Article Text

FRI0578 Is the Localisation of Erosions on MRI of Diagnostic Interest in Rheumatoid Arthritis?
  1. A. Hermet1,
  2. L. Gossec1,
  3. V. Foltz1,
  4. J. Renoux2,
  5. N. Costedoat-Chalumeau3,
  6. G. Mercy2,
  7. Z. Amoura4,
  8. J.-C. Piette4,
  9. N. Morel4,
  10. P. Bourgeois1,
  11. B. Fautrel1,
  12. F. Gandjbakhch1
  1. 1Rheumatology
  2. 2Radiology, CHU Pitie-Salpetriere
  3. 3Internal medicine, CHU Cochin
  4. 4Internal medicine, CHU Pitie-Salpetriere, Paris, France


Background MRI is a sensitive imaging tool to assess synovitis and erosions but its specificity is low to moderate1,2 with a subsequent risk of falsely positive RA diagnosis.

Objectives To evaluate the sensitivity and specificity of MRI erosion in different hand localisations in RA patients versus controls in order to determine which localisations of MRI erosion are the most specific.

Methods The GRALE study was a cross-sectional study including patients between 2009 and 20113. Patients had either: (a)RA (according to ACR 87 classification criteria and with established disease ≥2 years disease duration); (b)systemic lupus erythematosus, SLE, or (c)primary Sjögren's syndrome, pSS, both according to ACR classification criteria and without RA association; or (d)healthy controls without any history of rheumatic disease. In the present study RA patients were compared to all other patients. MRI examination was performed on wrist and MCP 2 to 5 using a dedicated MRI with T1 and STIR sequences in coronal and axial planes. Two independent readers with good inter and intra-reliability scored erosions according to the RAMRIS system, blinded to clinical and radiographic data. Each localisation of the RAMRIS was evaluated as erosive yes/no and sensitivity and specificity were assessed. Fisher's exact test was applied to compare frequencies of erosion.

Results 90 subjects were included: 30 with RA patients and 60 as controls (21 SLE, 19 pSS, 20 healthy controls). In the RA patients, the mean disease duration was 6.6 years, 72% were positive for ACPA, and the mean DAS28 was 2.97. On plain radiographs of hands and feet, 83% of the RA patients and 0% of the controls had erosions. When assessing MRI erosions, the most sensitive and specific localisations were the inferior extremity of the radius (Se=0.50, Spe=0.92) and the inferior extremity of the ulna (Se=0.50, Spe=0.88) (Table). The sensitivity was higher at the wrist (range 0.43-0.90 excluding pisiform bone) than on MCP (0.20-0.53). Specificity was lower for wrist as compared to MCP joints with specificity <80% for all bones of the wrist except for trapezium and pisiform. In particular erosions were very frequently observed in the capitatum (97%) and lunatum (93%) bones in the control group. MCP bases had intermediate sensitivity (0.17-0.67) with high specificity (0.50-0.98).

Conclusions These results suggest the sensitivity and importantly, the specificity of MRI erosions for the diagnosis of RA vary widely across localisations. We propose MRI evaluation in this context should focus on the most specific sites, ie, MCP, basal metacarpal, radius and ulna bones rather than wrist.


  1. Colenbatch et al. Ann Rheum Dis 2013:72(6),804-814

  2. Suter et al. Arthritis Care & Research 2011:63(5),675-688

  3. Gandjbakhch et al. Ann Rheum Dis. 2014:73(Suppl2),656-657

Disclosure of Interest None declared

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.