New radiographic bone erosions in the wrists of patients with rheumatoid arthritis are detectable with magnetic resonance imaging a median of two years earlier

Authors


Abstract

Objective

In a 5-year followup study, we investigated the temporal relationship between development of wrist joint erosions as visualized by magnetic resonance imaging (MRI) versus conventional radiography (CR), in patients with rheumatoid arthritis. We also evaluated the risk of erosive progression on CR associated with the presence of MRI erosions.

Methods

In 10 patients with rheumatoid arthritis, MRI and CR of the dominant wrist were performed annually for 5 years. In each image set, each wrist bone (metacarpal bases, carpal bones, radius, and ulna) was assessed for the absence or presence of bone erosions.

Results

Nine bones showed radiographic erosions at baseline. Twenty-seven new radiographic erosions developed during the 5-year followup period. Of these 27 new erosions, 21 were detected 1–5 years earlier by MRI than by CR, 3 were simultaneously detected by both methods, 2 were detected 1–2 years later by MRI than by CR, and 1 erosion (radiographically detected at 5-year followup) was not visualized with MRI. MRI detection of new radiographic erosions preceded CR detection by a median of 2 years. In bones with MRI erosions at baseline, the relative risk of radiographic erosions at 5-year followup was 4.5 (95% confidence interval [95% CI] 2.6–7.6), compared with bones without baseline MRI erosions. If bones with baseline radiographic erosions were excluded from the analysis, the relative risk was 4.1 (95% CI 2.2–7.5).

Conclusion

Most new radiographic bone erosions (78%) were visualized at least 1 year earlier by MRI than by CR. This illustrates that the information on joint destruction provided by CR is considerably delayed compared with that provided by MRI. A significantly increased risk of progression of radiographic erosion in bones with baseline MRI erosions was observed, demonstrating a prognostic value of MRI with respect to long-term radiographic outcome.

Conventional radiography (CR) is the standard outcome measure of structural joint damage in rheumatoid arthritis (RA), in clinical trials as well as in clinical practice (1–3). Furthermore, the presence of radiographic bone erosions is one of the diagnostic criteria for RA (4). The modern treatment strategy in RA involves use of aggressive therapy to suppress inflammation and minimize joint destruction, followed by prompt therapy adjustments if it is recognized that these treatment goals are not being achieved (1, 5). Optimal implementation of this strategy requires sensitive methods for diagnosis and monitoring of the disease, as well as for determining the prognosis.

Several studies of patients with RA have shown that magnetic resonance imaging (MRI) detects more bone erosions than does CR (6–8). Due to the lack of long-term MRI followup studies, however, the time advantage offered by MRI compared with CR is unknown. In addition, the predictive value of MRI findings with respect to long-term progression of radiographic erosions has not been established.

The aim of the present 5-year followup study was to investigate the temporal relationship between development of wrist joint erosions in patients with RA as visualized by CR versus MRI. Furthermore, we wanted to assess the risk of erosive progression on radiographs associated with the presence of erosions on MRI.

PATIENTS AND METHODS

Ten patients fulfilling the American College of Rheumatology (formerly, the American Rheumatism Association) 1987 classification criteria for RA (4), ranging in age from 20 years to 76 years (median age 62 years) and with disease duration of 0.5–9 years (median 1.5 years), were followed up for 5 years. MRI and CR of the dominant wrist were performed every year during the 5-year followup period. All patients were receiving disease-modifying antirheumatic drugs (DMARDs) at baseline (7 received sulfasalazine, 2 received methotrexate, and 1 received penicillamine) and 7 patients were receiving DMARDs at 5-year followup (2 received sulfasalazine, 3 received methotrexate, 1 received penicillamine, and 1 received myocrisin). No patient received biologic therapy.

For MRI, continuous coronal and axial T1-weighted spin-echo images (repetition time 480–600 msec, time to echo 15–17 msec, slice thickness 3 mm) were obtained before and after intravenous injection of contrast agent (gadolinium diethylenetriaminepentaacetic acid [Schering, Berlin, Germany], 0.05 mmoles per kg body weight). Radiographs were obtained in the posteroanterior view. In each image set, each wrist bone (metacarpal bases, carpal bones, distal radius, distal ulna) was assessed for the absence or presence of bone erosions. To avoid overestimation of MRI findings and in accordance with international recommendations (9, 10), MRI bone erosions were required to be visible in 2 planes, with a cortical break seen in at least 1 plane, in order to be registered. For the same reason, erosions were evaluated as the absence or presence of erosion in each bone (i.e., erosion enlargement or development of more erosions in a previously eroded bone was not taken into account). The observers were blinded to results obtained with the other imaging technique. The relative risk for radiographic erosions at 5-year followup in bones without versus bones with baseline MRI erosions was calculated according to the method described by Altman (11).

RESULTS

Of the 150 wrist bones examined, 9 had radiographic erosions at baseline. During the 5-year followup period, new radiographic erosions developed in 27 bones. In these 27 bones, 21 of the new radiographic erosions (78%) were detected 1–5 years earlier by MRI than by CR, 3 were visualized at the same time by both methods, 2 were detected 1–2 years later by MRI than by radiography, and 1 erosion (radiographically detected at 5-year followup) was not visualized by MRI (Table 1). As a whole, MRI detection of bone erosions preceded visualization with CR by a median of 2 years.

Table 1. Timing of detection of new radiographic bone erosions by MRI compared with conventional radiography*
 New radiographic erosions
  • *

    Values are the number (%) of erosions. MRI = magnetic resonance imaging.

Detection by MRI compared with conventional radiography 
 5 years earlier1 (4)
 4 years earlier3 (11)
 3 years earlier4 (15)
 2 years earlier6 (22)
 1 year earlier7 (26)
 Simultaneous3 (11)
 1 year later1 (4)
 2 years later1 (4)
 Not detected1 (4)
Total27 (100)

During the 5-year followup period, new MRI erosions developed in 47 bones that did not have erosions at baseline. The total numbers of bones with erosions detected by MRI and radiography, respectively, at the various time points were as follows: at baseline, 37 and 9 (CR detected 24% of the erosions detected by MRI); at 1 year, 51 and 15 (29%); at 2 years, 65 and 22 (34%); at 3 years, 72 and 27 (38%); at 4 years, 79 and 29 (37%); and at 5 years, 83 and 36 (43%). Of the 37 erosions visualized on MRI at baseline, 19 (51%) were visualized by CR 5 years later.

For the bones in which erosions were visualized with MRI at baseline, the relative risk of having radiographic bone erosions at 5-year followup was 4.5 (95% confidence interval [95% CI] 2.6–7.6), compared with the bones without baseline MRI erosions. When the bones with baseline radiographic erosions were excluded from the analysis, the relative risk was 4.1 (95% CI 2.2–7.5).

DISCUSSION

In the present study, the majority (78%) of new radiographic bone erosions were visualized at least 1 year earlier with MRI than with CR. This illustrates that the information on joint destruction provided by CR is associated with a considerable delay. In accordance with results of earlier studies (6–8, 12), we observed that the sensitivity of MRI with respect to visualization of bone erosions was superior to that of CR. Furthermore, MRI detected erosive progression in more bones than did CR (i.e., MRI was more sensitive to change).

A >4-fold increased risk of progression of radiographic erosion was observed in bones in which erosions were visualized by MRI at baseline compared with bones without MRI erosions at baseline. The increased risk persisted when bones with radiographic erosions at baseline were excluded from the analysis (Table 2). This demonstrates a significant prognostic value of MRI with respect to long-term radiographic erosive damage.

Table 2. Relationship between bone erosions detected by magnetic resonance imaging (MRI) at baseline and by conventional radiography (CR) at 5-year followup*
CR at 5 yearsMRI at baseline
Bones without erosionBones with erosion
  • *

    Values are the number of bones when bones with radiographic erosions at baseline are excluded; values in parentheses are the number of bones when bones with radiographic erosions at baseline are included. The relative risk of radiographic erosions at 5 years in bones with baseline MRI erosions versus bones without baseline MRI erosions, when bones with baseline radiographic erosions are excluded ([16/31]/[14/112]), is 4.1 (95% confidence interval 2.2–7.5). The relative risk of radiographic erosions at 5-year followup in bones with baseline MRI erosions versus bones without baseline MRI erosions, when bones with baseline radiographic erosions are included ([22/37]/[15/113]) is 4.5 (95% confidence interval 2.6–7.6).

Bones without erosion98 (98)15 (15)
Bones with erosion14 (15)16 (22)

Approximately half of the erosions visualized by MRI at baseline did not become detectable with CR during the 5-year followup period. Because MRI erosions are not invariably followed by development of radiographic erosions, and because the MRI signal in erosions reflects tissue characteristics different from the bone calcium loss visualized by radiography, it is relevant to discuss whether MRI erosions are true representations of rheumatoid bone destruction. However, in a given time period, not all radiographic erosions will increase in size. Consequently, not all MRI erosions will obviously become so much larger that they become detectable by radiography. In fact, during optimal treatment, it should be expected that erosions do not enlarge and, consequently, that they will never become detectable on radiographs. Furthermore, the markedly increased risk of progression of radiographic erosion associated with the presence of MRI erosions, which was demonstrated in this study, proves the close relationship between erosions detected by MRI and those detected by radiography.

Even though verification of the relationship between MRI and CR (the standard measure of structural joint damage) is very important, it should be remembered that CR is not an optimal “gold standard” outcome measure of joint damage. In early RA, radiographic status is not related to functional outcome measures such as the Health Assessment Questionnaire (13) score, and in established disease (duration 5–8 years) changes visualized by radiography explain only ∼25% of disability in patients with established RA (14). It appears likely that the more detailed information provided by MRI could capture disease manifestations responsible for a higher proportion of patient disability.

Nevertheless, validation of the impact of MRI erosions should, obviously, be encouraged. Relevant research topics include examination of the occurrence of MRI bone erosions in other diseases and in healthy controls, development and validation of evaluation systems, and comparison with reliable references (e.g., histopathology, miniarthroscopy, and computed tomography). Although preliminary data obtained by ultrasonography-guided biopsies (15) and miniarthroscopy (16) indicated that MRI erosions represent true bone damage, more data are warranted. Large followup studies are needed to confirm the prognostic value of MRI bone erosions in early RA and in early persistent polyarthritis.

In comparison with CR, which is the traditional imaging modality, MRI offers the ability to assess destructive bone damage with improved sensitivities to both early pathology and change. The superior sensitivity of MRI can be expected to be of major significance in both clinical trials and the clinical management of patients with RA. In trials, using MRI may allow reductions in sample size and study length, due to more sensitive separation of responders from nonresponders. In clinical practice, treating physicians who use MRI may make quicker, more appropriate decisions due to the availability of more accurate information.

In conclusion, this long-term followup study is the first to demonstrate that bone erosions in patients with RA often are detected years earlier by MRI than by radiography, and that the presence of MRI bone erosions is significantly related to a poor long-term radiographic outcome.

Ancillary