Olaf Wiesner, 1 Kimberly A. Russell, 1 Augustine S. Lee, 1 Dieter E. Jenne, 2 Matteo Trimarchi, 3 Gina Gregorini, 3 and Ulrich Specks 1

Transcription

1 ARTHRITIS & RHEUMATISM Vol. 50, No. 9, September 2004, pp DOI /art , American College of Rheumatology Antineutrophil Cytoplasmic Antibodies Reacting With Human Neutrophil Elastase as a Diagnostic Marker for Cocaine-Induced Midline Destructive Lesions but Not Autoimmune Vasculitis Olaf Wiesner, 1 Kimberly A. Russell, 1 Augustine S. Lee, 1 Dieter E. Jenne, 2 Matteo Trimarchi, 3 Gina Gregorini, 3 and Ulrich Specks 1 Supported by the Mayo Foundation. Dr. Wiesner s work was supported by the Division of Pulmonology, Otto-von-Guericke University, Magdeburg, Germany. Dr. Russell s work was supported by NIH training grant HL Dr. Jenne s work was supported by project grant SFB571 from the German Research Council. Dr. Specks work was supported by NIH grants R21-AI and R01-AR and the Robert N. Brewer Foundation. 1 Olaf Wiesner, MD, Kimberly A. Russell, MD, Augustine S. Lee, MD, Ulrich Specks, MD: Mayo Clinic, Rochester, MN; 2 Dieter E. Jenne, MD: Max Planck Institute of Neurobiology, Martinsried, Germany; 3 Matteo Trimarchi, MD, Gina Gregorini, MD: University of Brescia, Brescia, Italy. Address correspondence and reprint requests to Ulrich Specks, MD, Division of Pulmonary and Critical Care Medicine, Mayo Clinic and Foundation, 200 First Street SW, Rochester, MN specks.ulrich@mayo.edu. Submitted for publication October 10, 2003; accepted in revised form May 20, Objective. Human neutrophil elastase (HNE) and proteinase 3 (PR3) are structurally and functionally related. PR3 is the prominent target antigen for antineutrophil cytoplasmic antibodies (ANCAs) in Wegener s granulomatosis (WG). Reported frequencies of HNE ANCAs in WG and other autoimmune diseases range from 0% to 20%. We previously detected HNE ANCAs in patients with cocaine-induced midline destructive lesions (CIMDL). We tested the hypothesis that discrepancies in the reported frequencies of HNE ANCAs in patients with vasculitis may be related to differences in detection methods, and that HNE ANCA may be a marker for CIMDL. Methods. HNE ANCA reactivity in 25 patients with CIMDL was characterized and compared with that in a control cohort of 604 consecutive patients (64 with WG, 14 with microscopic polyangiitis [MPA], and 526 others) and 45 healthy volunteers. HNE ANCAs were measured by indirect immunofluorescence using a previously undescribed expression system for recombinant HNE and by direct and capture enzyme-linked immunosorbent assays using purified native HNE as target antigen. Results. Among patients with CIMDL, HNE ANCAs were detectable by 1 assay in 84%, by 2 assays in 68%, and by all 3 assays in 36%. Fifty-seven percent of HNE ANCA positive CIMDL sera were also PR3 ANCA positive by at least 1 assay. In contrast, only 8 (1.3%) of 604 control sera reacted with HNE in at least 1 assay, 3 (0.5%) reacted in 2 assays, and only 1 serum sample (0.16%) reacted in all 3 assays. Sera obtained from patients with WG or MPA were universally HNE ANCA negative, as were sera obtained from healthy controls. Conclusion. Optimal sensitivity for HNE ANCA requires multimodality testing. HNE ANCAs are frequent in CIMDL but not in other autoimmune diseases, including classic ANCA-associated vasculitis. HNE ANCAs may discriminate between CIMDL and WG, whereas a positive test result for PR3 ANCA may not. Antineutrophil cytoplasmic antibodies (ANCAs) were first described 2 decades ago (1,2). By indirect immunofluorescence (IIF), 2 fluorescence patterns can be distinguished on ethanol-fixed neutrophil cytospin preparations (3,4). The classic cytoplasmic pattern (canca) is caused almost exclusively by antibodies against proteinase 3 (PR3) (5 7). In contrast, the perinuclear fluorescence pattern (panca) can be caused by antibodies reacting with a variety of different neutrophil granule constituents, including myeloperoxidase (MPO), lactoferrin, human neutrophil elastase (HNE), and others (8). Classic ANCA reacting with PR3 and panca reacting with MPO have become accepted diagnostic tools in the evaluation of patients with Wegener s granulomatosis (WG) and patients with microscopic polyangiitis (MPA) (9). Evidence is mounting 2954

2 HNE ANCA IN CIMDL AND AUTOIMMUNE DISEASES 2955 that these specific antibodies not only are disease markers but also are pathogenic of small-vessel vasculitis, the characteristic histopathologic hallmark of these disorders (10,11). PR3 and HNE belong to the chymotrypsin family of serine proteinases. They share gene localization, amino acid sequence homology, posttranslational intracellular processing, and substantial structural and functional characteristics (12). Despite these similarities and the prominence of PR3 ANCAs in patients with WG and patients with MPA, information about HNE ANCAs remains sparse and inconsistent (13 18). Among patients with autoimmune diseases (including vasculitis), the frequency of HNE ANCAs is thought to be low. However, reported frequencies are widely discrepant, ranging from 0% to 20% in patients with WG or MPA (14,15,17). Methodologic issues have been implicated but have not been addressed formally (17). We have expressed recombinant HNE (rhne) in the human mast cell line HMC-1 (HMC-1/HNE). Using HMC-1/HNE cells as substrate for HNE ANCA detection by IIF, we found a surprisingly high frequency (28%) of HNE ANCAs in a cohort of patients with cocaine-induced midline destructive lesions (CIMDL) (19). The present study was designed to characterize the presence of HNE ANCAs in patients with CIMDL in comparison with that in a clinically well-characterized cohort of consecutive patients undergoing an evaluation for possible vasculitis (20). HNE ANCA reactivity was determined by 3 different methods, complemented by standard ANCA testing with an IIF method using ethanol-fixed neutrophils as substrate, by PR3 ANCA testing using 3 different assays, and by MPO ANCA testing. PATIENTS AND METHODS Materials. Unless specified otherwise, all reagents were obtained from Sigma (St. Louis, MO). The polyclonal sheep and rabbit anti-hne antibodies were obtained from Biodesign (Saco, ME). The 2 mouse monoclonal anti-hne antibodies were from Dako (Carpenteria, CA) and PharMingen (San Diego, CA), respectively. Purified polymorphonuclear (PMN) HNE and PMN PR3 were purchased from Athens Research and Technology (Athens, GA). Functional expression of rhne in HMC-1 cells. The human mast cell line HMC-1, a kind gift from Dr. J. H. Butterfield (Mayo Clinic, Rochester, MN), was cultured in RPMI, 10% bovine calf serum, and 1 g/ml monothioglycerol, as previously described (21). The full-length complementary DNA (cdna) for wild-type HNE was amplified from total cdna of human bone marrow cells using the forward primer DJ568 (5 -CCGGATCCCCAGCCCCACCAT-3 ), the backward primer DJ569 (5 -GGCAGCCCTTCTCAGTGGGTC- 3 ), and Pfu DNA polymerase. The amplified product was digested with Bam HI and ligated into the Bam HI/Eco RV restricted expression vector pcdna3.1/zeo (Invitrogen, Carlsbad, CA). The cdna sequence was verified by comparison with the published sequence (GenBank accession number Y00477). The plasmid was transfected into HMC-1 cells by electroporation (21), and rhne-expressing HMC-1 cell populations were selected using Zeocin (100 g/ml) and cloning by limiting dilution. Sham-transfected (plasmid pcdna3 without insert) control cells (HMC-1/VEC) were cultured and selected in parallel under the same conditions. Enzymatic activity assay. The enzymatic activity of rhne in comparison with that of purified native HNE was determined by measuring the hydrolysis of the elastase substrate N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide (N- MeOSucc-AAPV-pNA), as described in detail elsewhere (21). Inhibitors, when used, were added during the first 30-minute incubation, prior to the addition of substrate, in the following concentrations: for 1 -plasmin inhibitor ( 1 PI), 50 g/ml; for eglin C, 0.1 M; and for aprotinin, 0.1 trypsin-inhibiting units/ml. Activity measurements of 20 ng of PMN HNE were performed in the presence of equivalent amounts of shamtransfected HMC-1/VEC cell lysate. IIF, biosynthetic labeling, and immunoprecipitation. IIF was performed using ethanol-fixed cytospin preparations of HMC-1/HNE, HMC-1/PR3, and HMC-1/VEC cells, as previously described (22). Serum samples were assayed in parallel on HMC-1/HNE or HMC-1/PR3 cells and their respective HMC-1/VEC control cells. Nonspecific fluorescence that was identical on HMC-1/VEC and HMC-1/HNE or HMC-1/PR3 cells was interpreted as a negative reaction (22). The standard IIF technique for ANCA detection was performed using ethanol-fixed neutrophil cytospin preparations (23). Patient sera were evaluated using 1:4 and 1:16 dilutions, as previously described and validated (20,22,23). Biosynthetic labeling of HMC-1/HNE and HMC-1/ VEC cells using 35 S-methionine/ 35 S-cysteine (ICN Biomedicals, Costa Mesa, CA), labeling of purified HNE with 3 H- diisopropylfluorophosphate (DFP), and immunoprecipitation using anti-hne antibodies and HNE ANCA were performed according to previously published procedures (21,24). Enzyme-linked immunosorbent assay (ELISA) methods for ANCA detection. PR3 ANCA and MPO ANCA were measured using capture and direct ELISA methods, as previously described (20,25). For HNE ANCA detection, a capture ELISA and a direct ELISA method were used. Purified PMN HNE, pretreated 1:1 with 1M DFP, was used as target antigen in both assays. Following the guidelines put forth by the National Committee for Clinical Laboratory Standards, intraand interassay coefficients of variation (CV) were determined by testing a low-positive and a high-positive serum sample 10 times within 1 assay and in 20 different consecutive assays, respectively (26). To determine the analytical sensitivity of the assays, serial dilutions of a positive and a negative control sample were examined. For the capture ELISA, microtiter wells (Immulon I; Dynatech, Chantilly, VA) were coated with polyclonal sheep anti-hne (0.1 mg/ml phosphate buffered saline [PBS]) for 18 hours at 4 C, washed, and blocked with PBS/0.1% bovine serum albumin (incubation buffer) for 1 hour at room temper-

3 2956 WIESNER ET AL Figure 1. Characterization of the recombinant expression system for human neutrophil elastase (HNE). A, Indirect immunofluorescence of ethanol-fixed HMC-1/HNE cells using 2 different monoclonal antibodies, M1 and M2. M1 generates granular cytoplasmic and perinuclear fluorescence. M2 recognizes only the recombinant HNE (rhne) in granules. B, Immunoprecipitation of 35 S-methionine/ 35 S-cysteine labeled cell extracts and of 3 H-diisopropylfluorophosphate ( 3 H-DFP) labeled purified polymorphonuclear (PMN) HNE. Cell lysates of HMC-1/HNE cells (lane 1) and HMC-1/VEC cells (lane 2) and 0.5 g of 3 H-DFP labeled purified PMN HNE (lane 3) were immunoprecipitated with 5 l of polyclonal rabbit anti-hne antibody. The antibody immunoprecipitated a kd protein from HMC-1/HNE cells but not from HNE/VEC cells. Immunoprecipitated proteins were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis on the same 12% gel. Autoradiography exposure times were 2 days for lanes 1 and 2, and 7 days for lane 3. Arrow indicates the gel front. C, Enzymatic activity of rhne. Recombinant HNE expressed in HMC-1 cells was processed to an active enzyme hydrolyzing the small synthetic peptide N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide (solid circles). Hydrolysis of the substrate was inhibited by 95% by 1.85 M 1 -plasmin inhibitor (open circles) and 0.1 M eglin C (open triangles), and by 83% by 0.1 trypsin-inhibiting units/ml aprotinin (open squares). Parallel control experiments were performed with purified PMN HNE. Purified PMN HNE (20 ng) was mixed with equivalent amounts of cell lysate from sham-transfected HMC-1/VEC cells ( cells/well). HMC-1/VEC cell lysates alone did not hydrolyze this substrate (21, and data not shown). Values are the mean SEM of 3 separate experiments, each of which was performed in triplicate. O.D. optical density. ature. Half of the wells were incubated with HNE (1 g/ml of incubation buffer) for 2 hours at room temperature. The other half of the wells was incubated with incubation buffer alone. Test samples diluted 1:20 in incubation buffer were incubated in duplicate HNE-saturated wells and in duplicate wells lacking antigen (background) for 1 hour at room temperature. Positive and negative control sera were included on each plate. Bound antibody was detected with an alkaline phosphatase conjugated goat anti-human IgG (1:10,000) followed by the color reaction using p-nitrophenyl phosphate (1 mg/ml in 0.1M Tris HCl, 5 mm MgCl 2, 0.1M NaCl [ph 9.6]). The absorbance (at 405 nm) was determined after 30 minutes. Results are expressed as the net absorbance, as determined by subtracting the background values from the values obtained in wells containing PMN HNE for each individual sample. This capture ELISA has an intraassay CV of % and an interassay CV of %. The cutoff value of 0.11 was based on the highest net absorbance obtained from the 45 normal control sera (median 0.01, range ) and was equivalent to 4 standard deviations above the mean. Even at a dilution of 2 11, the positive control serum yielded a net absorbance of For the direct ELISA, HNE (1 g/ml PBS) was coated to microtiter wells for 18 hours at 4 C. The remainder of the assay was performed as described for the capture ELISA. Because many sera displayed nonspecific binding to plastic, each serum sample was incubated in parallel in uncoated wells, and data are expressed as the net absorbance, as for the capture ELISA. The intraassay and interassay CVs of the direct ELISA were % and %, respectively. The cutoff value for this assay was determined to be This, again, was based on the highest net absorbance obtained from the healthy control group (median 0, range ). At a dilution of 2 10, the positive control serum still showed positivity above this cutoff value. Patient sera. Ninety-five serum samples obtained from 25 consecutive patients in whom CIMDL was detected between January 1991 and December 2001 were analyzed. ANCA frequencies were determined based on the first serum sample available from each patient. The clinical and histopathologic features of 18 of the patients with CIMDL included in this analysis have been described in detail in a previous report (19). Seven patients with CIMDL were identified subsequently, based on their clinicopathologic presentation and admitted cocaine abuse. One of these admitted cocaine users (CIMDL patient 19) also had WG, based on an independent histopathologic review of the nasal biopsy specimen (Facchetti F, Colby TV: personal communication) and a partial clinical response to immunosuppressive therapy. A total of 604 sera from a cohort of 615 consecutive patients undergoing routine ANCA testing as part of an evaluation for suspected vasculitis or autoimmune disease at the Mayo Clinic during a 10-month period (20) were also analyzed using the 3 different HNE ANCA determination methods. Samples from the remaining 11 patients were not sufficient in quantity to perform the HNE ANCA assays. Patients diagnosed with WG (n 64) fulfilled both the American College of Rheumatology 1990 criteria for WG (27) and the Chapel Hill Consensus Conference definition of WG (28). In patients with MPA (n 14), the diagnosis was made according to the definition of the Chapel Hill Consensus Conference (28). A clear diagnosis of autoimmune disease was

4 HNE ANCA IN CIMDL AND AUTOIMMUNE DISEASES 2957 error method for nonparametric data, using this same software (29). Figure 2. Human neutrophil elastase (HNE) antineutrophil cytoplasmic antibody detection by indirect immunofluorescence using HMC-1/HNE cells. A, Serum obtained from a patient with cocaineinduced midline destructive lesions (patient 6) was incubated on ethanol-fixed cytospin preparations of HMC-1/HNE cells and caused a granular immunofluorescence pattern. B, No immunofluorescence was generated by this serum sample on sham-transfected HMC-1/VEC cells. C, On ethanol-fixed neutrophil cytospin preparations, this particular sample caused a combined cytoplasmic and perinuclear staining pattern. D, Specific reactivity of this serum sample with HNE was verified by immunoprecipitation. Two micrograms of 3 H- diisopropylfluorophosphate labeled purified polymorphonuclear (PMN) HNE was immunoprecipitated with 65 g of polyclonal rabbit anti-hne IgG (lane 1) and 70 g of purified IgG from the same patient sample (lane 2). Lane 3 shows PMN HNE immunoprecipitated by IgG from patient 14 under the same conditions. The autoradiography exposure time was 6 days. established for 118 of the remaining 526 patients; these diagnoses included a variety of collagen vascular diseases, giant cell arteritis, drug-induced vasculitis, unclassified vasculitis, inflammatory bowel disease, autoimmune liver disease, autoimmune thyroid disease, and others. The routine ANCA test results in this cohort have been described in detail elsewhere (20). Sera obtained from 45 healthy volunteers served as negative controls. Use of the serum samples for the purpose of this study was approved by the Institutional Review Board. Statistical analysis. Correlations between values obtained with different assays were examined using Spearman s correlation test. The Mann-Whitney U test was used to determine the statistical significance of differences between groups. P values (2-tailed) less than 0.05 were considered significant. Receiver operating characteristic curves, plotted as sensitivity versus 1 specificity for every possible cutoff value, were generated using S-Plus version 6.12 statistical software (Insightful, Seattle, WA). The 95% confidence intervals for the area under the curve were calculated by applying the DeLong RESULTS To provide more definitive data about the frequency of HNE ANCAs in CIMDL and autoimmune vasculitis, we used 3 methods of HNE ANCA detection: an IIF method using a recombinant expression system for HNE, which has not been previously described in detail, a capture ELISA, and a direct ELISA. Purified native HNE was used as target antigen in both solidphase assays. Characterization of the expression system for rhne. The expression of rhne protein in stable clones of HMC-1/HNE cells was evaluated by IIF on ethanolfixed cytospin preparations. Both granular and perinuclear cytoplasmic fluorescence were detected in HMC- 1/HNE cells with monoclonal and polyclonal anti-hne antibodies, indicating partial rearrangement of rhne around the nucleus. Perinuclear rearrangement is a known artifact of ethanol fixation of neutrophils and seems to be a factor for HMC-1 cells as well (8). Anti-HNE antibodies generated no fluorescence signal in sham-transfected HMC-1/VEC control cells or HMC- 1/PR3 cells expressing rpr3 (22). Different monoclonal anti-hne antibodies yielded different fluorescence patterns (Figure 1A). The rabbit anti-human HNE antibody specifically immunoprecipitated rhne from extracts of biosynthetically labeled HMC-1/HNE cells as a kd band (Figure 1B). Pulse chase and deglycosylation experiments revealed that the larger molecular components of the immunoprecipitated band represent unprocessed pro-hne with larger sugar side chains (data not shown). To determine whether HMC-1 cells process rhne to an active enzyme, we measured the ability of HMC-1/HNE cell lysates to cleave the substrate N-MeOSucc-AAPV-pNA (Figure 1C). As previously reported (21,30), lysates of HMC-1 and HMC-1/VEC cells displayed no endogenous hydrolytic activity against this sensitive substrate for HNE. The hydrolysis of N-MeOSucc-AAPV-pNA by HMC-1/HNE lysates of 10 6 cells was equivalent to that of 40 ng of purified PMN HNE mixed in with cell lysate of 10 6 sham-transfected HMC-1/VEC cells (Figure 1C). The activity of rhne against this substrate was inhibited 95% by eglin C and 1 PI; this level of inhibition was similar to that observed with purified PMN HNE. Aprotinin was slightly less effective against rhne compared with purified PMN HNE, inhibiting the activity of rhne by 83% (Figure 1C).

5 2958 WIESNER ET AL Table 1. ANCA test results for 25 patients with CIMDL* HNE ANCA PR3 ANCA Patient IIF, PMN cells IIF, HMC-1/HNE cells Capture ELISA Direct ELISA IIF, HMC-1/PR3 cells Capture ELISA Direct ELISA 1 panca (0.26) (0.19) (43.8) 2 (0.23) (0.50) ND 3 panca (0.30) (1.22) 4 panca (0.44) 5 panca (0.38) (2.68) 6 panca (0.30) (0.65) (257.9) 7 canca (1.13) (2.30) 8 panca (0.98) 9 panca (0.63) (1.18) 10 panca (0.25) (0.17) (181.7) 11 panca (0.95) (0.37) 12 panca (0.86) ND 13 (0.32) ND (38.1) 14 panca (0.34) (0.36) (8.4) 15 ND ND ND ND 18 panca (0.64) (0.67) ND (13.0) 19 canca (0.64) (58.0) 20 (0.40) 21 panca (1.41) (0.90) (0.14) (11.5) 22 panca (0.35) (2.35) (0.71) 23 panca (1.74) (2.45) 24 panca (1.64) (3.00) 25 panca (0.32) (66.0) * ANCA antineutrophil cytoplasmic antibodies; CIMDL cocaine-induced midline destructive lesion; HNE human neutrophil elastase; PR3 proteinase 3; IIF indirect immunofluorescence; PMN polymorphonuclear neutrophil; ELISA enzyme-linked immunosorbent assay; panca perinuclear ANCA; ND not done; canca classic ANCA. Numbers in parentheses are net absorbance values. Numbers in parentheses are units/ml (normal 5). Using ethanol-fixed HMC-1/HNE cells, we previously detected HNE ANCA reactivity by IIF in patients with CIMDL (19). An example of HNE ANCA reactivity in a serum sample obtained from one of the CIMDL patients is shown in Figure 2. To confirm that sera yielding a strong IIF signal on HMC-1/HNE cells did indeed contain HNE ANCAs, IgG preparations from 5 of these sera were used to immunoprecipitate 3 H-DFP labeled purified PMN HNE (Figure 2D). The fluorescence pattern caused by HNE ANCA positive sera on HMC-1/HNE cells may vary. A few sera generated only granular cytoplasmic staining, some exhibited only perinuclear staining, and others displayed both patterns simultaneously. Taken together with the IIF results obtained with the 2 different monoclonal anti-hne antibodies, this difference in fluorescence pattern suggests that HNE ANCAs from different patients are directed against several different HNE epitopes. HNE ANCA are common in CIMDL. Table 1 shows the ANCA test results for the serum samples obtained at first presentation from the 25 patients classified as having CIMDL. By IIF on HMC-1/HNE cells, 14 patients (56%) had detectable HNE ANCAs. The direct ELISA also detected HNE ANCAs in 14 (56%) of these samples (median net absorbance 0.275, range 0 3.0) (Figure 3B). The capture ELISA was the most sensitive method, detecting HNE ANCAs in 19 (76%) of these sera (median net absorbance 0.275, range ) (Figure 3A). None of the methods identified all patients with detectable HNE ANCAs (n 21; 84%). HNE ANCAs were detectable by at least 2 methods in 17 patients (68%), but concordance between all 3 HNE ANCA detection methods was found in only 9 patients (36%). Even though most sera yielded a panca pattern on routine clinical ANCA testing using ethanol-fixed neutrophils as substrate, the IIF pattern fluctuated be-

6 HNE ANCA IN CIMDL AND AUTOIMMUNE DISEASES 2959 Figure 3. Human neutrophil elastase antineutrophil cytoplasmic antibody detection by capture and direct enzyme-linked immunosorbent assays (ELISAs). The net absorbance readings of the capture (A) and direct (B) ELISA for sera from patients with cocaine-induced midline destructive lesions (CIMDL; n 25), Wegener s granulomatosis (WG; n 64), microscopic polyangiitis (MPA; n 14), other autoimmune disorders including vasculitides (AID/VC; n 526), and healthy control volunteers (Co; n 45). Data in A and B are presented as box plots, where the boxes represent the interquartile range, the lines within the boxes represent the median, the open squares represent the means, the lines outside the boxes represent the 10th and 90th percentiles, represents the 1st and 99th percentiles, represents the maximum values of each category, and dotted line represents the cutoff values of the assays. The CIMDL samples yielded significantly higher values when compared with the samples from all other disease categories in both assays. In the direct ELISA, the control samples yielded significantly lower readings than all other groups. P 0.01; P Receiver operating characteristic curves for the capture ELISA (C) and direct ELISA (D). Arrows show the position of the cutoff values on the curve. For the capture ELISA, the corresponding sensitivity and specificity were 76% and 99%, respectively. For the direct ELISA, the sensitivity and specificity at the cutoff value were 40% and 90%, respectively. The validity of these 2 tests for the diagnosis of CIMDL was confirmed by area under the curve (AUC) values of 0.5. For the capture ELISA the AUC was 0.84 (95% confidence interval [95% CI] ), and for the direct ELISA the AUC was 0.69 (95% CI ). tween perinuclear and cytoplasmic staining on serial samples in some patients. Because the IIF pattern for each serum sample was stable in replicate assays, the pattern fluctuations of serial samples from the same patient indicated changes of antibody type over time. None of the 95 serum samples obtained from these 25 patients had MPO ANCAs. One of the patients with CIMDL (patient 19) consistently had a canca pattern on IIF testing using ethanol-fixed neutrophils as substrate and tested positive for PR3 ANCAs in all 3 target antigen specific assays. This patient was HNE ANCA negative by all 3 methods on the initial test and on 2 subsequent occasions. One of the serial serum samples obtained 16 months after this patient s initial clinical presentation tested weakly positive (0.17) in the HNE ANCA capture ELISA. In addition to the necrotizing inflammation seen in all of the biopsy specimens obtained from patients with CIMDL, the nasal biopsy for this patient revealed giant cells and microabscesses indicative of WG. HNE ANCA are rare in autoimmune disease and vasculitis. To determine whether HNE ANCAs could be detected in patients with vasculitis and other autoimmune diseases, we tested sera from 604 consecutive patients evaluated over a 10-month period (20), using all 3 methods. None of the patients with WG or MPA had positive results for HNE ANCAs in any of the assays. Among the remaining 526 patients, 118 had a clear-cut diagnosis of autoimmune disease/vasculitis. HNE ANCAs were identified by IIF using HMC-1/HNE cells in only 3 of these patients. Two of these patients had ulcerative colitis (UC), which was complicated by idiopathic thrombocytopenic purpura (ITP) in 1 patient and by sclerosing cholangitis in the other. The third patient had ulcerative esophagitis. Only the patient with UC/ITP showed positive reactivity in both solid-phase HNE ANCA assays and had a panca pattern on standard IIF ANCA testing using neutrophils. The other 2 patients showed positivity in only 1 of the ELISAs and lacked immunofluorescence on neutrophils (Table 2). Four (0.76%) of the 526 sera obtained from patients in the autoimmune/vasculitis disease control group had HNE ANCAs detectable by capture ELISA (median net absorbance 0.008, range ) (Figure 3A and Table 2). Two of these samples were obtained from the patients with UC. Only 1 of these patients was positive for pancas on neutrophils, both displayed perinuclear staining on HMC-1/HNE cells, and 1 showed weak cross-reactivity with HMC-1/PR3 cells but no reactivity in PR3 ANCA solid-phase assays. The clinical diagnoses of the 2 other patients were facial paresthesias of indeterminate etiology and Cogan s syndrome, respectively.

7 2960 WIESNER ET AL Table 2. Clinical diagnoses and ANCA test results for 8 patients with autoimmune disease and HNE ANCAs* HNE ANCA PR3 ANCA Patient Diagnosis IIF PMN IIF HMC-1/HNE cells Capture ELISA Direct ELISA IIF HMC-1/PR3 cells Capture ELISA 1 Ulcerative esophagitis (0.67) 2 Urticaria (0.34) 3 Renal insufficiency (0.25) (0.29) 4 Giant cell arteritis (0.34) (0.17) 5 UC/ITP panca (0.24) (0.27) 6 Facial paresthesias (0.16) 7 UC/SC (0.12) 8 Cogan s syndrome (0.24) *UC ulcerative colitis; ITP idiopathic thrombocytopenic purpura; SC sclerosing cholangitis (see Table 1 for other definitions). Numbers in parentheses are net absorbance values. Numbers in parentheses are units/ml (normal 5). Direct ELISA Five (0.95%) of the 526 sera tested positive for HNE ANCAs by direct ELISA (median net absorbance 0.033, range ) (Figure 3B and Table 2). One of these patients had giant cell arteritis, 1 had renal insufficiency, 1 had urticaria, and 1 had ulcerative esophagitis (Table 2). None of these sera from these 4 patients tested positive by capture ELISA. Only the 1 patient with UC/ITP showed positive reactivity in both the direct and the capture ELISAs. Sera from 2 of these patients also showed reactivity with PR3 by capture ELISA but not by other methods. To further validate the reactivity of these sera with HNE, we performed inhibition assays by preincubating the sera with excess purified PMN HNE. This resulted in a 50 94% signal reduction in the direct ELISA for 4 of these sera. The reactivity of the 1 serum sample obtained from the patient with renal insufficiency could not be inhibited by preincubation with HNE. Correlation of HNE ANCA results determined by the 2 ELISA methods. The solid-phase assay results for all samples that showed positive HNE ANCA reactivity in at least 1 assay (n 28) are plotted in Figure 4. The correlation did not reach statistical significance (r 0.207, P 0.29). However, analysis of serial samples from individual CIMDL patients (Figures 5B, C, and E) revealed a close relationship between net absorbance values obtained by the 2 methods in some patients, whereas in others the levels seemed to fluctuate in opposite directions (Figure 5D). Reactivity of HNE ANCA positive samples with PR3. PR3 ANCA test results are shown in Tables 1 and 2. Twelve of 21 (57%) of the HNE ANCA positive CIMDL samples and 3 of the HNE ANCA positive control samples reacted with PR3, as determined by different methods. Analysis of the serial CIMDL samples revealed that PR3 ANCA levels did not parallel HNE ANCA levels (Figures 5D and E). To determine whether reactivity with both HNE and PR3 was caused by cross-reactivity or by coexisting separate reactivities, cross-inhibition experiments were performed on serum samples obtained from 3 of the CIMDL patients, using the direct ELISA. Preincubation Figure 4. Correlation of net absorbance readings of positive human neutrophil elastase (HNE) antineutrophil cytoplasmic antibody (ANCA) sera obtained by capture and direct ELISA. The values shown are for the 20 samples obtained from patients with CIMDL (circles) and the 8 samples from the group with AID/VC that yielded a positive result in at least 1 of the ELISAs (squares). Positive reactivity by indirect immunofluorescence on HMC-1/HNE cells is represented by open circles and open squares, respectively. Solid lines represent the cutoff values for both assays. See Figure 3 for other definitions.

8 HNE ANCA IN CIMDL AND AUTOIMMUNE DISEASES 2961 Figure 5. Examples of serial proteinase 3 (PR3) antineutrophil cytoplasmic antibody (ANCA) and human neutrophil elastase (HNE) ANCA determinations in patients with cocaine-induced midline destructive lesions. Open circles represent results obtained by capture enzyme-linked immunosorbent assay (ELISA), and solid circles represent results obtained by direct ELISA. Solid horizontal lines indicate the cutoff value for a positive capture ELISA result, and dotted lines represent those for direct ELISA results. For many patients the HNE ANCA results obtained by the 2 methods correlated well with each other (B, C, and E). However, in some patients the HNE ANCA results seemed to fluctuate in opposite directions (D). No correlation between PR3 ANCA and HNE ANCA results was detectable (A E).

9 2962 WIESNER ET AL with DFP-inactivated PR3 resulted in a significant reduction of the HNE ANCA signal generated by the sera from 2 CIMDL patients. The reactivity of another serum sample was not affected. This suggests that some patients have at least partially cross-reacting antibodies, whereas others may have coexisting independent HNE ANCAs and PR3 ANCAs. DISCUSSION In this report we describe an unexpectedly high frequency (84%) of HNE ANCAs in patients presenting with CIMDL. In contrast, no HNE ANCAs were detected in patients with WG or MPA, and HNE ANCAs were detected only rarely in patients with other autoimmune diseases or vasculitis. Many of the sera obtained from patients with CIMDL also reacted with PR3. Consequently, HNE ANCAs occurring in patients with midline destructive lesions may be discriminatory between CIMDL and WG, whereas PR3 ANCA testing alone may not be relied upon for this purpose. Our findings, which were obtained with 3 different HNE ANCA detection methods, indicate that differences in the reported frequencies of HNE ANCAs in patients with autoimmune disorders and vasculitis are, in all likelihood, the result of differences in the detection methods applied. In order to achieve optimal analytical sensitivity for the detection of HNE ANCAs, multimodality testing appears preferable. We previously used HMC-1 cells for the expression of functional human and murine recombinant PR3 (21,30). Here, we demonstrate that HMC-1 cells can also be used to express functional rhne with a molecular mass similar to that of PMN HNE (31). Like purified native HNE, the rhne hydrolyses the substrate N-MeOSucc-AAPV-pNA and is inhibited by 1 PI, eglin C, and aprotinin. This indicates that HMC-1 cells process rhne amino-terminally, allowing the molecule to assume the proper conformation required for enzymatic function. Furthermore, this report represents the first direct documentation of enzymatic activity of rhne expressed in hematopoietic cells. Other investigators have previously expressed rhne in the rat basophilic cell line, RBL-1, a cell line also used for the expression of recombinant human cathepsin G and PR3 (32,33). Our choice of the human HMC-1 cell line over hematopoietic cell lines from other species is of particular importance when the cells expressing recombinant ANCA target antigens are used as substrate for target antigen specific ANCA testing by IIF. Many patients have antibodies reacting with rodent proteins that generate false-positive IIF results on rodent cells. The use of a human cell line for expression of rhne circumvents this problem. HMC-1/HNE cells can be used as target antigen specific substrate for the detection of HNE ANCAs by IIF, as previously demonstrated for PR3 ANCA detection (22). In a first application of this system we had found HNE ANCAs in 5 sera (28%) obtained from the first 18 patients in this CIMDL cohort, who were part of a clinicopathologic comparison of CIMDL and WG with nasal involvement (19). This initial observation prompted the present systematic and detailed analysis of ANCA reactivity in patients with CIMDL. In contrast to the first study, here we took a substantially different approach to the serologic testing. First, we extended the cohort of CIMDL patients by including 7 additional consecutive CIMDL patients who had presented to the participating centers since January Second, we based the ANCA frequency analysis solely on the first serum samples obtained at the time of presentation of the CIMDL patients. Third, we included an analysis of all available serial serum samples from this CIMDL cohort in order to identify any shifts in ANCA reactivity over time. Fourth, because of widely discrepant HNE ANCA frequency reports in other disorders, we used 3 different HNE ANCA test methods with 2 different antigen sources in parallel. Fifth, all sera were also tested in parallel by standard IIF on ethanol-fixed neutrophil cytospin preparations, for PR3 ANCAs using 3 different target antigen specific methods, and for MPA ANCAs by standard direct ELISA. Last, we chose a different control population. The choice of control populations is always an issue when reports on the diagnostic utility of ANCA testing are interpreted (34). The control population described in the original report included only patients with WG and nasal involvement (19). We retested the serum samples obtained from these 21 WG patients for HNE ANCAs by direct ELISA and capture ELISA and did not detect any HNE ANCAs (data not shown). Because other investigators have previously described HNE ANCAs in patients with a variety of inflammatory conditions (14,15,17,35), we wanted to include a different control population that contained additional WG patients with more heterogeneous organ involvement as well as patients with other autoimmune disorders. Furthermore, we wanted to minimize selection bias. A

10 HNE ANCA IN CIMDL AND AUTOIMMUNE DISEASES 2963 cohort that fulfilled all of these criteria was readily available from another study (20). By applying all 3 methods of HNE ANCA testing to the first serum samples obtained at the time of presentation, we found a surprisingly high frequency of HNE ANCAs (84%) in the 25 patients with CIMDL. The capture ELISA was the most sensitive (76%), whereas both the IIF method using HMC-1/HNE cells and the direct ELISA had a sensitivity of 56%. Overall, the correlation between HNE ANCA absorbance readings determined by capture and direct ELISA was poor (Figure 4). However, the analysis of serial samples obtained over the course of several years in some of the CIMDL patients indicates that in some patients these values change in parallel (Figures 5B, C, and E), whereas in others they fluctuate in opposite directions (Figure 5D). The IIF pattern on neutrophils and HMC- 1/HNE cells may also vary over time (data not shown). Together, these observations suggest that the HNE ANCA response in CIMDL patients is oligoclonal or polyclonal and variable over time. The high frequency of HNE ANCAs observed in patients with CIMDL is in stark contrast to the frequency in patients with WG, in whom HNE ANCAs were not detectable by any method. In contrast to only 1 serum sample from the control population (patient 5, who had ITP and UC), sera from many patients with CIMDL showed universal HNE ANCA reactivity across all assays. Another 7 samples from the control population showed positive reactivity for HNE ANCAs in at least 1, but not all, target antigen specific assays and lacked panca reactivity on neutrophils. This might suggest that they represented method-intrinsic falsepositive test results. However, this is unlikely, because the reactivity of 4 of the 5 sera that were positive in the direct ELISA could be inhibited specifically by preincubation of the sample with excess HNE. We could not confirm the previously reported high frequency of HNE ANCAs in patients with MPA, and we did not find any HNE ANCAs in patients with WG (15,17). It is of interest that in the report by Tervaert et al, 1 of the 2 HNE ANCA positive patients identified among a cohort of 315 patients with vasculitis and its mimickers had so-called chronic midline destructive disease (15). Several groups of investigators have reported observing HNE ANCAs in patients with systemic lupus erythematosus (SLE) (13,14,17,35). However, this finding could not be confirmed by others (15). Only 6 patients with a firm diagnosis of SLE were included in our cohort, but all of them were HNE ANCA negative in all assays. Some of the reported HNE ANCAs in patients with SLE may represent drug-induced ANCAs (13,16,18). In contrast to patients with WG and MPA, patients with drug-induced vasculitis commonly have multiple simultaneous ANCA specificities (16,18). Unlike the observations reported in other studies, we did not observe simultaneous reactivity with MPO in any of the HNE ANCA positive serum samples (14,16 18). However, sera from many of the patients with CIMDL and 3 of the other HNE ANCA positive sera also reacted positively with PR3 in at least 1 target antigen specific assay. The 2 patients with ulcerative colitis were receiving mesalamine, the therapeutically active part of the sulfasalazine molecule. For some of the CIMDL samples we could detect cross-inhibition of HNE ANCA reactivity by preincubation with PR3, suggesting that some ANCAs cross-react because they recognize antigenic determinants that are conserved on HNE and PR3 (36). Other CIMDL sera seem to contain separate HNE ANCA and PR3 ANCA species. This is also supported by the divergent serial titers (Figures 5D and E). How the exact PR3 epitopes recognized by double-positive sera compare with those recognized by single-positive PR3 ANCA sera from patients with WG, and whether they represent crossreacting or coexisting antibody species, may be of pathogenic relevance. This is being investigated separately. The fact that many HNE ANCA positive CIMDL sera also reacted with PR3, but none of the sera obtained from patients with WG or MPA showed any reactivity with HNE, is of clinical significance. First, it seems that the presence of HNE ANCAs is discriminatory, whereas the presence of PR3 ANCAs is not. Consequently, in the clinical setting of necrotizing inflammation of the upper respiratory tract, additional testing for HNE ANCAs may be useful for differentiating CIMDL from limited WG. Finally, our findings also illustrate the importance of adhering to current consensus guidelines for ANCA testing calling for corroboration of IIF results by target antigen specific assays or vice versa; only the canca/pr3 ANCA and the panca/mpo ANCA combinations have a clinically useful positive predictive value for ANCA-associated vasculitis (9,20,37). In this context, it is of interest that the only cocaine abuser with nasal septal perforation, who was thought to have WG based on the presence of microabscesses and giant cells in the nasal biopsy specimen (patient 19), had a canca pattern on neutrophils by IIF and showed positive reactivity in all 3 PR3 ANCA

11 2964 WIESNER ET AL assays and negative reactivity in all 3 HNE ANCA assays. Most of the patients with CIMDL had a panca pattern, but none reacted with MPO. ACKNOWLEDGMENTS We thank Amber M. Hummel, Margaret A. Viss, and Cari J. McDonald for excellent technical assistance with the transfection, cell culture, and immunoprecipitation procedures, Dr. James Lymp for statistical support, and Dr. David N. Fass for careful review of the manuscript and many helpful discussions. REFERENCES 1. Davies DJ, Moran JE, Niall JF, Ryan GB. Segmental necrotising glomerulonephritis with antineutrophil antibody: possible arbovirus aetiology. Br Med J (Clin Res Ed) 1982;285: Van der Woude FJ, Rasmussen N, Lobatto S, Wiik A, Permin H, van Es LA, et al. Autoantibodies against neutrophils and monocytes: tool for diagnosis and marker of disease activity in Wegener s granulomatosis. Lancet 1985;1: Falk RJ, Jennette JC. Anti-neutrophil cytoplasmic autoantibodies with specificity for myeloperoxidase in patients with systemic vasculitis and idiopathic necrotizing and crescentic glomerulonephritis. N Engl J Med 1988;25: Wiik A. Delineation of a standard procedure for indirect immunofluorescence detection of ANCA. APMIS 1989;97 Suppl 6: Jenne DE, Tschopp J, Ludemann J, Utecht B, Gross WL. Wegener s autoantigen decoded [letter]. Nature 1990;346: Ludemann J, Utecht B, Gross WL. Anti-neutrophil cytoplasm antibodies in Wegener s granulomatosis recognize an elastinolytic enzyme. J Exp Med 1990;171: Gupta SK, Niles JL, McCluskey RT, Arnaout MA. Identity of Wegener s autoantigen (p29) with proteinase 3 and myeloblastin [letter]. Blood 1990;76: Hoffman GS, Specks U. Anti-neutrophil cytoplasmic antibodies. Arthritis Rheum 1998;41: Savige J, Gillis D, Benson E, Davies D, Esnault V, Falk RJ, et al. International consensus statement on testing and reporting of antineutrophil cytoplasmic antibodies (ANCA). Am J Clin Pathol 1999;111: Russell KA, Specks U. Are antineutrophil cytoplasmic antibodies pathogenic? Experimental approaches to understand the antineutrophil cytoplasmic antibody phenomenon. Rheum Dis Clin North Am 2001;27: Xiao H, Heeringa P, Hu P, Liu Z, Zhao M, Aratani Y, et al. Antineutrophil cytoplasmic autoantibodies specific for myeloperoxidase cause glomerulonephritis and vasculitis in mice. J Clin Invest 2002;110: Jenne DE. Structure of the azurocidin, proteinase 3, and neutrophil elastase genes: implications for inflammation and vasculitis. Am J Respir Crit Care Med 1994;150:S Nassberger L, Sjoholm AG, Jonsson H, Sturfelt G, Akesson A. Autoantibodies against neutrophil cytoplasm components in systemic lupus erythematosus and in hydralazine-induced lupus. Clin Exp Immunol 1990;81: Gallicchio MC, Savige JA. Detection of anti-myeloperoxidase and anti-elastase antibodies in vasculitides and infections. Clin Exp Immunol 1991;84: Tervaert JW, Mulder L, Stegeman C, Elema J, Huitema M, The H, et al. Occurrence of autoantibodies to human leucocyte elastase in Wegener s granulomatosis and other inflammatory disorders. Ann Rheum Dis 1993;52: Short AK, Lockwood CM. Antigen specificity in hydralazine associated ANCA positive systemic vasculitis. QJM 1995;88: Apenberg S, Andrassy K, Worner I, Hansch GM, Roland J, Morcos M, et al. Antibodies to neutrophil elastase: a study in patients with vasculitis. Am J Kidney Dis 1996;28: Choi HK, Merkel PA, Walker AM, Niles JL. Drug-associated antineutrophil cytoplasmic antibody positive vasculitis: prevalence among patients with high titers of antimyeloperoxidase antibodies. Arthritis Rheum 2000;43: Trimarchi M, Gregorini G, Facchetti F, Morassi ML, Manfredini C, Maroldi R, et al. Cocaine-induced midline destructive lesions: clinical, radiographic, histopathologic, and serologic features and their differentiation from Wegener granulomatosis. Medicine (Baltimore) 2001;80: Russell KA, Wiegert E, Schroeder DR, Homburger HA, Specks U. Detection of anti-neutrophil cytoplasmic antibodies under actual clinical testing conditions. Clin Immunol 2002;103: Specks U, Fass DN, Fautsch MP, Hummel AM, Viss MA. Recombinant human proteinase 3, the Wegener s autoantigen, expressed in HMC-1 cells is enzymatically active and recognized by c-anca. FEBS Lett 1996;390: Specks U, Wiegert EM, Homburger HA. Human mast cells expressing recombinant proteinase 3 (PR3) as substrate for clinical testing for anti-neutrophil cytoplasmic antibodies (ANCA). Clin Exp Immunol 1997;109: Specks U, Wheatley CL, McDonald TJ, Rohrbach MS, DeRemee RA. Anticytoplasmic autoantibodies in the diagnosis and follow-up of Wegener s granulomatosis. Mayo Clin Proc 1989;64: Sun J, Fass DN, Viss MA, Hummel AM, Tang H, Homburger HA, et al. A proportion of proteinase 3-specific anti-neutrophil cytoplasmic antibodies only react with proteinase 3 after cleavage of its N-terminal activation dipeptide. Clin Exp Immunol 1998;114: Sun J, Fass DN, Hudson JA, Viss MA, Wieslander J, Homburger HA, et al. Capture-ELISA based on recombinant PR3 is sensitive for PR3-ANCA testing and allows detection of PR3 and PR3- ANCA/PR3 immunecomplexes. J Immunol Methods 1998;211: National Committee for Clinical Laboratory Standards. Evaluation of precision performance of clinical chemistry devices: approved guideline. NCCLS publication EP5-A. Villanova (PA): NCCLS; Leavitt RY, Fauci AS, Bloch DA, Michel BA, Hunder GG, Arend WP, et al. The American College of Rheumatology 1990 criteria for the classification of Wegener s granulomatosis. Arthritis Rheum 1990;33: Jennette JC, Falk RJ, Andrassy K, Bacon PA, Churg J, Gross WL, et al. Nomenclature of systemic vasculitides: proposal of an international consensus conference. Arthritis Rheum 1994;37: Hajian-Tilaki KO, Hanley JA. Comparison of three methods for estimating the standard error of the area under the curve in ROC analysis of quantitative data. Acad Radiol 2002;9: Jenne DE, Frohlich L, Hummel AM, Specks U. Cloning and functional expression of the murine homologue of proteinase 3: implications for the design of murine models of vasculitis. FEBS Lett 1997;408: Baugh RJ, Travis J. Human leukocyte granule elastase: rapid isolation and characterization. Biochemistry 1976;15: Gullberg U, Lindmark A, Lindgren G, Persson A-M, Nilsson E, Olsson I. Carboxy-terminal prodomain-deleted human leukocyte elastase and cathepsin G are efficiently targeted to granules and

12 HNE ANCA IN CIMDL AND AUTOIMMUNE DISEASES 2965 enzymatically activated in the rat basophilic/mast cell line RBL. J Biol Chem 1995;270: Garwicz D, Lindmark A, Hellmark T, Gladh M, Jogl J, Gullberg U. Characterization of the processing and granular targeting of human proteinase 3 after transfection to the rat RBL or the murine 32D leukemic cell lines. J Leukoc Biol 1997;61: Langford CA. The diagnostic utility of c-anca in Wegener s granulomatosis. Cleve Clin J Med 1998;65: Nassberger L, Jonsson H, Sjoholm AG, Sturfelt G. Circulating anti-elastase in systemic lupus erythematosus [letter]. Lancet 1989;1: Dalpe G, Beliveau A, Menard HA, Boire G. Cross-reactivity of an anti-proteinase 3 antibody to elastase. Autoimmunity 1994;17: Stone JH, Talor M, Stebbing J, Uhlfelder ML, Rose NR, Carson KA, et al. Test characteristics of immunofluorescence and ELISA tests in 856 consecutive patients with possible ANCA-associated conditions. Arthritis Care Res 2000;13: