An explanation for the apparent dissociation between clinical remission and continued structural deterioration in rheumatoid arthritis

Patients and controls.

In this prospective longitudinal cohort study, consultant rheumatologists used their clinical judgment to identify RA patients from their outpatient clinics whose disease was considered to be in remission while taking conventional DMARDs. In addition, all patients satisfied the following 6 criteria: RA classified according to the ACR criteria (13), age >18 years, disease duration of at least 12 months, no disease flare within the preceding 6 months, stable therapy for 6 months, and no clinical indication for a change in treatment.

A control group of 17 sex-matched normal subjects was also studied. None of these subjects had a history of joint disease, nor did they have any musculoskeletal symptoms or signs at the time of evaluation. All control subjects underwent a clinical assessment and an MRI scan of their dominant hand and wrist at baseline, using the same technique as used in the RA patients (see below).

Ethical approval for the project was obtained from the Ethics Committee of the Leeds Teaching Hospitals National Health Service Trust. Written informed consent was obtained from all subjects prior to study inclusion.

Study assessments.

Inclusion criteria were applied at –2 months and at 0 months (baseline). At baseline and 12 months, all patients underwent complete clinical, laboratory, radiographic, functional, quality of life, and imaging assessments. In addition, at 3, 6, and 9 months clinical, laboratory, functional, and quality of life assessments were performed.

Assessment of demographic and clinical characteristics.

Demographic characteristics recorded at baseline included age, sex, duration of current remission, current therapy, and previous DMARDs received. Clinical data were recorded at every visit and included the duration of morning stiffness; Likert and visual analog scales (VAS) for fatigue, joint pain, physician's assessment of disease activity, and patient's global impression of health and disease activity; and the number of painful, tender, and swollen joints as assessed by an independent trained metrologist. The Health Assessment Questionnaire (HAQ) (14) and the RA Quality of Life (RAQoL) questionnaire (15) were completed at every visit, and the scores were calculated. Remission and disease activity scores at each visit were calculated using the ACR remission criteria (8) (5 of 6 criteria satisfied on separate occasions 2 months apart) and the DAS28 criteria (10).

Laboratory assessment.

A complete blood cell count, erythrocyte sedimentation rate, and C-reactive protein level were measured at every visit. Rheumatoid factor (RF) titers and HLA type were determined at baseline to assess the presence of RA susceptibility (shared epitope) alleles.

Radiographic assessment.

Posteroanterior radiographs of the hands, wrists, and feet were obtained at baseline and at 12 months using standardized techniques. Radiographic joint damage was scored according to the Genant-modified Sharp scoring method (16). A single experienced reader (CGP) who was blinded to all other imaging and clinical findings and the chronological sequence of the baseline and followup radiographs performed this assessment.

Ultrasonographic assessment.

Each patient underwent a musculoskeletal US assessment of the joints of the dominant hand and wrist at baseline and at 12 months using GS and PD techniques. Eight joint regions were imaged by musculoskeletal US: metacarpophalangeal (MCP) joints 2–5 and the radiocarpal, ulnar-carpal, distal radioulnar, and intercarpal compartments of the wrist. Musculoskeletal US was performed by a single experienced sonographer (ZK) who was blinded to all other study findings, using an ATL HDI 3000 machine (ATL Ultrasound, Bothell, WA) with a 10–5 MHz linear array “hockey stick” transducer, according to the European League Against Rheumatism (EULAR) guidelines (17). The presence and location of any synovial hypertrophy and erosions were recorded with reference to standardized definitions subsequently agreed upon by the Outcome Measures in Rheumatology (OMERACT) group (18).

Synovial hypertrophy on GS images was graded according to a semiquantitative scoring method (0–3 scale, where 0 = no synovial hypertrophy, 1 = mild synovial hypertrophy, 2 = moderate, and 3 = severe) (19, 20). The area of maximum enhancement on PD images was recorded using a previously described semiquantitative technique (0–3 scale, where 0 = normal/minimal vascularity, 1 = mild hyperemia, 2 = moderate, and 3 = marked) (21). Erosions were scored according to their location and severity/size using a similar 0–3 semiquantitative scale (22). Each patient's total musculoskeletal US score for each pathologic feature (GS/PD/erosion) was calculated by summing the corresponding scores for each joint region.

Magnetic resonance imaging assessment.

MRI of the dominant hand and wrist was performed at baseline and 12 months using a conventional 1.5T Phillips MRI scanner (Philips, Surrey, UK). The same joint regions were imaged by MRI as for musculoskeletal US. The subjects were placed in a supine position with their arm outstretched in a “superman” position. A dual-coil approach was used in order to image the hand and wrist in a single sitting. T1 and T1 spectral presaturation with inversion recovery (SPIR; a method of fat suppression) sequences were recorded in the axial and coronal planes after intravenous administration of gadolinium.

Synovitis, bone marrow edema (BME), and erosions were defined according to OMERACT guidelines (23). Images were graded by a single experienced reader (PGC) who was blinded to all other imaging and clinical findings, using the OMERACT RA MRI Scoring (RAMRIS) system (23). Synovitis was scored semiquantitatively (0–3 scale, where 0 = normal, 1 = mild, 2 = moderate, and 3 = severe), with each point representing one-third of the maximum volume of enhancing tissue in the synovial compartment. A similar method was used to score BME, corresponding to the proportion of each bone containing edema (0–3 scale, where 0 = no edema, 1 = 1–33% of bone edematous, 2 = 34–66%, and 3 = 67–100%). Erosions were graded on a scale of 0–10, with each point representing a 10% loss of bone volume. Each patient's total MRI score for each pathologic feature (synovitis/BME/erosion) was calculated by summing the corresponding scores for each joint region.

Statistical analysis.

Data evaluation and statistical analysis were performed using SPSS version 14.0 software (SPSS, Chicago, IL). Analyses were conducted at the patient and individual joint level. Patients were also stratified according to standard clinical remission criteria (ACR and DAS28) as well as signs and symptoms (“asymptomatic” patients had clinically normal joints without any evidence of pain, tenderness, or swelling). Normally distributed continuous data were analyzed using parametric tests (independent t-test) and were summarized with means and 95% confidence intervals (95% CIs). Non-normally distributed and ordinal data were analyzed using nonparametric tests (Wilcoxon's signed rank test, Mann-Whitney U test) and were summarized with medians and interquartile ranges (IQRs). Categorical data were analyzed using chi-square tests, with continuity correction where necessary if the expected frequency for a cell in a cross-table was <5. Binary logistic regression was used to examine which baseline factors were associated with radiologic progression. HAQ and RAQoL data were Rasch-transformed to linearity (24). Intrareader reliability of radiographic scoring was assessed using the Bland-Altman plot technique (25). The smallest detectable change (SDC) for paired radiographic change scores was calculated according to the following formula:

where k represents the number of readings over which one wants to average the analyses of the trial (in this case, k = 1) (26).

Patient population.

A total of 102 RA patients who satisfied the inclusion criteria were evaluated. Full sets of radiographs at both time points (baseline and 12 months) were available for 90 patients; these formed the core dataset for the imaging analysis. Patients who did not have a complete set of films did not differ significantly from those with complete data in terms of age, sex, shared epitope, RF positivity at baseline, or ACR/DAS28 remission status throughout the study. Due to patient migration and contraindications or intolerance of scanning procedures, complete MRI and musculoskeletal US data at both time points were not available for all patients.

Demographic, clinical, and laboratory characteristics at baseline.

The baseline demographic, clinical, and laboratory characteristics of the RA patients are shown in Table 1. The study population was predominantly female (67%) and had a mean age of 57 years; 64% of were RF positive, and 61% were shared epitope positive. The median disease duration was 7 years, and the median period of remission at the baseline assessment was almost 2 years. During the course of their disease, 99% of the patients had received DMARD therapy, and the majority had received >1 such drug. At baseline, 91% of the cohort were currently taking DMARDs, with oral methotrexate and sulfasalazine being the most commonly used medications. Twenty-one percent of the patients were taking combination DMARD therapy, and 2 of the patients had previously received biologic agents. Thirty-nine percent were currently receiving a nonsteroidal antiinflammatory drug (NSAID), and 2% were taking oral prednisolone (all ≤5 mg/day).

Clinical and laboratory measures of disease activity all showed low activity, as measured by VAS assessments, the duration of early morning stiffness, the formal joint counts, and levels of acute-phase reactants. The study group had generally low levels of functional impairment, as measured by the HAQ (median 0.31 [IQR 0.14–0.57]), and had mild-to-moderate quality of life impairment, as measured by the RAQoL (median 7.5 [IQR 3.5–13.5]). A total of 54% fulfilled the ACR remission criteria, and 56% satisfied the DAS28 remission criteria (DAS28 score <2.6). The mean ± SD DAS28 score in the entire cohort was 2.53 ± 0.96 (95% CI 2.34, 2.72).

Findings of imaging assessments at baseline.

The majority of patients (57 of 90 [63%]) had evidence of radiographic joint damage, as assessed using the Genant-modified Sharp technique (60% with erosions and 51% with joint space narrowing [JSN]; median total joint damage score 2.000 [IQR 0.000–12.625]). On musculoskeletal US (n = 88), 68% of patients had erosions, 89% had evidence of synovial hypertrophy, and 63% had increased PD signal. Using MRI (n = 76), 96% of patients had erosions, 92% had synovitis, and 53% had BME.

Of the patients who satisfied the ACR remission criteria, 87% (39 of 45) had synovial hypertrophy and 56% (25 of 45) had increased PD signal on musculoskeletal US, and 95% (36 of 38) had detectable synovitis on MRI. Similarly, of the patients who fulfilled the DAS28 remission criteria, 84% (41 of 49) and 51% (25 of 49) had synovial hypertrophy and increased PD signal, respectively, on musculoskeletal US, and 88% (36 of 41) had demonstrable synovitis on MRI.

Considering the completely asymptomatic and clinically normal joints (i.e., no pain, tenderness, or swelling) in the dominant hand or wrist, 38% (145 of 378) had synovial hypertrophy and 15% (56 of 378) had increased PD signal on musculoskeletal US, and 50% (160 of 322) had detectable synovitis on MRI, despite the normal clinical findings. High levels of intra- and interobserver reliability for imaging scoring have previously been reported in this cohort (12).

Findings of clinical and imaging assessments in the control group.

All control subjects had normal findings on clinical examination. Three control subjects (18%) had evidence of synovitis on MRI. This comprised a total of 4 joints (3%), all of which were graded as mild. BME was not seen in any joints of the control subjects.

Change in RA disease activity over 12 months.

There were no statistically significant differences in the results of disease activity measures between the baseline and outcome assessments. At 12 months, 45% of the patients (38 of 85) fulfilled the ACR remission criteria and 61% (52 of 85) fulfilled the DAS28 remission criteria. The mean change in the DAS28 score was –0.016 (95% CI –0.255, 0.223) (t = 0.134, P = 0.894 Student's paired t-test). Only 5% of patients required a change in therapy due to increased disease activity throughout the study period, despite regular clinical assessments at 3-month intervals.

Progression of structural changes in each patient over 12 months.

The calculated SDC for the erosion score (both hands and feet) and for the total radiographic score (combined erosion and JSN scores in both hands and feet) was 1. The SDC for the JSN score (both hands and feet) and for the erosion/JSN/combined score (dominant-hand MCP joints 2–5) did not exceed the smallest unit of change in the scale, which was 0.5.

A total of 19% of patients (17 of 90) experienced a deterioration in their total radiographic joint damage scores that exceeded the SDC over the study period. The quantitative change in the total radiographic damage score was statistically significant (Z = –3.26, P = 0.001 by Wilcoxon's signed rank test) (Table 2). The probability plot shown in Figure 1 illustrates the proportion of patients with a change in the total radiographic score, as well as a quantitative measure of the difference in scores, over the 12-month study period.

Musculoskeletal US and MRI scanning detected changes in the total erosion score in the dominant-hand MCP joints in a larger proportion of patients (24% [23 of 95] and 20% [15 of 74], respectively) than did radiography (6% [5 of 90]) (Table 2). There were significant quantitative changes in the total erosion scores in the dominant hand and wrist, as measured by musculoskeletal US (n = 82) (median change 0 [IQR 0–1]; Z = −2.41, P = 0.016 by Wilcoxon's signed rank test) and by MRI (n = 66) (median change 0 [IQR 0–1]; Z = −3.37, P = 0.001). The change in dominant-hand MCP score was statistically significant only for the MRI assessment (n = 74) (median change 0 [IQR 0–0]; Z = −3.12, P = 0.002), although US showed a trend toward significance (n = 95) (median change 0 [IQR 0–0]; Z = −1.74, P = 0.082) (Table 2).

We performed univariate binary logistic regression analyses to investigate the association between baseline factors and radiographic progression and found that in addition to known predictive factors, such as RF (borderline significant; odds ratio [OR] 2.95, P = 0.054) and ACR remission (borderline significant; OR 0.33, P = 0.054), the total baseline PD score was also associated with an adverse structural outcome radiographically (increase in OR per unit increase in score 1.36, P = 0.038). The PD score for the dominant-hand MCP joints was also significantly associated with radiographic progression in any joint anywhere in the hands and feet (increase in OR per unit increase in score 1.64, P = 0.036) (Table 3).

In the completely asymptomatic patients, who had no painful, tender, or swollen joints, 16% (4 of 25) experienced radiographic progression, as reflected by an increased total joint damage score exceeding the SDC in the hands and feet. In the dominant-hand MCP joints, 7% (2 of 30) had a worse erosion score on radiography, 22% (6 of 27) had a worse erosion score on musculoskeletal US, and 25% (5 of 20) on MRI (Table 2 and Figure 1).

At 12 months, 11% of the patients who satisfied the ACR remission criteria and 12% of the patients who satisfied the DAS28 remission criteria also showed radiographic progression (data not shown).

Progression of structural changes in individual joints and association with baseline parameters.

SDC values were calculated for the radiographic erosion score, JSN score, and the combined score for individual dominant-hand MCP joints. For each of the scores, the calculated SDC did not exceed the smallest unit of change in the scale (0.5).

Associations between baseline clinical joint assessments and structural progression in individual joints were evaluated. The dominant-hand MCP joints were used for this purpose since comparable data for each imaging modality were available for each patient at these anatomical sites. While the clinical characteristics were associated with an increased likelihood ratio of structural progression, particularly for joint pain and tenderness (OR 3.32 and OR 2.17, respectively), none of these parameters reached statistical significance (Table 4). In contrast, baseline imaging findings were associated with a statistically significant likelihood of subsequent structural deterioration in the MCP joints. In particular, a positive PD signal (OR = 12.21, P < 0.001) as well as the scores for the PD (OR = 4.00, P < 0.001) and synovial hypertrophy (OR = 2.31, P = 0.032) by musculoskeletal US, as well as synovitis (OR = 2.98, P = 0.002) by MRI, were associated with significantly higher odds of progression (Table 4).

Progression of structural changes in asymptomatic joints and association with findings of imaging assessments at baseline.

As shown in Figure 2, similar proportions of asymptomatic and symptomatic MCP joints demonstrated structural progression. We assessed whether baseline imaging data were associated with longitudinal radiographic progression in these asymptomatic joints. Despite a relatively small number of joints with radiographic structural deterioration, PD findings were significantly associated with radiographic progression, as measured either by the difference in the PD score (median score 1 [IQR 1–1] in joints without progression and median score 1 [IQR 1–2] in joints with progression; increase in OR per unit increase in score 6.41 [95% CI 1.81, 22.74], P = 0.004) or by the number of joints with increased PD signal (4% of joints without progression versus 29% of joints with progression; OR 8.77 [95% CI 1.54, 49.89], P = 0.014) (data not shown).