaining zones. This observation corresponds to previous findings that localized osteolysis may occur with or without loosening of the femoral stem (18); therefore, it may be important to analyze the distribution of MR signal intensity patterns when MR images are used to assess the stability of the femoral prosthesis.
Regardless of design features of the prosthesis, nonfocal inflammatory bone resorption can occur in the femur after hip replacement with or without cement (2). Membranes of cemented implants contain many macrophages and giant cells, and there is evidence of frequent granuloma formation. Histiocytosis, fibrosis, and necrosis are seen within this membrane (19). Membranes around the cementless prosthesis usually consist of poorly vascularized, dense fibrous tissue (20). In our study, type I and II patterns were associated with either foreign body granuloma or fibrosis.
There were several study limitations. First, the study was not prospective, and the number of patients studied was relatively small. Second, since the same observers interpreted both MR images and radiographs, there could have been some bias. There was a relatively long interval between the interpretation of MR images and radiographs, however, which minimized the likelihood of recall bias. Third, because MR imaging was performed with the frequency-encoding read-out gradient oriented along the long axis of the femoral stem, susceptibility artifact would be present along the long axis of the femoral stem. Since the artifact was notable at the corner of the distal end of the prosthesis, however, the signal intensity patterns seemed to be reliably assessed in our patients. Fourth, although pathologic association was investigated, the number of patients in whom pathologic correlation was possible is limited. Further, specimens were small and there could have been other tissues in each Gruen zone; therefore, associations between MR signal intensity characteristics and pathologic findings were somewhat unreliable. Nonetheless, our results are encouraging. As of now, orthopedic surgeons in our hospital rarely use arthrography and scintigraphy, but they rely on MR imaging in the preoperative investigation of hip prosthesis failure. Although the present study was not designed to assess whether MR imaging adds useful information prior to revision surgery, orthopedic surgeons at our institution believe that information provided by MR imaging regarding ease or difficulty in the extraction of the femoral stem is helpful in planning revision surgery.
In conclusion, type I and II MR signal intensity characteristics are associated with radiographic focal or nonfocal lucency, an unstable prosthesis, and foreign body granuloma or fibrosis. Type III patterns are associated with radiographically normal bone, a stable prosthesis, and normal bone tissue. Low-field-strength MR imaging yields unique information regarding periprosthetic tissue and may be useful for preoperative assessment of stem loosening.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, H.S.; study concepts and design, H.S., I.H.; literature research, H.S.; clinical studies, H.S., I.H., A.F., Y.K., W.Y., Y.I.; experimental studies, I.H., Y.I.; data acquisition and analysis/interpretation, H.S., I.H.; statistical analysis, H.S., E.M.; manuscript preparation, H.S.; manuscript definition of intellectual content, H.S., I.H.; manuscript editing, H.S.; manuscript revision/review, H.S., I.H.; manuscript final version approval, H.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  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]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
