Tom Eirik Mollnes Department of Immunology and Transfusion Medicine, Nordland Central Hospital, N-8017 Bodø, Norway
Several complement modulating effects of high-dose intravenous immunoglobulins (IVIG) have been proposed from in vitro studies and experimental animal models. However, the in vivo effects of IVIG on plasma complement in humans are yet not known. We have investigated the in vivo effects of IVIG on complement in seven women with unexplained recurrent spontaneous abortion who were without evidence of autoimmune disease. Samples were obtained before and after the very first infusion of IVIG. There was a marked increase in immunoglobulin G (IgG) from (median and range) 12.4 (9.4–15.9) to 26.8 (22.4–30.0) g/l but no change in immunoglobulin A (IgA) or immunoglobulin M (IgM). A significantly increased complement activation was demonstrated using neoepitope-specific enzyme immunoassays to the activation products C3bc (median increased from 9.8 to 31.2 AU/ml), Bb (0.66–1.66 g/ml), C5a (10.5–12.7 ng/ml), and TCC (0.81–2.19 AU/ml) (P = 0.015 for all). There were no changes in antigenic concentrations of individual complement components or regulators (C1q, C4, C3, C1-inhibitor, C4b-binding protein) and no decrease in complement haemolytic activity (classical and alternative CH50), which were all within the normal range. The classical pathway activation products C1rs/C1-inhibitor complexes, C4bc and C4d were elevated in all patients before IVIG treatment and did not change significantly during treatment. In conclusion, IVIG induced a significant activation of complement in vivo.
High-dose intravenous immunoglobulins (IVIG) are used as treatment for an increasing number of autoimmune and inflammatory conditions [1, 2], but this treatment is generally accepted for routine use in only a few conditions, such as idiopathic thrombocytopenic purpura, Kawasaki's syndrome and steroid-resistant dermatomyositis. Several reports have documented a beneficial effect of IVIG on both autoimmune and alloimmune recurrent abortions [3–10] but contradictory results have been reported [11–13]. The data were recently reviewed  and it can be concluded that larger controlled studies are required before definite conclusions can be drawn.
The aim of the present study was, however, not to investigate the effect of IVIG on the outcome of pregnancies in women with recurrent abortions, but rather to study the effect of IVIG on complement in vivo in patients undergoing such a treatment. Modulation of complement is one of the several documented immunomodulatory properties of IVIG that may contribute to its clinical effects. It has been shown that IVIG treatment in patients with dermatomyositis reduces the deposition of complement in muscle tissue [15, 16]. This strongly suggests a modulation of complement action by IVIG in the fluid phase, but to our knowledge such studies have not been performed in humans. We therefore investigated the effect of IVIG on complement by using specific assays for complement activation products, and by measuring individual complement components and functional complement activities.
MATERIALS AND METHODS
Seven women were selected for first-time treatment with IVIG because of unexplained recurrent spontaneous abortion. They were all treated at the Department of Gynaecology and Obstetrics, University Tor Vergata, Rome . The treatment protocol was as follows: 0.5 g IVIG (Alphaglobin®; see below for details) per kg body weight were infused during about 3 h. The concentration was 5 g/100 ml and the total volumes of infusions were thus in the range of 500–700 ml. The patients remained in the hospital during the day of infusion and were discharged in the evening as no adverse reactions were observed.
The patients were considered to have unexplained recurrent spontaneous abortion without evidence of an autoimmune condition, including the presence of circulating immune complexes, antiphospholipid antibodies and/or other autoantibodies (screened as part of the routine testing before inclusion in the study). The panel of autoantibodies was as follows: anti-nuclear antibodies (ANA, ENA), antibodies to phospholipids (anticardiolipin immunoglobulins G and M (IgG and IgM) and lupus anticoagulant), smooth muscle, mitochondria, thyreoperoxidase and thyreoglobulin. Other possible causes of recurrent spontaneous abortion (anatomical, endocrinological, genetical, infective, etc) had been excluded as far as possible. No patient received any medication before IVIG infusion and no drugs were administered, except for IVIG, during the study period.
Characterization of Alphaglobin®
Alphaglobin, manufactured by Alpha Therapeutic Corporation (Los Angeles, CA, USA), by a process involving liquid heat treatment (10 h at 60°C) for optimal inactivation of both lipid and non-lipid-coated virus, was supplied by Alpha Therapeutic Italia S.p.A. as a sterile 5% solution ready for use. It was kept stored at 2–8°C and warmed to room temperature before infusion. The preparation of Alphaglobulin is performed by means of Cohn-Oncley fractionation, polyethylene glycol precipitation and DEAE Sephadex filtration. The preparation contains > 97% IgG (mean 99.23% in the product imported to Italy and used in this work). The content in monomers and dimers is nearly 100% with the absence of fragments and of aggregates. Immunoglobulin A (IgA) content is < 0.008 mg/ml and IgM < 0.011 mg/ml. The four IgG subclasses were represented with a pattern similar to that found in normal human plasma.
Blood samples were taken immediately before and 30 min after the end of infusion of IVIG as complement activation would occur immediately after infusion and most of the activation products tested have half-lives around this time limit. EDTA-plasma was obtained by drawing blood directly into a tube containing EDTA. The tube was immediately cooled on crushed ice, centrifuged at 2000 g for 10 min and plasma was aliquoted and stored at −70°C until analysed in one batch. The EDTA-plasma samples were used for analysis of complement activation products. Serum was obtained by taking blood into empty tubes, allowing it to clot at room temperature for 2 h and then centrifuged. After centrifugation, serum was removed, aliquoted and stored at − 70°C until analysed in one batch. Serum was used for quantification of serum proteins and for the functional complement analyses.
Serum albumin and total serum protein were analysed in the routine laboratory at the Institute of Clinical Biochemistry, The National Hospital, Oslo, using standard techniques.
Immunoglobulins (IgG, IgA, IgM) were measured in the routine laboratory at the Institute of Immunology and Rheumatology, Oslo, using a standard nephelometric technique.
The antigen concentration of C1q was measured using enzyme immunoassay (EIA) as described previously . Briefly, goat antiserum to C1q (Quidel, San Diego, CA, USA) was used as capture antibody at a dilution of 1:10 000. A human serum pool from healthy blood donors was used as a reference, defining 100%. A monoclonal antibody (MoAb) to C1q (Quidel) was used as secondary antibody at a concentration of 0.1 mg/l.
The antigen concentration of C1-inh was measured by single radial immunodiffusion (NOR-Partigen, Behringwerke A/G, Marburg, Germany) and performed according to the manufacturer's instructions. C1-inh function was measured using a chromogene substrate assay, described in detail previously .
The antigen concentration of C3 and C4 was measured in the routine laboratory at the Institute of Immunology and Rheumatology, Oslo, using a standard nephelometric technique.
The antigen concentration of C4b-binding protein was measured using EIA, as described previously . Briefly, sheep antiserum to C4bp (Nordic Immunological Laboratories, Tilburg, the Netherlands) was used as capture antibody, diluted 1:10 000. A human serum pool from healthy blood donors was used as a reference, defining 100%. A MoAb to C4bp (Quidel) was used as a secondary antibody at a concentration of 0.1 mg/l.
Complement haemolytic activity
Classical and alternative CH50 were measured using microwell techniques, as described in detail previously .
Complement activation products
All assays for complement activation products were EIAs based on neoepitope-specific MoAbs to the activation products.
C1rs/C1-inhibitor complexes (C1rs/C1inh) were measured as described in detail previously  using the KOK-12 MoAb that is specific for a neoepitope in C1-inh when it is in complex with the protease . Human serum activated with heat-aggregated IgG was used as a standard and defined as 1000 arbitrary units (AU)/ml.
C4bc (i.e. the sum of C4b, iC4b and C4c) was measured mainly as described in detail previously . The same standard was used as for the C1rs/C1inh assay, defined to contain 1000 AU/ml. The MoAbs to C1-inh and C4bc were a kind gift from Prof. C. E. Hack, Amsterdam, the Netherlands.
C4d and Bb were quantified using the commercial kits from Quidel, which were used according to the manufacturer's instructions.
The C3bc (i.e. the sum of C3b, iC3b and C3c) and TCC (terminal SC5b-9 complement complex) assays have been developed in our own laboratory and were performed mainly as described . The standard in the C3bc and TCC assays was normal human serum activated with zymosan, defined to contain 1000 AU/ml.
C5a was quantified by EIA, as described previously, using the MoAb 4A2E10E2 as capture antibody . The zymosan-activated standard used for the C3bc and TCC assays was calibrated to ng/ml with respect to C5a and used as a standard in this assay. The MoAb was a kind gift from Prof. Kåre Bergh, Trondheim, Norway.
Correction for dilution of serum proteins
To correct for the dilution observed for the serum proteins, we used the changes in albumin according to the following formula: actual value = observed value × (albumin before treatment ÷ albumin after treatment). To ensure that the albumin value was representative for the dilution effect in all patients, we correlated the changes (value after minus value before) in IgA and IgM with the changes in albumin as an internal control (the IVIG preparation did not contain IgA or IgM) and found almost complete correlation (IgA versus albumin, r = 0.92, P = 0.006; IgM versus albumin, r = 0.82, P = 0.034). Thus, we relied on albumin as a correction factor for the further analyses.
Non-parametric tests were used: Spearman's rank test for correlations and Wilcoxon's signed rank test for comparison of values before and after treatment.
Patient characteristics and pregnancy outcome
Six of the patients were pregnant when IVIG treatment was started. One patient was treated during the preovulatory period, but did not become pregnant. No adverse effects of IVIG treatment were observed. Three patients spontaneously aborted between weeks 8 and 14. Three patients delivered spontaneously and uneventfully at term.
Effect of IVIG infusion on serum proteins
Albumin. A marked dilution of serum proteins, except for IgG and total serum protein, was observed during IVIG infusion. The albumin concentration decreased 17% from 48 (45–51) g/l (median and range) to 40 (36–46) g/l (P = 0.015) (Table 1). The individual albumin changes in the seven patients are shown in Fig. 1 (upper left panel). All other data presented have been corrected for this dilution effect as described in the Materials and methods.
Table 1. . Effect on peripheral blood complement and immunoglobulins of treatment with high-dose intravenous immunoglobulins (IVIG) in seven patients with recurrent habitual abortions *ND, not determined.†All other values in the table were corrected for dilution using albumin as correction factor (see Materials and methods).AU, arbitrary units; TCC, terminal complement complex.
Immunoglobulins and total serum protein. The IgG concentration increased markedly from 12.4 (9.4–15.9) to 26.8 (22.4–30.0) g/l (P = 0.015) and total protein from 74 (70–80) to 89 (83–101) g/l (P = 0.015) (Table 1). There were no changes in the concentrations of IgA and IgM. The individual values for IgG, IgA and IgM for the seven patients are shown in Fig. 1.
Complement activation induced by IVIG
Complement activation products. Complement activation induced by IVIG was revealed by a marked and significant increase in the complement activation products Bb, C3bc, C5a and TCC during treatment (Table 1). Taking into account the relatively low number of patients examined, the increase was highly significant. The theoretically lowest possible P-value using non-parametric intergroup statistics with this number of observations is 0.015, the value observed for all these activation products. All the seven patients showed an increase of these activation products, as shown by the individual changes in Fig. 2. There was a close correlation between the increases in C3bc and TCC (r = 0.95; P = 0.0006). The classical pathway activation products (C1rs/C1-inhibitor complexes, C4bc and C4d) were elevated in the baseline sample from six of the seven patients and did not increase further during IVIG treatment. The baseline level of these three activation products correlated very closely: C1rs/C1inh versus C4bc (r = 0.95; P = 0.001), C1rs/C1inh versus C4d (r = 0.77; P = 0.041) and C4bc versus C4d (r = 0.89; P = 0.006). The patient who was not pregnant at the time of the study had the same degree of classical activation as the others (98, 118 and 7.28 were the values for C1rs/C1inh, C4bc and C4d, respectively).
Complement components. The antigen concentration of individual complement components (C1q, C3, C4) and classical complement regulators (C1-inhibitor, C4b-binding protein) did not change during treatment with IVIG and were all within the normal range at both sampling points (Table 1). A slightly significant increase in C1-inhibitor function and in alternative CH50 was observed, whereas classical CH50 did not change.
Investigation of the fluid phase effect of high-dose IVIG on complement in vivo has not, to our knowledge, been performed in humans. Ideally, healthy individuals should be used to study such an effect on a normally functioning complement system. However, as IVIG have certain documented adverse effects [25, 26] and are a biological product, although virus inactivated and regarded as relatively safe clinically, it is doubtful whether such a protocol in healthy volunteers could be justified ethically. For the purpose of the present study we therefore selected a group of women, undergoing IVIG therapy because of unexplained recurrent spontaneous abortion, on the assumption that they had a normal complement system. However, we found a substantial increase in all the three classical pathway activation products studied (C1rs/C1-inh, C4bc and C4d) as well as a close correlation between them, clearly indicating that the classical complement pathway was pathologically activated in these patients. As the patient who was not pregnant at the time of treatment had the same degree of activation as the other six, it is reasonable to suggest that this pathological activation could be related to the condition of recurrent abortion and not to the pregnancy per se. The reason for this activation is unknown and will be a subject for further investigation.
The aim of the present study was, however, to investigate the effect of IVIG on complement in vivo. An activation of the complement system was clearly demonstrated. At least three different effects of IVIG on complement have previously been proposed. First, activated C3 and C4 may bind to immunoglobulin molecules, which then serve as scavengers to avoid in situ deposition of these fragments [27, 28]. Second, C1q may bind to immunoglobulin leading to a deviation of C1 binding from the target to the IVIG [19, 29]. Third, IVIG may enhance the inactivation of C3 in complex with immunoglobulins and thus down-regulate the C3 convertase activity . The present study does not exclude any of these theories, but supports IVIG as an activator of complement. If this occurs via C1, it may divert the activation away from the target, which may be a cell membrane, onto the IVIG in the fluid phase. Another intriguing aspect of IVIG as a regulator at the C1q level may be a change from an active, pathologically destructive ongoing process, to a more benign or physiological association between C1q and immunoglobulins. Such an association with C1q has been proposed as a physiological complement regulatory mechanism of IgG . An increased amount, and probably even slightly modified, IgG may enhance this effect and may be one of the mechanisms by which complement is modified during treatment with IVIG. The concentration of IgG increased, in fact, more than twofold during treatment in the present study.
Activation of complement by intravenously administered substances is potentially harmful. The presence of aggregates of IgG in the intramuscular IgG preparations leads to a marked activation of complement with a risk of anaphylactic reaction and are therefore contraindicated for use intravenously. The present study shows that although a significant activation was found in all patients, this activation was in fact modest. This is in accordance with in vitro studies showing a modest formation of complement activation products after adding IVIG to normal serum compared with intramuscular preparations or aggregated IgG [19, 32]. According to the C1q diverting theory, such a minor and controlled activation, which does not cause a detrimental effect per se, may be beneficial as it may reduce a pathological and detrimental activation occurring as part of the disease. It is important to bear in mind that the amount of activation products detected in plasma does not necessarily reflect how detrimental the activation is, as the activation stimulus at the target, which is important for the pathogenesis, is not always reflected in a corresponding level of circulating activation products.
In conclusion, we have demonstrated a significant increase in complement activation in patients with unexplained recurrent spontaneous abortion during IVIG treatment. Thus, IVIG has a complement-modulating potential which, under controlled conditions, may be beneficial in diseases where complement participates in the pathogenesis.
We thank Iren Hatlehol Andreassen, Grethe Bergseth, Hilde Fure and Bente Falang Hoaas for excellent technical assistance. Analyses of serum albumin and total serum protein were kindly carried out by the staff at the Institute of Clinical Biochemistry, The National Hospital, Oslo. The study was financially supported by Gythfeldt's legacy and Norsk Revmatiker Forbund.