HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Zanetti, M. Articles by Hodler, J. Search for Related Content PubMed PubMed Citation Articles by Zanetti, M. Articles by Hodler, J.

Bone Marrow Edema Pattern in Osteoarthritic Knees: Correlation between MR Imaging and Histologic Findings¹

¹ From the Departments of Radiology (M.Z., J.H.) and Orthopedic Surgery (J.R.), Orthopedic University Hospital Balgrist, Forchstrasse 340, CH-8008 Zurich, Switzerland; and the Department of Pathology, University Hospital of Zurich, Switzerland (E.B.). From the 1998 RSNA scientific assembly. Received March 1, 1999; revision requested May 12; final revision received September 14; accepted October 4. Address correspondence to M.Z. (e-mail: mzanetti@balgrist.unizh.ch).

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To correlate magnetic resonance (MR) images of a bone marrow edema pattern with histologic findings in osteoarthritic knees.

MATERIALS AND METHODS: Sixteen consecutive patients (age range, 43–79 years; mean, 67 years) referred for total knee replacement were examined with sagittal short inversion time inversion-recovery (STIR) and T1- and T2-weighted turbo spin-echo MR imaging 1–4 days before surgery. Tibial plateau abnormalities on MR images were compared quantitatively with those on histologic maps.

RESULTS: The bone marrow edema pattern zone (ill-defined and hyperintense on STIR images and hypointense on T1-weighted MR images) mainly consisted of normal tissue (53% of the area was fatty marrow, 16% was intact trabeculae, and 2% was blood vessels) and a smaller proportion of several abnormalities (bone marrow necrosis [11% of area], abnormal [necrotic or remodeled] trabeculae [8%], bone marrow fibrosis [4%], bone marrow edema [4%], and bone marrow bleeding [2%]). The bone marrow edema pattern zone and the zone with a normal MR imaging appearance differed significantly in the presence of bone marrow necrosis (P = .021), bone marrow fibrosis (P = .014), and abnormal trabeculae (P = .011) but not in the prevalence of bone marrow edema (P = .069). Bone marrow edema also was found in zones with an unremarkable MR appearance (perifocal zone, 5% edema; control zone, 2% edema).

CONCLUSION: A bone marrow edema pattern in osteoarthritic knees represents a number of noncharacteristic histologic abnormalities. Edema is not a major constituent of MR imaging signal intensity abnormalities in such knees.

Index terms: Bone marrow, abnormalities, 452.833 • Bone marrow, MR, 452.121411, 452.121413, 452.121415 • Knee, abnormalities, 452.77 • Knee, MR, 452.121411, 452.121413

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Ill-defined hyperintensities seen on short inversion time inversion-recovery (STIR) images and on T2-weighted, fat-suppressed magnetic resonance (MR) images commonly are considered to represent bone marrow edema (1–8). There are two arguments in favor of this theory: Bone marrow changes commonly are reversible, such as those observed after jogging, or are due to altered biomechanics (9,10). Bone marrow edema also has been found at histologic examinations performed after core decompression in proximal femora with abnormal MR imaging findings (11,12).

However, in results of MR imaging–histologic correlation studies of such bone marrow edema patterns, fat cell destruction and fibrovascular regeneration also have been demonstrated (11,12). Histologic samples obtained in two patients with edemalike MR imaging findings in the tibia revealed focal marrow fibrosis and new bone formation, with foci of devitalized bone (5). Therefore, the expression "bone marrow edema" may be used too commonly in the radiology literature (1–8).

It is difficult to obtain additional histologic data on the bone marrow edema pattern because surgical treatment or biopsy usually is not performed in situations associated with this MR imaging appearance, such as bone bruise, fatigue fracture, or early avascular necrosis. One possibility for obtaining information in a larger series of patients, however, is the analysis of tibial plateaus removed during arthroplasty in knees with advanced osteoarthritis. MR imaging findings consistent with bone marrow edema are not uncommon in such knees.

The purpose of this investigation was to correlate MR images that showed a bone marrow edema pattern with histologic findings in osteoarthritic knees.

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Sixteen consecutive patients (15 women and one man; age range, 43–79 years; mean, 67 years) were included prospectively during a period of 6 months. All patients from the Orthopedic University Hospital Balgrist, Zurich, Switzerland, who were scheduled to undergo total knee replacement and who had advanced osteoarthritis were included. Two additional patients refused to participate in the study. The study was approved by the hospital's ethics committee. Written informed consent was obtained from each patient.

MR Imaging
MR imaging of the knee was performed 1–4 days before surgery. A 1.0-T system (Expert; Siemens Medical Systems, Erlangen, Germany) was used. A send-receive extremity coil was used. Three sagittal, standard MR imaging sequences were performed: STIR (4,800/30/150 [repetition time msec/effective echo time msec/inversion time msec]); T2-weighted, turbo spin-echo (3,800/96 [effective]); and T1-weighted, turbo spin-echo (850/12 [effective]) imaging. The echo train length was seven for STIR and T2-weighted imaging and three for T1-weighted imaging. Two signals were acquired with STIR and T2-weighted imaging; three, with T1-weighted imaging. The section thickness was 3.0 mm and the intersection gap was 0.9 mm for all imaging sequences. Imaging times were 6 minutes 11 seconds for STIR imaging, 4 minutes 57 seconds for T2-weighted imaging, and 3 minutes 48 seconds for T1-weighted imaging. The field of view was 140 x 160 mm, and the matrix was 256 x 224 for all sequences.

In three patients, the resected, formalin–fixed specimen obtained at surgery was imaged 12 hours after surgery to document that no substantial change in the MR imaging appearance of the tibial plateaus had occurred. MR imaging parameters for these specimens were identical to the parameters used preoperatively.

Two radiologists (J.H., M.Z.) experienced in musculoskeletal MR imaging reviewed the images in conference, with images obtained with each of the three sequences in each patient placed side by side. Regions of increased signal intensity in the subchondral bone were identified initially on the STIR images.

In each patient, the one MR imaging section with the most extensive abnormality was chosen for further evaluation. On this section, maps were created that noted signal intensity abnormalities as follows: (a) zone 1 had high signal intensity compared with normal fatty bone marrow signal intensity on STIR images, intermediate signal intensity on T2-weighted images, and ill-defined (no clear border between normal and abnormal signal intensity) low signal intensity on T1-weighted images (edemalike pattern) (Fig 1); (b) zone 2 had high signal intensity on STIR images, intermediate signal intensity on T2-weighted images, and well-defined (with a distinct border between two signal intensities) low signal intensity on T1-weighted images (Fig 2); (c) zone 3 had a rounded well-demarcated cystlike appearance, with pronounced hyperintensity on STIR and T2-weighted images (Fig 3) and with low signal intensity on T1-weighted images; (d) zone 4 had low signal intensity on images obtained with each of the three sequences that was surrounded by high signal intensity on STIR images (Fig 2); (e) zone 5 was in a perifocal area that was normal appearing (with normal MR imaging signal intensity adjacent to the abnormality [Fig 2d]); and (f) zone 6 had normal MR imaging signal intensity on a separate section as far as possible from the abnormality in the opposite tibial plateau. A 1:1 drawing (Fig 2d) noting zones 1–6 on the MR image was made on translucent paper.

View larger version (102K): [in this window] [in a new window] [Download PPT slide]   Figure 1a. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows zone 1, with edemalike pattern (arrowheads) in the tibial plateau. (b) The corresponding sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) shows that the edemalike zone 1 has intermediate signal intensity and cannot be differentiated from fatty marrow. (c) Sagittal, T1-weighted, turbo spin-echo (850/12 [effective]) MR image shows that the edemalike zone 1 (arrows) is ill defined and has intermediate to low signal intensity.

View larger version (85K): [in this window] [in a new window] [Download PPT slide]   Figure 1b. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows zone 1, with edemalike pattern (arrowheads) in the tibial plateau. (b) The corresponding sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) shows that the edemalike zone 1 has intermediate signal intensity and cannot be differentiated from fatty marrow. (c) Sagittal, T1-weighted, turbo spin-echo (850/12 [effective]) MR image shows that the edemalike zone 1 (arrows) is ill defined and has intermediate to low signal intensity.

View larger version (85K): [in this window] [in a new window] [Download PPT slide]   Figure 1c. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows zone 1, with edemalike pattern (arrowheads) in the tibial plateau. (b) The corresponding sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) shows that the edemalike zone 1 has intermediate signal intensity and cannot be differentiated from fatty marrow. (c) Sagittal, T1-weighted, turbo spin-echo (850/12 [effective]) MR image shows that the edemalike zone 1 (arrows) is ill defined and has intermediate to low signal intensity.

View larger version (85K): [in this window] [in a new window] [Download PPT slide]   Figure 2a. Zone 1 is edemalike with high signal intensity on the STIR image and is ill defined with low signal intensity on the T1-weighted image. Zone 2 has well-defined, low signal intensity on the T1-weighted image and high signal intensity on the STIR image. Zone 4 has low signal intensity with all three sequences. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows a high-signal-intensity abnormality. Differentiation between the well-defined zone 2 (arrowheads) and the ill-defined zone 1 (arrows) is better seen in c. Zone 4, with low signal intensity, is less conspicuous than in b. (b) Sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) demonstrates well the subchondral zone 4, with low signal intensity (arrowheads). The well-defined zone 2 is only partially visible (arrows). (c) Sagittal, T1-weighted, turbo spin-echo MR image (850/12 [effective]) shows the well-defined subchondral hypointense zone 2 (arrows). (d) Corresponding schematic drawing demonstrates zone 1, with an edemalike pattern; zone 2, with a well-defined high-signal-intensity abnormality; zone 4, with a low-signal-intensity abnormality surrounded by a high-signal-intensity abnormality; and perifocal zone 5, with a normal MR imaging appearance. The dotted and dashed line represents the osteotomy line. (e) Photomicrograph shows corresponding sagittal histologic slice. Only increased subchondral bone sclerosis (arrowheads) can be seen. Histologic details (bone marrow necrosis, bone marrow fibrosis, trabeculae abnormality, bone marrow edema) are not visible with this low-power view. (Hematoxylin-eosin stain, original magnification, x1.)

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure 2b. Zone 1 is edemalike with high signal intensity on the STIR image and is ill defined with low signal intensity on the T1-weighted image. Zone 2 has well-defined, low signal intensity on the T1-weighted image and high signal intensity on the STIR image. Zone 4 has low signal intensity with all three sequences. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows a high-signal-intensity abnormality. Differentiation between the well-defined zone 2 (arrowheads) and the ill-defined zone 1 (arrows) is better seen in c. Zone 4, with low signal intensity, is less conspicuous than in b. (b) Sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) demonstrates well the subchondral zone 4, with low signal intensity (arrowheads). The well-defined zone 2 is only partially visible (arrows). (c) Sagittal, T1-weighted, turbo spin-echo MR image (850/12 [effective]) shows the well-defined subchondral hypointense zone 2 (arrows). (d) Corresponding schematic drawing demonstrates zone 1, with an edemalike pattern; zone 2, with a well-defined high-signal-intensity abnormality; zone 4, with a low-signal-intensity abnormality surrounded by a high-signal-intensity abnormality; and perifocal zone 5, with a normal MR imaging appearance. The dotted and dashed line represents the osteotomy line. (e) Photomicrograph shows corresponding sagittal histologic slice. Only increased subchondral bone sclerosis (arrowheads) can be seen. Histologic details (bone marrow necrosis, bone marrow fibrosis, trabeculae abnormality, bone marrow edema) are not visible with this low-power view. (Hematoxylin-eosin stain, original magnification, x1.)

View larger version (84K): [in this window] [in a new window] [Download PPT slide]   Figure 2c. Zone 1 is edemalike with high signal intensity on the STIR image and is ill defined with low signal intensity on the T1-weighted image. Zone 2 has well-defined, low signal intensity on the T1-weighted image and high signal intensity on the STIR image. Zone 4 has low signal intensity with all three sequences. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows a high-signal-intensity abnormality. Differentiation between the well-defined zone 2 (arrowheads) and the ill-defined zone 1 (arrows) is better seen in c. Zone 4, with low signal intensity, is less conspicuous than in b. (b) Sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) demonstrates well the subchondral zone 4, with low signal intensity (arrowheads). The well-defined zone 2 is only partially visible (arrows). (c) Sagittal, T1-weighted, turbo spin-echo MR image (850/12 [effective]) shows the well-defined subchondral hypointense zone 2 (arrows). (d) Corresponding schematic drawing demonstrates zone 1, with an edemalike pattern; zone 2, with a well-defined high-signal-intensity abnormality; zone 4, with a low-signal-intensity abnormality surrounded by a high-signal-intensity abnormality; and perifocal zone 5, with a normal MR imaging appearance. The dotted and dashed line represents the osteotomy line. (e) Photomicrograph shows corresponding sagittal histologic slice. Only increased subchondral bone sclerosis (arrowheads) can be seen. Histologic details (bone marrow necrosis, bone marrow fibrosis, trabeculae abnormality, bone marrow edema) are not visible with this low-power view. (Hematoxylin-eosin stain, original magnification, x1.)

View larger version (20K): [in this window] [in a new window] [Download PPT slide]   Figure 2d. Zone 1 is edemalike with high signal intensity on the STIR image and is ill defined with low signal intensity on the T1-weighted image. Zone 2 has well-defined, low signal intensity on the T1-weighted image and high signal intensity on the STIR image. Zone 4 has low signal intensity with all three sequences. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows a high-signal-intensity abnormality. Differentiation between the well-defined zone 2 (arrowheads) and the ill-defined zone 1 (arrows) is better seen in c. Zone 4, with low signal intensity, is less conspicuous than in b. (b) Sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) demonstrates well the subchondral zone 4, with low signal intensity (arrowheads). The well-defined zone 2 is only partially visible (arrows). (c) Sagittal, T1-weighted, turbo spin-echo MR image (850/12 [effective]) shows the well-defined subchondral hypointense zone 2 (arrows). (d) Corresponding schematic drawing demonstrates zone 1, with an edemalike pattern; zone 2, with a well-defined high-signal-intensity abnormality; zone 4, with a low-signal-intensity abnormality surrounded by a high-signal-intensity abnormality; and perifocal zone 5, with a normal MR imaging appearance. The dotted and dashed line represents the osteotomy line. (e) Photomicrograph shows corresponding sagittal histologic slice. Only increased subchondral bone sclerosis (arrowheads) can be seen. Histologic details (bone marrow necrosis, bone marrow fibrosis, trabeculae abnormality, bone marrow edema) are not visible with this low-power view. (Hematoxylin-eosin stain, original magnification, x1.)

View larger version (63K): [in this window] [in a new window] [Download PPT slide]   Figure 2e. Zone 1 is edemalike with high signal intensity on the STIR image and is ill defined with low signal intensity on the T1-weighted image. Zone 2 has well-defined, low signal intensity on the T1-weighted image and high signal intensity on the STIR image. Zone 4 has low signal intensity with all three sequences. (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows a high-signal-intensity abnormality. Differentiation between the well-defined zone 2 (arrowheads) and the ill-defined zone 1 (arrows) is better seen in c. Zone 4, with low signal intensity, is less conspicuous than in b. (b) Sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) demonstrates well the subchondral zone 4, with low signal intensity (arrowheads). The well-defined zone 2 is only partially visible (arrows). (c) Sagittal, T1-weighted, turbo spin-echo MR image (850/12 [effective]) shows the well-defined subchondral hypointense zone 2 (arrows). (d) Corresponding schematic drawing demonstrates zone 1, with an edemalike pattern; zone 2, with a well-defined high-signal-intensity abnormality; zone 4, with a low-signal-intensity abnormality surrounded by a high-signal-intensity abnormality; and perifocal zone 5, with a normal MR imaging appearance. The dotted and dashed line represents the osteotomy line. (e) Photomicrograph shows corresponding sagittal histologic slice. Only increased subchondral bone sclerosis (arrowheads) can be seen. Histologic details (bone marrow necrosis, bone marrow fibrosis, trabeculae abnormality, bone marrow edema) are not visible with this low-power view. (Hematoxylin-eosin stain, original magnification, x1.)

View larger version (96K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Sagittal, T2-weighted, turbo spin-echo MR image (3,800/96 [effective]) shows rounded, well-demarcated, cystlike zone 3 (arrow), with high signal intensity.

After decalcification, the specimens were embedded in paraffin (Paraplast; Sherwood Medical, St Louis, Mo). Sagittal histologic slices were obtained every 4 mm with an Autocut 1140 microtome (Reichert-Jung, Vienna, Austria). Staining was performed with hematoxylin-eosin. The slice corresponding to the MR image chosen for evaluation was determined by measuring the distance from the medial to the lateral border of the tibial plateau. In addition, to guarantee that the histologic slice had the same orientation and position as the MR imaging section, the macroscopic contours of the histologic slice were compared with the contours on the MR image.

The slices then were evaluated by a staff pathologist (E.B.) who specializes in bone abnormalities. Histologic mapping was performed on the basis of the drawing provided by the radiologist (Fig 2d). Each zone was assessed by the pathologist for bone marrow and trabeculae abnormalities. Areas of similar abnormalities were grouped together. Their contributions to the total area were estimated semiquantitatively in mean percentage areas adding up to 100%, as described in the assessment of tumor necrosis in osteosarcomas after chemotherapy (13,14).

The pathologist was blinded to MR imaging findings in that she was not informed about the MR imaging appearance that related to each of the presented zones. Each zone marked on the drawing was evaluated by estimating the percentage areas for each of the following abnormalities: (a) bone marrow edema characterized at histologic examination by the accumulation of eosinophilic extracellular fluid, by swollen fat cells, and by the incipient disintegration of fat cells (Fig 4); (b) bone marrow necrosis defined by foam cell formation and by swollen fat cells with loss of nuclei (Fig 5); (c) bone marrow fibrosis characterized by the replacement of fatty marrow with collagenous fibers and with spindle-cell formations (Fig 6); (d) bone marrow bleeding; (e) cysts with central, basophilic, mucinous material; and (f) trabeculae abnormalities (necrosis characterized by increased sclerosis and loss of nuclei; new bone formation characterized by signs of repeated remodeling, with reversal lines [Fig 7]; and bone resorption, with increased osteoclastic activity [Fig 8]). In addition, normal bone constituents—(g) normal fatty marrow, (h) normal hematopoietic marrow, (i) normal blood vessels, and (k) intact trabeculae—were noted.

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Photomicrograph shows bone marrow edema with swollen fat cells () and accumulation of eosinophilic extracellular fluid (arrows). (Hematoxylin-eosin stain; original magnification, x220.)

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Photomicrograph shows bone marrow necrosis with foam cell formation (arrows) and swollen fat cells with focal loss of nuclei (area adjacent to arrowheads). (Hematoxylin-eosin stain; original magnification, x220.)

View larger version (122K): [in this window] [in a new window] [Download PPT slide]   Figure 6. Photomicrograph shows bone marrow fibrosis (F) with sparse collagen texture and occasional thin-walled blood vessels (arrow). (Hematoxylin-eosin stain; original magnification, x220.)

View larger version (120K): [in this window] [in a new window] [Download PPT slide]   Figure 7. Photomicrograph shows trabeculae abnormality with sclerosis (black arrows); with necrosis (white arrows); and with signs of repeated remodeling, with reversal lines (arrowheads). (Hematoxylin-eosin stain; original magnification, x220.)

View larger version (117K): [in this window] [in a new window] [Download PPT slide]   Figure 8. Photomicrograph shows trabeculae abnormality with increased osteoclastic activity and bone resorption (white arrows). Marrow fat cells are swollen (arrowheads). Eosinophilic extracellular fluid is also present (black arrows). (Hematoxylin-eosin stain; original magnification, x220.)

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Results are summarized in the Table. On MR images, the edemalike zone (zone 1) was found in 11 of 16 patients. This zone was composed mainly of normal tissue (53% fatty bone marrow, 16% normal trabeculae, and 2% blood vessels) and of smaller proportions of several histologic abnormalities. Bone marrow edema was included in only a small proportion (4% of the area). Bone marrow edema was also found in the well-defined (3% of the area), cystlike (1%), hypointense (3%), perifocal (5%), and control (2%) zones.

The normal-appearing, perifocal zone 5 could be evaluated on MR images in only 12 of 16 patients (no abnormality was seen in one patient; abnormal signal intensity that involved the entire section, in three patients). No control zone (zone 6) was found in two patients because the entire specimen was abnormal on MR images.

Significant differences between the zone with the edema pattern (zone 1) and the control zone (zone 6) were found for bone marrow fibrosis (P = .014), bone marrow necrosis (P = .021), and trabeculae abnormalities (P = .011) but not for bone marrow edema (P = .069), cysts (P = .343), or bone marrow bleeding (P = .696).

The abnormalities at MR imaging in the three formalin-fixed specimens correlated well with the preoperative MR image appearance (Fig 9).

View larger version (98K): [in this window] [in a new window] [Download PPT slide]   Figure 9a. Comparison of pre- and postoperative MR images with edemalike abnormalities (zone 1). (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows an edemalike zone (arrowheads) with high signal intensity. (b) Corresponding postoperative, sagittal STIR MR image (4,800/30 [effective]/150) of the formalin-fixed specimen demonstrates the same extent of signal intensity abnormalities (arrowheads).

View larger version (57K): [in this window] [in a new window] [Download PPT slide]   Figure 9b. Comparison of pre- and postoperative MR images with edemalike abnormalities (zone 1). (a) Sagittal STIR MR image (4,800/30 [effective]/150) shows an edemalike zone (arrowheads) with high signal intensity. (b) Corresponding postoperative, sagittal STIR MR image (4,800/30 [effective]/150) of the formalin-fixed specimen demonstrates the same extent of signal intensity abnormalities (arrowheads).

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

To our knowledge, there are only a few pathologic-radiologic correlation articles or case reports describing patients with ill-defined bone marrow abnormalities visible on MR images (5,11,12). Plenk et al (12) investigated specimens from core decompression in patients with hip pain and with MR imaging signal intensity abnormalities consistent with transient osteoporosis or with early avascular necrosis. They found interstitial edema (diffuse or spotty accumulations of a homogeneous, pale-to-dark–staining fluid) between bone marrow elements in the intertrabecular spaces that is comparable to our findings (Fig 4) (12). However, they also found a certain amount of fragmented necrotic fat cells. Moreover, the remnants of hematopoietic marrow included regions with necrosis, which led to the discussion of a continuum between bone marrow edema and avascular necrosis (12).

Ill-defined high-signal-intensity zones on STIR images are described with increasing frequency as bone marrow edema by radiologists (2,4,9,10,15) despite warnings that such wording may not be sufficiently precise or may even be misleading (4). We are aware that our histologic findings corresponding to ill-defined signal intensity abnormalities in knees with advanced osteoarthritis are not necessarily present in similar-appearing signal intensity abnormalities associated with other diseases. However, it should be acknowledged that use of the term "bone marrow edema" for other diseases such as transient osteoporosis rests also on few MR imaging-histologic finding correlations (5,8,11,12). Given this fact, although there are difficulties in extrapolating from our population to other populations with similar abnormalities, it is probably just as difficult to extrapolate from existing studies (5,8,11,12), given the very few MR imaging-histologic correlations in those studies.

Our results demonstrate that, at least in osteoarthritic knees, MR imaging changes that would be described as bone marrow edema by many radiologists are explained insufficiently. These hyperintense MR imaging abnormalities are rather an expression of a number of noncharacteristic histologic abnormalities that include bone marrow necrosis, bone marrow fibrosis, and trabeculae abnormalities. Bone marrow edema was not significantly more common in abnormal zones than in normal perifocal or control zones and was rare as an isolated finding. The indiscriminately used term "bone marrow edema" probably should be described only as "ill-defined signal intensity abnormality" or alternatively as "edemalike MR imaging abnormality."

However, a more precise diagnosis often can be made on the basis of associated MR imaging abnormalities. In osteoarthritic knees, these ill-defined signal intensity abnormalities represent noncharacteristic histologic abnormalities. However, with other characteristic MR imaging findings—cartilage loss, subchondral cysts, contour irregularities, and subchondral sclerosis—the diagnosis of osteoarthritis is evident. In the presence of a hypointense fracture line, the description probably should be "ill-defined signal intensity abnormality accompanying a well-demarcated hypointense line representing a [stress] fracture" (16). Similar expressions could be used for avascular necrosis, early infection, and peritumoral signal intensity abnormalities.

Bone marrow necrosis and trabecular abnormalities similar to those associated with avascular necrosis of bone appear to be relatively common in patients with osteoarthritis. This indicates that histologic signs of avascular necrosis should not be translated automatically into the clinical diagnosis of avascular necrosis. For this diagnosis, additional signs should be present, such as sudden onset of severe pain; predisposing factors such as female sex or advanced age; typical MR imaging appearance such as extensive and well-demarcated signal intensity abnormalities; and increasing collapse of the cortical bone during the follow-up time of a few months (17). Articles in which the continuum between diseases such as transient osteoporosis and avascular necrosis is discussed also should present the degree and extent of histologic signs of necrosis to allow comparison with other data, such as those presented for the osteoarthritic knees in our study.

We have found that fibrosis is a relatively important constituent (19% of area) of well-defined subchondral lesions despite high signal intensity on STIR images (Fig 2). The presence of bone marrow fibrosis in zone 2 in our study is also compatible with that in a previously published article on an MR imaging-histologic correlation study (18), in which fatty marrow was replaced with bone marrow fibrosis in well-defined subchondral zones of osteoarthritis. Trabecular necrosis and other abnormalities—small trabecular cysts and cartilage fragments—also were present in these regions (18). Therefore, it is not surprising that such zones may be hyperintense on images obtained with highly sensitive STIR imaging. The hyperintensity in our well-defined zones can be explained further by bone marrow necrosis and by a small proportion of bone marrow edema. Even fibrotic tissue can appear as high signal intensity on STIR images, depending on its maturity, as demonstrated previously in the musculoskeletal literature (5,19,20). Hyperemia, another possible reason for high-signal-intensity abnormalities on STIR images, was not found in our specimens.

As expected in specimens originating from the epiphysis of an extremity bone, no hematopoietic marrow was found in our specimens. The reconversion of fatty marrow into hematopoietic marrow was not responsible for hyperintensities on STIR images.

Although the method of using area percentages for the characterization of the various tissue components also is used for the precise description of bone tumors (13,14), there may have been an error in the estimation of percentage areas. However, we believe that such an error would not have changed the substance of our results. Moreover, by relying on means obtained from 16 specimens, we reduced the influence of outliers.

The influence of postoperative changes in the specimens was another feasible source of failure. This possible problem and the difficulty of guaranteeing that MR imaging sections correspond exactly with histologic slices are inherent drawbacks of every MR imaging-histologic correlation study. With demonstration that the MR images of the three additional specimens correlated well with the preoperative MR images, postoperative changes seemed not to be crucial. For example, if processing removed a large proportion of water that corresponded to edema from the specimens, one would expect to see dramatic differences in the postprocessing appearances of the specimens.

In conclusion, the bone marrow edema pattern in osteoarthritic knees represents histologically noncharacteristic abnormalities that include bone marrow necrosis, bone marrow fibrosis, and trabeculae abnormalities but represents only a small amount of bone marrow edema. Therefore, this pattern should not be attributed routinely to bone marrow edema.

	Footnotes

 
Abbreviation: STIR = short inversion time inversion recovery

Author contributions: Guarantor of integrity of entire study, M.Z.; study concepts, M.Z.; study design, M.Z., J.H.; definition of intellectual content, M.Z., J.H.; literature research, M.Z., E.B.; clinical studies, J.R., E.B.; data acquisition, M.Z., E.B.; data analysis, J.H., M.Z., E.B.; statistical analysis, M.Z.; manuscript preparation, M.Z.; manuscript editing, J.H.; manuscript review, J.R., E.B.

	References

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 

1. Yu JS, Cook PA. Magnetic resonance imaging (MRI) of the knee: a pattern approach for evaluating bone marrow edema. Crit Rev Diagn Imaging 1996; 37:261-303.[Medline]
2. Lecouvet FE, van de Berg BC, Maldague BE, et al. Early irreversible osteonecrosis versus transient lesions of the femoral condyles: prognostic value of subchondral bone and marrow changes on MR imaging. AJR Am J Roentgenol 1998; 170:71-77.[Abstract/Free Full Text]
3. MacDougall L, Conway WF. Controversies in magnetic resonance imaging of the hip. Top Magn Res Imaging 1996; 8:44-50.
4. Palmer WE, Levine SM, Dupuy DE. Knee and shoulder fractures: association of fracture detection and marrow edema on MR images with mechanism of injury. Radiology 1997; 204:395-401.[Abstract/Free Full Text]
5. Reinus WR, Fischer KC, Ritter JH. Painful transient tibial edema. Radiology 1994; 192:195-199.[Abstract/Free Full Text]
6. Vahlensieck M, Reiser M. Bone marrow edema in MRT. Radiologe 1992; 32:509-515.[Medline]
7. Vande Berg BE, Malghem JJ, Labaisse MA, Noel HM, Maldague BE. MR imaging of avascular necrosis and transient marrow edema of the femoral head. RadioGraphics 1993; 13:501-520.[Abstract]
8. Wilson AJ, Murphy WA, Hardy DC, Totty WG. Transient osteoporosis: transient bone marrow edema?. Radiology 1988; 167:757-760.[Abstract/Free Full Text]
9. Schweitzer ME, White LE. Does altered biomechanics cause marrow edema?. Radiology 1996; 198:851-853.[Abstract/Free Full Text]
10. Lazzarini KM, Troiano RN, Smith RC. Can running cause the appearance of marrow edema on MR images of the foot and ankle?. Radiology 1997; 202:540-542.[Abstract/Free Full Text]
11. Neuhold A, Hofmann S, Engel A, et al. Bone marrow edema of the hip: MR findings after core decompression. J Comput Assist Tomogr 1992; 16:951-955.[Medline]
12. Plenk H, Jr, Hofmann S, Eschberger J, et al. Histomorphology and bone morphometry of the bone marrow edema syndrome of the hip. Clin Orthop 1997; 73-84.
13. Glasser DB, Lane JM, Huvos AG, Marcove RC, Rosen G. Survival, prognosis, and therapeutic response in osteogenic sarcoma: the Memorial Hospital experience. Cancer 1992; 69:698-708.[Medline]
14. Huvos AG, Rosen G, Marcove RC. Primary osteogenic sarcoma: pathologic aspects in 20 patients after treatment with chemotherapy en bloc resection, and prosthetic bone replacement. Arch Pathol Lab Med 1977; 101:14-18.[Medline]
15. Hayes CW, Conway WF, Daniel WW. MR imaging of bone marrow edema pattern: transient osteoporosis, transient bone marrow edema syndrome, or osteonecrosis. RadioGraphics 1993; 13:1001-1011.[Abstract/Free Full Text]
16. Lee JK, Yao L. Stress fractures: MR imaging. Radiology 1988; 169:217-220.[Abstract/Free Full Text]
17. Lotke PA, Ecker ML. Osteonecrosis of the knee. J Bone Joint Surg Am 1988; 70:470-473.[Free Full Text]
18. Bergman AG, Willen HK, Lindstrand AL, Pettersson HT. Osteoarthritis of the knee: correlation of subchondral MR signal abnormalities with histopathologic and radiographic features. Skeletal Radiol 1994; 23:445-448.[Medline]
19. Quinn SF, Erickson SJ, Dee PM, et al. MR imaging in fibromatosis: results in 26 patients with pathologic correlation. AJR Am J Roentgenol 1991; 156:539-542.[Abstract/Free Full Text]
20. Morrison WB, Schweitzer ME, Wapner KL, Lackman RD. Plantar fibromatosis: a benign aggressive neoplasm with a characteristic appearance on MR images. Radiology 1994; 193:841-845.[Abstract/Free Full Text]

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Zanetti, M. Articles by Hodler, J. Search for Related Content PubMed PubMed Citation Articles by Zanetti, M. Articles by Hodler, J.