1643 - B Cell Repopulation and Rheumatoid Factor Predict Duration of Response to Rituximab in Rheumatoid Arthritis

Monday, November 7, 2011: 3:00 PM
W375c (McCormick Place West)
Edward M. Vital1, Sudipto Das2, Shouvik Dass1, Maya H. Buch3, Frederique Ponchel1, Andrew Rawstron1 and Paul Emery4, 1NIHR Leeds Biomedical Research Unit, University of Leeds and Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom, 2NIHR Leeds Musculoskeletal Biomedical Research Unit, University of Leeds and Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom, 3University of Leeds, Leeds, United Kingdom, 4Chapel Allerton Hospital, Leeds, United Kingdom
Presentation Number: 1643


Duration of response to rituximab in rheumatoid arthritis is variable.  It is currently unclear whether fixed or on-demand retreatment produces better overall long term outcomes.  Predictors of time to relapse may therefore be valuable. We investigated known predictors of response to rituximab to test whether they could predict sustained response to 12 months.


104 rituximab-treated patients pooled from 3 clinical trials were analysed and results from 78 patients with EULAR response at 6 months used.  All protocols restricted NSAID and corticosteroid use similarly.  DAS28 and EULAR response criteria were measured at 0, 14, 26 and 40 weeks.  Patients who had relapsed by worsening EULAR criteria were retreated.  Naive and memory B and plasmablast subsets were measured by highly sensitive flow cytometry (as previously described) at 0, 2, 6, 14, 26 and 40 weeks.


Of the 78 patients with response at 6 months, 57% had moderate response (MOD) and 43% had good response (GOOD).  In patients with detectable B cells at 6 months DAS28 worsened by mean(SD)0.45(1.06), whereas in patients with undetectable B cells DAS28 improved by mean 0.44(0.45), p<0.001 between groups.  At 12 months NR/RT, MOD, GOOD rates were 51, 28, 21%.  We tested a binary logistic regression model including B cell detection, level and RF, IgG and DAS28 for prediction of continued EULAR response at 12 months, which produced a significant model (p=0.001) with overall accuracy 70.4% (see table)

Prediction of EULAR Moderate or Good Response at 12 months


Odds Ratio



Rheumatoid Factor (per IU/mL)




IgG (per g/L)








B cells detectable (Y/N)




Plasmablast count




Based on these results we also tested a simplified two-variable categorial model with RF and DAS28 using thresholds of 108 IU/mL and 3.7 (derived from ROC analysis) as predictive of relapse, which also produced a significant model with overall accuracy 69%(p=0.001).  Response rates at 12 months by these two criteria were as follows: DAS28<3.7 and RF<108 (n=30): 73%; DAS28>3.7 and RF<108 (n=15):53%; DAS28<3.7 and RF>108:35%; DAS28>3.7 and RF>108:8%.


Following an initial response to rituximab at 6 months, many patients lose response over the next 6 months.  This may be predicted at 6 months by B cell repopulation and rheumatoid factor, after correcting for disease activity.  These findings need to be validated, but may allow for selection of retreatment regime based on B cell biomarkers.

Keywords: B cells, autoantibodies, biomarkers, rheumatoid arthritis (RA) and rituximab

Disclosure: E. M. Vital, Roche Pharmaceuticals, 8 ; S. Das, None; S. Dass, Roche Pharmaceuticals, 8 ; M. H. Buch, Abbott Immunology Pharmaceuticals, 5, UCB , 8, Roche Pharmaceuticals, 5, Pfizer Inc, 8 ; F. Ponchel, None; A. Rawstron, None; P. Emery, None.