Dr. Schönermarck and colleagues raise 3 interesting points. The first is that relapses differ from one another in terms of severity and clinical importance. We agree. However, defining a relapse that is clinically important and/or deemed important by the patient is not straightforward for several reasons. There is limited evidence to suggest what types of relapse portend a poor prognosis and to what degree a relapse modifies the risk of death, end-stage renal disease, or other clinically important outcomes (Walsh M, Tonelli M, Jayne D, Manns B. Surrogate end points in clinical trials: the case of anti-neutrophil cytoplasm antibody-associated vasculitis. J Nephrol 2007;20:119–29). Furthermore, clinicians' perceptions of the most clinically important relapses are frequently not congruent with what patients believe is important (e.g., renal relapses defined on the basis of rising creatinine levels are often considered important to clinicians but not to patients) (Herlyn K, Hellmich B, Seo P, Merkel PA. Patient-reported outcome assessment in vasculitis may provide important data and a unique perspective. Arthritis Care Res [Hoboken] 2010;62:1639–45). Patients with ANCA-associated vasculitis are often under intense scrutiny for relapse and have disease-modifying treatment instituted early in the course of a relapse. As such, the clinically important risks associated with relapse are often related to the treatment rather than directly to the vasculitis itself. It is also relevant to note that although nonsevere relapses may not directly increase the risk of death or organ failure, if missed, they may progress to more severe relapses, and patients with nonsevere relapses are more likely to have further relapses and therefore further exposure to treatment. Since our definition of relapse included disease activity that required a treatment change, we felt this was the most appropriate primary outcome for the study.
The second related point raised was whether detailed analyses of the organ systems involved at relapse explain the risk associated with cardiovascular manifestations at baseline. We agree that documenting the relationship between original disease manifestations and subsequent relapse manifestations is of interest. However, it is likely that cardiovascular manifestations may be a surrogate for overall disease severity rather than a manifestation that relapses at a particularly high rate itself. Given that cardiovascular manifestations are infrequently present at diagnosis of ANCA-associated vasculitis, they do not explain the large absolute difference in the risk of relapse between patients with differing numbers of risk factors (at 2 years the difference in first relapses between those with 1 versus those with 0 risk factors was already double the absolute number of patients with a cardiovascular manifestation at diagnosis). Furthermore, relapses are diagnosed more quickly than initial disease, and it seems likely that fewer patients will develop severe manifestations (such as cardiovascular manifestations) or undergo the diagnostic tests required to document cardiovascular manifestations once the diagnosis is already known.
Schönermarck and coworkers also suggest that the relationship between baseline characteristics and the relative hazard of a first relapse may differ after 24 months. This would suggest a time-varying relationship between the risk factors and first relapse. Such a relationship was not seen. We agree, however, that treatment-related factors after initial therapy likely change the risk of relapse and that these need to be carefully studied to refine our understanding and estimation of relapse risk, but this does not imply that the risk associated with baseline characteristics changes over time.