Article Text

SAT0039 Development and Validation of A Prognostic Clinical Model for RAPID Radiographic Progression in Patients with RA
  1. E. Alemao1,
  2. S. Joo2,
  3. P. Allison3,
  4. M. Al4,
  5. M. Rutten-van Molken4,
  6. G. L'Italien1,
  7. C. Iannaccone5,
  8. M. Frits5,
  9. N. Shadick5,
  10. M. Weinblatt5,
  11. K. Liao5
  1. 1Bristol-Myers Squibb, Princeton
  2. 2Bristol-Myers Squibb, Hopewell
  3. 3University of Pennsylvania, Philadelphia, United States
  4. 4Erasmus University, Rotterdam, Netherlands
  5. 5Brigham and Women's Hospital, Boston, United States


Background Identification of factors predictive of rapid radiographic progression (RRP) in RA pts will enable clinicians and policymakers to target appropriate treatment strategies to at-risk pts.

Objectives To evaluate baseline (BL) factors associated with RRP in a longitudinal RA cohort study and to develop and validate a prognostic model for RRP in RA pts.

Methods Data from the Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study (BRASS) registry were used to develop the model for RRP. Pts in the BRASS registry underwent physical and laboratory assessments at BL and annually thereafter. Joint X-rays were conducted at BL and Yr 2 and scored by the van der Heijde modified Sharp score (TSS). Pts with both BL and Yr 2 X-ray data were included in the analysis. Annualized change in TSS was calculated and pts having a 5+ unit change were considered rapid progressors. Univariate association of BL factors with RRP was conducted using either a chi-square (χ2) or a Fisher's exact test. Baseline factors with p<0.1 were included in a logistic regression model for binary responses of RRP vs no radiographic progression (defined as annualized change of +1 to –1 units). Weight was categorized based on distribution (75th percentile) by sex; males ≥97 kg and females ≥77 kg were considered overweight. The model developed using BRASS data was validated using BL and 1-yr X-ray data from the Abatacept versus Adalimumab Comparison in Biologic-Naive RA Subjects with Background Methotrexate (AMPLE) study. The model's discriminative properties were measured by ROC, and calibration by the Hosmer & Lemeshow goodness-of-fit χ2.

Results Of 1297 pts in the BRASS registry, 644 had BL and Yr 2 TSS. There were no significant differences seen between pts with and without available TSS. Overall, 82% of pts were female; mean (SD) age was 57 (14) yrs and mean symptom duration was 15 yrs. Baseline mean (SD) DAS28 (CRP) was 3.9 (1.6), total swollen tender joints was 16.1 (14.1), TSS was 48.6 (61.1) and 71% of pts were seropositive (RF or ACPA positive). RRP was experienced by 67 (10%) pts in the sample and 302 (46.9%) did not experience radiographic progression. Most (95%) patients were exposed to DMARDs, with 45% of these exposed to biologic DMARDs. BL factors associated with RRP in logistic regression were seropositivity (OR=3.35; 95% CI 1.41, 7.99), duration of RA <2 yrs (OR=0.22; 95% CI 0.07, 0.64), under to normal weight (OR=4.88; 95% CI 1.82, 13.11), DAS28 (CRP) (OR=1.24; 95% CI 1.02, 1.52) and BL TSS (OR=1.01; 95% CI 1.00, 1.01). The model had an ROC of 0.80 (95% CI 0.75, 0.85) and χ2 of 10.47 with 8 DF (p=0.233). The AMPLE validation dataset comprised 579 RA pts with 36 (6.2%) pts experiencing RRP. In this dataset, the model had good external discrimination (ROC=0.73; 95% CI 0.63, 0.82); however, the model overestimated RRP in AMPLE (χ2 of 164.9 with 8 DF; p<0.001). On adjusting for 3.8% lower BL RRP rates in AMPLE, the model calibration improved (χ2 of 18.4 with 8 DF; p=0.02).

Conclusions RRP in RA can be predicted based on BL seropositivity, body weight, disease duration, DAS28 (CRP) and TSS. Further validation of the model with other datasets is required to confirm the findings.

Disclosure of Interest E. Alemao Shareholder of: BMS, Employee of: BMS, S. Joo Shareholder of: BMS, Employee of: BMS, P. Allison: None declared, M. Al: None declared, M. Rutten-van Molken: None declared, G. L'Italien Shareholder of: BMS, Employee of: BMS, C. Iannaccone: None declared, M. Frits: None declared, N. Shadick Grant/research support: AbbVie, AMGEN, BMS, Crescendo Bioscience, Genentech, UCB, M. Weinblatt Grant/research support: BMS, Crescendo Bioscience, UCB, Consultant for: BMS, Crescendo Bioscience, UCB, AbbVie, Roche, Janssen, K. Liao: None declared

DOI 10.1136/annrheumdis-2014-eular.2233

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.