In recent years, magnetic resonance imaging (MRI) has been increasingly used to evaluate the joints of patients with rheumatoid arthritis (RA) (1, 2). Erosions, a pathognomonic finding in RA, are visualized much better by MRI than by conventional radiography, thanks to multiplanar imaging (3). However, erosions are a relatively late event in the course of RA. A major breakthrough associated with MRI is that this technique can improve appreciation of rheumatoid lesions that occur in the early phase of disease through damage to the soft tissues, such as synovium, tendons, and menisci. The synovial membrane is the principal site of inflammation in early RA. The sequence of pathologic events at this site includes penetration by and presentation of an unknown antigen, its recognition by CD4+ T cell clones, production of cytokines, adhesion molecules, proteases, neuropeptides, and nitric oxide, and, finally, neoangiogenesis (4).
An increased number of capillaries has been observed in the RA synovial membrane as compared with the normal synovial membrane. In addition, the inflammatory process enhances capillary perfusion and permeability. Increased vascularization fosters the trafficking of antigens and cells to the sites of inflammation and the production of adhesion molecules in the vessel wall. The synovial cells receive abundant nutrients and proliferate to form the rheumatoid pannus (5). These considerations indicate that evaluating the synovial vascular bed could be a valuable method to diagnose early-phase RA and follow its course.
Gadolinium–diethylenetriamine pentaacetic acid (Gd-DTPA), a paramagnetic contrast agent, is transported in the plasma as an unlinked molecule and diffuses freely to the interstitial space. It allows differentiation of the synovial membrane from the surrounding tissues through a marked increase of signal intensity on T1-weighted images. In addition, registration of the time-dependent changes in signal intensity of the synovial membrane on Gd-DTPA–contrasted images could represent a marker of synovial inflammation and disease activity.
This study was conducted to evaluate changes in signal intensity of the synovial membrane of the wrist on MRI images, after Gd-DTPA infusion, in patients with RA compared with controls. This technique has been used to study RA lesions of the knee (6–10). The wrist was chosen for the current study because it is the presenting site of RA in approximately half of patients with RA (11). An additional rationale for this study was that developing an objective method to discriminate patients in different stages of RA would represent a prerequisite for the followup of patients undergoing treatment.
- Top of page
- PATIENTS AND METHODS
Group A comprised 12 patients, group B comprised 14 patients, and group C comprised 10 patients. The demographic, clinical, and laboratory data of the 3 groups of patients are shown in Table 1. Pharmacologic treatment was not statistically different in the 3 subgroups of patients, with the exception of patients in group C, who were more frequently treated with hydroxychloroquine than were patients in group A (P = 0.05). Dynamic MRI was performed in all subjects, without any discomfort or adverse events. The mean duration of the complete examination was 15 minutes.
Figure 2 shows a typical sequence in patients with different levels of disease activity and in a control subject. Figure 3 compares the mean values of the curves in the 4 subgroups. Before Gd-DTPA infusion and at the time of infusion, the 4 curves were almost identical. However, a highly significant difference in enhancement was seen by ANOVA at t = 138 seconds (P = 0.003), t = 156 seconds (P = 0.0001), and thereafter (P < 0.00001). The curves identified the following 2 groups of patients: controls and RA patients in remission, and RA patients with active disease or disease of intermediate activity.
Figure 2. Dynamic gadolinium-diethylenetriamine pentaacetic acid (Gd-DTPA)-enhanced magnetic resonance images of the wrist. A, Patient with active rheumatoid arthritis (RA). B, RA patient with intermediate level of disease activity. C, RA patient in remission. D, Normal control. The first column (far left) shows precontrast images; the following columns (left to right) show images acquired after 90 seconds, 180 seconds, and 360 seconds, respectively. Note the different degree of synovial membrane Gd-DTPA enhancement in the 4 groups.
Download figure to PowerPoint
Figure 3. Slope of the mean enhancement curves in patients with highly active rheumatoid arthritis (RA) (▴), patients with intermediately active RA (▪), RA patients in remission (♦), and controls (×). The arrow indicates the time of gadolinium–diethylenetriamine pentaacetic acid infusion.
Download figure to PowerPoint
Figure 4 and Table 2 show the REE and RE values in patients and controls. Both the REE and the RE were significantly higher in patients with active RA than in controls or RA patients in clinical remission. In controls and RA patients in remission, the REE ranged from 0.02% per second to 0.37% per second, and from 0.09% per second to 1.32% per second, respectively. In RA patients in groups A and B, the REE ranged from 1.13% per second to 3.49% per second and from 1.08% per second to 2.79% per second, respectively. Groups A and B did not differ significantly. Similarly, groups C and D did not differ significantly in terms of REE and RE. These variables were significantly correlated with the number of swollen joints (P < 0.00001 and P = 0.003, respectively), number of tender joints (P < 0.00001 and P = 0.004, respectively), the Ritchie index (P = 0.0002 for both RE and REE), the HAQ (P = 0.0002 and P = 0.0007, respectively), the HAQ items for hand function (P < 0.00001 and P = 0.0001, respectively), the DAS (P = 0.0004 and P = 0.0008, respectively), early morning stiffness (P = 0.001 and P = 0.009, respectively), CRP (P = 0.015 and P = 0.03, respectively), ESR (P = 0.03 for RE only), and α2 globulins (P = 0.036 and P = 0.028, respectively). No correlation was demonstrated between REE or RE and type and dosage of the previously described treatment.
Figure 4. Individual values of the rate of early enhancement and relative enhancement in patients with active rheumatoid arthritis (RA) (A), patients with intermediately active RA (B), RA patients in remission (C), and normal controls (D). Triangles indicate the mean values. Vertical bars indicate standard deviations.
Download figure to PowerPoint
Table 2. Comparison of the mean ± SD rate of early enhancement (REE) and relative enhancement (RE) in patients with rheumatoid arthritis (groups A, B, and C) and controls (group D)*
| ||Group A (n = 12)||Group B (n = 14)||Group C (n = 10)||Group D (n = 5)||P|
|REE||2.1 ± 0.66||1.7 ± 0.57||0.58 ± 0.42||0.12 ± 0.15||<0.0001†|
|RE||139.9 ± 31.4||141.6 ± 39.5||66.2 ± 30.5||20.2 ± 23.6||<0.00001‡|
- Top of page
- PATIENTS AND METHODS
This study shows that dynamic MRI is a promising method for the evaluation of patients with RA. Based on our experience, it can differentiate patients with active disease from those in remission or from controls. After intravenous injection, Gd-DTPA, a relatively small sized molecule, rapidly diffuses to highly vascularized tissues, such as the inflamed synovial membrane of the knee, and reaches the synovial fluid in ∼10–15 minutes (17, 18). At this stage, the interface between synovial membrane and synovial fluid becomes blurred. For this reason, we evaluated rapidly acquired serial images within 1.5 minutes (REE) and 6 minutes (RE) of administering the intravenous injection of Gd-DTPA. The contrast agent diffusion reached its steady state after 60–120 seconds (Figure 2) and remained stable thereafter. This indicates that the number of acquisitions needed could be safely reduced, with savings in terms of the duration of the examination.
Standardization of the infusion technique is crucial, because Gd-DTPA diffusion and pharmacokinetics are influenced by several determinants. In an attempt to control variability, previous physical activity involving the wrists, diameter of the needle, size of the vein, and duration of infusion were standardized in our study. Blood pressure, cardiac rate, and intraarticular pressure due to fluid effusion are unpredictable variables that may affect results but can hardly be standardized. Apart from these considerations on system standardization, changes in diffusion of the contrast agent should be related to the number, size, and permeability of the synovial vessels, the variables investigated, as well as the volume of the synovial membrane.
Inflammation in the knee synovial membrane is often patchy in RA (19). As a result, it is theoretically necessary to study as much synovium as possible to generate a reproducible standard. In contrast, we used a small circular area of the wrist synovial membrane to measure enhancement, a protocol that ignored most of the tissue. We thought that measuring the area of highest enhancement could be sufficient for comparison between patients and the clinical and laboratory findings. In preliminary studies, we measured enhancement on predefined, analogous circular areas of interest on 3 sequentially acquired wrist slices from the same subject and found that the resulting values were highly significantly associated. The same association was observed when the small region of interest was compared with an area obtained by manual outlining of the whole synovial membrane (data not shown). These data suggest that the choice of a small region of interest, provided that it is located in the area of maximal enhancement, is both practical and reliable. A possible explanation for this observation is that synovial membrane inflammation could be more homogeneous in the wrist than in the knee.
A frequent problem was difficulty in imaging of the synovial membrane in healthy controls. In such cases, the area where the synovium was presumptively located was chosen, on the basis of anatomic landmarks and comparison of pre- and postenhancement images. For this reason, the circular region of interest was usually smaller in controls than in patients with RA.
The comparison of the REE and the RE allowed discrimination between patients with active RA and those who were in remission or controls. In contrast, within groups A and B, differences between patients with different levels of disease activity were not significant, despite the apparent gradient shown in Figure 2. This observation may indicate that our method is not useful for followup studies of patients undergoing different treatment protocols, a task requiring a high-sensitivity technique. However, we favor another explanation. In our study, the definition of disease activity was arbitrary and might therefore not have reflected the real degree of joint inflammation. This view is supported by the fact that a significant correlation was observed between the REE and the RE and several clinical and laboratory data of continuous distribution related to disease activity, as well as with the DAS. Another consideration supporting this concept is that continuous variables, such as the DAS and the HAQ score, could fully differentiate the 3 groups of patients (P < 0.00001 by ANOVA). We believe that the usefulness of dynamic MRI for future prospective studies of pharmacologic treatment of RA, which depends on its sensitivity to change, should be evaluated in properly designed clinical trials. In a previous study of the knee, dynamic MRI could measure changes in synovial membrane inflammation after intraarticular injection of methylprednisolone (20).
The REE and RE were significantly correlated with several clinical and laboratory parameters. This could be surprising, considering that we examined a single joint of patients with RA, a systemic polyarticular disease. In previous studies, the rate of synovial enhancement of the rheumatoid knee did not correlate with clinical and laboratory assessments (7, 20). Accordingly, synovial enhancement of the metacarpophalangeal joints was not correlated with the ESR, the CRP level, or the presence of IgM rheumatoid factor (21). There are 2 possible reasons for the correlation between Gd-DTPA enhancement and clinical and laboratory findings observed in our study. First, the wrist joint could be a better indicator of the situation of the whole body in RA. This observation, which has no clear explanation, has also been made in other studies. McQueen et al (22) found a significant correlation between a composite MRI index and the CRP level. Lindegaard et al (3) observed a significant positive association between MRI-evaluated synovial membrane hypertrophy of the wrist (but not of the metacarpophalangeal joints) and the CRP level. Second, we studied a relatively large group of RA patients with different degrees of disease severity, whereas most of the previous studies with dynamic MRI were performed in small groups of patients.
Another interesting finding is the correlation between the REE and RE and the HAQ. As expected, this correlation was even higher with the HAQ items related to hand function. Dynamic MRI, being correlated not only with disease activity but also with a patient's function, shows some of the characteristics of the ideal outcome measure (23).
In our patients, correlation with clinical and laboratory features was present for both the REE and RE, but was higher for the REE (except for ESR and α2 globulins). The REE was associated with enhancement during the first 1.5 minutes after Gd-DTPA administration and reflected the initial distribution of Gd-DTPA. It provided more detailed information on the inflammatory activity of the synovial membrane than did data obtained on late images, where the curves reach a plateau. Further studies will assess whether the REE and RE indicate the same or different aspects of inflammation in the arthritic joint.
This study did not investigate whether dynamic MRI of the wrist could also reflect microscopic changes of the joint. In contrast, results of dynamic MRI of the rheumatoid knee have been compared with the histology of the corresponding synovial membrane. In one study, dynamic MRI was able to differentiate between fibrous, slightly hypervascular, and hypervascular pannus (7). Tamai et al showed that specimens with a high degree of inflammation had significantly more enhancement than those with minimal changes (10). In particular, significantly higher RE was observed in specimens with high proliferation of blood vessels, a fact that may indicate that dynamic MRI has the potential to reflect angiogenesis. In animal studies, angiogenesis has been shown to be a better predictor of the development of cartilage erosions than is joint swelling (24).
In conclusion, our data support the use of dynamic MRI for discriminating active from inactive RA. This technique is easy to perform and devoid of associated adverse events. It can be repeated frequently and is an excellent candidate for the ideal method for long-term followup of patients with RA. Enhancement curves are associated not only with laboratory and clinical indicators of inflammation, but also with the HAQ, a functional index that is one of the best predictors of outcome in RA. These results were reproducible, although they were obtained with a 0.2T magnet, which entails a lower signal-to-noise ratio and more random fluctuation of the baseline signal compared with high-field magnets. We believe that results of at least comparable value should be expected with other MRI systems equipped with higher-field magnets.