A Comparison Between Magnetic Resonance Imaging, Pathology, and Radiology in 34 Limbs with Navicular Syndrome and 25 Control Limbs

The pathology of navicular syndrome/palmar foot pain predominantly affects the structures that define the boundaries of the navicular bursa. Alterations in magnetic resonance signal intensity and tissue contour of these structures represent changes in tissue structure detectable using gross and histopathological examination. Magnetic resonance imaging (MRI) has a high sensitivity and specificity for abnormalities of the navicular bone, deep flexor tendon, collateral sesamoidean ligaments, and distal sesamoidean impar ligament. MRI had a poor to fair sensitivity for osteophytes and cartilage damage in the distal interphalangeal joint and for early fibrocartilage loss from the flexor surface of the navicular bone. Although there was a good correlation between MRI and the radiological navicular bone grading system, MRI was also able to show medullary and flexor border abnormalities of the navicular bone that were not visible on radiographs.

1. Introduction

Navicular syndrome/palmar foot pain has been defined as a clinical manifestation of a number of different disease processes in the foot [1]. It has become apparent that in addition to degeneration of the navicular bone, soft-tissue injuries are also an important cause of pain and lameness in the palmar aspect of the foot [2,3]. To understand the nature of the pathology responsible for palmar foot pain, the incidence of lameness must be correlated with the anatomy and pathology of these structures: deep digital flexor tendon (DDFT), collateral sesamoidean ligaments (CSL), distal sesamoidean impar ligament (DSIL), collateral ligaments (CL) of the distal interphalangeal joint (DIPJ), fibrocartilage of the navicular bone (NB), NB, synovial membrane of the DIPJ, and synovial membrane of the navicular bursa. The identification of structural tissue abnormalities in the foot has largely been limited to bone-related pathology with the use of radiography and nuclear scintigraphy. However, Wright et al. [3] found that fibrillation of the DDFT and partial-thickness fibrocartilage loss were the two most common lesions of significance in horses with navicular disease. Neither of these lesions is detectable by radiography. Consequently, it is not surprising that a large number of horses with navicular syndrome have no radiological abnormalities or various patterns of radiopharmaceutical uptake in the foot. Magnetic resonance imaging (MRI) is becoming increasingly important in the diagnosis of the cause of lameness in these patients because of its ability to show superior soft-tissue contrast and detail. However, experience of equine clinicians with MRI is limited, and image interpretation is generally extrapolated from knowledge of human MRI. An indispensable part of the learning process in equine MRI must be the critical evaluation of the significance of alterations in signal intensity and signal patterns in the horse's foot.

We hypothesized that MRI would represent changes in tissue structure detectable on macroscopic and microscopic examination of the entire foot and radiographic examination of the navicular bone in isolation. The objectives of the study were to compare the findings in feet from horses with foot lameness and feet from horses without foot lameness and to determine the value of MRI for each of the pathological and radiological abnormalities found to be significantly associated with the presence of foot lameness. This paper summarizes a comparison between the pathology, radiology, and MRI of the navicular bone and the soft-tissue structures in the foot from horses with clinically diagnosed navicular syndrome/palmar foot pain and age-matched control horses.

2. Materials and Methods

Thirty-four limbs were collected from 18 horses (Group L) in which a diagnosis of "navicular syndrome" or palmar foot pain was made based on the results of clinical examination, response to nerve blocks, radiography, and in some cases, scintigraphy. The minimum inclusion criteria were lameness of 2-mo duration, elimination of lameness by a palmar digital nerve block, and absence of obvious other causes of foot pain. Twenty-five limbs were collected from 15 horses without evidence or history of front foot lameness (Group N). All limbs were labeled and stored frozen at -20°C, and then, they were thawed for up to 24 h before examination. Magnetic resonance (MR) images were acquired using a 1.5-T GE Signa Echospeed MR system [a] with a medical extremity radio frequency coil, and MR image analysis software [b] was used for image interpretation and cross-referencing as previously described [4]. All interpretation was performed by two experienced analysts blinded with respect to lameness group. Structures evaluated included the DDFT, NB, distal phalanx, DSIL, CSL, CLs of the DIPJ, DIPJ, and navicular bursa. Radiographic examination was performed on isolated navicular bones obtained after dissection using lateromedial (LM), palmaroproximal-palmarodistal oblique, and dorsoproximal-palmarodistal oblique (upright pedal) projections and a 150-kV, 1250-mA tube [c] with single emulsion film [d] in cassettes with a single 400-speed screen [e]. The isolated bones were correctly positioned as described by De Clercq et al. [5]. All NBs were analyzed for evidence of radiological abnormalities associated with "navicular disease" [5-7]. To enable comparison of the MR grade of each NB with its radiological appearance, each NB was given an overall radiological grade (1 - 4) based on the criteria used by Dik and Van den Broek [6]. Dissection of each foot was performed immediately after MRI, and macroscopic findings were recorded and saved photographically. Eight units of tissue were collected for histological examination corresponding to structures examined using MRI. Histological preparation was performed as previously described [8,9].

The occurrence of macroscopic pathological findings was compared between groups using a 2 x 2 table association test, taking a significance level of p < 0.05. Radiological grades were compared using a Mann Whitney U test, and histological grading was taken using a Mann Whitney U test. Lesions detected with MRI were compared with lesions detected on pathological and radiological assessment. A sensitivity and specificity comparison of MR imaging with macroscopic and microscopic pathological examination was performed. A Spearman Correlation was used to test for associations between MR and radiological grading.

3. Results

MRI

The grading and distribution of MR signal irregularities, according to the first author, in the limbs of both groups of this study are summarized in Table 1 and Table 2 and have been discussed in greater detail elsewhere [4].

In Group N, mild and even moderate signal irregularities were not uncommonly present, but severe abnormalities were absent. In the NBs in Group N, focal fluid accumulation in the navicular bursa was frequently visible on short inversion time inversion recovery (STIR) images at the site of the palmar depression in the sagittal ridge, and mild to moderate increase in signal intensity in the medullary cavity was frequently observed on STIR images. The DDFT in Group N occasionally contained incomplete linear areas of signal increase in the sagittal plane of the DDFT, small focal areas of bright signal in the core of the DDFT, mild irregularity of the dorsal border, and moderate asymmetry between the lateral and medial tendon lobes. The normal MRI appearance of the DSIL had a uniform distribution of low signal interspersed with more or less uniformly distributed areas of high signal. There was often mild endosteal or palmar cortical irregularity at the insertion of the DSIL and a consistent area of axial adherence of variable size between the DSIL and the dorsal surface of the DDFT. There was also an area of axial adherence of variable size between the CSL and the dorsal surface of the DDFT, and focal areas of increased signal were often seen within the CSL. Mild heterogeneities in signal intensity were common in articular cartilage of the DIPJ. Mild to moderate asymmetry of the DIPJ CL was seen in limbs from both Groups L and N.

The most common findings that distinguished limbs in Group L were all graded as severe abnormalities. In the NB, they most commonly consisted of irregular signal at the level of the flexor surface consistent with flexor fibrocartilage or cortical defects (Fig. 1a and Fig. 2a), irregular areas of decreased signal in the medulla or along the endosteal surface consistent with medullary or endosteal mineralization, and increased signal intensity on fat-suppressed images (Fig. 3a) consistent with medullary edema. Focal low signal at the distal border of the NB suggestive of distal border fragments or mineralization within the proximal aspect of the DSIL was also common (Fig. 4a).

Figure 1. (A) Sagittal three-dimensional (3-D) T2*-GRE image of the NB from a lame limb with a focal accumulation of synovial fluid indicating a depression in the flexor surface (white arrow). (B) Palmar view of the same NB as A with a mid-ridge depression and two areas of fibrocartilage loss on both sides of the sagittal ridge (black arrows).

Figure 2. (A) Transverse 3-D T2*-GRE image with fat-saturation of a lame foot. There is discontinuity of the flexor cortex of the NB with medium high-intensity signal in the subchondral bone (white arrow). There are continuous strands of medium high signal between the dorsal surface of the DDFT and the palmar surface of the NB (black arrows). (B) Proximodistal view of the navicular bursa of the same foot as A with a full-thickness flexor cortex erosion and adhesions between the palmar surface of the NB and the dorsal surface of the DDFT (white arrow).

Figure 3. (A) Sagittal STIR image of the NB of a lame limb shows diffuse high-intensity signal in the medulla close to the subchondral bone plate of the flexor cortex. (B) H&E stain (×100) of flexor surface and medulla of the same bone as A. There is evidence of focal osteonecrosis (wide arrow) and fibrosis in the intertrabecular spaces (narrow arrow) adjacent to a defect in the flexor surface.

Figure 4. (A) Frontal 3-D T1-weighted SPGR image of the NB of a lame foot. There are two focal areas of low-signal intensity at the angles of the horizontal distal border with the sloping borders of the NB. (B) Palmar view of the NB in the same foot after removal of the DDFT. There is an osseous body (distal border fragment) within the insertion of the DSIL at each angle of the distal border of the NB with the sloping borders.

In the DDFT, abnormalities mainly included signal irregularities to the dorsal surface (Fig. 5a) and to the core region (Fig. 6a and Fig. 7a) as well as sagittal and parasagittal linear areas of high signal (Fig. 8a). Additionally, adherence between the DDFT and the flexor surface of the NB (Fig. 2a) between the DDFT and DSIL and between the DDFT and the palmar aspect of the CSL were also common. Different abnormality types were frequently present concurrently at different levels of the DDFT. Severe grade signal irregularities in the DDFT most commonly spanned multiple levels, extending from the insertion to a level proximal to the navicular bursa. The most common abnormalities involving the DSIL were distal border fragments or focal mineralization near the origin of the DSIL, prominent pockets of fluid signal (Fig. 9a), and enlargement of the extent of adherence between the DSIL and the DDFT. Almost all higher grade abnormalities in the CSL consisted of symmetrical or asymmetrical enlargement with increased adherence to the DDFT. Soft-tissue proliferation within the proximal or distal aspects of the navicular bursa, with or without the presence of synovial effusion, was the main bursal abnormality observed. Few horses presented with severe abnormalities of the DIPJ. There were no severe MRI abnormalities in the DIP CL of any limbs in this study.

Figure 5. (A) Detail of transverse 3-D T2*-GRE image with fat saturation at the level of the NB of a lame foot. There are multiple, short linear areas of high-signal intensity arising from the dorsal surface of the DDFT and coursing into the body of the DDFT in a palmar direction. The dorsal surface of the DDFT is irregular (white arrows). (B) Frontal view of the DDFT at the level of the navicular bursa in the same foot as A. There are two vertical linear fibrillations/erosions on the dorsal surface of the DDFT with some loose tendon fibers on the exposed surface of the fibrillations (black arrows). Remnants of severed adhesions between the dorsal surface of the DDFT and the palmar surface of the NB are visible in the center of the DDFT.

Figure 6. (A) Transverse 3-D T1-weighted SPGR image of the DDFT of a horse that underwent a palmar digital neurectomy 3 mo previously. There is a focal area of high-intensity signal in the core of the lateral lobe of the tendon (white arrow). (B) Transected surfaces of the DDFT of the same foot as in figure 5A, proximal to the level of the navicular bursa. There is evidence of a core lesion in the lateral lobe of the DDFT, which is characterized by dark discoloration (black arrows).

Figure 7. (A) Transverse 3-D T2*-GRE image at the level of the insertion of the DDFT to the flexor surface of the distal phalanx in a lame left foot. There is a small focal area of high intensity signal in the insertion of the DDFT just lateral to the midline. (B) Palmar view of the transected surface of the DDFT within 1 cm of its insertion to the flexor surface of the distal phalanx in the same foot as A. There is asymmetric thickening of the DDFT centered on a focal "core lesion" just lateral to the midline that is surrounded by red discoloration of the adjacent tendon tissue.

Figure 8. (A) Detail of transverse 3-D T2*-GRE image at the level of the proximal border of the NB of a lame right foot. There are two (para)sagittal linear areas of high-intensity signal in the DDFT, one of which is the normal axial division between both lobes of the DDFT (narrow white arrow); the other forms an abnormal abaxial split in the lateral lobe (wide white arrow). (B) Frontal view of the DDFT at the level of the navicular bursa in the same foot as A. There are fibrillations on the dorsal surface of both tendon lobes. There is also a full thickness parasagittal split in the lateral lobe through which forceps can be passed to the palmar surface of the tendon (black arrows).

Figure 9. (A) Detail of transverse 3-D T2*-GRE image at the level of the insertion of the DSIL to the flexor surface of the distal phalanx in a control limb without lameness. There are prominent focal areas of high-signal intensity (fluid) throughout the impar ligament (white arrows). (B) Palmar view of the transected surfaces of the DSIL and DDFT within 1 cm of their insertions to the flexor surface of the distal phalanx. There are multiple empty spaces between the concentrated fiber bundles of the DSIL into which synovial outpouchings of the DIPJ and the navicular bursa protrude (black arrows).

There was a significant effect of group (N or L) on the MRI grade and type of lesion in the DDFT, on the MRI grade of the navicular bone in all its different components, and on the MRI grade of the DSIL, DIPJ, navicular bursa, and CSL. There was no effect of group on MRI grade of the distal phalanx or the DIPCL.

Radiology of the NB

Forty-seven percent of lame NBs were graded as "normal" [6]. Radiological evidence of erosions of the flexor cortex, blurring of the cortico-medullary interface, diffuse osteosclerosis of the medullary cavity, focal medullary osteosclerosis adjacent to the palmar subchondral bone plate, and distal border fragments were all significantly associated with the presence of lameness. Widened conical, rounded, and mushroom-shaped synovial invaginations were not associated with lameness, but narrow, deeply penetrating synovial invaginations were. Smooth-walled or irregular new bone and spurring of the wings occurred equally at the proximal border of NBs of both groups.

Pathology

The histopathological findings of this study have been discussed in greater detail elsewhere [8,9].

NB

A round or oval depression in the middle third of the sagittal ridge and covered with fibrocartilage was frequently encountered in both groups. However, loss of palmar fibrocartilage (Fig. 1b), either partial or full thickness, was significantly more common in lame horses (41%) than in normal horses (16%), and it occurred mostly on the sagittal ridge or either side of the sagittal ridge, approximately one-third proximal to the distal border of the bone. Flexor cortex erosions (Fig. 2b) were present in 44% of lame limbs only, and 66% of those had adhesions to the DDFT.

Histologically, only 4 of 59 bones were considered to have a normal flexor fibrocartilage and cortex. Medullary lesions generally extended into the medulla from subchondral bone necrosis adjacent to lesions of the flexor fibrocartilage (Fig. 3b). Only one horse from Group L had severe primary medullary necrosis. No feet showed evidence of thrombosis of blood vessels within the medulla. Distal border fragments were identified grossly as osseous fragments embedded in the DSIL near a crater-like defect in the distal border of the NB at the angle between the horizontal and sloping borders (Fig. 4b) in 10 lame feet and 3 control feet. Fibrocartilaginous metaplasia at the interface between the distal border of the NB and the DSIL was typically more marked in the lame limbs and accompanied moderate to marked enlargement of the synovial invaginations. Osteophyte formation at the proximal articular margin of the NB was encountered frequently and equally in both groups. The flexor aspect, proximal border, and distal border of the NB in Group L had a higher histological grade than feet in Group N.

DDFT

Gross abnormalities of the DDFT were observed in 62% of lame limbs and were subdivided into five types: core lesions (C; Fig. 5b), parasagittal splits (S; Fig. 6b), insertional lesions (I; Fig. 7b), dorsal surface fibrillations and erosions (D; Fig. 8b), and adhesions of the DDFT to the NB (A; Fig. 2b). None of these lesion types were observed in control limbs. Core lesions were present in seven lame limbs, three of which had undergone palmar digital neurectomy previously for treatment of navicular disease. Dorsal fibrillations and erosions were nearly always associated with macroscopic degenerative changes of the fibrocartilage or flexor cortex of the NB. A combination of lesion types was present in 15 of 34 lame limbs. Histologically, abnormalities of the dorsal surface of the DDFT consisted of parasagittal crevices and splits of >20% depth from the surface with prominent fibrillation and sometimes focal fibroplasia. Deeper in the DDFT, there was frequently evidence of fibrocartilaginous metaplasia and ghosting of blood vessels in the interfascicular septa. Focal core necrosis of the tendon was not observed in any feet, although several showed marked focal fibroplasia of ~25% of the cross-sectional area of the tendon, possibly indicating healing of an earlier core lesion. The DDFT of feet in Group L had significantly higher histological grade than feet in Group N, except for at the DDFT insertion where there was no difference.

DSIL

The DSIL consisted of well-delineated fiber bundles running longitudinally between the NB and the distal phalanx that are separated by interstitial pockets lined by synovial membrane (Fig. 9b). In some ligaments, these pockets were more prominent than in others. Sub-synovial and perivascular hemosiderin deposition was present in 31% of lame feet, indicating previous haemorrhage.

The DSIL had a central attachment to the DDFT at the distal extent of the navicular bursa, which was noticeably more extensive in many lame limbs. Fibrocartilaginous metaplasia in the DSIL was common in both groups, but more so at the insertion to the distal phalanx in lame limbs. The DSIL in Group L had a significantly higher histological grade than feet in Group N.

CSL

Soft-tissue thickening in the area of the CSL, caused by bursal synovial hypertrophy and congestion, occurred in 29% of lame limbs. Resulting adhesions between the DDFT and the CSL only occurred in lame limbs. In three limbs with thickening of the CSL, an encapsulated, fluid-filled cyst was associated with the proximal surface of the CSL. Fibrocartilaginous metaplasia was common in both groups, but severe abnormalities were not encountered in any foot; there was no difference in histological grade between the groups.

Navicular Bursa Synovium

There was significantly more evidence of synovial hypertrophy in the proximal recess of the navicular bursa in Group L with lymphoplasmacytic infiltration and focal subsynovial hemosiderin deposition.

DIPJ

Osteophytes, variable degrees of articular cartilage damage, sub-synovial lymphocytic inflammatory infiltration, and hemosiderin deposition were seen in both groups. There were no significant differences in the histological score of the DIPJ synovial membrane between groups.

DIPJ CL

No gross abnormalities were detected in these ligaments. Histologically, severe fibrocartilaginous metaplasia within the ligaments was observed in five CL (three in Group L and two in Group N), but there was no difference in histological grade between the groups.

Comparison of MRI and Pathology

NB

MRI assessment by the first author only had a poor sensitivity (36%) but high specificity (100%) for gross evidence of partial or complete loss of fibrocartilage from the flexor surface of the NB. The sensitivity (100%) and specificity (97%) for the presence of a palmar mid-ridge depression was high. The sensitivity and specificity for the presence of macroscopic partial or complete flexor cortex erosions was excellent. When MRI was compared with histology, all focal, saucer-like lesions of >50% depth of the fibrocartilage on histological examination were characterized by adjacent fluid accumulation, endosteal irregularity, and irregularity of the chondro-osseous junction on MRI. More superficial lesions were less consistently identified. Evidence of adhesion to the DDFT was observed as loss of navicular bursa fluid signal, cortical defects, and apparent continuity of tissue between the DDFT and NB. The overall sensitivity of MRI for all histological abnormalities of the flexor surface was good (83%), but the specificity was only fair (65%). All NBs with severely high medullary signal intensity on fat-suppressed images and low signal on spoiled gradient (SPGR) and gradient-recalled echo (GRE) images had corresponding histological evidence of focal or generalized medullary osteonecrosis and fibrosis. The sensitivity of MRI for histological abnormalities of the navicular medulla was excellent (94%), and the specificity was good (85%). The sensitivity (92%) and specificity (93%) of MRI for macroscopically visible distal border fragments was high.

DDFT

MR examination had a fair sensitivity (74%) but an excellent specificity (93%) for fibrillations/erosions of the dorsal surface of the tendon in the area of the navicular bursa identified on necropsy. The sensitivity and specificity of MRI was excellent or good for all other lesions of the DDFT. Histologically, all feet with mild, moderate, or severe irregularity of the dorsal tendon surface on MR images were characterized by superficial dorsal fibrillation, crevicing, or splitting. Parasagittal splits on MRI matched with histological crevices and dorsal ridges of the appropriate depth with respect to the MR images. Core lesions on MR imaging were seen histologically as obliteration of normal tendon fascicle structure, increased cellularity, and increased vascularization. The overall sensitivity and specificity of MRI for histological abnormalities of the DDFT was excellent (95% and 100%).

DSIL

Sensitivity and specificity of MRI for detection of adhesions between the DSIL and the DDFT was fair. Histologically, limbs with moderate to severe signal irregularities of the DSIL on MRI were characterized by large intraligamentous synovial pockets, prominent vascularity, and fibrocartilaginous metaplasia. The overall sensitivity of MRI for histological abnormalities of the DSIL was good (80%), but the specificity was poor (50%).

CSL

MR detected gross periligamentous tissue proliferation, adhesion formation between the DDFT and the CSL, and cysts in the CSL with high levels of sensitivity. Histologically, moderate to severe signal heterogeneity was generally associated with fibrocartilaginous metaplasia within the CSL. Focal high signal in CSL was associated with synovial in-pouchings and hyperplasia in three limbs. The overall sensitivity of MRI for histological abnormalities of the DSIL was fair (73%), but the specificity was excellent (97%).

Navicular Bursa

There was an excellent correlation between the appearance of synovial proliferation in the proximal recess on MRI and on pathology.

DIPJ

MRI had a poor sensitivity for gross evidence of periarticular osteophytes and a fair sensitivity for articular cartilage abnormalities in the DIPJ.

Comparison MRI and Radiology of the NB

The overall navicular MR grade (resulting grade of flexor surface grade, proximal border grade, distal border grade, dorsal surface grade, and medulla grade) correlated well with the radiological grade (p = 0.0001); the Spearman r coefficient was 0.68. The NB edema grade also correlated well with the radiological grade (p = 0.0001), but the Spearman rs coefficient of 0.53 suggested that the association was not as strong as with the overall navicular MR grade.

4. Discussion

The results of this study supported the hypothesis that alterations in MR signal represent changes in tissue structure detectable using gross and histopathological examination; a high sensitivity and specificity for abnormalities of the NB, DDFT, CLS, and DSIL were detected, but only a poor to fair sensitivity, in the opinion of the first author, was detected for articular cartilage damage in the DIPJ, early fibrocartilage loss from the flexor surface of the NB, and the presence of osteophytes at the joint margins of the DIPJ. Furthermore, the hypothesis was confirmed that radiological abnormalities of the NB associated with lameness corresponded to alterations in MR signal. Additionally, MRI was able to show medullary and flexor border abnormalities of the NB that were not visible on radiographs.

As in a previous study [3] the significant changes associated with palmar foot pain were those of "classic navicular disease" [10]. Abnormalities of the DDFT were most commonly present concurrently with lesions of the flexor surface of the NB in this study. Recent clinical MRI studies of horses with palmar foot pain have proposed a much higher incidence of primary tendonitis of the DDFT independent from abnormalities of the NB [11,12]. Five distinct lesion types were observed in the DDFT in this study. As in previous clinical studies, MRI was clearly able to distinguish each lesion type.

The close relationship between lameness and radiographic changes of the flexor surface and adjacent medulla compared well with the gross necropsy findings of this study. The association between the presence of distal border fragments and lameness is not universally accepted [5,6,13], but these fragments have recently gained increasing support as a cause of palmar foot pain [3,14]. In addition, clinical experience with MRI suggests that these fragments frequently remain unidentified on standard foot radiographs. The poor association between the radiological features of navicular disease and the presence and degree of lameness is widely accepted [15,16]. A good correlation between the navicular grading system according to Dik and Van den Broek [6] and the overall MRI grading system used for the NB in this study at least showed that both imaging modalities resulted in a similar weighting of NBs of both lame limbs and control limbs. MRI can offer considerable additional information on medullary abnormalities and fibrocartilage degeneration as well as on pathology of the soft-tissue structures associated with the NB. Therefore, MRI is superior for imaging of a foot with diagnosis of navicular syndrome [17].

MRI was shown in this study to be a highly sensitive and specific imaging modality for the detection of macroscopic and microscopic abnormalities of the NB, the DDFT, the navicular bursa, and the CSL.

"NB edema" was a frequent finding in this study. Hyperintense signal on fat-saturated and STIR images, with hypointense signal on T1-weighted images, has been described in man as a "bone marrow edema pattern" [18]. The appearance of a "NB edema" pattern was recently reported as one of the earliest and most important MRI findings in navicular disease [19]. Nevertheless, it remains uncertain whether medullary edema is an acute phenomenon associated with medullary trauma and inflammation or whether it occurs as a consequence of chronic medullary remodeling and fibrosis with or without venous congestion [3,20]. In our study, a "bone marrow edema pattern" in the NB medulla was not associated with edema but with osteonecrosis, fibrosis, loss of trabecular structure with a moth-eaten appearance to the bone trabeculae, prominent capillary infiltration, and, in one horse, adipose tissue necrosis. This suggests a more chronic phenomenon than would normally be associated with edema. This agrees with evidence found in man where hyperintense signals in the medullary cavity on fat-suppressed images has been associated with acute bone and joint trauma [21]. However, "transient bone marrow edema syndrome" was unrelated to trauma [22,23] and chronic end-stage osteoarthritis [18,24]. It is possible that, both in man and horse, the same pattern of hyperintense medullary signal may reflect different tissue changes in each of these different scenarios (i.e., accumulation of fluid [edema and haemorrhage] in acute trauma and/or "bone marrow edema syndrome" and fibrosis of the marrow spaces in chronic degeneration).

Loss of fibrocartilage from the flexor surface of the NB has been considered one of the earliest pathological changes of navicular disease [17]. This should not be confused with the presence of a palmar depression in the middle third of the sagittal ridge, which has previously been described as a mid-ridge synovial fossa and considered a normal anatomical variation [3,13,20,25]. Therefore, identification of early fibrocartilage degeneration would be considerably helpful in early recognition of this disease, especially as radiography is unable to detect this change. However, in the opinion of the first author, the sensitivity of MRI for fibrocartilage loss was inconsistent. Focal accumulation of bursal fluid next to a depression in the flexor surface produces a high signal on T2-weighted or STIR MR images and allows for identification of flexor surface irregularities. The distinction between fibrocartilage loss and the presence of a normal palmar depression was often difficult on MRI, because the main MRI characteristic of both was an accumulation of bursal fluid at the indentation in the flexor surface of the NB. As a consequence, sensitivity of MRI was good for recognition of a palmar depression but not for fibrocartilage degeneration. Not surprisingly, the sensitivity and specificity of MRI could be improved if pathological fibrocartilage loss and sagittal ridge depression were grouped together as indentations of the flexor surface. Although we believed fibrocartilage to be evident as a layer of high-signal intensity on SPGR sequences and as a layer of medium high-signal intensity (brighter than the flexor cortex and the DDFT but less bright than synovial fluid) on T2*-GRE sequences, other authors have also commented that the navicular fibrocartilage could be difficult to image [17].

MRI was useful for detection of all lesion types of the DDFT in this study. MRI is the technique of choice for assessment of tendon injuries in man [26]. In the horse, MRI has been shown to reflect tissue abnormalities in acute and chronic superficial digital flexor tendon injuries [27]. In one previous study, acute injuries were associated with increased signal intensity on both T1- and T2-weighted images, whereas chronic tendonitis was associated with increased signal intensity on T1-weighted images but relatively less of an increase on T2-weighted images [28]. In our study, mild lesions were clearly observed on T1-weighted SPGR images but not consistently on T2*-GRE images. The histological changes of these mild lesions indicated low-grade, chronic change and supported previous findings [28]. More severe histological lesions in our study were associated with increased signal intensity on all MR sequences, which would suggest a combination of both acute and chronic characteristics; this was described as superimposition of acute injury over chronic tendonitis in a previous MRI study of horses with tendon injuries [28]. Although focal core necrosis of the DDFT was not observed in any feet of this study, collagen necrosis with matrix liquefaction and vacuolization have previously been associated with the presence of focal hyperintense signal in the core of the DDFT [29,30]. Several tendons in this study contained focal fibroplasia of ~25% of the cross-sectional area of the tendon. Although this was considered as evidence of a healed scar in the DDFT of several limbs, it was still associated with the presence of focal hyperintense MR signal. This finding suggests that it may be difficult to distinguish the focal necrosis in acute core lesions from fibrous tissue present in healing or healed tendons, because both produced focal hyperintense MR signal on T1-weighted images.

Fibrillations and erosions of the dorsal surface of the DDFT in the navicular bursa were the most frequently encountered abnormality in 16 of 34 lame limbs (47%). As for fibrocartilage loss, these lesions are undetectable by radiography, emphasizing the limitations of this imaging modality in the assessment of navicular disease. MRI was able to identify superficial fibrillation of the DDFT with a sensitivity of 74%, and thus, this lesion was missed in some lame limbs. In contrast to dorsal tendon fibrillations proximal to the NB, fibrillations at the level of the NB did not always seem to produce an irregular MR signal at the dorsal aspect of the DDFT, possibly because of its compression against the palmar surface of the NB.

Although sensitivity of MRI was good for identification of both macroscopic and microscopic abnormalities of the DSIL, the heterogeneous nature of MR signal intensity throughout this ligament was striking. The occurrence of voids in normal DSIL can be explained by the presence of both penetrating blood vessels [31] and normal fenestrations containing synovial invaginations from the navicular bursa and DIPJ [32]. Because these features are integral to the normal anatomy of the DSIL, it can be difficult to consistently determine when normal anatomical features become abnormal using MRI. Histologically, these pockets were prominent in some DSIL, suggesting synovial hyperplasia and invagination into the ligament, but it was unclear whether these were anatomical differences rather than reactive change. Although sensitivity of MRI for histological abnormalities of the DSIL was good (80%), the specificity was poor (50%), suggesting a tendency for false-positive diagnosis of DSIL desmitis.

MRI produces a more detailed image than any other imaging modality that has been used in equine veterinary practices so far. As a result of this overwhelming detail, signal irregularities associated with artefacts or unexpected anatomical variations are not uncommon and have to be distinguished from signal irregularities caused by tissue damage [4]. High detail of artefact and anatomical variation may also help to explain some of the few average sensitivity and specificity results. Specificity of MRI was excellent for all macroscopic abnormalities of the DIPJ and NB and most abnormalities of the DDFT, DSIL, and CSL.

The implication of high specificity is the low incidence of false-positive diagnoses, which is a major advantage of this imaging modality. The specificity of MRI was good but not excellent for core lesions and parasagittal splits of the DDFT, periligamentous tissue proliferation of the CSL, and adhesions between the DSIL and the DDFT; it was only fair for the presence of adhesions between the CSL and DDFT. Closer scrutiny of the grade distribution between true-positive and false-positive MR readings was performed for core lesions and parasagittal splits of the DDFT to elucidate if an association existed between the signal irregularity grade and the incidence of "false-positive lesions" for either abnormality. None of the grade 3 signal irregularities were "false-positive lesions", and none of the grade 1 signal irregularities were "true-positive lesions" for any lesion type in the DDFT (core lesions, parasagittal splits, insertional lesions, or dorsal erosions). Similar observations were made for the NB, where all "false-positive" distal border fragments and flexor cortex erosions were found to be grade 1 or grade 2 MR signal irregularities. These findings indicate that the risk for false-positive diagnosis of abnormalities of the DDFT can be significantly reduced by ignoring grade 1 signal irregularities in the DDFT. Consequently, the MR diagnostician may have to be aware that low-grade signal irregularities can be incidental findings of limited clinical significance.

The results of this study show the potential of MRI for visualization of structural changes within osseous and soft-tissue structures of the equine foot. However, further investigation is required to understand the significance of these findings and to improve understanding of progression of these lesions so that rational preventative and therapeutic strategies may be developed.

This study was funded by the Home of Rest for Horses. The authors are very grateful to the veterinary surgeons who provided limbs for this project and to Marion Branch, Julie Breingan, and Ray Wright for technical assistance.

Footnotes

1. GE Medical Systems, Slough, UK SL1 4ER.
2. GE Advantage Windows 3. 1, GE Medical Systems, Slough, UK SL1 4ER.
3. Polydoros 100 kW generator, Siemens AG, Munich, Germany.
4. Fuji MI-NH, Fujifilm GmbH, Düsseldorf, Germany.
5. Fuji HR Regular, Fujifilm GmbH, Düsseldorf, Germany.