To the Editor:

Reinhold-Keller and colleagues wondered whether all the 93 patients reported in our study had been accurately diagnosed with WG, presumably inferring that some of them might rather have had microscopic polyangiitis (MPA). All the patients analyzed in our study were classified according to the American College of Rheumatology criteria for WG (1), a tool that has proven to be 92% specific for this vasculitis and that, given the precise items on which it is based, might even yield higher specificity with respect to MPA. Their statement that granulomatosis had been observed in only 9% of our patients is incorrect. As emphasized in our report, the 8 individuals with histologic evidence of granulomatosis were among the 31 patients who had no histologic proof of vasculitis or glomerulonephritis. Moreover, our finding that 13% of the patients exhibited a P-/MPO-ANCA pattern does not preclude the diagnosis of WG for all these patients, since our value is in complete agreement with the 12% of MPO-ANCA recently published for a series of 180 WG patients (2).

Reinhold-Keller et al were further concerned about the marginally lower prevalence of ENT involvement observed in our series (84%), as compared with 90–93% in their WG cohort (3). In that respect, it should be recalled that other WG series had even lower frequencies of ENT involvement, notably 73–77% (2, 4, 5). In our study, ENT disease was defined based on clinical symptoms (e.g., nasal crusting), imaging findings (e.g., sinusitis) and/or those of an ENT examination (e.g., otitis media). Because the ENT symptoms of WG are so prominent, we are not convinced that only a highly trained ENT specialist can accurately recognize WG-related ENT involvement. Similarly, it is questionable to what extent magnetic resonance imaging offers better diagnostic sensitivity over computed tomography of the sinuses, which we consider the first-line imaging technique for patients suspected of having WG. As suggested by the lower rate of glomerulonephritis in the series reported by Reinhold-Keller and coworkers (54% versus 62% of our patients) (3), it can be hypothesized that the slightly diverging baseline characteristics between the 2 study populations might partly reflect the composition of those authors' cohort, as it included 15% of patients with localized disease, whereas we considered only systemic WG.

We agree with Reinhold-Keller and colleagues that further investigation of the factors determining WG prognosis is warranted. In contrast, we do not subscribe to the principle that such studies necessarily have to be based on single-center cohorts. Indeed, it can be put forth that large monocenter series are subject to selection bias, particularly by excluding those patients whose disease gravity leads to very early death or whose severe manifestations are incompatible with transportation to consult in distant referral centers. Analysis of the characteristics of the 12 individuals included in our study that had been referred to our institution during the disease course revealed that their survival tended to be longer than that of the 81 others who had been followed exclusively in our department or other care centers (P = 0.12 by log-rank test). Pertinently, these 12 individuals were also younger at disease onset (44.6 versus 51.7 years; P = 0.12 by Student's t-test), and more often had ENT involvement (92% versus 79%; P = 0.36 by Fischer's exact test), less frequent renal disease (50% versus 64%; P = 0.45 by Fischer's exact test), and a lower percentage of them had an initial serum creatinine >160 μMol/L (10% versus 39%; P = 0.09 by Fischer's exact test). These findings would tend to point out that, for rare and polymorphous diseases, like WG, case recruitment from multiple centers is useful to give a more accurate reflection of the phenotypic and prognostic spectrum of the disease.


  1. Top of page

Alfred Mahr MD*, Dominique Bligny MD*, Loïc Guillevin MD*, * Hôpital Cochin, Université Paris Paris, France.