Background/Purpose: Most patients with anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) achieve clinical remission after induction therapy. However, even among patients who achieve remission, induction therapy sometimes must be changed or prolonged, or prednisone may be required to maintain remission. A decision support tool to better define patients with harder-to-control disease would help guide clinicians when starting therapy.

Methods: Twenty-eight serum proteins representing diverse aspects of the biology of AAV were measured at screening (during active disease) in 181 patients enrolled in the Rituximab in ANCA-Associated Vasculitis (RAVE) clinical trial. Two outcomes were used to test the ability of a combination of these biomarkers to predict optimal response to therapy: 1) “complete remission,” the primary endpoint of the RAVE trial, defined as remission (Birmingham Vasculitis Activity Score for Wegener’s granulomatosis [BVAS/WG]=0) at month 6, remaining in the assigned treatment arm, and off prednisone, with all other patients defined as non-responders; 2) remission (BVAS/WG=0) at month 6, with analysis limited to patients who remained in the trial through month 6 in their original treatment arm, and did or did not re-start prednisone at or before month 6. Demographic and clinical variables included age, sex, ANCA specificity (MPO or PR3), new diagnosis at entry, treatment (rituximab vs cyclophosphamide), and presence of renal disease. Models were developed using least absolute shrinkage and selection operator (lasso) regressions with leave-one-out cross-validation. Variables included all biomarkers in addition to demographic and clinical information.

Results: All subjects had severe active vasculitis (median BVAS/WG= 8) at screening. Sixty-two percent (112/181) achieved complete remission at month 6. A combination of 10 serum biomarkers and treatment with rituximab provided the best predictive ability for complete remission at 6-month (areas of under the curve [AUC] = 0.73), see Figure. Among 155 patients who remained on study protocol at month 6 in their original treatment group, 130 were in remission. A combination of 7 biomarkers, new diagnosis at entry, and rituximab treatment provided highest accuracy in predicting remission at month 6 (AUC = 0.85). Forcing additional demographic and clinical information did not improve accuracy in either model (AUC = 0.74 and 0.85).

Conclusion: A combination of several biomarkers during active disease could provide a useful tool to predict patients with AAV who will achieve optimal response to induction therapy. Further validation studies are needed to test the clinical utility of this panel of biomarkers.

Figure.