Article Text

PDF
A5.05 Prediction of persistent inflammatory arthritis with ultrasound: A data-driven method
  1. I Sahbudin1,4,
  2. P de Pablo1,4,
  3. L Pickup1,4,
  4. Z Cader1,
  5. G Allen2,
  6. P Nightingale3,
  7. CD Buckley1,5,
  8. K Raza1,5,
  9. A Filer1,4
  1. 1Rheumatology Research Group, Institute of Inflammation and Ageing, University of Birmingham, UK
  2. 2Green Templeton College, University of Oxford, UK
  3. 3Wolfson Computer Laboratory, University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK
  4. 4Department of Rheumatology, University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK
  5. 5Department of Rheumatology, Sandwell and West Birmingham Hospitals NHS Trust, Birmingham, UK

Abstract

Background Prediction of disease persistence in early inflammatory arthritis is important to enable timely initiation of appropriate therapy. Currently available predictive algorithms for persistent arthritis do not include US variables. We used a data-driven method to identify the minimal core set of US, clinical and serological variables predicting persistent inflammatory arthritis in a cohort of patients with early arthritis.

Methods 107 patients [female n = 60, median age 51] with ≥1 clinically apparent synovitis and symptom duration ≤3 months underwent clinical, laboratory and US assessments. Final diagnosis was determined at 18-months follow-up. US assessment determined the presence of Grey scale (GS) and Power Doppler (PD) synovitis at 38 joints (bilateral MCPs 1–5, PIPs 1–5, wrists, shoulders, elbows, ankles and MTPs 2–5). First, univariate analysis was performed to identify clinical, serological and US variables significantly associated with persistent disease at 18-months. Second, principal component analysis (PCA) was performed on 1) clinical and serological variables (age, gender, symptom duration, ESR, CRP, RF, ACPA, duration of early morning stiffness, tender joint and swollen joint count), 2) US GS variables and 3) US PD variables to identify variables that reflected similar themes. Finally, one variable from each component was extracted and made available in a forward step-wise multivariate logistic regression analysis.

Results 63 patients developed persistent inflammatory arthritis and 44 patients had resolving disease. On PCA, three components were identified within the clinical and serological variables, four components were identified within the GS US variables and four components within the PD US variables. The final multivariate logistic regression model included RF positivity (OR = 5.83, p = 0.01), MCP2 PD positivity (OR = 4.33, p = 0.002) and MTP2 PD positivity (OR = 10.72, p = 0.030). In seronegative patients, the final multivariate logistic regression included early morning stiffness >60mins (OR=3.62, p = 0.017) and PIP 2 PD (OR = 8.44, p = 0.003).

Conclusion This is the first study using a data-driven method to show that US provides independent data beyond clinical and serological variables in the prediction of persistent arthritis, even in RF and ACPA-negative patients. Bilateral MCP2 and MTP2 is the minimal joint sub-set that provides independent predictive data in this cohort.

Statistics from Altmetric.com

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.