Prevalence and clinical significance of antineutrophil cytoplasmic antibodies in Churg-Strauss syndrome

Authors


Abstract

Objective

Churg-Strauss syndrome (CSS) is classified among the so-called antineutrophil cytoplasmic antibody–associated systemic vasculitides (AASVs) because of its clinicopathologic features that overlap with the other AASVs. However, while antineutrophil cytoplasmic antibodies (ANCAs) are consistently found in 75–95% of patients with Wegener's granulomatosis or microscopic polyangiitis, their prevalence in CSS varies widely and their clinical significance remains uncertain. We undertook this study to examine the prevalence and antigen specificity of ANCAs in a large cohort of patients with CSS. Moreover, we evaluated the relationship between ANCA positivity and clinicopathologic features.

Methods

Immunofluorescence and enzyme-linked immunosorbent assay were used to determine the presence or absence of ANCAs in 93 consecutive patients at the time of diagnosis. The main clinical and pathologic data, obtained by retrospective analysis, were correlated with ANCA status.

Results

ANCAs were present by immunofluorescence in 35 of 93 patients (37.6%). A perinuclear ANCA (pANCA) pattern was found in 26 of 35 patients (74.3%), with specificity for myeloperoxidase (MPO) in 24 patients, while a cytoplasmic ANCA pattern, with specificity for proteinase 3, was found in 3 of 35 patients (8.6%). Atypical patterns were found in 6 of 30 patients with anti-MPO antibodies (20.0%). ANCA positivity was associated with higher prevalences of renal disease (51.4% versus 12.1%; P < 0.001) and pulmonary hemorrhage (20.0% versus 0.0%; P = 0.001) and, to a lesser extent, with other organ system manifestations (purpura and mononeuritis multiplex), but with lower frequencies of lung disease (34.3% versus 60.3%; P = 0.019) and heart disease (5.7% versus 22.4%; P = 0.042).

Conclusion

ANCAs are present in ∼40% of patients with CSS. A pANCA pattern with specificity for MPO is found in most ANCA-positive patients. ANCA positivity is mainly associated with glomerular and alveolar capillaritis.

Churg-Strauss syndrome (CSS) is defined as an eosinophil-rich and granulomatous inflammation involving the respiratory tract, coupled with necrotizing vasculitis affecting small to medium-size vessels, and associated with asthma and eosinophilia (1, 2). CSS is a rare disorder with an incidence of 1.3–6.8 cases per 1,000,000 patients per year and an overall prevalence of 10.7–13 per 1,000,000 adults (3–5).

Three different phases can usually be recognized in CSS. Asthma and atopic allergies such as rhinitis may precede by months, and sometimes by several years, the development of an eosinophilic infiltrative disease with eosinophilic pneumonia or gastroenteritis, followed by the vasculitic phase (6–9). Nasal polyposis, sinusitis, and nonfixed pulmonary infiltrates are seen in the majority of patients. Vasculitis commonly affects the skin, peripheral nerves, gastrointestinal tract, and heart. Coronary arteritis and myocarditis are the main causes of morbidity and mortality (1, 6–9). Pathologic confirmation is based on the presence of extravascular granulomas in association with necrosis and predominant extravascular eosinophils, as well as necrotizing vasculitis (1, 2), but all of these lesions are rarely found together in biopsy specimens (6). The vast majority of patients have blood eosinophilia (>1,500 cells/mm3 or >10%), and the exceptions are those previously treated with glucocorticoids for asthma (7–9).

CSS is considered to be one of the so-called antineutrophil cytoplasmic antibody–associated systemic vasculitides (AASVs) because of its clinical and pathologic features that overlap with those of the other AASVs, Wegener's granulomatosis (WG) and microscopic polyangiitis (MPA) (10, 11). However, these 2 diseases differ from CSS clinically, by the absence of asthma, and pathologically, by the absence of eosinophilia and eosinophil-rich tissue infiltrates (9–12). Moreover, while antineutrophil cytoplasmic antibodies (ANCAs) are consistently found in 75–95% of patients with active WG or MPA (10, 11), the prevalence of ANCAs in CSS has been reported to be variable, ranging from ∼40% to >70% (7, 8, 12–15). This is probably due to the small number of patients studied in most series, to selection bias, to the different classification criteria used for CSS, and to the various methods used to test for ANCA positivity (for review, see ref. 8). The immunofluorescence pattern is usually perinuclear (pANCA) with specificity for myeloperoxidase (MPO) by enzyme-linked immunosorbent assay (ELISA). Only a minority of patients have cytoplasmic ANCAs (cANCA) with antibodies to proteinase 3 (PR3) (7, 14, 15). Data are scarce concerning the clinical significance, if any, of ANCA positivity.

The aim of the present study was to examine the prevalence, fluorescence patterns, and antigen specificity of ANCAs in a large cohort of consecutive unselected patients with CSS. Moreover, we examined the relationship between ANCA positivity and clinicopathologic features and long-term outcome.

PATIENTS AND METHODS

Patients.

We identified 106 unselected consecutive patients in whom CSS was diagnosed clinically at internal medicine departments (nephrology, clinical immunology and rheumatology, pulmonology, neurology, and others) in 4 general hospitals in northern Italy between 1985 and 2004. Ninety-three patients, in whom CSS was diagnosed between 1989 and 2004, were tested for the presence of ANCAs at the time of diagnosis and were therefore eligible for the present study. The clinical characteristics of the 13 excluded patients were similar to those of the included patients (data not shown). CSS was diagnosed, as described by Guillevin et al (7), when asthma, hypereosinophilia (>10% or >1,500 cells/mm3), and clinical manifestations consistent with systemic vasculitis, with or without histologic evidence, were present. The following exceptions were allowed: 1) hypereosinophilia was not considered mandatory in the case of a few patients previously treated with oral steroids for asthma in the presence of histologic confirmation (vasculitis plus extravascular eosinophils); 2) the absence of asthma was not considered an exclusion criterion in the case of 4 patients with hypereosinophilia (>1,500 cells/mm3) and histologic confirmation. We examined the medical records of all these patients. CSS was defined according to the Chapel Hill Consensus Conference nomenclature (2). The classification criteria for CSS of the American College of Rheumatology (ACR) (16) as well as Lanham's (Hammersmith) criteria (6) were retrospectively applied to the study population. All patients had direct (histologic) or indirect (using surrogate markers) evidence of vasculitis (14).

Organ system involvement was assessed using the Birmingham Vasculitis Activity Score (BVAS) item list (17). Clinical manifestations were usually confirmed by instrumental examinations as appropriate and/or tissue biopsy. Unusual manifestations not included in BVAS items (e.g., cholecystitis and genitourinary involvement) were also considered.

Definitions and pathologic studies.

A patient's disease was considered to be in full remission when there was a complete absence of clinical disease activity for at least 6 months according to the BVAS item list, with the exception of asthma or neurologic and renal sequelae due to scars (7). A relapse was defined as the occurrence or recurrence of a clinical manifestation attributable to CSS. Persistent asthma or an isolated increase of eosinophilia was not considered a relapse (7).

One or more biopsy specimens from affected tissues were obtained when considered necessary by the clinicians. A biopsy specimen was considered consistent with a diagnosis of CSS when eosinophilic tissue infiltration and/or vasculitis were found (9). The Disease Extent Index and the prognostic Five-Factors Score (FFS) were also calculated as described by their respective authors (18, 19).

Biochemical and serologic markers.

Routine laboratory tests were performed in all cases at the time of diagnosis and at followup. The presence of ANCAs was determined in all patients at the time of diagnosis, before starting immunosuppressive treatment, using indirect immunofluorescence on ethanol-fixed granulocytes and antigen-specific PR3 and MPO ELISAs, as previously described (20–22). Antigen-specific ELISAs were performed on all serum samples, including those that were ANCA negative by immunofluorescence. Because PR3 and MPO ELISAs were not available in 2 laboratories until 1995, these assays were performed on stored serum samples in 6 cases. ANCA positivity was also tested at the end of the followup period in 37 patients and at the time of a relapse of disease in 16 patients. ANCA serology was tested first in each hospital and rechecked centrally, on frozen serum samples, in a laboratory that participated in the EC/BCR study for ANCA assay standardization (23–25). The staining patterns were defined according to the nomenclature of the International Consensus Statement, as follows: cANCA = cytoplasmic fluorescence with interlobular accentuation; cANCA atypical = other types of cytoplasmic fluorescence (e.g., homogeneous); pANCA = perinuclear or granulocyte-specific nuclear fluorescence; atypical ANCAs = other, less common patterns, such as mixed cytoplasmic and perinuclear fluorescence (26).

Perinuclear ANCA– and atypical ANCA–positive serum samples were also tested in the central laboratory on formalin-fixed neutrophils, as previously described, to confirm that they truly contained ANCAs and not, for example, antinuclear antibodies (22). Antigen specificity was also confirmed by inhibition study in selected cases as previously described (20).

Treatment.

All patients were treated with corticosteroids (1 mg/kg/day for 3–4 weeks with subsequent tapering), preceded by methylprednisolone pulses in the case of 22 patients. Cyclophosphamide (daily oral or pulses), as induction treatment, was added for 3–6 months in 42 patients (45.2%), usually those with the most severe disease (27). Other induction treatment consisted of methotrexate (5 patients) and azathioprine (1 patient). Plasma exchange and intravenous immunoglobulin were added in 4 patients and in 1 patient, respectively. Maintenance treatment consisted of low doses of glucocorticoids (5–12.5 mg/day) in all patients plus immunosuppressive drugs in 33 patients (35.5%) (azathioprine in 14, methotrexate in 12, and cyclosporin A in 7). Long-term followup and outcome were established after the patient's last visit or death.

Statistical analysis.

All analyses were performed using Stata statistical software, release 8.2 (Stata Corporation, College Station, TX). The differences between ANCA-positive and ANCA-negative patients in continuous variables were tested by the Mann-Whitney U test and in categorical variables by Fisher's exact test. We also computed the prevalence ratio and the prevalence difference of organ involvement, together with their associated 95% confidence intervals, using the Stata program epitab (28,29). The cumulative risk of relapse and death was estimated by the life-table (actuarial) method since survival curves were calculated at yearly intervals, and therefore individual times at which the events (or censoring) occurred were not precisely known. The difference between the risk curves was tested by the likelihood ratio test. These latter computations were performed using the Stata program ltable (29). All reported P values are 2-sided. P values less than 0.05 were considered significant.

RESULTS

Clinical and histologic characteristics.

There were 39 male patients and 54 female patients with a mean age of 51.6 years (median 51 years, range 18–86 years). All but 4 patients had bronchial asthma, which usually preceded the diagnosis of CSS. Seven patients had been receiving low doses of oral corticosteroids (<15 mg/day of prednisone or equivalent) for asthma for a few weeks at the time of diagnosis. Eosinophilia (>10%) was present in 88 patients (94.6%), and the 5 patients without eosinophilia had received previous steroid treatment for asthma. Patients had a median of 4,400 eosinophils/mm3 (range 600–28,815). Of the 93 patients with CSS, 85 (91.4%) met the ACR classification criteria for CSS; Lanham's criteria were met by 77 patients (82.8%), and 90 patients (96.8%) met at least 1 of the 2 criteria. The main clinical features of these patients are summarized in Table 1. One or more tissue biopsies were performed in 63 patients (67.7%). Biopsy samples were suggestive of CSS in 57 patients. Six biopsy samples were negative or aspecific.

Table 1. Main clinical features of the 93 patients with Churg-Strauss syndrome*
 No. (%) of patients
  • *

    CNS = central nervous system.

Asthma89 (95.7)
Constitutional symptoms63 (67.7)
Sinusitis72 (77.4)
Skin involvement49 (52.7)
Lung involvement47 (50.5)
Heart involvement15 (16.1)
Gastrointestinal involvement20 (21.5)
Peripheral neuropathy60 (64.5)
CNS involvement13 (14.0)
Renal involvement25 (26.9)

Prevalence of ANCAs and ANCA antigen specificity.

ANCAs were present by immunofluorescence in 35 of 93 patients (37.6%) tested at the time of diagnosis (Table 2). Of the 7 patients treated with oral corticosteroids at the time of ANCA testing, 3 (42.9%) were ANCA positive. The prevalence of ANCA positivity varied largely according to the specialty unit of origin, ranging from 0% to 12.5% in patients from pulmonology units to 90–100% in patients from nephrology units. A pANCA pattern was found in 26 of 35 patients (74.3%), with specificity for MPO in 24 patients, while a cANCA pattern with specificity for PR3 was found in 3 of 35 patients (8.6%). Two pANCA-positive samples were negative by ELISA. Interestingly, a cANCA pattern and an atypical ANCA pattern (Figure 1) were found in 3 patients (8.6%) and in 3 patients (8.6%), respectively, with anti-MPO antibodies by ELISA (Table 3). When the immunofluorescence assay was repeated in the central laboratory, the 3 cANCA-positive sera (with anti-MPO specificity) were defined as cANCA atypical due to the absence of interlobular accentuation, usually seen with PR3 ANCAs (Figure 1). The same patterns were also demonstrated using commercially available kits (Inova; Menarini Diagnostics, Florence, Italy) as well as “home-made” granulocyte preparations (Table 3).

Table 2. ANCA assay results in Churg-Strauss syndrome*
Author, year (ref.)No. of patientsIndirect immunofluorescenceELISA
Total ANCAscANCApANCANot classifiable ANCAsAnti-PR3 ANCAsAnti-MPO ANCAs
  • *

    Values are the number (%) of antineutrophil cytoplasmic antibody (ANCA)–positive patients, the number of patients with a given type of ANCA/total number of ANCA-positive patients (%) within a given series in the case of immunofluorescence, or the number of ANCA-positive patients by enzyme-linked immunosorbent assay (ELISA)/total number of patients tested (%) in the case of ELISA. In the case of ELISA, denominators may differ from numbers of patients due to differences between studies in methods of reporting antibody positivity. Only series describing at least 15 patients are shown. cANCA = cytoplasmic ANCA; pANCA = perinuclear ANCA; PR3 = proteinase 3; MPO = myeloperoxidase; NR = not reported.

  • Atypical patterns.

  • Pooled data from patients tested at the time of diagnosis and at the time of relapse.

  • §

    The number of patients tested for anti-MPO antibodies was not stated.

Present study9335 (37.6)3/35 (8.6)26/35 (74.3)6/35 (17.1)3/93 (3.2)30/93 (32.3)
Keogh and Specks, 2003 (14)3022 (73.3)1/22 (4.5)21/22 (95.5)0/22 (0.0)NR28/37 (75.7)
Della Rossa et al, 2002 (13)18NRNRNRNR0/18 (0.0)7/18 (38.9)
Solans et al, 2001 (15)1814 (77.8)1/14 (7.1)13/14 (92.9)0/14 (0.0)1/14 (7.1)13/14 (92.9)
Guillevin et al, 1999 (7)4220 (47.6)1/20 (5.0)15/20 (75.0)4/20 (20.0)0/11 (0.0)10/11 (90.9)
Reid et al, 1998 (12)1710 (58.8)0/10 (0.0)0/10 (0.0)10/10 (100.0)NR5/?§
Schnabel et al, 1996 (33)177 (41.2)5/7 (71.4)2/7 (28.6)0/7 (0.0)5/17 (29.4)2/17 (11.8)
Figure 1.

Patterns of antineutrophil cytoplasmic antibodies (ANCAs). A, Cytoplasmic ANCA pattern on ethanol-fixed neutrophils obtained from a patient with Wegener's granulomatosis and anti–proteinase 3 antibodies. B, Perinuclear ANCA pattern on ethanol-fixed neutrophils obtained from a patient with microscopic polyangiitis and antimyeloperoxidase (anti-MPO) antibodies. C, Atypical ANCA pattern (perinuclear + cytoplasmic) on ethanol-fixed neutrophils obtained from a patient with Churg-Strauss syndrome (CSS) and anti-MPO antibodies. D, Cytoplasmic ANCA atypical pattern on ethanol-fixed neutrophils obtained from a patient with CSS and anti-MPO antibodies. See Patients and Methods for descriptions of staining patterns. (Original magnification × 400.)

Table 3. ANCA assay results in 6 patients showing a cANCA or atypical ANCA pattern and anti-MPO antibodies*
PatientIndirect immunofluorescence patternELISA, arbitrary units
Ethanol fixedFormalin fixedPR3 ANCAsMPO ANCAs
Original laboratoryCentral laboratory (home-made)Central laboratory (commercial kit)
  • *

    cANCA pattern = cytoplasmic fluorescence with interlobular accentuation; cANCA atypical pattern = other types of cytoplasmic fluorescence; atypical (C + P) pattern = other, less common patterns, such as mixed cytoplasmic and perinuclear fluorescence (see Table 2 for other definitions).

  • PR3 ANCAs = anti-PR3 antibodies (normal values <20, standard curve range 6–400); MPO ANCAs = anti-MPO antibodies (normal values <20, standard curve range 5–320).

1cANCAcANCA atypicalcANCA atypicalcANCA6100
2Atypical (C + P)cANCA atypicalcANCA atypicalcANCA<662
3Atypical (C + P)Atypical (C + P)Atypical (C + P)cANCA<6163
4cANCAAtypical (C + P)Atypical (C + P)cANCA8107
5Atypical (C + P)Atypical (C + P)Atypical (C + P)cANCA10129
6cANCAcANCA atypicalcANCA atypicalcANCA11200

Inhibition studies confirmed that the cANCA and atypical ANCA patterns were caused by antibodies to MPO. Preincubation with MPO, but not with an irrelevant protein, could inhibit the binding both in ELISA and in immunofluorescence. All pANCA and atypical ANCA serum samples showed a cANCA pattern when tested on formalin-fixed neutrophils (Table 3).

Twenty-five consecutive unselected serum samples, with various levels of anti-MPO antibodies, from patients with MPA and WG were also retested by immunofluorescence to exclude a casual finding (due to a particular batch of slides). None of these samples showed a cANCA or an atypical ANCA fluorescence pattern.

Clinical significance of ANCA status.

ANCA positivity was associated with a significantly higher prevalence of renal involvement (51.4% versus 12.1% in ANCA-negative patients; P < 0.001) and, in particular, with a clinical picture of rapidly progressive glomerulonephritis (28.6% versus 5.2%; P = 0.004) (Table 4 and Figure 2). Worth noting is the fact that all 11 patients with a histologic picture of pauciimmune necrotizing crescentic glomerulonephritis were ANCA positive (data not shown). Moreover, ANCA positivity correlated with constitutional symptoms (85.7% versus 56.9%; P = 0.006) and with certain organ system manifestations, such as mononeuritis multiplex (51.4% versus 24.1%; P = 0.013), purpura (25.7% versus 6.9%; P = 0.015), and pulmonary hemorrhage (20.0% versus 0.0%; P = 0.001). In contrast, ANCA positivity was associated with lower prevalences of lung involvement (with the exception of alveolar hemorrhage) (34.3% versus 60.3%; P = 0.019) and heart involvement (5.7% versus 22.4%; P = 0.042). ANCA-positive patients tended to have a higher BVAS, but the difference was not statistically significant. Other clinical and serologic parameters did not differ between ANCA-positive and ANCA-negative patients (Table 4 and Figure 2). The interval between onset of asthma and diagnosis was shorter in ANCA-positive patients, but the difference was not statistically significant (median 3.5 years, range 0–47 years versus median 7.5 years, range 0–53 years; P = 0.077).

Table 4. Clinical features in ANCA-positive and ANCA-negative patients*
 ANCA positive (n = 35)ANCA negative (n = 58)P
  • *

    Except where indicated otherwise, values are the number (%) of patients. ANCA = antineutrophil cytoplasmic antibody; CNS = central nervous system; RPGN = rapidly progressive glomerulonephritis; ACR = American College of Rheumatology; BVAS = Birmingham Vasculitis Activity Score; DEI = Disease Extent Index; VDI = Vasculitis Damage Index; FFS = Five-Factors Score.

  • Fisher's exact test for categorical variables and Mann-Whitney U test for continuous variables.

Asthma34 (97.1)55 (94.8)1.00
Constitutional symptoms30 (85.7)33 (56.9)0.006
Sinusitis27 (77.1)45 (77.6)1.00
Skin involvement21 (60.0)28 (48.3)0.29
Purpura9 (25.7)4 (6.9)0.015
Lung involvement, all kinds12 (34.3)35 (60.3)0.019
Pulmonary hemorrhage7 (20.0)0 (0.0)0.001
Heart involvement2 (5.7)13 (22.4)0.042
Gastrointestinal involvement7 (20.0)13 (22.4)1.00
Peripheral neuropathy, all kinds25 (71.4)35 (60.3)0.37
Mononeuritis multiplex18 (51.4)14 (24.1)0.013
CNS involvement6 (17.1)7 (12.1)0.54
Renal involvement18 (51.4)7 (12.1)<0.001
RPGN10 (28.6)3 (5.2)0.004
ACR criteria30 (85.7)55 (94.8)0.15
Lanham's criteria30 (85.7)47 (81.0)0.78
Eosinophilia >10%32 (91.4)56 (96.6)0.36
Eosinophils/mm3, median (range)4,881 (1,074–28,815)3,544 (600–25,637)0.51
BVAS, 0–63, median (range)22 (7–40)17 (6–40)0.15
DEI, 0–21, median (range)6 (3–10)6 (3–10)0.85
VDI, 0–11, median (range)0 (0–2)0 (0–5)0.30
FFS ≥29 (25.7)7 (12.1)0.15
Figure 2.

Ratios of and differences in prevalence of organ involvement in antineutrophil cytoplasmic antibody (ANCA)–positive and ANCA-negative patients. Left, Ratio of the prevalence of a given type of organ involvement in ANCA-positive patients to the corresponding prevalence in ANCA-negative patients. A prevalence ratio of 2 means that ANCA-positive patients are twice as likely to be affected as ANCA-negative patients, while a prevalence ratio of 0.5 means that ANCA-positive patients are half as likely to be affected as ANCA-negative patients. Right, Difference in the prevalence of a given type of organ involvement in ANCA-positive patients from the corresponding prevalence in ANCA-negative patients. A prevalence difference of +20% means that the prevalence of a given type of organ involvement in ANCA-positive patients is 20 percentage points higher than the corresponding prevalence in ANCA-negative patients. Solid circles represent prevalence ratios or prevalence differences. Horizontal bars represent 95% confidence intervals. Arrows indicate confidence interval upper bounds extending beyond the limits of the plot. The prevalence ratio for pulmonary hemorrhage, which cannot be estimated since this did not occur in any ANCA-negative patient, is artificially plotted as an open circle. Vertical dotted lines represent the “null” value, i.e., the number corresponding to identical prevalence of organ involvement in ANCA-positive and ANCA-negative patients, which is 1 for the prevalence ratio and 0 for the prevalence difference. CNS = central nervous system; RPGN = rapidly progressive glomerulonephritis.

Among the 63 patients in whom at least 1 tissue biopsy was performed, 57 had pathologic alterations compatible with a diagnosis of CSS. The most frequent histologic finding in ANCA-negative patients was eosinophilic tissue infiltration (58.8% versus 13.8%; P < 0.001), usually with a perivascular pattern, whereas a (necrotizing) small-vessel vasculitis and/or capillaritis (including necrotizing crescentic glomerulonephritis) was more commonly found in ANCA-positive patients (75.9% versus 32.4%; P < 0.001).

An FFS ≥2 was found more frequently in ANCA-positive patients, even though the difference was not statistically significant (Table 4). Moreover, we did not find statistical evidence that ANCA positivity carried a worse prognosis. In fact, the 5-year survival rate was 91.8% for ANCA-positive patients compared with 97.1% for ANCA-negative patients (P = 0.74), whereas the risks of relapse at 5 years were 46.3% and 35.4%, respectively (P = 0.20). It should be noted, however, that ANCA-positive patients were more likely to be treated with cyclophosphamide (65.7% versus 32.7%; P = 0.003). Patients with anti-MPO antibodies and atypical patterns by immunofluorescence did not differ from other ANCA-positive patients.

ANCAs were present in only 4 of 37 patients (10.8%) tested at the end of the followup period; however, 1 of these patients still had active disease and died of a complication of an infection. ANCAs were present in 3 of 16 patients (18.8%) tested at the time of a relapse.

DISCUSSION

The present study shows that ∼40% of patients with CSS are ANCA positive. Although a classic pANCA pattern with specificity for MPO is found in most patients, atypical patterns are frequently encountered, including a cANCA pattern with anti-MPO antibodies. ANCA-positive patients are more likely than ANCA-negative patients to present with the typical clinicopathologic picture of the other small-vessel vasculitides and are less likely to have heart and nonhemorrhagic lung involvement.

ANCAs were first detected in sera from patients with necrotizing glomerulonephritis and systemic vasculitis (30). Later, these autoantibodies were found in a high proportion of patients with WG or MPA, including their renal-limited form (idiopathic pauciimmune crescentic necrotizing glomerulonephritis) (31, 32). CSS is usually classified among the so-called AASVs (10, 11); however, data on the prevalence and clinical significance of ANCAs in CSS are scarce and conflicting (8).

Considering the studies with at least 15 patients, ANCAs were found by immunofluorescence in 41.2–77.7% of patients and in 73 of 124 patients (58.9%) reported overall in these studies (7, 12, 14, 15, 33). We found ANCAs in ∼40% of our population, which is approximately in the range previously reported by some investigators (38.9–47.6%) (7, 13, 33), but considerably below the range found by others (58.8–77.7%) (12, 14, 15). However, in the latter 3 studies (12, 14, 15), only 17, 30, and 18 patients, respectively, were tested for ANCAs. In our experience, ANCA positivity in CSS depends very much on the specialty department of origin, which in turn reflects the pattern of organ system involvement of patients. This might explain, at least in part, some discrepancies. Worth noting is the fact that a recent, large, multicenter study of >100 patients with CSS, reported only in abstract form, has found ANCA positivity in 38% of patients (34), which is exactly the prevalence we found in our population. We confirm that pANCA with specificity for MPO are found in approximately three-fourths (or more) of ANCA-positive patients. However, in addition to the “classical” MPO pANCA and PR3 cANCA patterns, we also found cANCA and atypical ANCA patterns due to anti-MPO antibodies (35, 36).

In order to exclude a casual finding and/or an artifact due to the fixation procedure that we used, ANCA testing was repeated in a central laboratory with “home-made” and commercial kits. Moreover, we tested a similar number of consecutive serum samples with anti-MPO antibodies from patients with MPA or WG. The MPO cANCA pattern is reported very rarely in MPA and WG, and it has been postulated that this particular staining pattern might be due to the different epitopes recognized by these antibodies (35). However, it has also been reported that atypical staining patterns may be found more frequently with some commercially available ethanol-fixed neutrophil substrates, including the one used by ourselves in the reference laboratory (37, 38). It has been speculated that this phenomenon might be caused by factors in the ethanol fixation conditions of these slides resulting in the differential redistribution of different MPO epitopes (38).

In our series of patients, we found that ANCA positivity was correlated with renal involvement, especially with the histologic picture of necrotizing crescentic glomerulonephritis, and, to a lesser extent, with constitutional symptoms. Moreover, ANCA-positive patients had a significantly higher frequency of certain organ system clinical manifestations, such as pulmonary hemorrhage, purpura, and mononeuritis multiplex. In contrast, ANCA-negative patients had a higher frequency of heart and (less severe) lung disease.

Limited data have been reported on the correlation between ANCA positivity and the clinical features in CSS, even though it should be noted that most (if not all) reported cases of necrotizing crescentic glomerulonephritis in CSS (39–43) involved ANCA-positive (usually MPO pANCA) patients, as in our cohort. Moreover, the results of studies of small series of patients have suggested that MPO ANCAs may be associated with the onset of glomerular disorder in CSS (44, 45).

In the largest series of CSS patients reported so far, Guillevin et al looked for ANCAs in the sera of 42 patients at the time of diagnosis. ANCAs were detected in 20 of 42 patients (47.6%), a percentage close to ours. No details were given about the clinical associations with ANCA positivity; however, a histologically proven, rapidly progressive necrotizing glomerulonephritis was documented in only 3 patients (7). More recently, the same group has reported, in abstract form, that ANCAs were detected in 43 of 112 patients with CSS (38%), and that ANCA positivity at the time of diagnosis was associated with a significantly higher frequency of renal involvement (P < 0.001) with a relative risk of 14.3, whereas ANCA-negative patients had higher rates of cardiac involvement (P < 0.001) (34). In addition, Booth et al found, in a retrospective multicenter study on ANCA-associated renal vasculitis, that ANCAs were present in 92% of 256 patients, including most of the 11 patients with CSS (46). Keogh and Specks found that central nervous system involvement was the only clinical manifestation that correlated with ANCA status (14). In that study, only 30 of 91 patients were tested at the time of diagnosis; moreover, the prevalence of ANCA positivity was much higher (73.3%) than that in our cohort, and details about renal disease definition and renal histology were not given (14).

In CSS, the histopathology and possibly the pathogenesis of tissue lesions can be different in the various organs and within the same target organ (6–9). Although the pulmonary infiltrates in the prodromal and vasculitic phases of the disease may be similar radiographically, the histologic picture in the former phase is usually that of extensive eosinophilic infiltration of alveoli and interstitium, whereas necrotizing vasculitis and granulomas are seen in the latter phase (6). Another example is provided by gastrointestinal involvement, which includes features of both eosinophilic gastroenteritis and necrotizing vasculitis (6, 9). Peripheral nerve involvement might also be due to either vasculitis of vasa nervorum or perineural eosinophilic infiltrate (6). The eosinophil may also be directly responsible for some of the classic disease features of CSS, by virtue of the release of its stored cationic proteins, such as eosinophilic cationic protein, which is implicated in the cardiotoxicity that is seen both in CSS and in the hypereosinophilic syndrome, and eosinophil-derived neurotoxin, which may contribute to the development of peripheral neuropathy (8).

The clinical manifestations resulting from these different pathogenic mechanisms cannot be easily distinguished from each other without histologic examination. It is tempting to speculate that the clinical manifestations might be due mainly to the eosinophilic infiltrative disease in ANCA-negative patients, whereas the necrotizing vasculitis component might prevail in ANCA-positive patients. Indeed, in our experience, the most frequent histologic finding in ANCA-negative patients was perivascular and/or tissue eosinophil infiltration, whereas an eosinophil-rich, necrotizing small-vessel vasculitis was found mainly in ANCA-positive patients.

Emerging clinical and in vivo (animal model) observations provide compelling evidence that ANCAs are primarily and directly involved in the pathogenesis of AASVs (47). In particular, 2 different groups of investigators have demonstrated that anti-MPO antibodies alone can cause necrotizing crescentic glomerulonephritis and pulmonary hemorrhage in experimental models (47, 48). Moreover, Fienberg et al have reported that of 6 patients with negative test results for ANCAs and histologically diagnosed WG, none had evidence of renal involvement (49). In contrast, Cohen Tervaert et al found that all patients with WG and anti-MPO antibodies had renal involvement (50). In MPA, which is usually MPO ANCA positive, renal disease is almost invariably present (10, 11, 32). The presence of ANCAs in CSS has suggested that these autoantibodies could be an integral part of the inflammatory diathesis of this disease (9). ANCAs are capable of activating neutrophils in a wide variety of ways, resulting in the release of reactive oxygen species, granule proteins, cytokines, chemokines, and adhesion molecules. Leukocytes that have been activated by ANCAs adhere to endothelium and cause endothelial damage (47). Although we do not think that they are the main cause of the disease, they might act by amplifying inflammation in a subset of patients, and they might contribute directly to the development of necrotizing glomerulonephritis, as our results would suggest.

In conclusion, ANCAs are found less frequently in CSS than in other ANCA-associated vasculitides, such as WG and MPA. The main fluorescence pattern is perinuclear with antibodies to MPO; however, PR3 cANCA can be found in a minority of patients. In addition to the “classical” patterns, cANCA and atypical ANCAs with MPO specificity are demonstrable in a significant proportion of patients (apparently a much higher proportion than in patients with MPA or WG). ANCA positivity correlates with renal involvement (in particular, with necrotizing crescentic glomerulonephritis), with constitutional symptoms, and with other particular organ system manifestations, such as mononeuritis multiplex, purpura, and pulmonary hemorrhage. In contrast, ANCA negativity correlates with higher frequencies of lung involvement (with the exception of alveolar hemorrhage) and heart involvement.

Ancillary