Early prediction of rheumatoid arthritis by serological variables and magnetic resonance imaging of the wrists and finger joints: results from prospective clinical examination

We aimed at characterising the serological variables and magnetic resonance imaging (MRI) early changes in the wrists and finger joints which would differentiate rheumatoid arthritis (RA) from rheumatic diseases other than RA (non‐RA) at the earliest stage.

Patients were referred from the Early Arthritis Clinic, started in 2001 at the First Department of Internal Medicine, Graduate School of Biomedical Sciences, Nagasaki University. After prospective follow up, a diagnosis was made according to international classification criteria, and in particular, RA was defined by 1987 criteria of the American College of Rheumatology for RA.¹ Informed consent was obtained from all the patients, and the protocol was approved by the Institutional Review Board of Nagasaki University.

Eighty consecutive patients with RA and 33 non‐RA patients were studied and a diagnosis evaluated 12 months after entry, by March 2005. The mean disease duration of the 80 patients with RA at entry was 4.8 months, and thus they were described as early stage RA.

MR images of both wrists and finger joints were acquired with a 1.5 T system (Sigma, GE Medical Systems, Milwaukee, WI, USA) with the use of an extremity coil. Coronal T₁ weighted spin echo (repetition time 450, echo time 13) and short time inversion recovery (repetition time 3000, echo time 12, T₁ 160) images were acquired. The images were evaluated for the presence or absence of bone marrow oedema, bone erosion, and synovitis in 15 joints in each finger and wrist—namely, distal radioulnar joint, radiocarpal joint, mid‐carpal joint, 1st carpometacarpal joint, 2nd–5th carpometacarpal joints (together), 1st–5th metacarpophalangeal joints separately, and 1st–5th proximal interphalangeal joints separately (total 30 joints from both hands). The extent of synovitis, bone marrow oedema, and bone erosion was determined, as previously described,²,³,⁴,⁵ by two experienced radiologists (MU and ST), and decisions were reached by consensus.

Symmetric arthritis is a characteristic feature of RA.¹ The presence of symmetric synovitis on MRI was defined as bilateral involvement of wrist sites, metacarpophalangeal joints, or proximal interphalangeal joints without absolute symmetry. Because we focused on the presence or absence of early joint changes on MRI for the differentiation, we did not use the OMERACT 5 RA‐MRI scoring system.⁴,⁵ As expected, the positivity of matrix metalloproteinase 3 (MMP‐3; measured by enzyme linked immunosorbent assay (ELISA; Daiichi Pure Chemicals, Fukuoka, Japan) (46.3% v 12.1%), anti‐cyclic citrullinated peptide antibody (anti‐CCP antibody; measured by ELISA; DIASTAT Anti‐CCP, Axis‐Shield, Dundee, UK) (67.5% v 12.1%), and IgM rheumatoid factor (IgM RF; measured by latex‐enhanced immunonephelometric assay; Dade Behring, Marburg, Germany) (67.5% v 30.3%) as well as the frequency of symmetric synovitis (81.3% v 36.4%), bone marrow oedema (56.3% v 12.1%), and bone erosion (45.0% v 9.1%) were higher in early stage RA than in non‐RA.

Logistic regression analysis using the statistical analysis system software demonstrated that the presence of anti‐CCP antibody and/or IgM RF, symmetric synovitis and bone marrow oedema and/or bone erosion at entry could discriminate between patients with RA and non‐RA patients (table 1).

Table 1 Serological variables and MRI findings for the discrimination between early stage RA and non‐RA.

Variables	Odds ratio	Coefficient	SE	p Value	Weighted score
Anti‐CCP antibody and/or IgM RF	7.42	2.00	0.57	0.0005	1
MMP‐3	2.87	1.05	0.72	0.14	0
Symmetric synovitis	4.37	1.47	0.57	0.009	1
Bone marrow oedema and/or bone erosion	5.48	1.70	0.63	0.007	1

Logistic regression analysis identified the presence of anti‐CCP antibody and/or IgM RF, symmetric synovitis on MRI, and bone marrow oedema and/or bone erosion on MRI as significant and independent measures for discrimination between early stage RA and non‐RA. The weighted score was calculated based on the regression coefficient for each variable as described in the text.

At the first visit, a total score of two or more of the three objective measures (anti‐CCP antibody and/or IgM‐RF: 1, symmetric synovitis on MRI: 1, bone marrow oedema and/or bone erosion on MRI: 1) allowed the prediction of RA with 82.5% sensitivity and 84.8% specificity, respectively (table 2). (Statistical weights of the variables were calculated based on the regression coefficient for each variable, standardised by dividing by the coefficient for symmetric synovitis; values were rounded off to yield integers.)

Table 2 Evaluation of the prediction score (⩾2) in early stage RA at the first visit.

Total score	Sensitivity (%)	Specificity (%)	Positive predictive value (%)	Negative predictive value (%)	Accuracy (%)
⩾1	96.3	30.3
⩾2	82.5	84.8	93.0	66.7	83.2
3	50.0	96.9

We calculated the sensitivity and specificity of our scoring system for the prediction of early stage RA according to the sum of weighted scores described in table 1.

Sensitivity and specificity are shown for patients classified as early stage RA according to the total score (sum of weighted score 1–3).

We evaluated the statistical character of prediction score (⩾2) for the present 113 patients for the prediction of RA at entry.

Our present data may indicate that the prediction of autoantibodies as well as MRI detection of early joint changes contribute to the accurate diagnosis of early stage RA.