Reanalysis of the Rituximab in ANCA-Associated Vasculitis trial identifies granulocyte subsets as a novel early marker of successful treatment

Abstract

Introduction: In the present study, we sought to identify markers in patients with anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) that distinguish those achieving remission at 6 months following rituximab or cyclophosphamide treatment from those for whom treatment failed in the Rituximab in ANCA-Associated Vasculitis (RAVE) trial.

Methods: Clinical and flow cytometry data from the RAVE trial were downloaded from the Immunology Database and Analysis Portal and Immune Tolerance Network TrialShare public repositories. Flow cytometry data were analyzed using validated automated gating and joined with clinical data. Lymphocyte and granulocyte populations were measured in patients who achieved or failed to achieve remission.

Results: There was no difference in lymphocyte subsets and treatment outcome with either treatment. We defined a Granularity Index (GI) that measures the difference between the percentage of hypergranular and hypogranular granulocytes. We found that rituximab-treated patients who achieved remission had a significantly higher GI at baseline than those who did not (p = 0.0085) and that this pattern was reversed in cyclophosphamide-treated patients (p = 0.037). We defined optimal cutoff values of the GI using the Youden index. Cyclophosphamide was superior to rituximab in inducing remission in patients with GI below -9.25% (67% vs. 30%, respectively; p = 0.033), whereas rituximab was superior to cyclophosphamide for patients with GI greater than 47.6% (83% vs. 33%, respectively; p = 0.0002).

Conclusions: We identified distinct subsets of granulocytes found at baseline in patients with AAV that predicted whether they were more likely to achieve remission with cyclophosphamide or rituximab. Profiling patients on the basis of the GI may lead to more successful trials and therapeutic courses in AAV.

Trial registration: ClinicalTrials.gov identifier (for original study from which data were obtained): NCT00104299 . Date of registration: 24 February 2005.

Figures

Fig. 1

Overview of the analytical process, starting with open access clinical trial data. Solid black arrows represent work carried out by the primary investigators. White arrows represent work done in the present study, enabled by the public accessibility of the original raw trial data. Dashed black arrows represent future work that could be triggered by the reanalysis process

Fig. 2

Validation of the Immunology Database and Analysis Portal flow cytometry clustering without K (ImmPort-FLOCK). Cell subset percentages by automated identification were validated against manual gating for the identification of immune cell populations on the basis of size and granularity (forward scatter and side scatter, respectively). One hundred random flow cytometry files were independently analyzed by two immunologists using standard FlowJo software. a Scatterplots between the two immunologists show significant concordance in the identification of different cell populations. b Similar correlation was seen for granulocyte and lymphocyte percentages between automated analysis and the average of the two immunologists (shaded area represents the 95 % confidence interval of the regression line; p values based on Pearson correlation test). c Originally published figure showing the drop in CD19+ B-cell counts with rituximab or control (cyclophosphamide) treatment generated using manual gating of flow cytometry results (Reproduced with permission from [2]. d Results obtained through automated identification of the CD19+ lymphocyte population. Results shown in (c) and (d) represent median cell counts. ANCA anti-neutrophil cytoplasmic antibodies, CD cluster of differentiation, MPO myeloperoxidase, PR3 proteinase 3

Fig. 3

Granulocyte subpopulations and treatment outcomes. a Representative bidimensional dot plot illustrating the granulocyte subpopulations identified in an automated manner. Left: Hypergranular granulocytes (definitions in Table 1). Right: Hypogranular granulocytes. FSC forward scatter, SSC side scatter. b and c Granularity Index at day 0 among patients receiving rituximab and cyclophosphamide, respectively, stratified by treatment outcome. Data distribution is shown as a notched boxplot (showing minimum, maximum, 25th percentile, median and 75th percentile). An unpaired Kruskal-Wallis rank-sum non-parametric test was used to calculate significance. WBC white blood cells

Fig. 4

Complete remission rates for patients in the Rituximab in ANCA-Associated Vasculitis (RAVE) trial. a Primary endpoint outcomes among RAVE trial subjects stratified by the Granularity Index (GI). b Primary endpoint outcomes among RAVE trial subjects treated with either rituximab or cyclophosphamide in the absence of stratification. Fisher’s exact test was used to calculate significance between rates of complete remission on rituximab and cyclophosphamide. *10 of the 197 original trial subjects did not have flow cytometry data at baseline

Fig. 5

Proposed “personalized” treatment algorithm for anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis on the basis of the Granularity Index