Power Doppler ultrasound, but not low-field magnetic resonance imaging, predicts relapse and radiographic disease progression in rheumatoid arthritis patients with low levels of disease activity

Authors


Abstract

Objective

Subclinical inflammation and radiographic progression have been described in rheumatoid arthritis (RA) patients whose disease is in remission or is showing a low level of activity. The aim of this study was to compare the ability of ultrasonography and magnetic resonance imaging (MRI) to predict relapse and radiographic progression in these patients.

Methods

Patients with RA of short or intermediate duration that was either in remission or exhibiting low levels of activity according to the Disease Activity Score (DAS) were included in the study. Over a period of 1 year, patients underwent clinical and biologic assessments every 3 months and radiographic assessments at baseline and 12 months. Radiographs were graded according to the modified Sharp/van der Heijde score (SHS). At baseline, patients underwent ultrasonography and MRI, which were graded using binary and semiquantitative scoring systems. Relapse was defined as a DAS of ≥2.4, and radiographic progression was defined as an increase in the SHS of ≥1. We tested the association of values by multivariate logistic regression.

Results

A total of 85 RA patients with a mean disease duration of 35.3 months were studied. RA was in remission in 47 of these patients, and 38 had low levels of disease activity. At 1 year, 26 of the 85 patients (30.6%) showed disease relapse, and 9 of the 85 patients (10.6%) showed radiographic progression. The baseline PD synovitis count (i.e., the number of joints at baseline for which the power Doppler [PD] signal indicated synovitis) predicted relapse (adjusted odds ratio [OR] 6.3; 95% confidence interval [95% CI] 2.0–20.3), and the baseline PD synovitis grade predicted disease progression (adjusted OR 1.4 [95% CI 1.1–1.9]). MRI was not predictive of outcomes.

Conclusion

For RA patients whose disease is in remission or who have low levels of disease activity, PD signals on ultrasonography could predict relapse or radiographic progression and identify those whose disease is adequately controlled, which is especially helpful when considering treatment tapering or interruption.

For patients with rheumatoid arthritis (RA), therapeutic objectives are remission or a low level of disease activity and no disease evolution (1–3). However, the concept of remission remains complex, and several definitions have been proposed, but no consensus has been reached (4–14).

Previous studies have reported on the possibility of radiographic evidence of disease progression in RA patients considered to be in clinical disease remission or with low levels of disease activity (15–19). Moreover, research has suggested the presence of subclinical residual inflammation in patients whose RA is in remission or is showing a low level of activity. This inflammation could explain the apparent discrepancy between clinical and structural evidence of disease evolution.

Ultrasonography (US) and magnetic resonance imaging (MRI) are considered superior to clinical examination and radiographic assessment of arthritis because of their direct, valid, sensitive, and specific visualization and objective assessment of inflammatory lesions and structural damage (20–22). Indeed, Brown et al (15) found that patients with disease considered to be in remission according to clinicobiologic indices showed persistent inflammation (synovitis, tenosynovitis, bone marrow edema) on US and MRI. So, these techniques can more accurately detect and measure RA activity and structural progression than can the established methods, particularly in patients with subclinical activity.

Because patients with disease in remission or with low levels of disease activity might show residual subclinical inflammation, we wondered whether such findings are predictive of future relapse or progression of structural damage. In the present study, we compared US and MRI findings in a prospective longitudinal evaluation of patients with RA in remission or with low levels of disease activity that was stable (i.e., for at least 2 months) according to the Disease Activity Score (DAS), whatever the treatment. We aimed to determine whether subclinical activity observed by US or MRI could predict disease relapse and/or radiographic evidence of disease progression in such patients.

PATIENTS AND METHODS

Study design.

This was a longitudinal observational study in which we prospectively followed a cohort of patients with RA whose disease was in remission or showing low levels of activity over 1 year to assess the ability of US and MRI to predict risk of relapse and radiographic evidence of disease progression at 1 year.

Ethical approval for the project was obtained from the ethics committee of the Hôpital Pitié Salpêtrière. Written informed consent was obtained from all patients.

Patient population.

To be included in the study, patients had to have established RA, which was defined according to the American College of Rheumatology (ACR) 1987 criteria (23), that was of short or intermediate duration (i.e., <6 years), and was in remission or exhibiting stable, low levels of activity according to the DAS (i.e., DAS <2.4 at 2-month measurements) (4). RA treatment needed to be stable for at least 2 months, and synthetic or biologic disease-modifying antirheumatic drugs (DMARDs) or low-dose corticosteroids (<10 mg/day) were allowed.

Patients were excluded if they had a contraindication to gadolinium or MRI, a DAS >2.4, a modification of DMARD therapy in the previous 2 months, or another significant comorbid condition that could interfere with the protocol.

Data collection.

At baseline and at 12 months, patients underwent complete clinical, biologic, and radiographic assessments. US and MRI were performed at baseline. In addition, at 3, 6, and 12 months, patients underwent clinicobiologic assessments.

Assessment of demographic and clinicobiologic characteristics.

The following demographic data were recorded at baseline: sex, age, medical history, and disease duration. Data on clinical and quality of life variables were collected, including the duration of morning stiffness, patient's assessment of pain by visual analog scale, patient's global assessment of health and disease activity by visual analog scale, count of 44 joints for tenderness and swelling (performed by a rheumatologist [BF] who was blinded with regard to current treatment and imaging findings), current treatment, and Health Assessment Questionnaire (HAQ) score (24). Laboratory assessments included a complete blood cell count, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, and serologic assessments for rheumatoid factor (RF) and anti–citrullinated protein antibody (ACPA).

Radiographic assessment.

All patients underwent posteroanterior radiography of the hands, wrists, and feet. Radiographic changes were graded anonymously by 2 trained readers (VF and FG), according to the modified Sharp/van der Heijde score (SHS) (25). Intraobserver and interobserver reliability was considered moderate to good; a preliminary exercise provided the smallest detectable change for defining structural disease progression (see Appendix A).

Ultrasonographic assessment.

All patients underwent US assessment of the wrists, metacarpophalangeal (MCP) joints 2, 3, and 5, and metatarsophalangeal (MTP) joints 2, 3, and 5. Each joint region was investigated in dorsal, ventral, and, when possible, lateral views in grayscale (GS) and power Doppler (PD) modes using a 7.5–13-MHz linear transducer (Esaote Technos).

The presence of synovitis in GS and PD modes and the presence of erosions were defined by the Outcome Measures in Rheumatology (OMERACT) guidelines (26). US was performed independently by 1 of 2 trained rheumatologists who were experienced in joint US (FE and CR) and were blinded with regard to the other data. Good interreader reliability was found (27).

The presence and location of synovitis in GS and PD modes, as well as the presence of erosions, were evaluated. Synovitis was scored using a binary scale (0 = absence/1 = presence; maximum score 14). Synovial blood flow in each joint was assessed in PD mode and was scored using a binary scale (0 = absence/1 = presence of PD signal) and the semiquantitative scoring system proposed by Szkudlarek et al (21, 28) (0–3 scale, where 0 = absent, no vascularization; 1 = mild, single-vessel signal or isolated signal; 2 = moderate, confluent vessels; and 3 = marked, vessel signals in more than half of the intraarticular area [maximum score 42]). The sum of the synovitis and PD scores was calculated for each patient. The number of erosions at the 30 locations examined were scored using a binary scale (0 = absence/1 = presence of 1 erosion; some sites had a score >1 because of the presence of several erosions).

MRI assessment.

All patients underwent low-field dedicated MRI (0.2T C-Scan; Esaote) of the dominant hand, exploring the wrist and MCP joints 2–5. Sequences were chosen according to OMERACT guidelines: 3-dimensional T1-weighted sequences in axial and coronal planes before and after intravenous administration of gadolinium and STIR sequences in axial and coronal planes. Synovitis, erosions, and bone marrow edema were scored according to the OMERACT RA MRI Scoring (RAMRIS) system (29–31). Synovitis was scored on a 4-point semiquantitative scale (0 = normal, 1 = mild, 2 = moderate, and 3 = severe), and a synovitis count was obtained. Bone marrow edema was graded on a 4-point scale (0 = no edema, 1 = 1–33% edema, 2 = 34–66% edema, and 3 = 67–100% edema). Erosions were graded on a 10-point scale, where each point represents a 10% loss of bone volume. Images were interpreted by 2 experienced readers (VF and FG), who were blinded with regard to all data, but were aware of the chronological sequence of the images (32). Intrareader and interreader reliability was assessed 3 months after the end of followup. Intraobserver reliability kappa values were 0.81, 0.82, and 0.90 for the RAMRIS semiquantitative scoring of synovitis, bone marrow edema, and erosions, respectively; interobserver kappa values were 0.69, 0.68, and 0.84, respectively.

Outcome measures.

Because of lack of a consensus definition of relapse or flare (33), we used a composite definition to identify disease relapse, as follows: 1) a DAS of >2.4 at one or more followup visits; and/or 2) a dosage increase or change in synthetic or biologic DMARDs because of exacerbation of RA activity (loss of efficacy), excluding change because of safety; and/or 3) an increase in the corticosteroid dosage >10 mg/day.

Radiographic evidence of disease progression was defined as an increase in the SHS greater than the smallest detectable change between baseline and 1 year (i.e., ΔSHS ≥1 degree).

Statistical analysis.

The sample size calculation was based on the ability of the regression model to discriminate between progression and nonprogression of disease by receiver operating characteristic (ROC) curve analysis. Assuming that 30% of patients would show radiographic evidence of disease progression (34), the inclusion of 84 patients would produce an area under the ROC curve of 0.75 with a P value of less than 0.05.

Univariate analysis involved Mann-Whitney tests for continuous variables and chi-square or Fisher's exact tests for categorical variables. Independent predictors of relapse and radiographic evidence of disease progression were identified by forward multivariate logistic regression. Variables with P values of less than 0.20 on univariate analysis were included in the model. P values less than 0.05 were considered statistically significant. All analyses were performed with the use of SAS v8.2 software (SAS Institute).

RESULTS

Patient characteristics at baseline.

We included 85 patients in the study (Figure 1). The main demographic, clinical, and laboratory data obtained at baseline are shown in Table 1. A total of 30 patients (35.3%) were receiving low-dose steroids (<10 mg/day), 82 (96.5%) were receiving conventional DMARD monotherapy, with methotrexate being the most common agent, and 17 (20%) were receiving combination therapy with biologic agents (15 taking tumor necrosis factor blockers, 1 taking anti–interleukin-6 receptor, and 1 taking rituximab). Three of the patients whose disease was in remission had not received DMARDs for several months. As expected, we found low levels of disease activity, with low values for the patient's global assessment of health and disease activity, the joint counts, the DAS, the ESR, and the CRP level. In addition, because of the low levels of disease activity and the short disease duration, the HAQ score was also low (mean ± SD 0.27 ± 0.33), which corresponded to no or mild disability, except for 2 patients, whose HAQ scores were >1.

Figure 1.

Flow chart showing the distribution of study patients throughout the trial.

Table 1. Baseline demographic, clinical, and laboratory characteristics of patients with RA in remission or expressing low levels of activity*
CharacteristicAll RA patients (n = 85)Patients with low levels of RA activity (n = 38)Patients with RA in remission (n = 47)
  • *

    RA = rheumatoid arthritis; VAS = visual analog scale (0–100 mm); DAS = Disease Activity Score; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; RF = rheumatoid factor; ACPA = anti–citrullinated protein antibody; SHS = modified Sharp/van der Heijde score; GS = grayscale image; PD = power Doppler; MRI = magnetic resonance imaging.

  • P < 0.01 versus patients with low levels of disease activity.

Age, mean ± SD years50.7 ± 13.549.1 ± 14.151.9 ± 13.1
No. (%) female69 (81.2)33 (86.8)36 (76.6)
Duration of RA, mean ± SD months35.3 ± 20.740.1 ± 24.431.4 ± 16.5
Patient's global assessment, by VAS, mean ± SD mm14.2 ± 13.919.1 ± 14.910.2 ± 11.7
Joint counts   
 No. of tender joints1.5 ± 1.82.8 ± 1.80.5 ± 0.9
 No. of swollen joints2.7 ± 1.93.1 ± 1.82.4 ± 1.9
DAS, mean ± SD (range 0–3)1.5 ± 0.542.02 ± 0.21.09 ± 0.3
Laboratory   
 ESR, mean ± SD mm/hour13.6 ± 12.116.1 ± 12.311.7 ± 11.7
 CRP, mean ± SD mg/liter5.5 ± 5.35.7 ± 5.55.33 ± 5.2
 No. (%) RF positive54 (63.5)
 No. (%) ACPA positive48 (56.5)24 (63.2)24 (51.1)
Radiography   
 No. (%) with erosions77 (90.6)35 (92.1)42 (89.4)
 No. (%) with SHS ≥185 (100)
Ultrasonography   
 No. (%) with GS-positive synovitis74 (87.1)33 (86.8)41 (87.2)
 No. (%) with PD-positive synovitis24 (28.2)17 (44.7)7 (14.9)
  Mean ± SD count per patient0.6 ± 1.31.0 ± 1.60.4 ± 1.0
  Mean ± SD grade per patient1.0 ± 2.01.4 ± 2.30.6 ± 1.6
  No. (%) with erosions68 (80.0)30 (78.9)38 (80.9)
MRI   
No. (%) with synovitis82 (96.5)37 (97.4)45 (95.7)
 No. (%) with bone marrow edema27 (31.8)13 (34.2)14 (29.8)
 No. (%) with erosions85 (100)38 (100)47 (100)

Evidence of residual inflammation.

Baseline US findings revealed 74 patients (87.1%) with GS-positive synovitis and 24 (28.2%) with PD-positive synovitis (Table 1 and Figure 2). GS-positive synovitis was predominantly observed at the wrist, MCP2, MTP2, and MTP3 joints (39.4%, 27.1%, 40.6%, and 40.9%, respectively). PD-positive synovitis was observed at the wrist (17.2% of patients) and equally at each MTP joint. The mean ± SD GS-positive synovitis count per patient was 3.4 ± 2.6, the mean ± SD PD-positive synovitis count was 0.6 ± 1.3, and the mean ± SD PD-positive grade was 1.0 ± 2.0. A total of 68 patients (80.0%) had at least 1 erosion, but the mean ± SD number of erosions per patient was low at 2.6 ± 2.6.

Figure 2.

Ultrasound evidence of synovitis at baseline. Synovitis is present in 2 different metacarpophalangeal joints, one with no power Doppler signal (A) and the other hypervascularized (i.e., positive power Doppler signal) (B).

Baseline MRI findings revealed 82 patients (96.5%) with synovitis. Synovitis was predominantly observed at the wrist, MCP2, and MCP3 joints (76.5%, 21.2%, and 22.4%, respectively). The mean ± SD synovitis count per patient was 4.5 ± 2.0 (maximum 7), and the mean ± SD synovitis grade was 5.6 ± 3.9. A total of 27 patients (31.8%) had at least 1 case of bone marrow edema, with a mean ± SD of 1.0 ± 2.6 bone marrow edemas (maximum 23) and a mean ± SD grade of 1.8 ± 4.4 (maximum 69) (Appendix B). Bone marrow edemas were predominantly observed at the wrist (21.2%) and MCP5 (10.6%) joints (Figure 3). MRI revealed erosions in all patients, with a mean ± SD number of 7.2 ± 3.7 erosions and a mean ± SD grade of 11.4 ± 7.9 erosions. The maximum inflammation (synovitis or bone marrow edema) was noted at the wrist at locations that were less accessible and less efficient for US evaluation (for synovitis, wrist 76.4% versus MCP joints from 9.4% to 21.2%; for bone marrow edema, wrist 21.2% versus MCP joints 0%).

Figure 3.

Magnetic resonance imaging evidence of structural progression of rheumatoid arthritis at baseline. Bone marrow edema appears as a hyposignal of the trapezium (white arrow) on the T1-weighted sequence image (A) and as a hypersignal of both the trapezium and the first metacarpal bone (black arrows) on the STIR sequence image (B).

Comparison of patients whose RA was in remission with those whose RA showed low levels of activity at baseline.

A subanalysis of patients with disease in remission or with low levels of activity according to the DAS at baseline (Table 1) revealed 38 patients with low levels of activity and 47 with disease in remission. The values for the clinical variables used for the DAS calculation were lower in patients whose RA was in remission than in patients with low levels of RA activity, and those with low levels of RA activity more often had at least 1 PD-positive case of synovitis (44.7% versus 14.9%; P = 0.005), a higher mean PD-positive synovitis count (1.0 ± 1.6 versus 0.4 ± 1.0; P = 0.01), and grade (1.4 ± 2.3 versus 0.6 ± 1.6; P = 0.01).

Relapse during followup.

Data concerning relapse were available for 81 patients (95.3%) (Figure 1). At 12 months, 26 patients (32.1%) experienced disease relapse: treatment had increased in 7 patients (i.e., dosage increase or change in synthetic or biologic DMARDs because of RA exacerbation), the DAS had increased to ≥2.4 in 15 patients (3 at 3 months, 4 at 6 months, and 8 at 12 months), and both the treatment and the DAS had increased in 4 patients (3 at 6 months) (Table 2). Eleven of these 26 patients had been in remission at baseline (2 receiving biotherapies) and 15 showed low levels of activity (4 receiving biotherapies); 3 of these 26 patients had radiographic evidence of disease progression at baseline.

Table 2. Baseline demographic, clinical, and laboratory characteristics of the RA patients, according to relapse or no relapse at 1 year*
CharacteristicRelapse (n = 26)No relapse (n = 55)
  • *

    RA = rheumatoid arthritis; DAS = Disease Activity Score; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; RF = rheumatoid factor; ACPA = anti–citrullinated protein antibody; GS = grayscale image; PD = power Doppler; MRI = magnetic resonance imaging.

  • P = 0.02 versus patients who experienced relapse.

  • P = 0.004 versus patients who experienced relapse.

  • §

    P = 0.005 versus patients who experienced relapse.

  • P = 0.03 versus patients who experienced relapse.

Age, mean ± SD years53.7 ± 12.750.0 ± 13.1
Duration of RA, mean ± SD months44.1 ± 27.132.5 ± 15.9
Treatment, no. (%)  
 Biologic agents6 (23.1)10 (18.2)
 Oral corticosteroids12 (46.2)15 (27.2)
DAS, mean ± SD (range 0–3)1.7 ± 0.61.5 ± 0.5
Laboratory  
 ESR, mean ± SD mm/hour15.5 ± 10.412.3 ± 12.6
 CRP, mean ± SD mg/liter5.5 ± 4.15.6 ± 6.0
 No. (%) RF positive20 (76.9)38 (69.1)
 No. (%) ACPA positive25 (96.2)19 (34.6)
Ultrasonography, mean ± SD per patient  
 GS-positive synovitis count3.6 ± 2.13.3 ± 2.8
 PD-positive synovitis  
  Count1.3 ± 1.80.4 ± 0.9
  Grade1.9 ± 2.70.6 ± 1.4§
 Erosion count2.3 ± 2.12.8 ± 2.9
MRI, mean ± SD per patient  
 Synovitis  
  Count4.8 ± 1.84.3 ± 2.1
  Grade8.4 ± 4.46.0 ± 4.4
 Bone marrow edema grade3.5 ± 8.42.0 ± 4.8
 Erosions grade12.2 ± 10.89.6 ± 7.1

Patients with relapse differed from those without relapse in terms of ACPA positivity (96.2% versus 35.2%; P = 0.02), the number of joints at baseline where a positive PD signal indicated synovitis (mean ± SD 1.3 ± 1.8 versus 0.4 ± 0.9; P = 0.004), the grade of joints at baseline where a positive PD signal indicated synovitis (mean ± SD 1.9 ± 2.7 versus 0.6 ± 1.4; P = 0.005), and the MRI-positive synovitis grade (8.4 ± 4.4 versus 6.0 ± 4.4; P = 0.03).

Multivariate analysis of variables found to be significant on univariate analysis (duration of RA, use of corticosteroids, the DAS, the ESR, RF and ACPA positivity, baseline PD-positive synovitis count and grade, and MRI-positive synovitis grade) revealed a significant association between only the baseline PD-positive synovitis count and relapse (adjusted odds ratio [OR] 6.3; 95% confidence interval [95% CI] 2.0–20.3).

Radiographic evidence of disease progression at 1 year.

Data concerning radiographic evidence of disease progression were available for 80 of the 85 patients (94%) (Figure 1). Among the 9 patients (11.3%) with radiographic evidence of disease progression at 1 year, at baseline 4 (5% of the entire population) had been in remission (P = 0.73) and 5 (6.2% of the entire population) had low levels of disease activity; 3 showed disease relapse. Only the mean GS-positive synovitis count per patient was significantly higher in patients with than in those without radiographic evidence of disease progression (4.8 ± 2.3 versus 3.2 ± 2.6; P = 0.04) (Table 3).

Table 3. Baseline demographic, clinical, and laboratory characteristics of the RA patients, according to radiographic evidence of disease progression at 1 year*
ChacteristicProgression (n = 9)No progression (n = 71)
  • *

    RA = rheumatoid arthritis; DAS = Disease Activity Score; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; RF = rheumatoid factor; ACPA = anti–citrullinated protein antibody; GS = grayscale image; PD = power Doppler; MRI = magnetic resonance imaging.

  • P = 0.04 versus patients who experienced progression.

Age, mean ± SD years51.9 ± 10.351.0 ± 13.5
Duration of RA, mean ± SD months29.4 ± 17.236.5 ± 20.6
Treatment, no. (%)  
 Biologic agents0 (0)15 (21.1)
 Oral corticosteroids3 (33.3)24 (33.8)
DAS, mean ± SD (range 0–3)1.6 ± 0.31.5 ± 0.6
Laboratory  
 ESR, mean ± SD mm/hour12.8 ± 7.113.4 ± 12.4
 CRP, mean ± SD mg/liter5.3 ± 2.45.6 ± 5.6
 No. (%) RF positive6 (66.7)51 (71.8)
 No. (%) ACPA positive6 (66.7)37 (52.1)
Ultrasonography, mean ± SD per patient  
 GS-positive synovitis count4.8 ± 2.33.2 ± 2.6
 PD-positive synovitis  
  Count1.8 ± 2.80.5 ± 0.9
  Grade2.8 ± 4.00.8 ± 1.5
 Erosion count3.0 ± 3.02.6 ± 2.6
MRI, mean ± SD per patient  
 Synovitis  
  Count5.0 ± 1.54.4 ± 2.1
  Grade8.1 ± 3.16.5 ± 4.7
 Bone marrow edema grade0.9 ± 2.02.5 ± 6.4
 Erosions grade12.1 ± 10.59.7 ± 6.7

Multivariate analysis of variables found to be significant on univariate analysis (therapy with biologic agents, GS-positive and baseline PD-positive synovitis count, baseline PD-positive synovitis grade, and MRI-positive synovitis grade) revealed a significant association between only the baseline PD-positive synovitis grade and radiographic evidence of disease progression (OR 1.4 [95% CI 1.1–1.9]).

DISCUSSION

Our study demonstrated that in a sample of 85 patients who were followed up prospectively over a period of 1 year, the baseline PD-positive synovitis count and grade were predictive of disease relapse and radiographic evidence of disease progression. In the same sample, MRI had no predictive value. To our knowledge, this is the first longitudinal study over 1 year conducted in RA patients with low levels of disease activity or remission according to validated tools, with homogenous disease duration and followup with classic clinicobiologic data as well as imaging data.

The main strength and originality of our study are the identification of factors that predict relapse. Previous studies have shown that up to 52% of patients with disease in remission experience disease exacerbation within 24 months, which suggests that residual inflammation may persist even in the presence of clinically undetectable disease activity (17). Nevertheless, no predictive factors for relapse have been identified until recently (35). In that study of RA patients with mean disease duration of 3.8 months, relapse was defined as a DAS ≥1.6 independently of potential modification of treatment: 43 patients achieved remission after 18 months, and 14 showed disease relapse during the next 6 months. PD-positive synovitis (at least at 1 site) was the main predictor of disease relapse, even after adjustment for steroid medication (dosage not reported). PD positivity therefore had a sensitivity of 85.7% and a specificity of 82.8% for detecting early relapse and a positive and negative predictive value of 70.6% and 92.3%, respectively.

Our study corroborates those results, showing a rate of relapse of ∼31% at 1 year and a positive predictive value of baseline PD-positive synovitis for disease relapse. These results underscore the sensitivity of PD analysis even at 1 joint and demonstrate that persistent PD-positive synovitis is a good predictor of unstable remission. US could thus be a useful examination to perform before reducing therapy in patients with adequately controlled RA.

Our results revealed that a high percentage of patients with RA in remission or with low levels of RA activity had subclinical inflammation in asymptomatic joints detected by US (87.1%) and MRI (96.5%). Moreover, we revealed PD-positive synovitis and bone marrow edema in more than one-fourth of the patients (28.2% and 31.8%, respectively). Indeed, these 2 lesions are considered to be “aggressive” for bone and increase the risk of future disease progression as seen on radiography (36, 37). The persistence of such subclinical inflammation observed in our study and in other studies (15, 17, 18) could explain why we found patients whose RA was in remission or exhibited low levels of activity to have radiographic evidence of disease progression. Radiographic evidence of disease progression could also be explained by the weak stringency of the indices used in practice (e.g., the DAS, the Clinical Disease Activity Index) in terms of cutoff values and their lack of stability over time (6, 13, 38). Whatever the index that is used, it reflects a transverse assessment state, rather than real disease activity, over a long period, which introduces the notion of a fluctuating disease activity state (39, 40).

Recent studies by Brown et al (15, 16) showed results similar to ours. Nevertheless, comparisons are difficult because the criteria for remission used in the previous studies were different, less stringent, and less precise: remission was based on the physician's judgment. Indeed, among the 102 patients included, the ACR and DAS 28-joint assessment (DAS28) criteria for remission applied to only 54% and 56% patients, respectively. Moreover, patients had a long disease duration (7 years), and the more advanced the disease, the more difficult it is to quantify progression of structural damage. Those investigators showed a disease progression rate of 11% at 12 months in patients who satisfied the ACR criteria for remission and 12% for those who satisfied the DAS28 criteria (16). Interestingly, among the 25 patients who were asymptomatic (no painful, tender, or swollen joints), 4 (16%) showed radiographic evidence of disease progression (16).

The results reported by Brown et al, along with our findings from the present study, support the hypothesis that radiographic evidence of disease progression is directly linked to persistent subclinical inflammation. We suggest that among all factors we studied, baseline PD-positive synovitis is the most robust determinant of future erosive damage; the presence of GS-positive synovitis is not sufficient (41).

The MRI findings were not associated with radiographic evidence of disease progression in our study. This lack of ability of MRI to predict disease progression in our study may be explained by several factors. MRI is known to be a sensitive examination, and in our cohort, a large number of patients with a weak grade of synovitis (grade 1) probably did not display “pathogenicity.” Patients were then classified as having either no synovitis (grade 0 or 1) or synovitis (grade 2 or 3). The analysis did not change our results. Second, on MRI, the maximum inflammation was detected at the wrist, which suggests that the maximum radiographic evidence of progression would be observed at this site. Nevertheless, we observed minor radiographic evidence of deterioration in the wrists over the 12 months. Indeed, as has previously been shown, radiography is insensitive in revealing or following up on erosions at the wrist (42). Third, the lack of an association between bone marrow edema and future lesions could be explained by the low sensitivity of low-field MRI for detecting bone marrow edema (43). Nevertheless, Brown et al (16) used high-field MRI and did not find an association between bone marrow edema and radiographic evidence of disease progression. Fourth, in light of the high sensitivity of MRI, the delay between the 2 radiographic assessments may have been too short. There is no consensus on the optimum delay between 2 radiographic assessments in observing disease progression in RA patients in remission (44). Finally, the lack of predictive power of MRI may be explained by a lack of statistical power. Indeed, we determined a need for a sample size of 85 patients in order to yield 30% radiographic evidence of disease progression. However, we found only 11% of patients with such evidence.

This study has some limitations. The first is related to the US examination itself. It is known to be operator dependent and was performed by 2 different operators. However, they performed several tests, and good reliability between them was confirmed, thus diminishing the risk of error in practice and interpretation. Second, we did not perform two US assessments as we did for the radiographic and MRI assessments. US readings were performed during examination of the patient because recording the images and reading them with complete blinding is technically difficult. Third, we used a low-field dedicated MRI, which is considered to have good sensitivity and specificity, as compared with high-field MRI, for detecting synovitis, tenosynovitis, and erosions (43, 45–47). Nevertheless, Ejbjerg et al (43) showed that low-field MRI could detect only about 40% of bone marrow edema lesions as compared with high-field MRI.

In conclusion, radiography remains the conventional measure of structural progression of disease in RA according to regulatory authorities. Increasing evidence supports the necessity of including US and MRI in the assessment of RA, especially when the disease is considered to be in remission or exhibiting low levels of activity. We revealed that US could be used to predict disease relapse and structural progression by detecting baseline PD-positive synovitis. In such patients, adjustment of therapy and/or tight control of RA are necessary, with no tapering of treatment. Future studies are necessary to confirm our results with other US operators and to determine whether US assessment should be included in routine practice.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Foltz had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Foltz, Gandjbakhch, Etchepare, Rosenberg, Bourgeois, Fautrel.

Acquisition of data. Foltz, Gandjbakhch, Etchepare, Rosenberg, Rozenberg, Bourgeois, Fautrel.

Analysis and interpretation of data. Foltz, Gandjbakhch, Etchepare, Rosenberg, Tanguy, Rozenberg, Bourgeois, Fautrel.

APPENDIX A

INTRAOBSERVER RELIABILITY AND RADIOGRAPHIC EVIDENCE OF DISEASE PROGRESSION

First, 2 experienced readers (VF and FG) who were blinded with regard to all other imaging and clinical findings, performed a pilot study of all radiographs. Both readers were aware of the chronological sequence of baseline and followup radiographs. The intraobserver reliability (intraclass correlation coefficient [ICC]) for each reader was, respectively, 0.96 and 0.96 for erosions, 0.94 and 0.71 for joint space narrowing, and 0.93 and 0.78 for the SHS. The ICCs were 0.77 for erosions, 0.6 for joint space narrowing, and 0.67 for the SHS. Interobserver reliability assessments were considered moderate to good.

Radiographic evidence of structural progression was defined as a variation in the SHS greater than the smallest detectable change (SDC), which was calculated using the following formula:

equation image

where k represents the total number of readings (48). For a trial with results based on the mean scores of k raters/readings, the measurement error diminishes by a factor of √2. SDΔ(change scores) represents the SD of the difference between the change scores from 2 reading sessions. This value reflects the measurement error of the difference between the 2 change scores, that is, the measurement error when discriminating between 2 change scores.

In this way, we obtained an SDC for an erosion score of 1. For 44 patients with progressive disease, or 51.8% of the cohort, the decrease in total radiographic joint damage scores exceeded the SDC over 1 year. Discrepancies in the readings were resolved by adding a consensus reading. A total of 9 patients (10.6%) showed structural disease progression on radiographs.

APPENDIX B

MAXIMUM SCORES FOR VARIABLES STUDIED BY US AND MRI

Table  . 
 USMRI
B modePD mode
  • *

    NA = not applicable.

Synovitis   
 Count14147
 GradeNA*4221
Erosions   
 CountNo limit23
 GradeNo limit230
Bone marrow edema   
 Count23
 Grade69

Ancillary