To investigate the role of metacarpophalangeal (MCP) joint anatomic and biomechanical factors in the distribution of synovitis and bone erosion in early rheumatoid arthritis (RA).
To investigate the role of metacarpophalangeal (MCP) joint anatomic and biomechanical factors in the distribution of synovitis and bone erosion in early rheumatoid arthritis (RA).
Thirty-three patients with early RA with clinically diagnosed MCP joint disease and 28 healthy controls were examined by magnetic resonance imaging of the second to fifth MCP joints of the dominant hand. T1 and T2 fat-suppressed coronal sequences were obtained to assess erosion, and dynamic contrast-enhanced images were acquired to assess synovitis in all of the RA patients and in 8 of the controls. Erosions were defined as bone defects with sharp margins observed using T1-weighted imaging in 2 planes, with a cortical break seen in at least 1 plane. The location of erosions was recorded. The volume of synovitis surrounding each MCP joint (divided into 8 regions) was calculated by summation of voxels derived from the maximal enhancement parameters. The synovial volumes adjacent to MCP joint collateral ligaments were determined by correcting synovial volumes for the positions of asymmetrically placed flexor tendons.
In patients with early RA in whom bone erosions were present, there was a propensity for involvement of the radial side of the second (P < 0.0001), third (P = 0.002), and fourth (P = 0.056) MCP joints, but not the fifth. Fifty-two of the 110 erosions (47.3%) occurred adjacent to the radial collateral ligaments of the second, third, and fourth MCP joints. The volume of synovitis was also greater on the radial side of the second (P < 0.0001) and third (P < 0.001) MCP joints. A predilection for synovitis in all of the MCP joints adjacent to the radial collateral ligaments was evident when the positional effects of the flexor tendon were considered. The position of radial collateral ligaments had an effect on erosion formation that was independent of synovitis. A predilection for radial bone damage was also evident in the controls, although lesions were 5-fold less frequent, were generally smaller, and had well-defined margins.
This study shows that there is a predilection for both synovitis and bone erosion formation on the radial side of the MCP joints in early RA, and that joint inflammation appears to drive the inherent tendency for bone damage on the radial side of joints. These findings have implications regarding the pathogenesis of joint damage in RA.
Rheumatoid arthritis (RA) is a symmetric polyarthritis characterized by synovitis and bone erosion. Recent clinical and magnetic resonance imaging (MRI) studies of the metacarpophalangeal (MCP) joints have confirmed the primacy of synovitis in early disease and the secondary nature of bone erosions (1–4). The pathogenesis of bone erosion in RA has been studied extensively in relation to immunologic factors (inflammatory cytokines, rheumatoid factors, HLA–DR4) and cellular mechanisms of damage (mediated by macrophages, transformed synovial fibroblasts, and osteoclasts) (5). However, the limited radiographic and MRI studies in RA suggest that anatomic and mechanical factors may be important, since a propensity for bone erosion over the radial (lateral) side of MCP joints and wrist joints has been noted (6, 7). The precise location of bone erosions and the basis for their asymmetric distribution and relationship to regional intraarticular synovitis has not been determined. A sonographic study of synovitis suggested a radial-side predilection for synovitis, but erosions were not assessed (8). Furthermore, radiography and sonography provide very limited views, so the significance of the reported changes is uncertain.
The influence of joint anatomic and biomechanical factors on the pathology in RA is recognized in chronic disease by the characteristic MCP joint deformity of ulnar drift, which results from local joint forces (9–14). Studies have shown that greater forces are exerted on the second and third MCP joints relative to the fourth and fifth joints in the course of normal hand function (9). Further, the MCP joint radial collateral ligaments provide the principal restraining force that maintains joint stability and limits palmar and ulnar movement of all of the MCP joints (9, 10). Recently, it has been suggested that factors related to joint biomechanics with associated injury may be involved in the pathogenesis of RA (15–17), but there is little direct supportive evidence for this. The MCP joints represent a good model system for studying synovitis and bone damage in RA because these joints have a comparatively simple anatomy, are invariably involved at disease onset, and are characteristic sites of bone erosion (18, 19). MRI is a validated method for detecting MCP joint synovitis in RA (20, 21), and it is also superior to radiography for detecting bone erosion in RA (22, 23).
In the present study we used MRI to investigate the relationship of synovitis and erosion to the known joint anatomic and biomechanical factors present at the radial collateral ligaments of the MCP joints. We confirmed the predilection for erosions on the radial side of the MCP joints in early RA and showed that this was associated with a predilection for synovitis adjacent to the radial collateral ligaments. These findings may be useful in leading to a better understanding of the pathogenesis of joint damage in early RA.
Thirty-three consecutive patients from an early arthritis clinic were recruited, all of whom fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for RA (24). Local research ethics committee approval for the study was obtained. Entry criteria included disease duration of <12 months and the presence of clinical MCP joint disease. Twenty patients were female and 13 were male. Patient ages ranged between 21 and 83 years (mean 54 years), and the disease duration ranged between 1.5 and 11 months. None of the patients had MCP ulnar drift or ulnar deviation.
A group of 28 healthy asymptomatic volunteers (19 women, 9 men) with no history of arthritis also had MRI of their MCP joints performed. The mean age of the control group was 40 years. Eight of these healthy subjects had gadolinium diethylenetriaminepentaacetic acid (Gd-DTPA) administered to assess the joints for synovitis.
MRI of the MCP joints of the dominant hand was performed using a 1.5T Gyroscan ACS-NT whole-body scanner (Philips Medical Systems, Best, The Netherlands), with a Philips 11-cm circular surface coil placed on the dorsum of the subject's hand.
Bone erosions were assessed using T1-weighted coronal and axial, T1-weighted coronal fat-suppressed (FS), and T2-weighted coronal FS images. The imaging parameters for all of the T1-weighted imaging sequences were as follows: repetition time (TR) 485 msec, echo time (TE) 20 msec, matrix 256 × 256, slice thickness 1.5 mm, number of signal averages (NSA) 2. The field of view (FOV) examined was 100 × 100 mm. For the T2-weighted coronal FS sequence, imaging conditions were as follows: TR 2,000 msec, TE 100 msec, matrix 256 × 256, slice thickness 2.0 mm, NSA 4, and FOV 100 × 100 mm.
For the assessment of synovitis, all patients were examined using a dynamic gadolinium-enhanced MRI (DEMRI) measurement as described in detail previously (25). The DEMRI data were acquired using a 3-dimensional T1-weighted spoiled gradient-echo imaging sequence with a 40° flip angle, a TE of 3.8 msec, a TR of 14 msec, and a 170 × 256 acquisition matrix. Images were obtained from 6 contiguous slices that covered almost all of the second to the fifth MCP joints. Each slice was recorded with an FOV of 50 × 100 mm and a thickness of 3 mm. The imaging protocol measured 20 consecutive data sets; each data set was acquired in 7.1 seconds. The contrast agent Gd-DTPA was injected at a standard dose of 0.1 mmole/kg body weight after the acquisition of the first data set. The DEMRI measurement was completed within a few minutes of the injection of the contrast agent; therefore, the synovial volumes calculated from this measurement were not affected by leakage of this agent into synovial fluid (26).
MRIs were scored for erosions by an experienced reader (PO'C) who was blinded to the clinical details and radiographic findings. Bone erosions were defined as bone defects with sharp margins visible on T1-weighted images in 2 planes, with a cortical break seen in at least 1 plane, according to recommendations of the Outcome Measures in Rheumatology Clinical Trials group (27). Each bone erosion was recorded according to its site within a given joint, i.e., radial, ulnar, dorsal, or volar.
The DEMRI data were processed using specialized software developed in-house as described previously (28). This software calculated the value of maximal enhancement (ME) on a pixel-by-pixel basis for each slice, using the 20 acquired “dynamic images.” The parametric images included all voxels where ME was more than 20% higher than baseline. The ME values were displayed in color as overlays superimposed on conventional gray-scale MR images of anatomy (Figure 1). Calculations of ME were made in the second to the fifth MCP joints, with data from the thumb being discarded because it was too far out of plane relative to the other 4 MCP joints and its orientation may vary when the hand is flat (unlike the other MCP joints).
To calculate the volume of synovitis and to compare the radial and ulnar sides of each MCP joint, the color images/overlays were then processed using a commercially available image analysis software package (Analyze; AnalyzeDirect, Lenexa, KS). This software allowed the generation of regions of interest (ROIs) that delineated the enhancing synovium within the 4 MCP joints studied. The ROIs were created by manually defining the edge of the synovial enhancement. Analyze software was then used to automatically divide this region of enhancement into 8 subregions (labeled 1–8) (Figure 1) about the center of each joint. The number of enhancing voxels of the 6 slices in the images of ME for subregions 1–4 was compared with the number of enhancing voxels in subregions 5–8 to determine the relative volumes of enhancing synovium in the ulnar and radial sides, respectively. Because the MR acquisition conditions remained fixed for all patients studied, the number of voxels corresponding to a particular region is proportional to the volume of that region. Occasionally, ROIs between 2 adjacent MCP joints coalesced. Under these circumstances, the midline between the 2 joints was used as the boundary of the respective ROIs.
A further analysis examined the relationship of the synovial volume to the collateral ligaments. For this analysis, regions 4 and 5 were omitted and regions 1–3 were compared with regions 6–8 (Figure 1). This calculation therefore excluded the volar segments of the joint, which are not in close anatomic proximity to the collateral ligaments. It also overcame the possible effects of the asymmetrically positioned MCP joint flexor tendons on synovial volumes (see Figure 1). The second and third MCP joint tendons are positioned on the ulnar side of joints, the fourth is central, and the fifth is positioned on the radial side of the joint (29) (Figure 1). The position of the flexor tendons or flexor tenosynovitis could compress intraarticular synovium, thus affecting synovial volume calculations at these respective sites. Since the origin of the collateral ligaments is near the dorsum of the joint, calculations of synovial volume were undertaken with regions 1 and 8 included and then omitted, to further evaluate synovitis adjacent to the collateral ligaments. Similar calculations were performed to compare relative volumes of synovitis within the dorsal (regions 1 and 8) and radial (regions 6 and 7) sides of the joint.
Student's paired t-test was used to compare the differences between the radial and ulnar sides of each joint for both erosions and synovitis. Although image analysis is fully automated and as such is 100% reproducible, the delineation of ROIs was subjective and was therefore analyzed for intra- and interobserver agreement. Intraobserver reproducibility was calculated from measurements made by one of the authors (ALT) in 5 patients. Interobserver agreement was obtained from repeat measurements made by a second observer (SFT) in 10 patients.
Pearson's correlation coefficient was calculated for observer readings on each side of the 4 MCP joints. This showed a significant mean correlation of 0.97 between the 2 observers, and a mean intraobserver correlation of 0.94.
A total of 110 erosions were detected in 67 of the 132 joints examined (Figure 2). The majority of erosions (85%) occurred in either the radial or ulnar side of the joints adjacent to the collateral ligaments, with fewer volar or dorsal erosions (7 volar and 9 dorsal erosions; total 15%).
Bone erosions were found to have a significant predilection for the radial sides of the second (21 of 28 erosions; P < 0.0001) and third (23 of 36 erosions; P = 0.002) MCP joints (Figure 3). In total, 47.3% of the erosions occurred at the radial side of either the second, third, or fourth MCP joints. A greater number of erosions over the radial side of the fourth MCP joint was also noted (8 of 13 erosions; P = 0.056), but in the fifth MCP joint more erosions were evident on the ulnar side, although this was not statistically significant (P = 0.601). Compared with the fifth MCP joint, the fourth joint was relatively protected from erosions (Figure 3).
In the healthy control group, bone damage was evident in 9 of 28 subjects (20 erosions in 19 of 112 joints examined, which was 5-fold fewer than the percentage of erosions in the RA group). Even in the healthy controls, a predilection for the occurrence of bone damage on the radial side was detected on the second MCP joint (5 radial erosions, no ulnar erosions; P = 0.022) and the third MCP joint (4 radial erosions, no ulnar erosions; P = 0.043), but not on the fourth (2 radial erosions, 1 ulnar erosion; P = 0.573) or fifth joints (no radial erosions, 2 ulnar erosions; P = 0.161). The erosions in the control group were generally smaller and had well-demarcated margins (Figure 2B), in comparison with the erosions in the RA group, which were larger with less distinct borders.
The second and third MCP joints had the greatest total volumes of synovitis, with the third MCP joint showing the largest volume (mean voxel count per patient 6,884 in the second MCP joint and 7,725 in the third MCP joint). Both the fourth and the fifth MCP joints had smaller volumes of synovitis (mean voxel count per patient 3,977 and 3,260, respectively).
When the volume of synovitis over the radial and ulnar aspects of each joint was determined, a significantly larger volume of synovitis was observed on the radial side of the second (mean voxel count per patient 3,928 on the radial side, 2,956 on the ulnar side; P < 0.0001) and third (mean voxel count 4,186 on the radial side, 3,539 on the ulnar side; P < 0.001) MCP joints (Figure 4A). In the fourth and fifth MCP joints, no significant difference in volumes of synovitis between the radial and ulnar sides was observed (Figure 4A).
These differences could be related to the position of the flexor tendons on synovial volume estimation (see Figure 1). To account for this, and to determine the volumes of synovitis adjacent to the collateral ligaments, the volumes of synovitis were then calculated by comparing regions 1–3 with 6–8 and omitting regions 4 and 5 (Figure 1). When these factors were considered, synovitis for the second (P = 0.21 [not significant]) and third (P = 0.007) MCP joints was also greater on the radial side, and a predilection for synovitis on the radial side was evident in both the fourth (P = 0.04) and the fifth (P = 0.018) MCP joints (Figure 4B).
To account for any confounding effect of the extensor tendon positions on synovial volumes, the calculations were repeated with regions 1 and 8 also excluded, thereby comparing regions 2–3 with 6–7. The same findings of larger radial synovitis volume were evident (P = 0.005, P = 0.002, P = 0.098, and P = 0.26 for the second, third, fourth, and fifth MCP joints, respectively).
We also compared the volume of synovitis between the radial (regions 6 and 7) and the dorsal (regions 1 and 8) sides on each MCP joint. As before, this showed that the volume of synovitis was generally greater adjacent to the radial collateral ligaments. This was statistically significant for the second MCP joint (mean voxel count per patient 2,076 on the radial side, 1,732 on the dorsal side; P = 0.008), but not for the third (mean voxel count 2,217 and 2,083 on the radial and dorsal sides, respectively; P = 0.25), fourth (1,113, and 1,100, respectively; P = 0.88), or fifth (962 and 769, respectively; P = 0.054) MCP joints.
To investigate whether the volume of synovitis was influenced by the presence of adjacent erosions, the joints were then evaluated according to whether bone erosions were present or absent. Figure 4C plots the average number of enhancing synovial voxels in the radial and ulnar aspects of the 65 joints in which erosions were not discernible. The graphs show that in the absence of erosions the volume of synovitis in the second, third, and fifth MCP joints was larger on the radial aspects, but the difference was not statistically significant (P = 0.096, P = 0.169, and P = 0.22, respectively). The failure to reach significance may have reflected the small synovial volume in joints without erosions, but the findings suggest that the predilection for synovitis on the radial side predates the development of erosions.
The average ME per voxel was calculated in order to investigate whether there were qualitative differences in synovitis between the radial and ulnar aspects of MCP joints. There was no significant difference in ME between the radial and ulnar sides of the second and third MCP joints, indicating that although the volume of synovitis was greater on the radial side, it did not differ qualitatively between the radial and ulnar aspects of the joint.
When the MCP joints from the 8 healthy subjects who received Gd-DTPA were analyzed, low-grade enhancement was noted. The mean voxel count in patients with RA was 5,462 per joint and that in the healthy subjects was 859 per joint, i.e., there was a 6-fold increase in enhancing voxels in the patients with RA. Mild enhancement in joints of healthy volunteers has been reported previously (30).
The distribution of erosions and, to a lesser extent, synovitis in the second and third MCP joints could be predicted from the known anatomic and biomechanical factors pertaining to the radial collateral ligaments at these sites (9–14). However, there was only a negligible difference in synovitis between the radial side of the third MCP joint (greatest number of erosions at any site) and the dorsal side of the same joint (where erosions were very uncommon) (mean voxel count per patient 2,217 on the radial side, 2,083 on the dorsal side; P = 0.25). Other examples of this included the greater volume of synovitis over the dorsal aspect of the third MCP joint (mean 2,217 voxels per patient) compared with the smaller volume in the radial aspect of the fourth MCP joint (mean 1,113 voxels) whereas erosions were much more numerous in the latter (total of 1 erosion on the dorsal side of the third MCP joint for all patients, versus 8 on the radial side of the fourth MCP joint).
The differences in volume of synovitis and number of erosions in the fourth and fifth MCP joints were less striking and did not generally reach statistical significance, and definite conclusions about these could not be drawn. However, the fourth MCP joint had a slightly higher synovial volume than the fifth MCP joint although the number of erosions was higher at the fifth (Figures 3 and 4B). This suggests that the relatively protected position of the fourth MCP joint lessened the propensity toward erosion in the setting of a higher degree of synovitis. When the effects of the tendons were considered (Figure 4B), a higher synovial volume was evident on the radial than on the ulnar side of the fifth MCP joint, but erosions were more common on the ulnar side of that joint (Figure 3). Taken together, these findings suggest an anatomic and mechanical basis, rather than an immunologic basis, for different rates of progression of synovitis and damage reported in RA.
Bone erosion in RA has been considered in the context of primary immunologic abnormalities, or cellular abnormalities of synovial fibroblasts, osteoclasts, and other cells (5). The purpose of this study was to investigate whether local anatomic factors related to the radial collateral ligaments could be correlated with the location of synovitis and bone erosions of the MCP joints in early RA. The distribution of erosions in the second, third, and fourth MCP joints, but not the fifth MCP joint, correlated with the volume of synovitis adjacent to the radial collateral ligaments. While synovitis and erosion have traditionally been viewed in relation to immunologic factors, the evidence from this study suggests that local joint factors have a profound influence on synovitis and damage in early RA.
It is likely that the findings in the present study may be related to local biomechanical factors in the joint. The importance of biomechanical factors in the genesis of the characteristic clinical features is recognized in chronic RA, where ulnar deviation of the hands is due to the biomechanics of the MCP joints, with a net palmo–ulnar vector leading to deformity (9–14). The principal MCP joint stabilizer that counteracts this tendency is the radial collateral ligament (9, 10), with studies showing a lesser range of motion in the second and third MCP joints when compared with the fourth and fifth MCP joints, principally due to the restraining effects of these ligaments (29, 31–33). These anatomic factors contribute to the greater degree of tension that is recognized at the radial collateral ligament of the second and third MCP joints in particular (9). The radial collateral ligament's influence on synovitis and bone erosion in early RA, especially over the second and third MCP joints, is consistent with the known biomechanics of these joints. Similarly, the local biomechanical stressing related to these ligaments may also explain the distribution of bone defects over the radial aspect of normal joints, as noted in the present study.
There is a growing recognition in rheumatology that joint biomechanical factors, microtrauma, and injury may play an important role in the pathogenesis of early RA (15–17). Indeed, evidence for tissue microtrauma, as suggested by microscopic inflammation, in otherwise normal joints has been reported (34). It is likely that joint biomechanical factors exacerbate inflammation at sites of highest stressing, by undefined mechanisms. One possibility is that common molecular pathways linking the inflammatory cascade and cellular mechanical stress could be involved. For example, key proinflammatory transcription factors including nuclear factor κB, and kinases including p38 mitogen-activated protein kinase, both of which are pivotal in the inflammatory response, are also regulated by biomechanical stress (35, 36). Several other molecules share this dual property (for review, see ref. 35), suggesting that biomechanical factors could contribute to the severity of synovitis and the degree of erosion. Furthermore, the higher biomechanical forces at these sites could contribute to increased oxidative stress, thus exacerbating inflammation and damage (37).
It has long been considered that the progression of erosive disease in RA may be, to some extent, independent of synovitis, an observation that is supported by findings in clinical trials in RA, in which it has been shown that suppression of synovitis does not necessarily halt erosion (38, 39). Consequently, it has been suggested that synovitis and erosion are uncoupled in RA, with this phenomenon being tentatively ascribed to abnormal synovial fibroblasts in pannus tissue that are capable of autonomous joint damage. Although this study generally showed a correlation between synovitis and erosion, there was evidence that these processes were divergent, if not uncoupled, at certain sites. However, such differences were related to the position of the joint collateral ligaments in terms of propensity for bone damage. In particular, the predilection for erosion on the radial side was very strong in the third MCP joints whereas the difference in synovial volume between the radial and dorsal side of the same joint was negligible.
While this study generally showed a direct correlation between the severity of regional synovitis and adjacent erosion, this association was not evident for the fifth MCP joint. At that site, erosions were fairly evenly distributed between the radial and ulnar aspects of the joint but synovitis was more pronounced adjacent to the radial collateral ligament. Anatomic factors may explain this. The ulnar side of the fifth MCP joint is the only one that is not protected by an adjacent MCP joint. Furthermore, the volume of synovitis was greater in the fourth MCP compared with the fifth MCP joint but the latter joint had more erosions, suggesting that the fourth MCP joint may be protected by its anatomic location.
It therefore seems likely that the mechanisms of erosion are complex and could be affected by the direct effects of the collateral ligaments on the bone in addition to the effects of regional synovitis within joints. This assertion is supported by the finding of a predilection for bone damage on the radial aspect of the second and third MCP joints in normal subjects. These observations suggest that MCP joint inflammation in RA exaggerates the inherent tendency for joint damage adjacent to the radial collateral ligaments.
In conclusion, the predominance of both synovitis and bone damage on the radial side of the MCP joints suggests that local anatomic and biomechanical factors may be important in the pathogenesis of both synovitis and bone erosion in early RA. Therefore, the existing immunologically based models for joint damage may have to be modified. The findings of this study have important implications in the study of the pathogenic mechanics of joint damage in RA.
The authors are grateful to Amy Carey for statistical advice.