Autoantibodies against granulocyte colony ... - Wiley Online Library

ARTHRITIS & RHEUMATISM Vol. 46, No. 9, September 2002, pp 2384–2391 DOI 10.1002/art.10497 © 2002, American College of Rheumatology

Autoantibodies Against Granulocyte Colony-Stimulating Factor in Felty’s Syndrome and Neutropenic Systemic Lupus Erythematosus Bernhard Hellmich,1 Elena Csernok,2 Helmut Schatz,3 Wolfgang L. Gross,2 and Armin Schnabel2 Objective. Cytokines and growth factors can be a target of autoantibodies in systemic inflammatory diseases. We examined whether patients with neutropenia and either Felty’s syndrome (FS) or systemic lupus erythematosus (SLE) have autoantibodies against granulocyte colony-stimulating factor (G-CSF) and whether these autoantibodies are functionally relevant. Methods. Fifteen patients with neutropenia due to FS were matched for age, sex, and disease activity with 16 normocytic rheumatoid arthritis (RA) control patients. Sixteen patients with SLE and neutropenia were matched with 16 normocytic SLE control patients. Antibodies against G-CSF were measured by enzymelinked immunosorbent assay and Western blotting. Antibody specificity was verified by competitive inhibition using recombinant human G-CSF. The effect of anti–G-CSF antibodies on the functional activity of their target molecule was measured in a bioassay using G-CSF–sensitive murine 32D cells. Results. IgG anti–G-CSF was found in 11 FS patients, 6 SLE patients with neutropenia, 6 SLE control patients, and none of the RA control patients. IgM anti–G-CSF was found in 6 neutropenic and 3 normocytic SLE patients. Anti–G-CSF antibodies were associated with an exaggerated serum level of G-CSF and a

low neutrophil count. A neutralizing effect of anti–GCSF antibodies on its target molecule was found in 3 of the 9 patients tested. Irrespective of the presence or absence of anti–G-CSF antibodies, neutropenic patients with FS and SLE had exaggerated serum levels of G-CSF. Conclusion. Anti–G-CSF autoantibodies are common in neutropenia due to FS and SLE. In individual patients, these autoantibodies have a neutralizing capacity. In patients without neutralizing antibodies, hyposensitivity of the myeloid cells to G-CSF appears to be central to the pathogenesis of the neutropenia in FS and SLE. Felty’s syndrome (FS) is clinically defined by the coexistence of rheumatoid arthritis (RA), neutropenia, and splenomegaly (1). Neutropenia in FS is usually chronic and is often associated with substantial morbidity related to recurrent infections. Leukopenia also occurs in up to 50% of patients with systemic lupus erythematosus (SLE) at some point during the course of the disease (2). The pathophysiology of neutropenia in FS and SLE is complex and appears to involve both cellular and humoral immune mechanisms. Increased neutrophil margination and increased peripheral neutrophil destruction, possibly due to autoantibodies directed against polymorphonuclear neutrophils, may contribute to the development of neutropenia in some patients (3,4). In addition, suppression of neutrophil production and maturation by T lymphocytes (CD4⫹, CD8⫹, and CD57⫹) and autoantibodies against CD34⫹ hematopoietic progenitor cells have been demonstrated (5–8). If an immune-mediated pathogenesis is supposed, potential targets of immune injury are not only myeloid cells, but also mediators involved in the regula-

1 Bernhard Hellmich, MD: Medizinische Universita¨t zu Lu ¨beck, Lu ¨beck, Germany, and Universita¨tsklinik Bergmannsheil, RuhrUniversita¨t Bochum, Bochum, Germany; 2Elena Csernok, PhD, Wolfgang L. Gross, Professor of Medicine, Armin Schnabel, MD: Medizinische Universita¨t zu Lu ¨beck, Lu ¨beck, Germany; 3Helmut Schatz, Professor of Medicine, Universita¨tsklinik Bergmannsheil, Ruhr-Universita¨t Bochum, Bochum, Germany. Address correspondence and reprint requests to Bernhard Hellmich, MD, Medizinische Klinik und Poliklinik, Universita¨tsklinik Bergmannsheil, Ruhr-Universita¨t Bochum, Bu ¨rckle-de-la-Camp Platz 1, Bochum 44789, Germany. E-mail: [email protected]. Submitted for publication December 17, 2001; accepted in revised form May 17, 2002.

2384

ANTI–G-CSF ANTIBODIES IN FS AND SLE

tion of myelopoiesis. That cytokines and hematopoietic growth factors can indeed be the target of autoimmunity has been demonstrated for autoimmune diseases (9–12) as well for lymphoproliferative disorders (13,14). Granulocyte colony-stimulating factor (G-CSF) is an endogenous hematopoietic growth factor that stimulates the proliferation, differentiation, and maturation of neutrophil precursors in the bone marrow by binding to a specific cell-surface receptor. It is essential for the maintenance of a normal neutrophil count (15). Antibodies against G-CSF have been found in patients with malignant lymphomas and appear to include not only antibodies induced by therapeutically administered G-CSF, but also autoantibodies against endogenous G-CSF (13). The aim of this study, therefore, was to examine sera from patients with neutropenia due to FS or SLE for the presence of antibodies against G-CSF. In addition, serum levels of G-CSF were measured and the findings were correlated with clinical and hematologic parameters. PATIENTS AND METHODS Patients. Fifteen consecutive patients with FS and 16 consecutive SLE patients with chronic neutropenia, all of whom had a mean absolute neutrophil count of ⬍2,000/␮l on at least 3 days, were recruited for this study. One control group consisted of 16 patients with RA but without current or previous neutropenia, who were matched with the FS patients by age, sex, glucocorticoid dosage, and disease activity. Disease activity in the FS and RA patients was assessed by a count of swollen joints and serologic markers of inflammation (erythrocyte sedimentation rate [ESR] and C-reactive protein [CRP]). The SLE patients with neutropenia were matched by age, sex, glucocorticoid dosage, and disease activity with a control group of 16 SLE patients without current or previous neutropenia. Disease activity in the SLE patients was measured by the SLE Disease Activity Index (SLEDAI) (16) and the European Consensus Lupus Activity Measure (ECLAM) (16). All patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) classification criteria for RA (17) or SLE (18). Drug-induced neutropenia was ruled out by careful scrutiny of the patients’ histories and, if necessary, by withdrawal and rechallenge with any indictable agent. No patients had received recombinant human G-CSF (rHuG-CSF) for therapeutic purposes at or before the time of serum collection. Determination of hematologic parameters. A white blood cell count with differential cell count, hemoglobin level, hematocrit value, and platelet count were determined by an automated cell counting system. Blood smears were studied microscopically for the presence of large granular lymphocytes, and patients with evidence of that condition were excluded from study.

2385

Bone marrow smears and trephine biopsies were performed on all patients with FS and on 13 of the 16 SLE patients with neutropenia. The results disclosed no evidence of hematologic malignancies, myelodysplastic syndrome, or other causes of neutropenia unrelated to the rheumatic disease. Sera from most patients with FS and SLE were examined for autoantibodies against white blood cells by antigranulocyte immunofluorescence, antigranulocyte agglutination, and antilymphocyte immunofluorescence tests. Determination of G-CSF in serum. Serum samples were stored at ⫺80°C. G-CSF was determined using a highly sensitive quantitative sandwich enzyme-linked immunosorbent assay (ELISA) obtained from R&D Systems (Minneapolis, MN). In this assay, there is no cross-reactivity with granulocyte–macrophage colony-stimulating factor or with a broad array of other cytokines (interleukin-1 [IL-1], IL-13, tumor necrosis factor ␣, interferon-␥, transforming growth factor ␤, etc.). The intra- and interassay coefficients of variation were 6.5% and 7.0%, respectively. The concentration range of the assay is 0.5–80 pg/ml. Detection of antibodies to rHuG-CSF by ELISA. Serum samples were screened for the presence of anti–G-CSF antibodies by ELISA technique, as previously described (13). Briefly, 96-well microtiter plates were coated with rHuG-CSF (Neupogen; Amgen, Luzern, Switzerland) at a concentration of 0.5 ␮g/100 ␮l of carbonate buffer (pH 9.6) per well, blocked with 1.5% bovine gelatin, and incubated overnight at room temperature. Wells not coated with G-CSF were used as controls to exclude nonspecific binding. For detection of IgG antibodies, serum samples were diluted 1:20 in Tris HCl buffer, pH 8.1, containing lactoferrin, gelatin, bovine serum albumin (BSA), and 0.5% Tween 20. Because the presence of rheumatoid factor (RF) can give false-positive results in IgM assays, serum samples for the determination of IgM antibodies were preincubated with RF adsorbent (Behring Diagnostica, Marburg, Germany) for 15 minutes at room temperature. Determination of RF activity confirmed the complete removal of RF by this procedure. To remove antibodies against Escherichia coli that might crossreact with any residual E coli antigen in the rHuG-CSF preparation, serum samples were also preincubated with an E coli lysate (Sigma, Munich, Germany). After these preparatory steps, the coated microtiter plates were incubated with 100 ␮l of serum per well for 2 hours at room temperature. The wells were washed in phosphate buffered saline (PBS) containing 0.5% Tween 20. Peroxidaseconjugated anti-human IgM (1:40 dilution) or anti-human IgG (1:40 dilution) was added, and samples were incubated for 2 hours at room temperature. After washing 4 times, 0.1 ml of the substrate solution (tetramethylbenzidine) was added to each well, and incubation continued for 30 minutes. The enzyme reaction was stopped by adding 0.5% sulfuric acid. The plates were read at 405 nm using a Behring ELISA Processor (Behring Diagnostica). The mean plus 3 standard deviations of the value in 120 healthy donors served as the upper limit of normal. The intra- and interassay coefficients of variation were 8% and 16%, respectively. The specificity of the ELISA was confirmed by an inhibition assay and Western blotting. For inhibition, sera testing positive for anti–G-CSF antibodies (1:20 dilution) were incubated with increasing concentrations of rHuG-CSF at 37°C

2386

for 1 hour and then at 4°C overnight. The mixtures were then measured in the anti–G-CSF ELISA as described above. Inhibition was calculated according to the formula: [(OD inhibited ⫺ OD background)/(OD not inhibited ⫺ OD background)] ⫻ 100 ⫽ % inhibition, where OD represents optical density at 405 nm. Purification of IgG. Anti–G-CSF–positive sera from patients with FS (n ⫽ 9) and anti–G-CSF–negative sera from patients with RA (n ⫽ 2), SLE (n ⫽ 2), and healthy controls (n ⫽ 3) were chromatographed on protein G–Sepharose (Pharmacia, Freiburg, Germany). The protein concentration in the eluted fraction was measured using both chemical (BCA protein assay reagent; Pierce Europe, Heidelberg, Germany) and immunologic assays. IgG concentrations were always ⬎90%, and preparations were endotoxin-free, as assayed by the kinetic–okra leaf curl Limulus amebocyte cell lysate test (Boehringer Ingelheim, Heidelberg, Germany). The volume of remaining serum from the other anti–G-CSF–positive patients was too small to purify sufficient amounts of IgG for further in vitro analysis. Western blotting. Sera positive by ELISA were also examined by Western blotting. Protein gel electrophoresis was performed under both native and denatured conditions using rHuG-CSF (10 ␮g/lane) in a sodium dodecyl sulfate gel buffer and 12% acrylamide gel. The protein was electroblotted onto nitrocellulose membrane (Hybond C-Extra; Amersham, Braunschweig, Germany). Each blot was verified by AuroDye protein staining (Amersham). Immunoblotting was performed after blocking nonspecific binding with 3% BSA/PBS for 1 hour at room temperature. Incubation with the IgG preparations outlined above was performed overnight at room temperature, with continuous agitation. A polyclonal rabbit anti– G-CSF (10 ␮g/ml; R&D Systems, Wiesbaden, Germany) served as a positive control. After washing, the membranes were incubated with goat anti-human IgG or goat anti-rabbit IgG labeled with colloidal gold (Amersham). The signal was amplified using a silver enhancement kit according to the manufacturer’s instructions (Amersham). Bioassay for G-CSF activity. The G-CSF receptor– expressing 32D cell line (kindly provided by Prof. Iwo Touw, Erasmus University, Rotterdam, The Netherlands) was generated by inserting the complementary DNA encoding the receptor for human G-CSF into the murine IL-3–dependent myeloid cell line 32D (19). These cells stably express the human G-CSF receptor, as verified by flow cytometry. When cultured with G-CSF alone for 7–10 days, these cells differentiate to mature neutrophils and die. When cultured without IL-3 and G-CSF, these cells die within 24 hours. The 32D cells were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum, 1% penicillin/ streptomycin, and 10 ng/ml of recombinant murine IL-3 (rMuIL-3; R&D Systems) at 37°C. Cells were washed twice and cultured with increasing doses of rHuG-CSF (0.1–100 ng/ml) and 10 ng/ml of rMuIL-3. The effect of the IgG from the neutropenic FS patients on cell proliferation was assessed by examining preparations of IgG from each of 9 patients as well as from each normocytic healthy control (n ⫽ 3). In addition, sera from 2 normocytic SLE patients and 2 normocytic RA patients matched for age and sex were used as disease controls. IgG preparations were

HELLMICH ET AL

added to the culture medium to give a final concentration of 0.1–50 ␮g/ml. A control experiment in which IgG was not added was also performed. Proliferation was measured after 3 days using a bromodeoxyuridine cell proliferation ELISA (Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s instructions. Results were expressed as a stimulation index, which was calculated from the OD readings as follows: stimulation index ⫽ OD value experimental/OD value unstimulated background. Statistical analysis. The Kolmogorov-Smirnov Z test was used to analyze the distribution of the samples. Normally distributed data are expressed as the mean ⫾ SEM; nonparametrically distributed data are expressed as the median (25th, 75th percentiles). For normally distributed data, statistical significance was tested using Student’s t-test and using analysis of variance in cases of ⬎2 groups. For nonparametrically distributed data, the Kruskal-Wallis test was used to compare values between ⬎2 groups, and the Mann-Whitney U test was used to compare values between 2 groups. The chi-square test with Yates’ correction was applied, if appropriate. Regression analysis was used to calculate Spearman’s correlation coefficients. Statistical analysis was performed using SPSS for Windows, version 8.0 (SPSS, Chicago, IL).

RESULTS Characteristics of the study population. Table 1 shows the basic clinical and serologic characteristics of the 4 study groups. Age, sex, and glucocorticoid dosage were similar in all 4 groups. Clinical as well as serologic parameters of disease activity, such as the CRP, ESR, serum complement levels, and antinuclear antibody titers, were not statistically significantly different among the 4 groups (data not shown). Prevalence of antibodies against G-CSF. IgG antibodies against G-CSF were detected by ELISA in 11 of the 15 FS patients (73%). One patient also had IgM anti–G-CSF antibodies. In contrast, no patient in the RA control group had either IgG or IgM antibodies against G-CSF. The prevalence of anti–G-CSF antibodies was lower in SLE patients with neutropenia than in FS patients, and the same antibodies were also found in the SLE controls. IgG anti–G-CSF was present in 6 SLE patients with neutropenia and in 6 SLE controls, whereas IgM anti–G-CSF was present in 6 SLE patients with neutropenia (40%) but only 3 SLE controls (19%). This difference was statistically significant (␹2 ⫽ 5.4; P ⫽ 0.02). All sera that yielded positive results on ELISA were assayed by Western blotting (Figure 1) and by an inhibition ELISA. These confirmatory tests gave identical results for all sera tested. The specificity of the anti–G-CSF autoantibodies was verified by an inhibition ELISA, which demonstrated substantial attenuation of


2387

Table 1. Clinical and laboratory characteristics of the study patients* SLE patients

Age, mean ⫾ SEM years Sex, no. male/female WBC count, /␮l ANC, /␮l GC dosage, mg/day DMARD therapy, no.

FS patients (n ⫽ 15)

RA patients (n ⫽ 16)

P†

With neutropenia (n ⫽ 16)

Without neutropenia (n ⫽ 16)

P†

54 ⫾ 2.0 2/13 1,340 (900–2,450) 292 (165–857) 3 (2–5) 4

54 ⫾ 2.4 3/13 7,800 (6,175–10,975) 5,423 (4,704–8,310) 5 (0–9) 11

⬎0.2 ⬎0.2 0.001 0.001 ⬎0.2 0.01

56 ⫾ 2.2 3/13 2,510 (1,700–3,200) 1,450 (700–1,963) 5 (0–7.5) 10

51 ⫾ 2.7 2/14 5,085 (3,297–7,095) 3,151 (2,192–6,169) 5 (5–7.5) 12

⬎0.2 ⬎0.2 0.001 0.001 ⬎0.2 ⬎0.2

* Values for the white blood cell (WBC) count, the absolute neutrophil count (ANC), and the glucocorticoid (GC) dosage are the median (interquartile range). FS ⫽ Felty’s syndrome; RA ⫽ rheumatoid arthritis; SLE ⫽ systemic lupus erythematosus; DMARD ⫽ disease-modifying antirheumatic drug. † P values for age were determined by Student’s t-test, those for sex and DMARD therapy, by chi-square test, and those for all others, by Mann-Whitney U test.

the readings in response to the addition of exogenous rHuG-CSF. Serum levels of G-CSF. The median serum level of G-CSF was 82.6 pg/ml (interquartile range [IQR] 33.3–104.2) in the FS patients, compared with 15.1 pg/ml (IQR 11.4–26.5) in the RA control patients; the difference was statistically significant (P ⫽ 0.001) (data not shown). The median serum G-CSF level was 32.5 pg/ml (IQR 9.5–73.1) in the patients with SLE and neutropenia; this was somewhat higher, although not statistically significantly higher (P ⫽ 0.144), than the median in the SLE control patients (22.1 pg/ml [IQR 14.3–30.2]). The serum level of G-CSF correlated negatively with the absolute neutrophil count in the RA patients (r ⫽ ⫺0.48; P ⫽ 0.03) and in the SLE patients (r ⫽ ⫺0.41; P ⫽ 0.12) who did not have anti–G-CSF antibodies, although the correlation was not statistically significant in the SLE patients (possibly due to the smaller sample size) (data not shown). This inverse correlation between the G-CSF level and the absolute neutrophil count was not seen in either patients with FS (P ⫽ 0.16) or patients with SLE (P ⫽ 0.31) who had anti–G-CSF antibodies. There was no correlation of the serum G-CSF level with either the ESR, CRP level, hemoglobin value, platelet count, or ECLAM or SLEDAI scores in any of the 4 subgroups. Characteristics of patients with anti–G-CSF antibodies. To analyze the clinical significance of the anti–G-CSF antibodies, antibody-positive FS patients were compared with the combined group of antibodynegative FS patients and RA control patients (Table 2). An association between anti–G-CSF antibodies and high serum levels of G-CSF as well as low absolute neutrophil counts was found (Figure 2). Antibody-positive FS patients had a median serum G-CSF level of 90.2 pg/ml (IQR 63.4–113.1), which was significantly higher than

the 17.8 pg/ml (IQR 12.5–30.7) level found in the combined group of antibody-negative FS and RA patients (P ⫽ 0.001). The serum concentrations of G-CSF

Figure 1. Western blotting of anti–granulocyte colony-stimulating factor (anti–G-CSF) antibodies. Serum anti–G-CSF antibodies were detected in sera from Felty’s syndrome patients (lanes A and B) (indicated by the arrow at the right). A positive control (prestained with a rabbit anti–G-CSF antibody) (lane C) and a negative control (anti–G-CSF–negative serum by enzyme-linked immunosorbent assay) (lane D) were included. Mobilities of the molecular weight standards are indicated to the left: 1 ⫽ carbonic anhydrase at 36.5 kd; 2 ⫽ soybean trypsin inhibitor at 26.6 kd; 3 ⫽ lysozyme at 16.0 kd; 4 ⫽ aprotinin at 8.4 kd; and 5 ⫽ insulin chain B at 3.8 kd.

2388

HELLMICH ET AL

Table 2. Characteristics of patients with and without anti–G-CSF antibodies* Rheumatoid arthritis and Felty’s syndrome patients

ANC, /␮l WBC count, /␮l Serum G-CSF, pg/ml ESR, mm/hour CRP, mg/dl Hypocomplementemia, no. Antineutrophil antibodies, no. ANA titer ECLAM score SLEDAI score

Systemic lupus erythematosus patients and controls

Anti–G-CSF positive (n ⫽ 11)

Anti–G-CSF negative (n ⫽ 20)

P†

Anti–G-CSF positive (n ⫽ 16)

Anti–G-CSF negative (n ⫽ 16)

P†

292 (180–580) 1,330 (800–2,000) 90.2 (63.4–113.1) 48 (30–94) 1.7 (0.9–11.3) 5/11 7/7 0 (0–128) ND ND

5,244 (2,164–6,717) 6,760 (3,290–9,395) 17.8 (12.5–30.7) 28 (15–80) 0.9 (0.5–4.8) 1/21 0/1 0 (0–1,280) ND ND

0.001 0.001 0.001 0.05 0.04 0.005 ⬍0.001 ⬎0.2 – –

1,989 (1,136–2,255) 2,965 (2,425–4,397) 28.2 (14.3–55.7) 33 (30–61) 0.8 (0.5–2.8) 11/16 4/8 1,280 (320–2,560) 3.5 (3–4) 8 (7–15)

2,754 (1,504–6,192) 3,670 (2,950–7,200) 22.8 (9.5–32.5) 24 (12–50) 0.5 (0.5–0.6) 9/15 3/3 1,280 (640–2,560) 3 (2–5) 8 (3–13)

0.063 0.090 ⬎0.2 0.053 0.03 ⬎0.2 0.086 ⬎0.2 ⬎0.2 ⬎0.2

* Patients were considered to be anti–G-CSF antibody positive if sera contained IgM anti–G-CSF, IgG anti–G-CSF, or both. Values for the absolute neutrophil count (ANC), white blood cell (WBC) count, serum granulocyte colony-stimulating factor (G-CSF), erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), antinuclear antibody (ANA) titer, European Consensus Lupus Activity Measure (ECLAM), and the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) are the median (interquartile range). Values for hypocomplementemia and antineutrophil antibodies are the number positive/number tested. ND ⫽ not done. † P values were determined by Mann-Whitney U test, except those for hypocomplementemia and antineutrophil antibodies, which were determined by chi-square test.

in the antibody-negative FS patients ranged from 19.4 pg/ml to 44.5 pg/ml. The median absolute neutrophil count in the antibody-positive FS patients was 292/␮l (IQR 180–580) compared with a median of 5,244/␮l (IQR 2,164–6,717) in the antibody-negative patients (P ⫽ 0.001). The same analysis was performed on the SLE patients and disclosed a median serum G-CSF level of 42.2 pg/ml (IQR 19.0–76.5) in those with IgM anti–GCSF compared with 21.3 pg/ml (IQR 9.2–30.8) in those without IgM anti–G-CSF (P ⫽ 0.045) (Figure 2). The

difference in the serum G-CSF level was of smaller magnitude in the SLE patients with IgG anti–G-CSF antibodies (28.3 pg/ml [IQR 14.3–55.7] versus 21.3 pg/ml [IQR 9.2–30.8]; P ⬎ 0.2). The respective absolute neutrophil counts were 1,989/␮l (IQR 400–2,560) and 2,754/␮l (IQR 1,504–6,192), a difference that approached statistical significance (P ⫽ 0.063). Antibody-positive FS and SLE patients also had a significantly higher CRP level than the antibodynegative FS and RA patients (P ⫽ 0.04) and SLE patients (P ⫽ 0.03) (Table 2). In addition, a tendency

Figure 2. A, Absolute neutrophil counts and B, serum levels of granulocyte colony-stimulating factor (G-CSF) in 15 patients with Felty’s syndrome (FS), 16 patients with rheumatoid arthritis (RA), and 32 patients with systemic lupus erythematosus (SLE). The FS patients and RA control patients as well as the SLE patients with neutropenia and SLE control patients were grouped according to the presence and absence of serum antibodies (AB) to G-CSF. Patients were considered to be anti–G-CSF antibody positive if sera contained IgM anti–G-CSF, IgG anti–G-CSF, or both. The 4 lowest data points in the second column of A represent FS patients who were negative for anti–G-CSF antibodies; the 2 highest data points in the first column of B represent 2 of these FS patients, whereas the other 2 FS patients had only modestly elevated G-CSF levels. Bold horizontal bars show the median; lighter horizontal bars show the 25th and 75th percentiles.


Figure 3. Proliferation of the murine 32D cell line that overexpresses granulocyte colony-stimulating factor (G-CSF) receptor. The 32D cells were incubated with increasing doses of G-CSF (0.1–100 ng/ml) in the presence or absence of IgG isolated from Felty’s syndrome (FS) patients (n ⫽ 9) who tested positive for IgG anti–G-CSF by enzymelinked immunosorbent assay (ELISA), neutralization ELISA, and Western blotting. IgG was isolated as described in Patients and Methods. After 3 days of culture, cell proliferation was determined by a cell proliferation ELISA, and a stimulation index was calculated (see Patients and Methods). Values are the mean of 3 individual experiments for each FS patient, each performed separately for all G-CSF concentrations. Control experiments were performed without IgG as well as with IgG fractions from anti–G-CSF–negative rheumatoid arthritis (n ⫽ 2) and systemic lupus erythematosus (SLE; n ⫽ 2) patients, and healthy controls (n ⫽ 3). IgG from 2 FS patients blocked the increased proliferation when the G-CSF concentration was raised to 1 ng/ml (FS patient 8) or to 10 ng/ml and 100 ng/ml (FS patient 4), suggesting a neutralization of G-CSF activity in vitro.

toward a higher ESR in patients positive for anti–G-CSF was evident in both the SLE and the FS groups; this difference approached statistical significance (P ⫽ 0.053). The antinuclear antibody titer (Table 2), the prevalence of antibodies against double-stranded DNA, Sm, Ro/SSA, and cardiolipin (data not shown) and the ECLAM and SLEDAI scores (Table 2) were virtually identical between antibody-positive and antibodynegative SLE patients. Examination of bone marrow disclosed maturation arrest of the granulopoiesis in all anti–G-CSF– positive FS patients and in 9 of the 12 anti–G-CSF– positive SLE patients. Antineutrophil antibodies were detected in all 7 anti–G-CSF–positive FS patients tested and in 4 of the 8 anti–G-CSF–positive SLE patients tested (Table 2). Effect of G-CSF antibodies in the G-CSF bioassay. IgG preparations were produced from 9 FS sera that tested positive for IgG anti–G-CSF by ELISA and

2389

Western blot. These samples were evaluated for their ability to interfere with the proliferation-enhancing effect of G-CSF on 32D cells. Cells cultured without IgG showed a dose-dependent increase in proliferation in response to G-CSF concentrations of 0.1–100 ng/ml. The increase in the proliferation index at 100 ng/ml, expressed as a percentage of the stimulation index at 0.1 ng/ml of G-CSF, was taken as a measure of the responsiveness of 32D cells to the growth factor. In the presence of IgG preparations from healthy donors, the stimulation index was augmented by 150– 173% (mean 160%), which was very similar to the increase without IgG. In the presence of IgG from the neutropenic FS patients, the stimulation index was augmented by 3.8–218%. IgG from 3 FS patients (patients 2, 4, and 9) diminished the increased stimulation index to less than half of the mean value in the control patients (Figure 3). We considered this to be a biologically important effect. IgG preparations from anti–G-CSF– negative healthy controls and anti–G-CSF–negative patients with RA and SLE showed no attenuating effect on G-CSF–dependent proliferation, demonstrating that the inhibitory effect of the IgG from some of the anti–GCSF–positive sera was most likely due to a neutralizing effect on G-CSF and not to other disease-specific factors contained in the IgG preparation. DISCUSSION In the present study, we demonstrated the presence of autoantibodies against G-CSF in a substantial proportion of patients with neutropenia due to FS or SLE. Care was taken to avoid false-positive results due to nonspecific binding, cross-reactivity, or inadequate sensitivity and specificity. Since the rHuG-CSF used for the binding of anti–G-CSF in the ELISA system we used was generated by expression in genetically engineered strains of E coli, all serum samples were pretreated by absorption with E coli lysate. This step should have eliminated from the rHuG-CSF preparation any residual E coli material that might have bound to anti–E coli antibodies in the patient sera and thereby produced falsely elevated ELISA readings (20). Moreover, interference in the ELISA system by RF was precluded by preadsorption of RF. The specificity of the ELISA results was confirmed by inhibition of antibody binding by an excess of rHuG-CSF in a neutralization assay. The cutoff level for a positive ELISA result was set at 3 standard deviations above the mean in 120 healthy blood donors, which excludes low positive results, which have been reported to occur in up to 10% of healthy individ-

2390

uals (13). Having taken these precautions, we are confident that our measurements were specific for anti–GCSF. Since no patient had ever been treated with exogenous G-CSF, the antibodies we identified are autoantibodies. We found that anti–G-CSF autoantibodies were associated with a persistently low neutrophil count and an exaggerated serum level of G-CSF. Obviously, this suggests hyposensitivity of the myeloid cells for the growth factor, which led us to examine whether the anti–G-CSF antibodies had a neutralizing effect on their target molecule. Sufficient amounts of serum from which to prepare IgG fractions were available from 9 FS patients with anti–G-CSF antibodies, the neutralizing capacity of which was then tested in a bioassay. The bioassay utilized a G-CSF receptor–transfected 32D cell line that grows in a dose-dependent manner in response to G-CSF. IgG preparations obtained from 3 of the 9 anti–G-CSF–positive FS patients blunted the G-CSF– dependent proliferation by ⬎50%, compared with IgG from normocytic healthy controls. We considered this to be a biologically important effect. The finding of neutralizing antibodies against G-CSF in the presence of low neutrophil counts and a maturation arrest of granulopoiesis suggests that in these patients, the anti–GCSF antibodies contribute to the development of neutropenia. The IgG of the remaining 6 patients tested in the bioassay had no substantial effect on the response of the 32D cells to G-CSF. These anti–G-CSF–positive FS patients had exaggerated levels of G-CSF in their circulation, but the antibodies appeared not to affect the biologic activity of the G-CSF. A central pathogenetic feature of the pathogenesis of the neutropenia in these patients thus appears to be an impaired response of the myeloid cells to G-CSF. The reasonable assumption that a variety of cellular and humoral inflammatory mechanisms are involved in the hyporesponsiveness of the myeloid cells in both SLE and FS is supported by the findings of several other studies (7,8,21–23). Autoantibodies against myeloid precursors, such as CD34⫹ hematopoietic progenitor cells, have been reported to occur in a substantial proportion of FS and SLE patients with neutropenia (6,21). It was suggested that in SLE, T lymphocytes are involved in the dysregulation of hematopoietic progenitor cell growth (24). Suppression of neutrophil production and maturation by T lymphocytes (CD4⫹, CD8⫹, and CD57⫹) has recently been demonstrated (5). In some patients with FS, suppressor T cells appear to be involved in defective bone marrow maturation (8,22). The findings of the present study therefore

HELLMICH ET AL

support the view that the pathophysiology of the neutropenia in FS is heterogeneous, but in individual patients, autoantibodies that neutralize the biologic effect of G-CSF appear to be an important pathophysiologic factor. Our results extend the findings of previous studies demonstrating antibodies against cytokines and growth factors in generalized autoimmune diseases (9,11,25) as well as in lymphoproliferative diseases (13,20). Antibodies against G-CSF have been found in patients with malignant lymphomas and appear to include not only antibodies induced by therapeutically administered G-CSF, but also autoantibodies against endogenous G-CSF (13,26). As we have shown, the functional role of anticytokine and anti–growth factor antibodies appears to be heterogeneous. While some anticytokine antibodies block the activity of their target molecule, as has been shown for anti–IL-1␣ in patients with RA (10), other anticytokine antibodies appear to function as stabilizers of their respective cytokine (11,27). The presence of non-neutralizing anti–IL-6 antibodies in sera from patients with systemic sclerosis has been shown to be associated with increased serum levels of IL-6 and increased IL-6 activity of the IL-6– anti–IL-6 antibody complex (11). Neutralizing as well as non-neutralizing antibodies against granulocyte– macrophage colony-stimulating factor have been found in patients with myasthenia gravis (9). Antierythropoietin antibodies have been found in a subset of patients with SLE (12,28,29). While in one study, these antibodies appeared to be associated with anemia in SLE (28), another study was unable to confirm this association with anemia, but found decreased serum levels of erythropoietin (29). However, data on the neutralizing capacity of these antierythropoietin antibodies in a bioassay have not been reported to date. In summary, we found autoantibodies to G-CSF in the majority of our patients with FS and in a significant percentage of our SLE patients with neutropenia. In individual patients, the antibodies had a neutralizing effect on the biologic effect of their target molecule, which suggests that in these subjects, the antibodies are causally involved in the pathogenesis of the neutropenia. In the majority of our patients, the anti–G-CSF antibodies did not alter the biologic effect of their target molecule. The exaggerated serum levels of G-CSF, concomitant with a depressed absolute neutrophil count, suggest that in these patients, hyposensitivity of the myeloid cells to G-CSF is central to the pathogenesis of the neutropenia. Since previous observations suggest that immunosuppressive treatment can


alleviate the neutropenia in FS and SLE as well as enhance the therapeutic effect of exogenous G-CSF, the pathogenesis of the hyposensitivity to G-CSF is likely to be of immunologic origin. Mechanisms other than neutralizing anti–G-CSF autoantibodies need to be clarified in more detail. ACKNOWLEDGMENTS The authors wish to thank Professor Iwo Touw (Erasmus University, Rotterdam, The Netherlands) for providing the G-CSF receptor–expressing 32D cell line and Professor Albert von dem Borne and Dr. Masja de Haas (Central Laboratory of The Netherlands Red Cross Blood Transfusion Service, Amsterdam, The Netherlands) for advice on developing the G-CSF bioassay.

REFERENCES 1. Campion G, Maddison PJ, Goulding N, James I, Ahern MJ, Watt I, et al. The Felty syndrome: a case-matched study of clinical manifestations and outcome, serologic features, and immunogenetic associations. Medicine (Baltimore) 1990;69:69–80. 2. Keeling DM, Isenberg DA. Haematological manifestations of systemic lupus erythematosus. Blood Rev 1993;7:199–207. 3. Rustagi PK, Currie MS, Logue GL. Complement-activating antineutrophil antibody in systemic lupus erythematosus. Am J Med 1985;78:971–7. 4. Hartman K. Anti-neutrophil antibodies of the immunoglobulin M class in autoimmune neutropenia. Am J Med Sci 1994;308:102–5. 5. Coakley G, Iqbal M, Brooks D, Panayi GS, Lanchbury JS. CD8⫹,CD57⫹ T cells from healthy elderly subjects suppress neutrophil development in vitro: implications for the neutropenia of Felty’s and large granular lymphocyte syndromes. Arthritis Rheum 2000;43:834–43. 6. Liu H, Ozaki K, Matsuzaki Y, Abe M, Kosaka M, Saito S. Suppression of haematopoiesis by IgG autoantibodies from patients with systemic lupus erythematosus (SLE). Clin Exp Immunol 1995;100:480–5. 7. Starkebaum G, Price TH, Lee MY, Arend WP. Autoimmune neutropenia in systemic lupus erythematosus. Arthritis Rheum 1978;21:504–12. 8. Abdou N. Heterogeneity of bone marrow–directed immune mechanisms in the pathogenesis of neutropenia of Felty’s syndrome. Arthritis Rheum 1983;26:947–53. 9. Meager A, Wadhwa M, Bird C, Dilger P, Thorpe R, NewsomDavis J, et al. Spontaneously occurring neutralizing antibodies against granulocyte-macrophage colony-stimulating factor in patients with autoimmune disease. Immunology 1999;97:526–32. 10. Suzuki H, Ayabe K, Kamimura J. Anti-IL-1␣ autoantibodies in patients with rheumatic diseases and in healthy subjects. Scand J Immunol 1991;30:219–23. 11. Suzuki H, Takemura H, Yoshizaji K, Koishihara Y, Ohsugi Y, Okano A, et al. IL-6-anti-IL-6 autoantibody complexes with IL-6 activity in sera from some patients with systemic sclerosis. J Immunol 1994;152:935–42. 12. Tzioufas AG, Kokori SI, Petrovas CI, Moutsopoulos HM. Autoantibodies to human recombinant erythropoietin in patients with

2391

13.

14. 15. 16.

17.

18. 19.

20.

21. 22. 23. 24.

25. 26.

27. 28.

29.

systemic lupus erythematosus: correlation with anemia. Arthritis Rheum 1997;40:2212–6. Laricchia-Robbio L, Moscato S, Genua A, Liberati AM, Revoltella RP. Naturally occurring and therapy-induced antibodies to human granulocyte colony-stimulating (G-CSF) in human serum. J Cell Physiol 1997;173:219–26. Revoltella R. Natural and therapeutically induced antibodies to cytokines. Biotherapy 1998;10:321–31. Welte K, Gabrilove J, Bronchud MH, Plazer E, Morstyn G. Filgrastim (r-metHuG-CSF): the first 10 years. Blood 1996;88: 1907–29. Vitali C, Bencivelli W, Isenberg DA, Smolen JS, Snaith ML, Sciuto M, et al, and The European Consensus Study Group for Disease Activity in SLE. Disease activity in systemic lupus erythematosus: report of the Consensus Study Group of the European Workshop for Rheumatology Research. II. Identification of the variables indicative of disease activity and their use in the development of an activity score. Clin Exp Rheumatol 1992;10:541–7. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25:1271–7. Ward AC, Smith L, de Koning JP, van Aesch Y, Touw IP. Multiple signals mediate proliferation, differentiation, and survival from the granulocyte colony-stimulating factor receptor in myeloid 32D cells. J Biol Chem 1999;274:14956–62. Ragnhammar P, Friesen H, Fro ¨din J, Lefvert A, Hassan M, ¨ sterborg A, et al. Induction of anti-recombinant human granuO locyte-macrophage colony-stimulating factor (Escherichia coliderived) antibodies and clinical effects in non-immunocompromised patients. Blood 1994;84:4078–87. Starkebaum G, Singer J, Arend WP. Humoral and cellular immune mechanisms of neutropenia in patients with Felty’s syndrome. Clin Exp Immunol 1980;39:307–14. Abdou N, NaPombejara C, Balentine L, Abdou N. Suppressor cell-mediated neutropenia in Felty’s syndrome. J Clin Invest 1978;61:738–43. Breedveld F, Lafeber G, Doekes G, Claas F, Cats A. Felty syndrome: autoimmune neutropenia or immune-complex-mediated disease? Rheumatol Int 1985;5:253–8. Bailey FA, Lilly M, Bertoli LF, Ball GV. An antibody that inhibits in vitro bone marrow proliferation in a patient with systemic lupus erythematosus and aplastic anemia. Arthritis Rheum 1989;32: 901–5. Jouvenne P, Miossec P. Autoantibodies to cytokines in inflammatory joint disorders. Rev Rhum Engl Ed 1995;62:541–5. Revoltella R, Laricchia-Robbio L, Moscato S, Genua A, Liberati A. Natural and therapy-induced anti-GM-CSF and anti-G-CSF antibodies in human serum. Leukemia Lymphoma 1997;26 Suppl 1:29–34. Bendtzen K, Hansen M, Ross C, Svenson M. High avidity autoantibodies to cytokines. Immunol Today 1998;19:209–11. Voulgarelis M, Kokori S, Ioannidis J, Tzioufas A, Kyriaki D, Moutsopoulos H. Anemia in systemic lupus erythematosus: aetiological profile and a role of erythropoietin. Ann Rheum Dis 2000;59:217–22. Schett G, Firbas U, Fureder W, Hiesberger H, Winkler S, Wachauer D, et al. Decreased serum erythropoietin and its relation to anti-erythropoietin antibodies in anemia of systemic lupus erythematosus. Rheumatology 2001;40:424–31.